Surgical techniques for the removal of mandibular wisdom teeth

  • Review
  • Intervention

Authors


Abstract

Background

The surgical removal of mandibular wisdom teeth is one of the most common operations undertaken in oral and maxillofacial surgery. The most common indication for surgery is infection about a partially erupted tooth that is impacted against bone or soft tissues. Other indications include unrestorable caries, pulpal and periapical pathology, fracture of the tooth and cyst development, amongst others. Most commonly the benefits of surgical removal of a wisdom tooth include alleviation of the symptoms and signs of pericoronitis and its potential consequences. However, surgery is frequently associated with postoperative pain, swelling and trismus. Less commonly complications include infection, including dry socket, trigeminal nerve injuries and rarely fracture of the mandible.

Objectives

To compare the relative benefits and risks of different techniques for undertaking various aspects or stages of the surgical extraction of mandibular wisdom teeth.

Search methods

We searched the Cochrane Oral Health Group's Trials Register (to 21 March 2014), CENTRAL (The Cochrane Library 2014, Issue 1), MEDLINE (OVID) (1946 to 21 March 2014) and EMBASE (OVID) (1980 to 21 March 2014). We searched ClinicalTrials.gov and the WHO International Clinical Trials Registry Platform for ongoing trials. There were no restrictions regarding language or date of publication in the electronic searches.

Selection criteria

RCTs comparing surgical techniques for removal of mandibular wisdom teeth.

Data collection and analysis

Two review authors conducted assessment of relevance, risk of bias and data extraction. Study authors were contacted for additional information. RRs were used for dichotomous data and MDs for continuous data, unless the event rate was very low and Peto ORs were used. The pairing of the split-mouth studies was taken into account in the analysis for both dichotomous and continuous outcomes, and parallel group and split-mouth studies were combined using the generic inverse variance method. Random-effects models were used provided there were more than three studies (fixed-effect models otherwise).

Main results

A total of 35 trials (2569 patients) were included. The interventions under consideration fell into seven broad categories, with many comparisons including only a small number of trials. Twenty-one of the trials were assessed at high risk of bias, the remaining 14 as unclear. The results are described in the summary of findings tables.

Triangular flaps were associated with a 71% reduction in alveolar osteitis at one week (RR 0.29, 95% CI 0.11 to 0.78; three trials, moderate quality) and reduction in pain at 24 hours (MD -0.21, 95% CI -0.32 to -0.10; two trials, moderate quality) compared with envelope flaps. There was no evidence of a difference in overall infection rates, in maximum mouth opening or in permanent sensation. However, there was some evidence that residual swelling after one week was slightly increased in the triangular flap groups compared to envelope flap types (MD 0.66 mm, 95% CI 0.26 to 1.07; two trials, low quality). We found no data on temporary sensation, or adverse events.

There was low quality evidence from two studies, looking at the use of a retractor during third molar surgery, to indicate more cases of temporary altered sensation (up to one month) when a retractor was used (Peto OR 5.19, 95% CI 1.38 to 19.49; two trials, low quality). One study reported that this did not persist for more than six months in either group. We found no data for use of a retractor on other primary outcomes or adverse effects (including fracture of the mandible).

Due to the small number of studies, the different comparisons evaluated, the variable outcomes reported and the paucity of useful data for all primary outcomes we were not able to draw any conclusions concerning bone removal in third molar surgery.

There was insufficient evidence from single studies of very low quality on irrigation method (manual versus mechanical) or irrigation volume (low or high) to determine whether there were differences or not for the outcomes of alveolar osteitis or postoperative infection. We found no data for any of the other primary outcomes.

There was insufficient evidence (low to very low quality) that any wound closure technique (primary versus secondary) was superior to another for the outcomes of alveolar osteitis, postoperative infection or maximum mouth opening achieved after seven days, or reactionary bleeding. There was evidence that secondary wound closure was associated with reduced pain at 24 hours (MD 0.79, 95% CI 0.35 to 1.24; four trials, moderate quality) and slightly reduced swelling after one week (MD 0.33, 95% CI 0.09 to 0.57; seven trials, moderate quality).We found no data on other primary outcomes.

There was some evidence that the use of a surgical drain was associated with less postoperative swelling (MD -0.90, 95% CI -1.62 to -0.19; five trials, moderate quality) and greater maximum mouth opening one week after surgery (MD 3.72 mm, 95% CI 2.84 to 4.59; two trials, moderate quality). There was insufficient evidence from a single study (low quality) to determine whether the presence of a drain made any difference to pain at 24 hours postoperation. There were no data for the other primary outcomes.

Although two RCTs compared coronectomy with complete extraction, flaws in the design and the unit of analysis of these studies meant that there were no reliable data available for inclusion.

Authors' conclusions

The thirty-five included trials looked at a range of different surgical techniques. The comparisons related to seven broad aspects of the surgical procedures for impacted mandibular third molars: type of surgical flap raised, use of retractors, techniques for bone removal, wound irrigation, wound closure, wound drainage, and complete/incomplete tooth removal. The quality of the body of evidence for each of these comparisons was very low to moderate due to the small number of trials and patients, and the majority of the trials being at high risk of bias (65%) with the remainder at unclear risk of bias.

The evidence for making changes to surgical practice is therefore limited. However, it is useful to describe the state of the research evidence supporting practice so that surgeons can make an informed choice in adopting new techniques, or continuing with established techniques.

Résumé scientifique

Techniques chirurgicales pour l'extraction des dents de sagesse mandibulaires

Contexte

L'extraction chirurgicale des dents de sagesse mandibulaires est une des opérations les plus courantes réalisées en chirurgie buccale et maxillo-faciale. L'indication la plus courante de la chirurgie est une infection sur une dent en éruption partielle appuyant sur l'os ou des tissus mous. D'autres indications comprennent les caries non récupérables, les pathologies pulpaires et périapicales, les fractures dentaires et le développement de kystes, entre autres. Le plus souvent, les avantages de l'extraction chirurgicale d'une dent de sagesse incluent le soulagement des symptômes et des signes de péricoronarite et de ses conséquences potentielles. Cependant, la chirurgie est souvent associée à la douleur postopératoire, à l'œdème et au trismus. Les complications moins fréquentes comprennent l'infection, y compris l'alvéolite sèche, les lésions du nerf trijumeau et, rarement, la fracture de la mandibule.

Objectifs

Comparer les avantages et les risques relatifs des différentes techniques pour entreprendre divers aspects ou étapes de l'extraction chirurgicale des dents de sagesse mandibulaires.

Stratégie de recherche documentaire

Nous avons effectué des recherches dans le registre des essais du groupe Cochrane sur la santé bucco-dentaire (jusqu'au 21 mars 2014), CENTRAL (Bibliothèque Cochrane 2014, numéro 1), MEDLINE (Ovid) (de 1946 au 21 mars 2014) et EMBASE (Ovid) (de 1980 au 21 mars 2014). Nous avons cherché des essais en cours sur ClinicalTrials.gov et dans le système d'enregistrement international des essais cliniques de l'OMS. Aucune restriction concernant la langue ou la date de publication n'a été appliquée lors des recherches électroniques.

Critères de sélection

ECR comparant des techniques chirurgicales pour l'extraction des dents de sagesse mandibulaires.

Recueil et analyse des données

Deux auteurs de la revue ont effectué l'évaluation de la pertinence et des risques de biais ainsi que l'extraction de données. Des auteurs d'études ont été contactés pour obtenir des informations supplémentaires. Le risque relatif (RR) a été utilisé pour les données dichotomiques et la différence moyenne (DM) pour les données continues, sauf lorsque le taux d'événements était très faible et le rapport des cotes (RC) de Peto a été utilisé. L'appariement des études en bouche divisée a été pris en compte dans l'analyse des variables dichotomiques et continues, et les études en groupes parallèles et en bouche divisée ont été combinées en utilisant la méthode générique de l'inverse de la variance. Des modèles à effets aléatoires ont été utilisés à condition qu'il y avait plus de trois études (modèles à effets fixes dans le cas contraire).

Résultats principaux

Un total de 35 essais (2 569 patients) ont été inclus. Les interventions examinées relevaient de sept grandes catégories, et plusieurs comparaisons comportaient seulement un petit nombre d'essais. Vingt-et-un essais ont été jugés à risque élevé de biais, les 14 autres à risque incertain. Les résultats sont décrits dans les tableaux de résumé des résultats.

Les lambeaux triangulaires ont été associés à une réduction de 71 % de l'ostéite alvéolaire à une semaine (RR 0,29, IC à 95 % de 0,11 à 0,78 ; trois essais, qualité modérée) et à une réduction de la douleur à 24 heures (DM -0,21, IC à 95 % de -0.32 à 0,10 ; deux essais, qualité modérée) par rapport aux lambeaux en enveloppe. Il n'y avait aucune preuve d'une différence dans les taux d'infection globaux, l'ouverture maximale de la bouche ou la sensation permanente. Cependant, certaines preuves indiquaient que l'œdème résiduel après une semaine était légèrement augmenté dans les groupes de lambeaux triangulaires par rapport aux lambeaux de type enveloppe (DM 0,66 mm, IC à 95 % de 0,26 à 1,07 ; deux essais, faible qualité). Nous n'avons trouvé aucune donnée sur la sensation temporaire ou les effets indésirables.

Des preuves de faible qualité, issues de deux études portant sur l'utilisation d'un écarteur au cours d'une chirurgie de la troisième molaire, indiquaient davantage de cas d'altération des sensations temporaire (jusqu'à un mois) en utilisant un écarteur (RC Peto 5,19, IC à 95 % de 1,38 à 19,49 ; deux essais, faible qualité). Une étude a rapporté que l'altération n'avait pas persisté plus de six mois dans aucun des deux groupes. Nous n'avons trouvé aucune donnée sur les effets de l'utilisation d'un écarteur sur les autres critères de jugement principaux ou les effets indésirables (y compris la fracture de la mandibule).

En raison du petit nombre d'études, des différentes comparaisons évaluées, de la variété des critères rapportés et de la rareté des données utiles pour tous les critères principaux, nous n'avons pu tirer aucune conclusion sur le retrait de l'os dans la chirurgie de la troisième molaire.

Les preuves issues d'études isolées de très faible qualité sur la méthode (manuelle versus mécanique) ou le volume (faible ou élevé) d'irrigation étaient insuffisantes pour déterminer s'il y avait des différences pour les critères d'ostéite alvéolaire ou d'infection postopératoire. Nous n'avons trouvé aucune donnée pour les autres critères de jugement principaux.

Les preuves étaient insuffisantes (qualité faible à très faible) pour affirmer qu'une technique de fermeture de plaie (primaire versus secondaire) ait été supérieure à l'autre pour les critères d'ostéite alvéolaire, d'infection postopératoire ou d'ouverture maximale de la bouche atteinte après sept jours, ou d'hémorragie réactionnaire. Certaines preuves indiquaient que la fermeture secondaire de la plaie était associée à une réduction de la douleur à 24 heures (DM 0,79, IC à 95 % de 0,35 à 1,24 ; quatre essais, qualité modérée) et réduisait légèrement l’œdème après une semaine (DM 0,33, IC à 95 % de 0,09 à 0,57 ; sept essais, qualité modérée). Nous n'avons pas trouvé de données sur d'autres résultats primaires.

Il y avait des preuves que l'utilisation d'un drain chirurgical était associée à une diminution de l’œdème postopératoire (DM -0,90, IC à 95 % de -1,62 à -0,19 ; cinq essais, qualité modérée) et à une plus grande ouverture maximale de la bouche une semaine après la chirurgie (DM 3,72 mm , IC à 95 % de 2,84 à 4,59 ; deux essais, qualité modérée). Les preuves issues d'une seule étude (faible qualité) étaient insuffisantes pour déterminer si la présence d'un drain faisait une différence dans la douleur à 24 heures après l'opération. Il n'y avait pas de données pour les autres critères principaux.

Bien que deux ECR aient comparé la coronectomie à l'avulsion complète, en raison de failles dans la conception et l'unité d'analyse de ces études, aucune donnée fiable n'était disponible pour inclusion.

Conclusions des auteurs

Les trente-cinq essais inclus ont examiné une variété de différentes techniques chirurgicales. Les comparaisons portaient sur sept grands aspects des procédures chirurgicales pour les troisièmes molaires mandibulaires incluses : le type de lambeau chirurgical réalisé, l'utilisation d'écarteurs, les techniques pour le retrait de l'os ainsi que pour l'irrigation, la fermeture et le drainage de la plaie, et l'avulsion complète/incomplète de la dent. La qualité du corpus de preuves pour chacune de ces comparaisons allait de très faible à modérée, en raison du petit nombre d'essais et de patients, et du risque élevé de biais dans la majorité des essais (65 %), le reste étant à risque imprécis.

Ainsi, les preuves pour apporter des modifications à la pratique chirurgicale sont limitées. Cependant, il est utile de décrire l'état des preuves issues de la recherche à l'appui de la pratique, afin que les chirurgiens puissent faire des choix éclairés en adoptant de nouvelles techniques ou en poursuivant l'utilisation de techniques établies.

Notes de traduction

Traduction réalisée par le Centre Cochrane Français

Plain language summary

Surgical techniques for the removal of mandibular wisdom teeth

Review question

This review has been produced to assess the benefits and harms of the different surgical techniques in use for the removal of wisdom teeth from the lower jaw in order to reduce complications following surgery. The risks of surgery that are included in this review are: pain following surgery, swelling, infection, being unable to open the jaw fully (trismus), damage to the nerves supplying sensation to the tongue and skin of the lower lip and chin (lingual or inferior alveolar nerve damage) and fracture of the jaw.

Background

The removal of wisdom teeth from the lower jaw is one of the most common operations in dental surgery. Various techniques have been developed to enable the procedure to be carried out successfully. It is important to evaluate the best evidence about these techniques so that the risks and complications associated with the operation can be reduced and the patient experience improved.

Study characteristics

The Cochrane Oral Health Group carried out this review and the evidence on which it is based was up to date on 21 March 2014. Thirty-five studies with 2569 participants were included, but the quality of these studies was not ideal and unlikely to make surgeons change their practice.

Key results

-Slight changes to the position of the cut into the gum may reduce dry socket and pain after surgery.

-It is possible to protect a nerve to the tongue from damage by the placement of a special surgical tool, however there were no good quality trials available to show that this reduces permanent nerve injury and the consequent loss of or altered sensations.

-Some surgeons will just remove the top part of the wisdom tooth and leave the root in place when the roots are very close to a nerve in the jaw (supplying feeling to the lower lip and skin of the chin). Although it would seem likely that this should reduce the risk of nerve injury the two trials available had unreliable data to show this. They did however find later movement of the root in 13% to 24% of the patients within two years.

Quality of the evidence

The quality of the 35 included studies was variable. None were assessed as at low risk of bias in all domains so they were unlikely to make surgeons change their practice.

Résumé simplifié

Techniques chirurgicales pour l'extraction des dents de sagesse mandibulaires

Question de la revue

Cette revue a été menée afin d'évaluer les avantages et les inconvénients des différentes techniques chirurgicales utilisées pour l'extraction des dents de sagesse de la mâchoire inférieure pour réduire les complications après la chirurgie. Les risques de la chirurgie inclus dans cette revue sont : la douleur après la chirurgie, le gonflement, l'infection, l'impossibilité d'ouvrir complètement la mâchoire (trismus), les dommages aux nerfs responsables des sensations de la langue et de la peau de la lèvre inférieure et du menton (lésions du nerf lingual ou du nerf alvéolaire inférieur) et la fracture de la mâchoire.

Contexte

L'extraction des dents de sagesse de la mâchoire inférieure est une des opérations les plus courantes en chirurgie dentaire. Diverses techniques ont été mises au point pour permettre d'exécuter cette procédure avec succès. Il est important d'évaluer les meilleures preuves sur ces techniques afin de pouvoir réduire les complications et les risques associés à cette opération et améliorer l'expérience du patient.

Caractéristiques des études

Le groupe Cochrane sur la santé bucco-dentaire a mené cette revue, dont les éléments de preuve sont à jour jusqu'au 21 Mars 2014. Trente-cinq études avec 2 569 participants ont été incluses, mais la qualité de ces études n'était pas optimale et elles sont peu susceptibles de changer les pratiques des chirurgiens.

Principaux résultats

- Modifier légèrement le positionnement de l'incision dans la gencive peut réduire l'alvéolite sèche et la douleur après la chirurgie.

- Il est possible de protéger un nerf de la langue des dommages par la mise en place d'un outil chirurgical spécial, mais il n'y avait pas d'essais de bonne qualité disponibles pour montrer que cela réduit les lésions nerveuses permanentes et la perte ou altération de sensations qui en résultent.

- Certains chirurgiens retirent juste la partie supérieure de la dent de sagesse et laissent la racine en place lorsque celle-ci est très proche d'un nerf de la mâchoire (responsable des sensations dans la lèvre inférieure et la peau du menton). Bien que cela devrait probablement réduire le risque de lésions nerveuses, les données pour le démontrer issues des deux essais disponibles n'étaient pas fiables. Un déplacement ultérieur de la racine a toutefois été observé dans ces deux essais chez 13 % à 24 % des patients dans les deux ans.

Qualité des preuves

La qualité des 35 études incluses était variable. Aucune de ces études n'a été évaluée comme étant à faible risque de biais dans tous les domaines ; aussi, elles sont peu susceptibles de faire évoluer les pratiques des chirurgiens.

Notes de traduction

Traduction réalisée par le Centre Cochrane Français

Laički sažetak

Kirurške tehnike uklanjanja umnjaka donje čeljusti

Kliničko pitanje

Ovaj Cochrane sustavni pregled izrađen je u svrhu procjene koristi i štete različitih kirurških tehnika korištenih za uklanjanje umnjaka iz donje čeljusti da bi se smanjile komplikacije nakon operacije. Rizici operacije koji su uključeni u ovaj sustavni pregled su: bol nakon operacije, oteklina, infekcija, nemogućnost potpunog otvaranja usta (trizmus), oštećenje živaca koji opskrbljuju osjet za jezik te kožu donje usne i brade (oštećenje lingvalnog ili donjeg alveolarnog živca) i prijelom čeljusti.

Uvod

Uklanjanje umnjaka iz donje čeljusti jedna je od najčešćih operacija u dentalnoj kirurgiji. Razvijene su različite tehnike da bi se omogućila uspješna provedba postupka. Važno je procijeniti najbolje dokaze o tim tehnikama tako da se rizici i komplikacije povezani s operacijom mogu smanjiti i neugodno iskustvo pacijenta poboljšati.

Značajke istraživanja

Ovaj sustavni pregled napravila je Cochrane skupina za oralno zdravlje i dokazi na kojima se temelji rezultati su istraživanja objavljenih do 21. ožujka 2014. Bilo je uključeno trideset i pet istraživanja s 2569 sudionika, ali kvaliteta istraživanja nije bila idealna i malo je vjerojatno da će kirurzi promijeniti svoju praksu.

Ključni rezultati

-Male promjene u položaju rezanja u zubno meso mogu smanjiti upalu kosti i bol nakon operacije.

-Moguće je zaštititi živac jezika od oštećenja postavljanjem posebnog kirurškog alata, međutim nije bilo dostupnih istraživanja dobre kvalitete koja bi pokazala da to smanjuje trajnu ozljedu živca i posljedični gubitak ili promjenu osjeta.

-Neki kirurzi uklonit će samo gornji dio umnjaka, a ostaviti korijen na svom mjestu kada su korijeni vrlo blizu živcu u donjoj čeljusti (koji opskrbljuju osjet za donju usnu i kožu brade). Iako se čini vjerojatnim kako bi to trebalo smanjiti rizik od ozljede živca, dva dostupna istraživanja imala su nepouzdane podatke koji bi to pokazali. Međutim poslije je pronađeno pokretanje korijena u 13% do 24% pacijenata unutar dvije godine.

Kvaliteta dokaza

Kvaliteta 35 uključenih istraživanja bila je različita. Ni jedno istraživanje nije imalo nizak rizik od pristranosti u svim područjima tako da je malo vjerojatno da će kirurzi promijeniti svoju praksu.

Bilješke prijevoda

Hrvatski Cochrane
Prevela: Josipa Čolak
Ovaj sažetak preveden je u okviru volonterskog projekta prevođenja Cochrane sažetaka. Uključite se u projekt i pomozite nam u prevođenju brojnih preostalih Cochrane sažetaka koji su još uvijek dostupni samo na engleskom jeziku. Kontakt: cochrane_croatia@mefst.hr

Streszczenie prostym językiem

Techniki chirurgiczne usuwania zębów mądrości z żuchwy

Pytanie przeglądu

Ten przegląd został wykonany, aby ocenić korzyści i szkody związane z zastosowaniem różnych technik chirurgicznego usuwania zębów mądrości z żuchwy w celu zmniejszenia ryzyka powikłań pozabiegowych, Powikłania pozabiegowe uwzględnione w tym przeglądzie to: ból po zabiegu, obrzęk, zakażenie, niemożność pełnego rozwarcia szczęk (szczękościsk), uszkodzenie nerwów zaopatrujących czuciowo język, skórę wargi dolnej i brody (uszkodzenie n. językowego lub zębodołowego dolnego) oraz złamanie żuchwy.

Wprowadzenie

Usuwanie zębów mądrości z żuchwy jest jednym z najczęstszych zabiegów w chirurgii stomatologicznej. Opracowano różne techniki mające zapewnić jak najlepszy wynik zabiegu. Ocena najlepszych danych naukowych dotyczących tych technik jest ważna, aby zminimalizować ryzyko powikłań i poprawić komfort pacjenta związany z zabiegiem.

Charakterystyka badań

Przegląd stworzyła Grupa Cochrane Oral Health a dane na których się opiera są zgodne ze stanem wiedzy na dzień 21 marca 2014. Oceniono trzydzieści pięć badań obejmujących 2569 uczestników, ale jakość tych badań nie była najlepsza i jest mało prawdopodobne, aby wynik tego przeglądu zmienił praktykę chirurgiczną.

Najważniejsze wyniki

-Niewielkie zmiany w sposobie nacięcia dziąseł mogą zmniejszyć częstotliwość występowania suchego zębodołu i bólu po zabiegu.

-Istnieje możliwość ochrony nerwu językowego przed uszkodzeniem poprzez umieszczenie specjalnego narzędzia, jednak nie było dostępnych dobrych jakościowo badań potwierdzających, że metoda ta zmniejsza ryzyko trwałego uszkodzenia nerwu i w konsekwencji zmniejsza ryzyko utraty lub zaburzenia czucia.

-Niektórzy chirurdzy usuwają jedynie koronową część zęba mądrości i zostawiają korzeń w kości w sytuacjach, kiedy te korzenie znajdują się bardzo blisko przebiegu nerwu zaopatrującego czuciowo wargę dolną i skórę brody. Mimo, iż wydaje się prawdopodobnym, że takie postępowanie zmniejsza ryzyko uszkodzenia nerwu, to dwa dostępne badania okazały się za mało wiarygodne, aby to potwierdzić. w tych badaniach pokazano natomiast przemieszczanie się korzenia z czasem u 13% do 24% pacjentów w ciągu dwóch kolejnych lat.

Jakość danych naukowych

Jakość 35 badań była różna. Żadnego z badań nie oceniono jako obarczonego małym ryzykiem błędu systematycznego we wszystkich domenach zatem jest mało prawdopodobne aby ich wyniki zmieniły obecną praktykę chirurgiczną.

Uwagi do tłumaczenia

Tłumaczenie Piotr Krucki Redakcja Andrzej Komorowski

Ringkasan bahasa mudah

Teknik pembedahan untuk pembuangan gigi geraham bongsu mandibel

Soalan ulasan

Ulasan ini telah dihasilkan bagi menilai faedah-faedah dan kemudaratan-kemudaratan pelbagai teknik pembedahan yang digunakan untuk pembuangan gigi geraham bongsu rahang bawah bagi mengurangkan komplikasi selepas pembedahan. Risiko-risiko pembedahan dalam ulasan ini meliputi: sakit selepas pembedahan, bengkak, jangkitan, pembukaan mulut yang terhad (trismus), kerosakan saraf yang membekalkan deria rasa kepada lidah dan kulit sekitar bibir bawah dan dagu (kerosakan saraf lingual dan saraf alveolar inferior) dan kepatahan rahang.

Latarbelakang

Pembuangan gigi geraham bongsu rahang bawah adalah pembedahan yang paling lazim dalam pembedahan pergigian.Pelbagai kaedah telah dibangunkan bagi membolehkan prosedur tersebut dijalankan dengan jayanya. Adalah penting untuk menilai bukti terbaik mengenai kaedah-kaedah tersebut agar risiko-risiko dan komplikasi-komplikasi yang berkaitan dengan pembedahan tersebut dapat dikurangkan dan pengalaman pesakit dapat ditambah baik.

Ciri-ciri kajian

Kumpulan Kesihatan Oral Cochrane telah membuat ulasan ini dengan berdasarkan bukti terkini pada 21 Mac 2014.Tiga-puluh lima kajian dengan penglibatan 2569 peserta tetapi kualiti kajian-kajian ini tidaklah unggul dan tidak mungkin membuatkan pakar-pakar bedah mengubah amalan mereka.

Keputusan-keputusan utama

Sedikit perubahan kepada posisi bedah ke dalam gusi boleh mengurangkan soket kering dan sakit selepas pembedahan.

Alat pembedahan khas boleh diletakkan untuk melindungi kerosakan saraf, namun tidak terdapat sebarang kajian yang berkualiti bagus untuk menunjukkan bahawa ia mengurangan kecederaan kekal pada saraf dan kehilangan atau perubahan pada deria rasa.

-Sesetengah pakar bedah akan hanya membuang bahagaian atas gigi geraham bongsu dan meninggalkan akar gigi berhampiran dengan saraf di dalam rahang (yang membekal deria rasa kepada bibir bawah dan kulit dagu). Walaupun ia seolah-olah dapat mengurangkan risiko kecederaan saraf, kedua-dua kajian mempunyai data yang tidak menyakinkan untuk membuktikan ini.Walaubagaimanapun, mereka mendapati pergerakan lewat akar dalam 13% hingga 24% pesakit-pesakit dalam tempoh dua tahun.

Kualiti bukti

Kualiti 35 kajian yang dilibatkan adalah berbeza-beza. Tiada kajian yang dinilai mempunyai risiko bias yang rendah dalam semua bahagian, oleh itu ia tidak mungkin membuat pakar-pakar bedah mengubah amalan mereka. Tiada kajian yang dinilai mempunyai risiko bias yang rendah dalam semua bahagian, oleh itu ia tidak mungkin membuat pakar-pakar bedah mengubah amalan mereka.

Catatan terjemahan

Diterjemahkan oleh Noorliza Mastura Ismail (Melaka-Manipal Medical College) Disunting oleh Tuan Hairulnizam Tuan Kamauzaman (Universiti Sains Malaysia) Untuk pertanyaan tentang terjemahan ini,sila hubungi noorliza.mastura@manipal.edu.my.

Summary of findings(Explanation)

Summary of findings for the main comparison. Surgical flap type
  1. 1 A further 4 studies evaluated other flap design comparisons but did not report outcome data suitable for inclusion in meta-analysis. The narrative results of these studies are reported under Effects of interventions

    2 Incidence estimated by median of envelope flap group

    3 Quality of evidence downgraded because studies at high or unclear risk of bias

    4 Quality of evidence downgraded because estimate is imprecise

    5 Quality of evidence downgraded due to very small number of participants

Triangular flap compared with envelope flap for third molar extraction

Patient or population: Adults with mandibular third molars requiring removal

Settings: Oral surgery

Intervention: Triangular flap1

Comparison: Envelope flap

OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
Envelope flap Triangular Flap

Alveolar osteitis

(1 week follow-up)

100 per 1000229 per 1000 (11 to 78) RR 0.29 (0.11 to 0.78)94 participants
(3 studies)
⊕⊕⊕⊝
moderate 3
Favours triangular

Infection

(1 week follow-up)

30 per 1000247 per 1000 (2 to 1000) RR 1.55 (0.07 to 35.89)45 participants
(1 study)
⊕⊕⊝⊝
low 3,4
 
Permanent tongue, chin or lip sensation20 per 1000290 per 1000 (1 to 1000) Peto OR 4.48 (0.07 to 286.49)45 participants (1 study)⊕⊝⊝⊝
very low 2, 3, 4, 5
 
Temporary tongue, chin or lip sensation     Not reported
Mean pain at 24 hoursMean pain on VAS (0-10 scale) is 42

The mean pain at 24 hours was 3.79

0.21 less in the triangular flap group (0.32 to 0.1 lower)

MD -0.21 (-0.1 to-0.32)102 participants
(2 studies)
⊕⊕⊕⊝
moderate 3
 
Swelling (after 1 week) The mean swelling after 1 week was
0.66 more in the triangular flap group (0.26 to 1.07 more)
MD 0.66 (0.26, 1.07)39 participants
(2 studies)
⊕⊕⊝⊝
low 2,5
Favours envelope
Maximum mouth opening (after 1 week) The mean maximal mouth opening was0.81 mm more in the triangular flap group (3.32 mm less to 4.94 mm more) MD 0.81 (-3.32 to 4.94 mm)39 participants
(2 studies)
⊕⊕⊝⊝
low 2,5
Also trismus in another small trial showed no evidence of a difference

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

CI: confidence interval; MD: mean difference; OD: odds ratio; RR: risk ratio; VAS: visual analogue scale

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Summary of findings 2 Retractor for lingual nerve protection

Summary of findings 2. Retractor for lingual nerve protection
  1. 1 Quality of evidence downgraded because studies at high or unclear risk of bias

    2 Control event rate is 0/55 in 1 study and 1/158 in the other. Quality of evidence downgraded because estimate is imprecise

Retractor compared with no retractor during wisdom tooth extraction

Patient or population: Adults with mandibular third molars requiring removal

Settings: Oral surgery

Intervention: Retractor (subperiosteal or Free's)

Comparison: No retractor

OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
No retractorRetractor
Alveolar osteitis     Not reported
Infection     Not reported
Permanent tongue, chin or lip sensation     Not reported

Temporary altered tongue, chin or lip sensation

(up to 1 month postoperation)

4 per 100021 per 1000 (4 to 78) Peto OR 5.19 (1.38 to 19.49)355 participants (2 studies)⊕⊕⊝⊝
low 1,2
 
Postoperative pain     Not reported
Swelling     No reported
Maximum mouth opening     Not reported
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; OR: odds ratio
GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Summary of findings 3 Bone removal techniques for the removal of mandibular wisdom teeth

Summary of findings 3. Bone removal techniques for the removal of mandibular wisdom teeth
Bone removal techniques for the removal of mandibular wisdom teeth
Patient or population: Adults with mandibular third molars requiring removal
Intervention: Bone removal techniques
OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
Bone removal with burLingual split with chisel
All outcomes   290 (5) Due to the small number of studies, the different comparisons evaluated, the variable outcomes reported and the paucity of useful data we are not able to draw any conclusions concerning bone removal in third molar surgery. Only one study reported one of the primary outcomes
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval
GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Summary of findings 4 Irrigation techniques (A compared to B) for the removal of mandibular wisdom teeth

Summary of findings 4. Irrigation techniques (A compared to B) for the removal of mandibular wisdom teeth
  1. 1 Quality of evidence downgraded twice because there is a single study for each comparison at either high or unclear risk of bias

    2 Quality of evidence further downgraded because the estimate is imprecise (confidence interval includes both a reduction in adverse outcomes and a small increase)

    3 Asssumed risk from the single study

Irrigation techniques (A compared to B) following the removal of mandibular wisdom teeth
Patient or population: Adults with mandibular third molars requiring removal
Settings: Oral surgery
Intervention: Irrigation technique A
Comparison: Irrigation technique B
OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
Irrigation technique B Irrigation technique A
 Manual irrigation (low volume)Mechanical irrigation (high volume)    

Alveolar osteitis

(up to 1 week follow-up)

103 per 10003 per 1000 (0 to 81) RR 0.33 (0.01 to 8.09)103 participants
(1 study)
⊕⊝⊝⊝
very low 1,2
Single split-mouth study Insufficient evidence to claim either is better

Infection

(up to 1 week follow-up)

203 per 100010 per 1000 (10 to 109) RR 0.5 (0.05 to 5.43)103 participants
(1 study)
⊕⊝⊝⊝
very low 1,2
Insufficient evidence to claim either is better
Permanent tongue, chin or lip sensation     Not reported
Temporary tongue, chin or lip sensation     Not reported
Postoperative pain     Not reported
Swelling     Not reported
Maximum mouth opening     Not reported
 
 

Low volume

(approx 25 ml)

High volume

(approx 175 ml)

    

Alveolar osteitis

(up to 1 week follow-up)

103 per 10005 per 1000 (3 to 10) RR 0.52 (0.27 to 1.02)211 participants
(1 study)
⊕⊝⊝⊝
very low 1,2
 

Infection

(up to 1 week follow-up)

283 per 10005 per 1000 (1 to 38) RR 0.17 (0.02 to 1.37)211 participants
(1 study)
⊕⊝⊝⊝
very low 1,2
Insufficient evidence to claim either is better
Permanent tongue, chin or lip sensation     Not reported
Temporary tongue, chin or lip sensation     Not reported
Postoperative pain     Not reported
Swelling     Not reported
Maximum mouth opening     Not reported
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; RR: risk ratio
GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Summary of findings 5 Primary versus secondary wound closure after the removal of mandibular wisdom teeth

Summary of findings 5. Primary versus secondary wound closure after the removal of mandibular wisdom teeth
  1. 1 Quality of evidence has been downgraded because these studies are at unclear or high risk of bias

    2 Quality of evidence has been further downgraded because the estimate is imprecise (confidence intervals include both an increase and a decrease in risk)

    3 Quality of evidence has been further downgraded because the estimate is imprecise (confidence intervals include both a benefit and a possible harm)

    4 Quality of evidence has been downgraded 3 times due to unclear risk of bias, single study and imprecision

    5 Medium event rate for control group used

Primary versus secondary wound closure for the removal of mandibular wisdom teeth

Patient or population: Adults with mandibular third molars requiring removal
Settings: Oral surgery
Intervention: Primary (complete) wound closure

Comparison: Secondary (partial) wound closure

OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
Secondary wound closure Primary wound closure
Alveolar osteitis (up to 1 week follow-up)43 per 1000543 per 1000 (18 to 103) RR 0.99 (0.41 to 2.40)

374 participants

(3 studies)

⊕⊕⊝⊝
low 1,2
Insufficient evidence to claim either is better
Infection (up to 1 week follow-up)   RR 4.77 (0.24 to 96.34)

82 participants

(1 study)

⊕⊝⊝⊝
very low 4
Insufficient evidence to claim either is better
Permanent tongue, chin or lip sensation     Not reported
Temporary tongue, chin or lip sensation     Not reported
Pain (after 24 hours)Mean pain at 24 hours in secondary wound closure group is 5.2 on 0-10 scaleThe mean pain at 24 hours in the primary wound closure groups was
0.79 more (0.35 to 1.24 more)
MD 0.79 (0.35 to 1.24)414 participants
(4 studies)
⊕⊕⊕⊝
moderate 1
 

Swelling

(after 7 days)

 The mean swelling (after 7 days) in the primary wound closure groups was 0.10 more (0.05 to 0.16 mm more) MD 0.33 (0.09 to 0.57)

555 participants

(7 studies)

⊕⊕⊕⊝
moderate 1
 

Maximum mouth opening (after 7 days)

(different scales)

 The mean maximum mouth opening (after 7 days) in the primary wound closure groups was
0.14 standard deviations higher
(0.13 standard deviations lower to 0.41 standard deviations higher)
SMD -0.14 (-0.41 to 0.13)214 participants
(3 studies)
⊕⊕⊝⊝
low 1,3
Insufficient evidence to claim either is better
Adverse effects - reactionary bleeding 175 per 1000 72 per 1000
(19 to 257)
RR 0.41
(0.11 to 1.47)
82 participants
(1 study)
⊕⊝⊝⊝
very low 4
Insufficient evidence to claim either is better
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; MD: mean difference; RR: risk ratio; SMD: standardised mean difference
GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Summary of findings 6 Drain versus no drain after the removal of mandibular wisdom teeth

Summary of findings 6. Drain versus no drain after the removal of mandibular wisdom teeth
  1. 1 Quality of evidence downgraded (2 steps) because there was a single study at high risk of bias reporting this outcome

    2 Quality of evidence downgraded because all 7 studies were at either high or unclear risk of bias

    3 Quality of evidence downgraded (1 step) because both studies were at high risk of bias

Drain versus no drain after the removal of mandibular wisdom teeth

Patient or population: Adults with mandibular third molars requiring removal
Settings: Oral surgery
Intervention: Drain (tube drain or gauze drain)

Comparison: No drain

OutcomesIllustrative comparative risks* (95% CI)Relative effect (95% CI)No of participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
No drainDrain
Alveolar osteitis     Not reported
Infection     Not reported
Permanent tongue, chin or lip sensation     Not reported
Tempoary tongue, chin or lip sensation     Not reported
Pain (after 24 hours)   MD 0.00 (-0.34 to 0.34)100 participants
(1 study)
⊕⊕⊝⊝
low 1
Insufficient evidence to claim either is better
Swelling (after 7 days)The median swelling (after 7 days) in the groups without a drain was 3 mmThe mean swelling (after 7 days) in the groups with a drain was 0.90 less (0.19 to 1.62 mm less) MD -0.90 (-1.62 to -0.19)

201 participants

(5 studies)

⊕⊕⊕⊝
moderate 2
 

Maximum mouth opening

(after 7 days)

The median maximum mouth opening (after 7 days) in the groups without a drain was 36.8 mmThe mean maximum mouth opening (after 7 days) in the groups with a drain was
3.72 mm greater
(2.84 to 4.59 mm greater)
MD 3.72 mm (2.84 to 4.59)123 participants
(2 studies)
⊕⊕⊕⊝
moderate 3
 
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; MD: mean difference
GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Summary of findings 7 Coronectomy versus complete extraction

Summary of findings 7. Coronectomy versus complete extraction
  1. 1 Quality of evidence downgraded because both studies at high risk of bias and data presented had unit of analysis errors

Coronectomy versus complete extraction

Patient or population: Adults with mandibular third molars requiring removal

Settings: Oral surgery

Intervention: Coronectomy

Comparison: Complete extraction

OutcomesCoronectomyComplete extractionRelative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments

*Alveolar osteitis

*Infection

*Alveolar nerve damage

*Postoperative pain

*Swelling

*Maximum mouth opening

  No reliable data on any of these outcomes359 participants (2 studies)⊕⊕⊝⊝
low 1
 
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval
GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Background

Description of the condition

Surgical removal of mandibular third molars (lower wisdom teeth) is one of the most common operations undertaken in oral and maxillofacial surgery. There are many indications for their removal but the most common reason is recurrent infection about the tooth as it attempts to erupt but is impacted against bone or soft tissues. Other indications include unrestorable caries, pulpal and periapical pathology, fracture of the tooth and cyst development, amongst others. Most commonly the benefits of surgical removal of a wisdom tooth include alleviation of the symptoms and signs of pericoronitis and its potential consequences. The symptoms of pericoronitis are pain, bad taste, swelling of the associated soft tissues about the tooth but also of the face and restricted mouth opening (trismus). Local infection may spread and be associated with lymphadenopathy, pyrexia and malaise. More rarely, swelling may threaten airway patency. Surgery is frequently associated with postoperative pain, swelling and restricted mouth opening (trismus). Less common complications include infection, including alveolar osteitis (dry socket), trigeminal nerve injuries (inferior alveolar, lingual and mylohyoid nerves) and, rarely, fracture of the mandible.

Patients requiring the surgical removal of wisdom teeth are frequently anxious about the anticipated postoperative pain, which may be severe. Such is the predictability of postoperative pain that this type of surgery is used as a model of pain for the clinical evaluation of novel analgesics. The severity of pain usually peaks within several hours after surgery and may last for several days or more. Facial swelling may also alarm patients and typically peaks at around one or two days before subsiding over the subsequent days. Restricted mouth opening results from inflammation of the muscles that move the jaw and may be described initially as having a protective function by encouraging the patient to rest the surgical site and permit healing. However, it may persist for more than a few days and exceed this function, causing concern and difficulty in eating for about two weeks or more.

Alveolar osteitis (dry socket) has a reported incidence of 1% (Goldberg 1985) to 2.9% (Muhonen 1997). The socket has exposed bone which is extremely painful and sensitive to touch. This condition is difficult to manage and usually causes pain for up to two weeks. Less commonly an infection presents with pus in the surgical site and may be associated with signs such as lymphadenopathy or raised body temperature. Damage to the branches of the trigeminal nerve may arise because of their proximity to the mandibular third molar and consequent physical damage during surgery. Sensory disturbance may be temporary or permanent, and is usually described as temporary if recovery of normal sensation occurs within four to six months (Mason 1988). The degree and description of altered sensation is variable and includes reduced sensation (hypoaesthesia), abnormal sensation (paraesthesia) and unpleasant painful sensation (dysaesthesia) (Jones 1992). The incidence of temporary and permanent nerve damage following the surgical removal of third molar teeth varies considerably between reports and may be related to a number of factors including the difficulty of surgery, surgical technique and the skill of the surgeon. The incidence of temporary lingual (tongue) nerve disturbance has been reported to be 0% (Chiapasco 1993) to 15% (Rood 1983) and that of permanent disturbance to be 0% (Schultze-Mosgau 1993) to 2% (Rood 1992). Inferior alveolar nerve damage has been reported to occur in about 5% (temporary) and 0.2% (permanent) of the patients (Smith 1997) and this affects the sensation of the skin of the lower lip and chin. Fracture of the mandible may occur during the surgery or postoperatively. The incidence of this rare complication has been estimated to be about 1 in 28,000 operations (Libersa 2002). Whilst this complication may be managed very effectively with techniques such as reduction and fixation with mini plates, the patient may experience significantly more pain and swelling.

Description of the intervention

Various techniques have been developed in an attempt to permit the successful removal of the third molar whilst reducing the complications. Some surgeons gain surgical access to enable the removal of a wisdom tooth by raising only a soft tissue flap on the buccal aspect of the tooth. Others also raise a soft tissue flap lingual to the tooth that incorporates the lingual nerve so that this may be protected by special instruments. These different practices are in part geographic and related to the history of surgical development (Pell 1933). Another approach is to use different types of instruments placed beneath the periosteum to protect the lingual nerve in a reflected flap in an attempt to reduce damage (Greenwood 1994). Bone removal may be carried out using drills or chisels and the choice of surgical technique has been implicated in the incidence of nerve damage, and the severity of pain and swelling. The most common technique using a chisel is the 'lingual split bone technique' in which a section of bone about the wisdom tooth is fractured off to facilitate the removal of the impacted tooth. This technique in which the socket is saucerized was originally developed to reduce infection at a time when this was common and fatalities were not unknown. It was later modified when the surgical drill was introduced (Ward 1956).

Modifications of flap design and techniques for bone removal may be expected to change the outcome and experience for the patient. If a wisdom tooth has a particularly intimate relationship with the inferior alveolar nerve then injury is more likely and so some have advocated leaving a part of the tooth root in situ to reduce this risk. Some surgeons have advocated using larger volumes of solution to irrigate the surgical wound before closure to increase the chances of removing all bony debris and reduce the incidence of infection. Others have recommended placing a surgical drain to reduce the size of haematoma although in many countries this practice is rarely carried out. An alternative to the use of surgical drains is to allow drainage by not completely closing the surgical wound over the socket.

Why it is important to do this review

Hospital Episode Statistics from the National Health Service (NHS) in the UK estimate that 77,000 patients underwent wisdom tooth removal during the year 2009 to 2010 (McArdle 2012). World wide the number of surgical operations to remove wisdom teeth is huge. Research suggests that wisdom tooth removal has an immediate negative impact on patients' working and social lives. In one study patients took an average of 1.6 days off work, with over one third of patients stating that the surgery had affected their performance at work (Colorado-Bonnin 2006); and participation in social activities, sports and other hobbies is also negatively affected (Conrad 1999). For some patients quality of life (QoL) is reduced for one to two weeks after surgery (Savin 1997). For those who suffer nerve injury, then the effects may be present for considerably longer. It is important to summarise and present the best evidence so that every effort can be made to reduce the postoperative morbidity and improve the patient experience for this commonly performed surgical procedure.

Objectives

To compare the relative benefits and risks of different techniques for undertaking various aspects or stages of the surgical extraction of mandibular wisdom teeth.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials (RCTs) comparing surgical techniques for removal of mandibular wisdom teeth.

Types of participants

Patients requiring the surgical removal of mandibular wisdom teeth. Studies with participants also requiring the surgical removal of a maxillary third molar tooth at the same operation were excluded when reporting pain, swelling, restricted mouth opening unless the maxillary third molar was erupted and removed as a simple extraction with forceps or elevator. Patients requiring removal of a mandibular wisdom tooth with only elevators or forceps without elevating a flap were excluded from the review.

Types of interventions

Different surgical techniques to remove mandibular wisdom teeth, including surgical flap design, nerve protection, bone removal technique, irrigation technique, wound closure, use of drains and coronectomy.

Types of outcome measures

Primary outcome measures
  • Alveolar osteitis.

  • Wounds becoming infected (presence of pus).

  • Permanent (more than six months) altered tongue sensation.

  • Permanent (more than six months) altered chin skin or lower lip sensation.

  • Other adverse events such as fracture of the mandible.

Secondary outcome measures
  • Temporary altered tongue sensation (only the time point closer to one month postoperatively was used).

  • Temporary altered chin skin or lower lip sensation (only the time point closer to one month postoperatively was used).

  • Postoperative pain (only the time point closer to one day postoperatively was used).

  • Swelling (only the time point closer to one week postoperatively was used).

  • Restricted mouth opening (only the time point closer to one week postoperatively was used).

Studies looking at periodontal outcomes relating to the second permanent molar were not included in this review.

Search methods for identification of studies

For the identification of studies included or considered for this review, we developed detailed search strategies for each database searched. These were based on the search strategy developed for MEDLINE (OVID) but were revised appropriately for each database. The search strategy used a combination of controlled vocabulary and free text terms and was linked with the Cochrane Highly Sensitive Search Strategy (CHSSS) for identifying RCTs in MEDLINE: sensitivity maximising version (2008 revision) as referenced in Chapter 6.4.11.1 and detailed in box 6.4.c of the Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 (updated March 2011) (Higgins 2011). The search of EMBASE was linked to the Cochrane Oral Health Group filter for identifying RCTs.

We searched the following electronic databases:

  • The Cochrane Oral Health Group's Trials Register (to 21 March 2014) (Appendix 1);

  • Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2014, Issue 1) (Appendix 2);

  • MEDLINE via OVID (1946 to 21 March 2014) (Appendix 3);

  • EMBASE via OVID (1980 to 21 March 2014) (Appendix 4).

There were no language restrictions in the electronic searches.

Searching other resources

We searched the following databases for ongoing trials:

We wrote to authors of the RCTs identified and personal contacts in order to identify unpublished or ongoing studies. The bibliographies of papers and review articles were checked for studies outside the handsearched journals. Personal references were also searched.

Handsearching of relevant dental journals for this review was undertaken as part of the Cochrane worldwide handsearching programme which is co-ordinated by the US Cochrane Center. Results from handsearching are incorporated into CENTRAL, which is published as part of The Cochrane Library. For a current list of journals and dates that have been handsearched see http://us.cochrane.org/master-list.

Data collection and analysis

Selection of studies

The titles and abstracts (when available) of all reports identified through the electronic searches were scanned independently by two review authors. For studies appearing to meet the inclusion criteria, or for which there were insufficient data in the title and abstract to make a clear decision, the full report was obtained. The full reports obtained from all the electronic and other methods of searching were assessed independently by two review authors to establish whether the studies met the inclusion criteria or not. Any disagreements were resolved by discussion. A third review author was to be consulted if there was any unresolved disagreement. All studies meeting the inclusion criteria underwent a validity assessment and data extraction. Any studies rejected at this or subsequent stages were recorded in the table of excluded studies, and the reason for exclusion recorded.

Data extraction and management

Data were independently extracted by two review authors using specially designed data extraction forms. The data extraction forms were piloted on several papers and modified as required before use. Any disagreement was discussed and a third review author consulted where necessary. Authors were contacted for clarification or missing information whenever possible.

For each trial the following data were recorded:

  • year of publication, country of origin and source of study funding, design of the trial (split-mouth or parallel group);

  • details of the participants including demographic characteristics, source of recruitment and criteria for inclusion and exclusion;

  • details on the type of surgical intervention, and the control intervention;

  • details of the outcomes reported, including method of assessment and time intervals.

We also noted whether or not an a priori calculation had been undertaken for sample size.

Assessment of risk of bias in included studies

Assessment of risk of bias was done by using the Cochrane Collaboration's risk of bias tool as described in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). The tool addresses the following domains: sequence generation; allocation sequence concealment; blinding of patients, surgeons and assessors; incomplete outcome data; selective outcome reporting; and other bias. Blinding of patients was straightforward in some studies as they were under a general anaesthetic.

Each piece of information extracted for the risk of bias tool was recorded together with the precise source of this information. The review authors were not blinded to the names of the authors, institutions, journal or results of a study. The assessment of risk of bias was done independently by two review authors. Any disagreements were resolved by consensus, with the assistance of a third review author.

Risk of bias was tabulated for each included study (see Characteristics of included studies), along with a judgement of low, high or unclear risk of bias for each domain. A risk of bias summary is presented in Figure 1.

Figure 1.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Overall risk of bias for each trial

After taking into account the additional information provided by the authors of the trials, studies were grouped into the following categories.

  • Low risk of bias (plausible bias unlikely to seriously alter the results) if all criteria were met.

  • Moderate risk of bias (plausible bias that raises some doubt about the results) if one or more criteria were partly met (these criteria were categorised as 'partly' in cases where authors had responded that they had made some attempt to conceal the allocation of patients, to blind the assessors or to give an explanation for withdrawals, but these attempts were not judged to be ideal).

  • High risk of bias (plausible bias that seriously weakens confidence in the results) if one or more criteria were not met as described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).

Further quality assessment was carried out to assess the definition of the exclusion and inclusion criteria, adequate definition of success criteria and comparability of control and treatment groups at entry. The quality assessment criteria were pilot tested using several articles.

Measures of treatment effect

For dichotomous outcomes, the estimate of effect of an intervention was expressed as risk ratios (RRs) together with 95% confidence intervals (CIs). For continuous outcomes, means and standard deviations were used to summarise the data for each trial using mean differences (MDs) and 95% CIs. Peto odds ratios (ORs) were used when the event rate was very low.

For the split-mouth studies Stata was used to calculate the ln(RR) and standard error (SE) for the dichotomous data, unless there were zero events in one of the arms when the data were treated as parallel group data. When the continuous data were not reported as paired data an intra-patient correlation of 0.5 was assumed between the sites within patients.

Unit of analysis issues

Patients in these studies had either one or two mandibular wisdom teeth and these were frequently presented as independent data, so the CIs were slightly narrower than they should be, and P values slightly less. If we were unable to obtain data taking the clustering of the teeth within patients into account, we used these data but were careful about the interpretation.

As expected many of these trials were split-mouth studies, where the teeth in each patient were surgically removed by different methods. The data from split mouth-studies were analysed according to Elbourne 2002.

Dealing with missing data

Data from split-mouth studies are frequently presented omitting the 'paired' relationship of the data. Estimates were made for the continuous outcomes assuming a correlation coefficient of 0.5, however estimates were not made for dichotomous outcomes.

Missing standard deviations were estimated using the methods outlined in Chapter 7 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).

Assessment of heterogeneity

There were insufficient studies in any one comparison to investigate heterogeneity. If there had been sufficient studies statistical heterogeneity would have been tested by the Q test (Chi2) and I2 statistic. The Chi2 test resulting in a P value < 0.10 is interpreted as indicating statistically significant heterogeneity. In order to assess and quantify the possible magnitude of inconsistency (i.e. heterogeneity) across studies, we would use the I2 statistic.

Assessment of reporting biases

Possible reporting biases were assessed on two levels: within-study and between-study.

Within-study selective outcome reporting was examined as a part of the overall risk of bias assessment (see Assessment of risk of bias in included studies). Outcomes listed in the methods section of a publication were compared against those with results reported. Where some indications of reporting bias were found, study authors were contacted for clarification.

If there were at least 10 studies included in a meta-analysis in the review, a funnel plot of effect estimates against their standard errors was planned to assess a possible between-study reporting bias. If asymmetry of the funnel plot was found by inspection and confirmed by statistical tests, possible explanations were to be considered and taken into account in the interpretation of the overall estimates of treatment effects.

Data synthesis

Only if there were studies of similar comparisons reporting the same outcome measures was a meta-analysis attempted. RRs were combined for dichotomous data, unless the event rate was very low and Peto odds ratios (ORs) were used. Weighted mean differences (WMD) were used for continuous data. Random-effects models were used as we expected heterogeneity between studies (fixed-effect models were to be used if there were fewer than four trials in any comparison). The split-mouth studies were combined with parallel group studies using the generic inverse variance option in Review Manager (RevMan 2012).

Subgroup analysis and investigation of heterogeneity

Where possible, subgroup analyses were to be undertaken in respect of the different surgical techniques and different numbers of operators or types of operator. No trials included in the review had data available.

Clinical heterogeneity would have been assessed by examining the types of participants and interventions for all outcomes in each study if sufficient numbers of studies had been included.

Sensitivity analysis

Sensitivity analyses were undertaken to examine the effect of allocation concealment and blinded outcome assessment on the overall estimates of effect. In addition, the effect of including unpublished literature on the review's findings was to be examined.

Presentation of main results

Summary of findings tables were presented for the following outcomes: alveolar osteitis; infection; permanent (more than six months) altered tongue, chin skin or lower lip sensation; temporary altered tongue, chin skin or lower lip sensation; postoperative pain; swelling; and restricted mouth opening. We assessed the quality of the body of evidence by considering the overall risk of bias of the included studies, the directness of the evidence, the inconsistency of the results, the precision of the estimates, the risk of publication bias, the magnitude of the effect, and whether or not there was evidence of a dose response. We categorised the quality of the body of evidence for each of the primary outcomes as high, moderate, low or very low.

Results

Description of studies

Results of the search

After de-duplication there were 2047 references identified from the searches. Titles and abstracts were assessed by at least two review authors independently and in duplicate. A total of 1978 references were discarded at this stage as they were not relevant to this review. Full text copies of the remaining 69 articles were obtained. These papers were each assessed by at least two review authors and 27 studies (with 29 references) were excluded. Thirty-seven references to 35 studies were found to meet the inclusion criteria for this review (Figure 2). There were also two studies awaiting classification (Kumar 2013; Ozveri Koyuncu 2013) and one ongoing trial (ACTRN 2010).

Figure 2.

Study flow diagram.

Included studies

Characteristics of the trial setting and investigators

Of the 35 included studies, five were conducted in India (Danda 2010; Nageshwar 2002; Praveen 2007; Sandhu 2010; Srinivas 2006), four in Brazil (Briguglio 2011; Cerqueira 2004; Gomes 2005; Xavier 2008), four in Nigeria (Bello 2011; Chukwuneke 2008; Osunde 2011; Osunde 2012), three in Iran (Haraji 2010; Hashemi 2012; Refo'a 2011), three in Turkey (Erdogan 2011; Mocan 1996; Saglam 2003), three in the UK (Absi 1993; Greenwood 1994; Renton 2005), three in the USA (Butler 1977; de Brabander 1988; Sweet 1976), three in Italy (Barone 2010; Pasqualini 2005; Rullo 2013), two in New Zealand (Goldsmith 2012; Kirk 2007), and one each in China (Leung 2009), South Africa (Roode 2010), Spain (Gargallo-Albiol 2000) and Thailand (Rakprasitkul 1997).

All of the included studies took place in hospitals and no commercial sponsorships were reported, although one study (Renton 2005) did receive financial support from the British Association of Oral and Maxillofacial Surgeons.

Fifteen studies (Barone 2010; Bello 2011; Briguglio 2011; Chukwuneke 2008; de Brabander 1988; Gargallo-Albiol 2000; Leung 2009; Mocan 1996; Nageshwar 2002; Osunde 2011; Osunde 2012; Pasqualini 2005; Praveen 2007; Refo'a 2011; Renton 2005) were of parallel group design where some participants were randomly allocated to the experimental group and others were randomly allocated to the control group.

The remaining 20 studies were of split-mouth design where participants had one mandibular third molar randomly allocated to the experimental intervention and the other third molar allocated to the control group. In 15 of the 20 split-mouth studies (Absi 1993; Butler 1977; Cerqueira 2004; Danda 2010; Gomes 2005; Greenwood 1994; Haraji 2010; Hashemi 2012; Kirk 2007; Roode 2010; Saglam 2003; Sandhu 2010; Srinivas 2006; Sweet 1976; Xavier 2008) both mandibular third molars were extracted during the same operating session, with each side of the mouth allocated to either experimental or control treatment. In two of these studies (Absi 1993; Butler 1977) maxillary third molars were also extracted in the same session. In the 15 split-mouth studies where both mandibular third molars were extracted in a single session, we considered that the outcome of trismus (on seventh postoperative day) was not applicable since it was not possible to determine which side of the mouth was causing any difference in mouth opening. We considered that it was possible to ascribe differences in the outcomes of pain and swelling to the specific side of the mouth. The pairing of the data was taken into account in the analysis.

In the other five split-mouth studies there was an interval of either two weeks (Baqain 2012), three weeks (Erdogan 2011; Goldsmith 2012), four weeks (Rullo 2013) or two months (Rakprasitkul 1997) between the two extractions, which meant that pain and swelling had generally resolved prior to the extraction of the second tooth. However, the data in these studies were paired and this was accounted for in the analysis of the data from these studies.

Characteristics of the participants

A total of 2569 patients were included in these studies, with individual studies recruiting between 13 and 300 participants (mean of 75 participants per study). In all of the included studies participants were generally healthy and without any indication of infection or inflammation surrounding the mandibular third molars.

In the majority of included studies (21/35) the mean age of participants was between 20 and 29 years. In eight studies the inclusion criteria specified an age range of approximately 18 to 40 years, but the mean age of participants was not reported (Briguglio 2011; Butler 1977; Chukwuneke 2008; Goldsmith 2012; Pasqualini 2005; Srinivas 2006; Sweet 1976; Xavier 2008). Two studies included participants with a mean age of 19 years (Haraji 2010; Roode 2010), two had participants in their 30s (Barone 2010; Praveen 2007) and in two studies the age of participants was not reported (Gomes 2005; Greenwood 1994).

Generally the participants in trials were healthy, and there was no indication of pain or infection at baseline. However, in one study (Baqain 2012) all participants had chronic pericoronitis.

Characteristics of the interventions

Most of the included studies used a local anaesthetic only (26 studies). In four studies all the procedures were done under general anaesthetic (GA) (Absi 1993; Greenwood 1994; Roode 2010; Sweet 1976), two studies used intravenous (IV) sedation (Butler 1977; Kirk 2007) and three studies used either GA or local anaesthetic (LA) (Gomes 2005; Leung 2009; Renton 2005).

The studies covered a wide range of interventions and comparisons, which we have combined into seven groups that follow the steps of the surgical removal of third molars.

1. Surgical flap type

Nine of the included studies evaluated different types of incision or types of flap. In each of the trials the incision was described slightly differently. We have grouped together similar comparisons where possible.

Triangular or modified triangular flap versus envelope or modified envelope flap (five trials)

Kirk 2007 and Haraji 2010 compared a modified triangular flap with an envelope flap design. The modified triangular flap involved an incision from the distobuccal edge of the second molar dropping at a slight oblique angle and curving forward into the mandibular vestibule, and a second part was a relieving incision from the ramus to the distobuccal aspect of the second molar. The envelope flap involved a sulcular incision from the first to the second mandibular molar and a distal relieving incision along the external oblique ridge to the ramus. The Kirk 2007 study used a split-mouth design in 32 patients who underwent surgery under local anaesthesia and intravenous conscious sedation. The Haraji 2010 study used a split-mouth design in 17 patients who underwent surgery under local anaesthesia alone.

The Baqain 2012; Briguglio 2011; Erdogan 2011 studies compared a standard triangular flap with an envelope flap. The triangular flap technique was as described above as the modified triangular flap design but without the curving forward of the buccal oblique incision. The envelope flap design was the same as described in the Kirk 2007 and Haraji 2010 studies except that the Briguglio 2011 study described a modified envelope design in which the incision finished at the mesial aspect of the second molar rather than continuing to the first molar. These minor modifications were determined to be of little clinical significance compared to the differences between triangular and envelope flaps and so we grouped all five of these studies together. The Briguglio 2011 study compared two different minor modifications of an envelope flap with a triangular flap in 45 patients who underwent surgery under local anaesthesia. The Erdogan 2011 study used a split-mouth design in 20 patients who underwent surgery under local anaesthesia with an interval of three weeks between one side and the other. The Baqain 2012 study used a split-mouth design under local anaesthesia in 19 patients.

The other four trials each made slightly different comparisons. Goldsmith 2012 compared a pedicle flap with an envelope flap, Roode 2010 compared the modified triangular flap with a single incision flap, Sandhu 2010 compared the bayonet flap with an envelope flap, and Nageshwar 2002 compared a modified envelope incision with a comma incision.

Pedicle flap versus envelope flap (one trial)

Goldsmith 2012 compared a pedicle flap design with the envelope flap. For the pedicle flap an incision distal to the third molar was extended approximately 1 cm and then curved towards the buccal sulcus allowing for rotation of the flap and primary closure over sound bone. Prior to closure, the gingival papilla distal to the second molar was removed and the apex of the pedicle de-epithelialised. A lingual flap was raised in the subperiosteal plane irrespective of flap design and the lingual nerve protected using a Howarth retractor. This study used a split-mouth cross-over design in 52 patients who underwent surgery under local anaesthesia and intravenous conscious sedation.

Modified triangular flap versus alternative single incision flap (one trial)

Roode 2010 compared the modified triangular flap as described above with a newly described alternative design using only a single straight incision about 15 mm in length and beginning 5 mm distal of the second molar and running mesio-bucally towards the sulcus adjacent to the second molar. Thirty-three patients underwent surgery under general anaesthesia in this split-mouth study.

Bayonet flap versus envelope flap (one trial)

Sandhu 2010 compared the bayonet flap, a type of modified triangular design, with an envelope flap. For the bayonet flap the incision was made as per a triangular flap but followed around the second molar sulcus until its buccal midpoint with the envelope flap. More specifically, the bayonet flap design incision started on the ascending ramus, following the centre of the third molar shelf to the disto-buccal surface of the second molar and then was extended as a sulcular incision up to the midpoint of the buccal sulcus of the second molar, followed by an oblique vestibular extension. This study used a split-mouth cross-over design in 20 patients who underwent surgery under local anaesthesia.

Buccal comma shaped incision versus modified envelope flap (one trial)

Nageshwar 2002 compared a newly described buccal comma shaped incision with the modified envelope flap described above. Before starting the comma shaped incision, the buccal vestibule below the adjacent second molar was stretched down as far as possible with the index finger or thumb of the hand not holding the scalpel to stretch the buccinator beyond its origin on the mandible. Starting from a point at the depth of this stretched vestibule reflection posterior to the distal aspect of the preceding second molar, the incision was made in an anterior direction. The incision was made to a point below the second molar, from where it was smoothly curved up to meet the gingival crest at the disto-buccal line angle of the second molar. The incision was continued around as a crevicular incision around the distal aspect of the second molar. This study used a parallel group design in 100 patients who underwent surgery under local anaesthesia.

2. Lingual nerve protection

Instruments may be used during third molar extraction with the intention of protecting the lingual nerve from damage. However, it is unclear whether the use of a retractor is effective in this role or might even introduce a higher risk of nerve damage.

There were three studies in this group; two (Gargallo-Albiol 2000; Gomes 2005) compared the use of a retractor with no retractor, and in the other study (Greenwood 1994) two types of retractor were compared.

Lingual nerve retractor versus no retractor (two trials)

Two studies compared the surgical removal of wisdom teeth with or without the use of a retractor for protection of the lingual nerve (Gargallo-Albiol 2000; Gomes 2005). An instrument was placed between the lingual flap and the bone lingual to the wisdom tooth with the intention of protecting the lingual nerve within the raised flap. In one of the studies (Gargallo-Albiol 2000) the surgery was undertaken under local anaesthesia alone at the University of Barcelona, Spain, and the type of retractor used was not specified. In Gomes 2005, the surgery was undertaken under local anaesthesia or general anaesthesia with local anaesthesia at the University of Pernambuco, Camaragibe, Brazil, and a Free's elevator was used.

Broad retractor versus conventional (Howarth) lingual flap retractor (one trial)

Greenwood 1994 compared the use a wider retractor with a conventional Howarth periosteal elevator for protection of the lingual nerve within the lingual flap. All participants in this study underwent surgery under day-case general anaesthesia.

3. Bone removal techniques

Five studies reported comparisons of different bone removal surgical techniques for the removal of mandibular wisdom teeth.

Bone removal with lingual split with chisel versus bur (three trials)

Distolingual bone obstructing the surgical removal of an impacted wisdom tooth may be removed with a chisel and surgical mallet in a 'distal split' technique or by using a surgical bur. Three studies compared these different surgical techniques (Absi 1993; Mocan 1996; Praveen 2007). The lingual split technique involves making a vertical cut in the buccal bone with a monobevel chisel distal to the second molar. A second horizontal cut is made to join the first posteriorly. The bone removed allows access to the mesio-buccal aspect of the impacted tooth and permits a point of application with an elevator. The bevel of the chisel is then turned lingually and a 'lingual split' is made, removing a distolingual piece of bone, which allows the tooth to be elevated. The depth is controlled by angling the chisel between vertical and 45° buccolingual during removal of the lingual cortex. When bone is removed with a bur, a rose-head bur or similar bur is electrically powered usually at a speed of 40,000 rpm and irrigation is used. In the Absi 1993; Mocan 1996; and Praveen 2007 studies the lingual nerve was protected by a Howarth's periosteal elevator.

The Absi 1993 study used a split-mouth cross-over design in 52 patients who underwent surgery under general anaesthesia. The Mocan 1996 study used a parallel group design in 20 patients who underwent surgery under local anaesthesia. The Praveen 2007 study used a parallel group design in 90 patients who underwent surgery under local anaesthesia.

Bone removal with ultrasonic tools versus surgical bur (two trials)

Two studies compared the use of ultrasonic surgery with traditional rotary instruments (drill and surgical bur) for bone removal in lower third molar surgery. Surgical fissure burs were used to section the teeth, where necessary, in both the control and experimental groups. All surgery was carried out by the same surgeon, under local anaesthesia. The Barone 2010 study used a parallel group design in 26 patients who underwent surgery under local anaesthesia. The Rullo 2013 study used a split-mouth cross-over design in 52 patients who underwent surgery under local anaesthesia.

4. Irrigation techniques

Two studies were conducted and reported during the 1970s.

Mechanical versus manual surgical wound irrigation (one trial)

Sweet 1976 compared two different modes of application of postsurgical lavage: an electrically driven, mechanical irrigator (Water Pik Model 47, 120 volt, 60 cycle) at a preset pressure of 45 psi versus a conventional hand syringe (50 ml disposable syringe with a 15 gauge needle). Both used the same volume (350 ml) of sterile saline, and the sockets were irrigated immediately after extraction.

Different manual irrigation volumes (one trial)

A single study (Butler 1977a) compared the use of 175 ml sterile saline after an extraction with a much smaller volume of "no more than 25 ml" sterile saline in a split-mouth trial; 211 participants were included, each having bilateral mandibular wisdom teeth removed under intravenous conscious sedation by the same operator.

5. Wound closure
Primary versus secondary closure of surgical wound (eight trials)

Wound closure techniques can consist of total closure in which the mucoperiosteum is hermetically sealed and healing occurs by primary intention or partial closure in which a window exists or is created to allow healing by secondary intention. With the latter technique, some sutures may be required or no sutures depending on the flap design. Eight included studies evaluated primary versus secondary wound closure techniques (Bello 2011; Danda 2010; Hashemi 2012; Osunde 2011; Osunde 2012; Pasqualini 2005; Refo'a 2011; Xavier 2008).

In the Bello 2011 study, primary closure was obtained after repositioning the flap and suturing completely with five sutures and secondary closure by leaving a window in the wound and placing four sutures. This study used a parallel group design in 82 patients who underwent surgery under local anaesthesia. In the Danda 2010 study, primary closure was obtained using two sutures on the distal arm of the incision and secondary closure by removing a wedge of mucosa distal to the second molar and by placing a single suture on the mesial arm of the incision and another on the distal arm. This study used a split-mouth cross-over design in 93 patients who underwent surgery under local anaesthesia. In the Hashemi 2012 study, primary closure was obtained after repositioning the flap and suturing completely with three sutures and secondary closure by placing no sutures. This study used a split-mouth design in 30 patients who underwent surgery under local anaesthesia. In the Osunde 2011 study, primary closure was obtained after repositioning the flap and suturing completely with multiple sutures and secondary closure by leaving a window in the wound and placing a single suture in the distal relieving incision. This study used a parallel group design in 50 patients who underwent surgery under local anaesthesia. In the Osunde 2012 study, primary closure was obtained after repositioning the flap and suturing completely with multiple sutures and secondary closure by placing no sutures. This study used a parallel group design in 80 patients who underwent surgery under local anaesthesia. In the Pasqualini 2005 study, primary closure was obtained after repositioning the flap and suturing hermetically and secondary closure by removing a wedge of mucosa distal to the second molar and by suturing. This study used a parallel group design in 200 patients who underwent surgery under local anaesthesia. In the Refo'a 2011 study, primary closure was obtained after repositioning the flap and suturing completely and secondary closure by leaving open the distal extension to the second molar but suturing the other parts of the flap. This study used a parallel group design in 32 patients who underwent surgery under local anaesthesia. In the Xavier 2008 study, primary closure was obtained after repositioning the flap and suturing completely and secondary closure by placing sutures for partial wound closure. This study used a parallel group design in 40 patients who underwent surgery under local anaesthesia.

6. Surgical drain versus no drain (six studies)

Five studies compared a tube drain with no drain (Cerqueira 2004; Chukwuneke 2008; Rakprasitkul 1997; Saglam 2003; Srinivas 2006) and one study compared the use of a gauze drain with no drain (de Brabander 1988).

In de Brabander 1988, a simple drain made of Vaseline coated gauze was placed into the socket and this was compared with using no drain in a parallel group study.

7. Coronectomy versus complete tooth removal (two trials)

Two studies (Leung 2009; Renton 2005) compared the extraction of lower third molars with coronectomy. Coronectomy involved transection of the tooth 3 to 4 mm below the enamel of the crown into the dentine. The pulp was then left in place after the crown had been levered off, and received no treatment other than a saline rinse and the re-apposition of the muco-periosteal flap. Coronectomy is designed to leave the apices of lower third molars intact if they are in immediate proximity to the inferior alveolar nerve as predicted by radiographic features. In both studies, patients were chosen due to radiographic signs of a close proximity of the tooth to the inferior alveolar nerve using plain radiographs (orthopantomographs).

Characteristics of outcome measures
Primary outcome measures

In split-mouth studies where the two interventions were delivered during the same surgical session and the outcome of alveolar osteitis (AO) was reported, we assumed events were unilateral, unless otherwise stated (Danda 2010; Haraji 2010; Kirk 2007). In four studies (Goldsmith 2012; Roode 2010; Sandhu 2010; Sweet 1976) we received confirmation that AO was unilateral.

Likewise in Baqain 2012 and Sandhu 2010, we assumed that wound dehiscence was unilateral, but we were unable to confirm this.

Secondary outcome measures

In split-mouth studies where a different intervention was used on each side of the mouth during the same surgical session we considered that outcomes of pain and trismus, or the presence or absence of swelling, could not be reliably ascribed to an intervention so we did not use these data (Cerqueira 2004; Hashemi 2012; Saglam 2003; Srinivas 2006; Sweet 1976). In split-mouth studies where the sides of the mouth were treated in two separate sessions at least two weeks apart, we used the pain and trismus outcome data that were reported.

Excluded studies

Twenty-seven studies (30 references) were excluded from this review after assessment of the full papers by at least two review authors. Eight studies were found not be to randomised controlled trials (RCTs) (Dubois 1982; Goyal 2012; Jakse 2002; Kerdvongbundit 1989; Rosa 2002; Sortino 2008; Strukmeier 1980; Suddhasthira 1991) and for two studies (Ayad 1995; Sweet 1978) the study design was unclear and attempts to contact the authors for clarification were unsuccessful. The reference Gonzalez 2001 was for an abstract and contained insufficient information to include the study in this review and a subsequent publication was not found. In five studies there was confounding due to two or more concurrent interventions (Abu-Serriah 2004; Akota 1998; Genu 2008; Holland 1984; Tuffin 1990) and in two studies there was surgical removal of both maxillary and mandibular third molars (Shevel 2001; Suarez-Cunqueiro 2003). In four studies some participants did not have a flap raised (Clauser 1994; Ding 2000; Robinson 1996; Zhang 1997) and in two the only outcomes were periodontal indices in the months following surgery (Cetinkaya 2009; Quee 1985). Germectomy rather than third molar extraction was the intervention in three studies (Chossegros 2002; Finne 1981; Sivolella 2011).

Risk of bias in included studies

See Figure 1.

Allocation

Sequence generation

Sixteen studies (46%) described clearly the methods used to generate the randomised sequence and these were assessed at low risk of bias for this domain (Baqain 2012; Barone 2010; Bello 2011; Briguglio 2011; Erdogan 2011; Goldsmith 2012; Greenwood 1994; Leung 2009; Nageshwar 2002; Pasqualini 2005; Refo'a 2011; Renton 2005; Roode 2010; Rullo 2013; Sandhu 2010; Sweet 1976). The remaining 19 studies did not report details concerning the methods of randomisation and were therefore assessed at unclear risk of bias for this domain.

Allocation concealment

Eleven studies (32%) described adequate concealment of allocation and were assessed at low risk of bias for this domain (Absi 1993; Barone 2010; Bello 2011; Briguglio 2011; de Brabander 1988; Goldsmith 2012; Greenwood 1994; Leung 2009; Pasqualini 2005; Refo'a 2011; Renton 2005). One study (Sandhu 2010) provided information that the coin toss was done by the operating surgeon immediately prior to the procedure so this study was assessed at high risk of selection bias. The remaining 23 studies provided insufficient information to enable a clear judgement and were therefore assessed at unclear risk of bias for this domain.

Blinding

It is important to note that participant blinding was unrealistic in many of these studies due to the surgery being carried out under local anaesthetic or with only light sedation. We considered that the blinding of outcome assessment was both possible and important to reduce the risk of detection bias. Where trials were described as double blind we interpreted this as meaning that both the participant and outcome assessor were blinded to the allocated intervention. Where the person assessing the outcomes was the same surgeon who performed the procedure, or where blinded outcome assessment was not mentioned, we considered the risk of detection bias to be high.

In 24 studies (71%) participants were blinded to the allocated intervention or it was considered that lack of participant blinding was not associated with a risk of bias due to the nature of the intervention. These 24 trials were assessed at low risk of performance bias (Absi 1993; Barone 2010; Bello 2011; Briguglio 2011; Butler 1977; Erdogan 2011; Goldsmith 2012; Greenwood 1994; Haraji 2010; Hashemi 2012; Kirk 2007; Leung 2009; Mocan 1996; Nageshwar 2002; Osunde 2011; Osunde 2012; Pasqualini 2005; Refo'a 2011; Renton 2005; Roode 2010; Saglam 2003; Sandhu 2010; Sweet 1976; Xavier 2008). In seven trials (Cerqueira 2004; Chukwuneke 2008; Danda 2010; de Brabander 1988; Gargallo-Albiol 2000; Rakprasitkul 1997; Srinivas 2006) the risk of performance bias was assessed as high due to the lack of participant blinding and self assessment by participants of pain outcomes. In the remaining four trials participant blinding was assessed as unclear (Baqain 2012; Gomes 2005; Praveen 2007; Rullo 2013).

Only 14 studies (40%) clearly described blinded outcome assessment (or provided information on request) and these were assessed at low risk of bias for this domain (Absi 1993; Briguglio 2011; Erdogan 2011; Goldsmith 2012; Gomes 2005; Greenwood 1994; Haraji 2010; Nageshwar 2002; Osunde 2011; Osunde 2012; Pasqualini 2005; Sandhu 2010; Sweet 1976; Xavier 2008). In four studies there was insufficient information to determine whether outcome assessors were blinded and these studies were assessed at unclear risk of bias for this domain (Barone 2010; Bello 2011; Chukwuneke 2008; de Brabander 1988). The remaining 18 studies (Baqain 2012; Butler 1977; Cerqueira 2004; Danda 2010; Gargallo-Albiol 2000; Goldsmith 2012; Hashemi 2012; Kirk 2007; Leung 2009; Mocan 1996; Praveen 2007; Rakprasitkul 1997; Refo'a 2011; Renton 2005; Roode 2010; Rullo 2013; Saglam 2003; Srinivas 2006) were assessed at high risk of bias for this domain because outcome assessors were not blinded to allocated interventions.

Incomplete outcome data

In 25 studies (71%) outcome data were complete or the numbers lost were less than 10% in split-mouth studies and so the risk of attrition bias was assessed as low (Absi 1993; Baqain 2012; Barone 2010; Bello 2011; Briguglio 2011; Butler 1977; Cerqueira 2004; Chukwuneke 2008; Erdogan 2011; Gargallo-Albiol 2000; Goldsmith 2012; Gomes 2005; Greenwood 1994; Hashemi 2012; Kirk 2007; Nageshwar 2002; Osunde 2012; Pasqualini 2005; Rakprasitkul 1997; Roode 2010; Saglam 2003; Sandhu 2010; Srinivas 2006; Sweet 1976; Xavier 2008). Eight studies (Danda 2010; de Brabander 1988; Haraji 2010; Mocan 1996; Osunde 2011; Praveen 2007; Refo'a 2011; Rullo 2013) did not clearly report the number of participants included in the outcome assessment and these studies were therefore assessed at unclear risk of attrition bias. In Leung 2009 9% of teeth in the coronectomy group were excluded from the outcome assessments due to 'failed coronectomy'. The authors supplied some of these missing data but we assessed the risk of attrition bias as unclear in this study. In Renton 2005 the numbers of teeth included in the outcome assessment varied, due to the exclusion of the 'failed coronectomy group' despite planned intention-to-treat analysis, and there was a significant loss to follow-up which was not explained. We considered the risk of attrition bias in this study to be high.

Selective reporting

The risk of reporting bias was assessed as low in 20 studies (Absi 1993; Baqain 2012; Barone 2010; Bello 2011; Cerqueira 2004; Chukwuneke 2008; Danda 2010; de Brabander 1988; Erdogan 2011; Gargallo-Albiol 2000; Greenwood 1994; Kirk 2007; Osunde 2012; Pasqualini 2005; Rakprasitkul 1997; Roode 2010; Saglam 2003; Sandhu 2010; Srinivas 2006; Sweet 1976) as planned outcomes listed in the methods sections were reported in full, or this information was supplied by the authors.

In two studies outcomes were reported incompletely, as graphs without numerical data and estimates of variance (Mocan 1996; Refo'a 2011), and in a further two studies (Haraji 2010; Hashemi 2012) some outcomes were not reported and it appeared likely that the paired nature of the data had not been accounted for in the analysis. Based on the published report it seems likely that attempts were made to measure swelling and trismus in Rullo 2013 but these outcomes were not reported in full because the measures were "not reproducible". In Renton 2005 and Leung 2009 outcomes were not reported for each randomised group of participants; instead the denominator was teeth, and the effects of paired teeth were not accounted for in the analysis. These seven studies were assessed at high risk of reporting bias.

In the remaining seven studies there was insufficient information presented on which to base a judgement, so these were assessed at unclear risk of reporting bias.

Other potential sources of bias

We identified other sources of bias in four studies in this review. In Butler 1977 the irrigation fluid was delivered by either a mechanical irrigation device or by hand, a confounding factor which may have introduced a bias. Also several patients in this study underwent concurrent extraction of maxillary third molars. It was likely that multiple tooth extraction carried a higher risk of alveolar osteitis and should therefore be considered as a confounding factor, in an unknown number of participants in each group.

In Greenwood 1994 the method of bone removal was not standardised: "the tooth was then removed employing either drill or chisel for bone removal, according to the operator's personal preference". The bone removal technique may possibly confound the results, and it was not recorded how many in each group had bone removal by each technique.

Mocan 1996 had a small sample size with 10 participants in each group, and a different distribution of impactions in the two intervention groups at baseline. The study report by Rullo 2013 contained contradictory information in the text and the tables. These four studies were assessed at high risk of other bias.

Twenty-two studies had a low risk of other bias (Baqain 2012; Barone 2010; Bello 2011; Briguglio 2011; Cerqueira 2004; Danda 2010; Erdogan 2011; Gargallo-Albiol 2000; Goldsmith 2012; Hashemi 2012; Kirk 2007; Nageshwar 2002; Osunde 2011; Osunde 2012; Pasqualini 2005; Praveen 2007; Refo'a 2011; Roode 2010; Saglam 2003; Sandhu 2010; Srinivas 2006; Sweet 1976) and in the remaining nine studies the risk of other bias was assessed as unclear.

Overall risk of bias

None of the studies included in this review were assessed at low risk of bias across all domains. In the majority of studies (21 studies, 65%) at least one domain was assessed at high risk of bias (Baqain 2012; Butler 1977; Cerqueira 2004; Chukwuneke 2008; Danda 2010; de Brabander 1988; Gargallo-Albiol 2000; Goldsmith 2012; Greenwood 1994; Haraji 2010; Hashemi 2012; Kirk 2007; Leung 2009; Mocan 1996; Praveen 2007; Rakprasitkul 1997; Refo'a 2011; Renton 2005; Roode 2010; Saglam 2003; Sandhu 2010; Srinivas 2006), and in the remaining 14 studies the overall risk of bias was unclear.

Effects of interventions

See: Summary of findings for the main comparison Surgical flap type; Summary of findings 2 Retractor for lingual nerve protection; Summary of findings 3 Bone removal techniques for the removal of mandibular wisdom teeth; Summary of findings 4 Irrigation techniques (A compared to B) for the removal of mandibular wisdom teeth; Summary of findings 5 Primary versus secondary wound closure after the removal of mandibular wisdom teeth; Summary of findings 6 Drain versus no drain after the removal of mandibular wisdom teeth; Summary of findings 7 Coronectomy versus complete extraction

1. Surgical flap type

Nine of the included studies compared different flap designs. Five of these studies were at high risk of bias (Baqain 2012; Haraji 2010; Kirk 2007; Roode 2010; Sandhu 2010) and in the remaining four the risk of bias was unclear (Briguglio 2011; Erdogan 2011; Goldsmith 2012; Nageshwar 2002). Five studies, one parallel group RCT (Briguglio 2011) and four split-mouth RCTs (total 132 participants), compared a triangular flap design with an envelope flap (Baqain 2012; Briguglio 2011; Erdogan 2011; Haraji 2010; Kirk 2007) and we attempted to pool the outcome data. The remaining four studies in this group each made a slightly different comparison and so data were reported as a separate subgroup for each of these studies.

Alveolar osteitis

Three studies (Briguglio 2011; Haraji 2010; Kirk 2007) compared a type of triangular flap with a type of envelope flap and found reduced risk of alveolar osteitis associated with the use of triangular flaps (risk ratio (RR) 0.29, 95% confidence interval (CI) 0.11 to 0.78, P value = 0.01, I2 = 0%; Analysis 1.1). A fourth study with 20 participants (Baqain 2012) made the same comparison and recorded no cases of alveolar osteitis in either group.

In a study of 57 participants (Goldsmith 2012), a pedicle flap design (similar to a triangular flap) was compared to an envelope flap and the incidence of alveolar osteitis in this study was also higher in the envelope flap group (Analysis 1.1). The risk of alveolar osteitis based on the pooled data from four studies (Briguglio 2011; Goldsmith 2012; Haraji 2010; Kirk 2007) was reduced with the triangular flap (RR 0.26, 95% CI 0.10 to 0.65, P value = 0.004, I2 = 0%; Analysis 1.1),

None of the other five studies evaluating flap design reported the outcome of alveolar osteitis.

Wound infection

Five of the nine studies evaluating different flap designs (Baqain 2012; Briguglio 2011; Goldsmith 2012; Roode 2010; Sandhu 2010) reported the outcome of postoperative infection. One study (Baqain 2012) found no infections in either group. In the other four trials, each made a different comparison between flap designs and there was no evidence that one design was better than another (Analysis 1.2)

Permanent (longer than six months) altered tongue or chin or lower lip sensation

Only one of the nine trials (Briguglio 2011) (45 participants) reported this outcome and found no apparent difference in the number of cases of altered tongue sensation persisting for six or more months between the group with a triangular flap and those patients with an envelope flap (Analysis 1.3).

Pain (after 24 hours)

One study (Briguglio 2011) (45 participants) reported no difference between the two different flap types with regard to the proportion of patients reporting pain 24 hours after surgery (Analysis 1.4).

Four studies reported mean pain scores for each group at 24 hours postsurgery (Baqain 2012; Kirk 2007; Nageshwar 2002; Roode 2010).

In two studies (Baqain 2012; Kirk 2007) (total of 51 participants) comparing a triangular flap with an envelope flap there was a small reduction in mean pain score at 24 hours favouring the triangular flap (mean difference (MD) -0.21, 95% CI -0.32 to -0.10, P value = 0.0003; Analysis 1.5, Subgroup 1.5.1).

The other two studies (Nageshwar 2002; Roode 2010) with 100 and 33 participants, respectively, made a slightly different comparison of flap designs and both found a difference in pain scores at 24 hours between the two flaps compared. However, because the risk of bias in these studies was unclear (Nageshwar 2002) or high, and the findings have not yet been independently replicated, the reductions in pain associated with different flap designs should be interpreted with caution (Analysis 1.5, Subgroups 1.5.2 and 1.5.3).

Swelling (after seven days)

Four studies comparing different flap designs reported the outcome of postoperative facial swelling at seven days. Briguglio 2011 found no difference between the proportion of patients in each group with swelling at seven days (Analysis 1.6). Two studies compared triangular flap versus envelope flap (Baqain 2012; Erdogan 2011:) (total of 39 participants) reported mean swelling in each group. We pooled the data from these two small studies and found a reduction in swelling favouring the envelope flap (MD 0.66, 95% CI 0.26 to 1.07, P value = 0.001, I2 = 0%; Analysis 1.7).

The other three studies (Nageshwar 2002; Roode 2010; Sandhu 2010) which reported swelling after seven days each made slightly different comparisons: two studies (Nageshwar 2002; Sandhu 2010) found some evidence of a difference in swelling between the two types of flap compared, and the other study (Roode 2010) found no difference in swelling between the experimental flap and the envelope flap (Analysis 1.8). The findings of these three studies should be interpreted with caution as each comparison has only been evaluated by a single study.

Trismus (after seven days)

One study (Briguglio 2011) found no difference in the proportion of participants in each group with trismus seven days after surgery (Analysis 1.9). Three other studies (Baqain 2012; Erdogan 2011; Nageshwar 2002) reported mean maximum mouth opening in each group, and one study (Nageshwar 2002) found some evidence of a difference in maximum mouth opening between the different flap types favouring the modified envelope flap (Analysis 1.10). We were unable to pool the data from these studies.

Adverse effects

Only two studies reported any other adverse effects due to the surgery. Wound dehiscence was not a risk in the studies but in two studies (Baqain 2012; Sandhu 2010) this was reported and neither study found a statistically significant difference between the groups. As these studies were small (20 participants per study) and at high risk of bias, this finding should be interpreted with caution pending further research.

Temporary (approximately one month postoperation) altered tongue or chin or lower lip sensation

None of the trials reported this outcome.

Overall there was some weak evidence of a small reduction in pain after 24 hours and a reduction in the development of alveolar osteitis following surgical extraction of mandibular third molars using a triangular flap, compared to envelope flap.

2. Lingual nerve protection

Three studies evaluated the use of retractors for lingual nerve protection. Gargallo-Albiol 2000 (300 participants in a parallel group study at high risk of bias) compared a subperiosteal retractor with no retractor, and Gomes 2005 (55 participants in a split-mouth study at unclear risk of bias) compared a Free's retractor with no retractor. The third study (Greenwood 1994) compared the use of a Howard retractor with a broad retractor.

The meta-analysis of the two studies (Gargallo-Albiol 2000; Gomes 2005) (300 and 55 participants, at high and unclear risk of bias respectively) which compared a retractor with no retractor found more cases of temporary altered sensation associated with the use of retractors at one month (Peto odds ratio (OR) 5.19, 95% CI 1.38 to 19.49, P value = 0.01, I2 = 40%; Analysis 2.1). Gargallo-Albiol 2000 found that there were no cases of lingual nerve disturbance persisting for more than six months in either group.

The direct comparison of two different retractors (Greenwood 1994) (150 participants, high risk of bias) showed evidence of more cases of temporary altered sensation for the Howarth's retractor for lingual nerve protection compared to the Broad retractor (Peto OR 5.83, 95% CI 1.92 to 17.69; Analysis 2.2).

None of these three studies reported the outcomes of alveolar osteitis, infection, pain, swelling or trismus or other adverse effects.

3. Bone removal techniques

Five included studies (total 240 participants) evaluated different bone removal techniques (Absi 1993; Barone 2010; Mocan 1996; Praveen 2007; Rullo 2013). Two studies were at unclear risk of bias (Absi 1993; Barone 2010) and three were assessed at high risk of bias.

The study by Rullo 2013 was a split-mouth study with 52 participants comparing a piezoelectric bone removal device used in the removal of a mandibular third molar from one side of the mouth with the use of conventional rotative instruments used in a separate session to remove the other mandibular third molar. The data were not reported for each randomised group but were analysed according to whether the procedure was simple or complex. We were unsuccessful in obtaining usable data from the authors of this study.

The study by Mocan 1996 randomly allocated 20 participants to either lingual split or buccal bone removal. However, no data were reported for each group for the outcomes of pain, swelling or trismus and we were unable to obtain these data from the authors. The paper stated in the text that there was no difference in pain and swelling between the two groups but "reduced trismus with the lingual split technique".

Alveolar osteitis

None of the studies of bone removal techniques reported the outcome of alveolar osteitis.

Infection

Absi 1993 was the only study in this group which reported incidence of infection and found no difference between the lingual split with chisel and lingual split with bur. There were three unilateral infections in each group in this split-mouth study of 52 participants, in which most of the participants (83%) also had maxillary third molars removed during the same treatment session.

Temporary (approximately one month postoperation) altered tongue or chin or lower lip sensation

Likewise Absi 1993 was the only study which reported data for the outcome of altered tongue or chin sensation within the first month following surgery. There was no evidence of a difference in these outcomes from this single split-mouth study of 52 participants (Analysis 3.2; Analysis 3.3).

Pain (after 24 hours)

Two studies (Barone 2010; Praveen 2007) (26 and 90 participants, respectively) reported data for the outcome of pain after 24 hours. Praveen 2007 conducted a three-arm parallel group study and found reduced pain in the lingual split with bur group compared to lingual split with a chisel, but no difference between lingual split (either bur or chisel) and a 'simplified bone removal technique' (Analysis 3.4).

There was no evidence of a difference in the outcome of mean pain after 24 hours between bone removal by ultrasound compared to a bur in Barone 2010. All these findings should be interpreted with caution until the single trial evaluating each comparison has been independently replicated.

Swelling (after seven days)

Two studies (Barone 2010; Praveen 2007) (26 and 90 participants, respectively) reported data for the outcome of swelling after seven days. Praveen 2007 asked participants to rate swelling on a 'swelling scale' and found no apparent difference in swelling associated with the lingual split with bur group compared to the lingual split with chisel (Analysis 3.5), and no apparent difference in swelling between lingual split (either bur or chisel) and a 'simplified bone removal technique'. There was evidence of significantly reduced swelling for bone removal using ultrasound compared with using a bur in Barone 2010 (MD 0.37, 95% CI 0.20 to 0.54, P value < 0.001). These findings should be interpreted with caution until the single trial evaluating each comparison has been independently replicated.

Trismus (after seven days)

Barone 2010 found no evidence of a difference in maximum mouth opening between the group who had bone removal using ultrasound compared to those patients where bone was removed with a bur.

None of the studies in this group reported the outcomes of persistent lingual nerve damage or other adverse effects.

Due to the small number of studies, the different comparisons evaluated, the variable outcomes reported, and the paucity of useful data we are not able to draw any conclusions concerning bone removal in third molar surgery.

4. Irrigation techniques

The two studies (Butler 1977; Sweet 1976) of irrigation techniques included in this review were both conducted in the 1970s. Both were large split-mouth studies (211 and 99 participants, assessed as high and unclear risk of bias, respectively). Butler 1977 compared high volume saline irrigation with low volume and Sweet 1976 compared mechanical irrigation with manual irrigation, with both groups receiving high volume of irrigant.

Alveolar osteitis

Both studies reported alveolar osteitis, and neither found a statistically significant difference between the two interventions compared (Analysis 4.1).

Infection

Likewise neither study showed a statistically significant difference between the two interventions compared with regard to the outcome of postoperative infection (Analysis 4.2).

Neither of these studies reported outcome data measured at the time points of interest for the other outcomes of this review.

5. Wound closure

Eight included studies compared primary versus secondary wound closure techniques (Bello 2011; Danda 2010; Hashemi 2012; Osunde 2011; Osunde 2012; Pasqualini 2005; Refo'a 2011; Xavier 2008). Three were split-mouth studies (Danda 2010; Hashemi 2012; Xavier 2008); three were assessed at high risk of bias (Danda 2010; Hashemi 2012; Refo'a 2011) and the remainder at unclear risk of bias

Alveolar osteitis

Four studies (Bello 2011; Danda 2010; Hashemi 2012; Pasqualini 2005) reported the outcome of alveolar osteitis. Hashemi 2012 reported no alveolar osteitis in either group and pooled data from the other three trials found no evidence of a difference in the incidence of alveolar osteitis (RR 0.99, 95% CI 0.41 to 2.40, P value = 0.98, with no heterogeneity; Analysis 5.1).

Infection

Two studies also reported the outcome of postoperative infection (Bello 2011; Hashemi 2012). Hashemi 2012 reported no infections in either group and Bello 2011 showed no statistically significant difference between the primary and secondary wound closure groups (Analysis 5.2).

Permanent (longer than six months) altered tongue or chin or lower lip sensation

No studies reported this outcome.

Temporary (approximately one month postoperation) altered tongue or chin or lower lip sensation

No studies reported this outcome.

Pain (after 24 hours)

Four parallel group studies, with a combined total of 414 participants, reported the outcome of mean pain in each group after 24 hours (Bello 2011; Osunde 2011; Osunde 2012; Pasqualini 2005) and found evidence of a reduction in pain favouring the secondary closure group (MD 0.79, 95% CI 0.35 to 1.24, P value = 0.0005, I2 = 85%; Analysis 5.3). There was substantial heterogeneity within these studies likely to be due to differences between the studies in the type of incision used and whether the secondary closure group had some or no sutures.

Swelling (after seven days)

Data for mean swelling in each group were reported by seven studies (Bello 2011; Danda 2010; Hashemi 2012; Osunde 2011; Osunde 2012; Pasqualini 2005; Xavier 2008) with a combined total of 555 participants. Meta-analysis of these data showed a reduction in swelling favouring the secondary closure groups (MD 0.33, 95% CI 0.09 to 0.57, P value = 0.007, I2 = 89%; Analysis 5.4). The substantial heterogeneity in this estimate was likely to be due to differences in both study design (three split-mouth studies and five parallel group studies) and also in the incision shape and number of sutures used to close the wound in the secondary closure groups.

Trismus (after seven days)

Three parallel group studies (Bello 2011; Osunde 2011; Osunde 2012) with a combined total of 212 participants reported the outcome of mean maximum mouth opening in each group after seven postoperative days. Meta-analysis of these data showed no evidence of a difference in maximum mouth opening between the two groups (MD -0.14, 95% CI -0.41 to 0.13, P value = 0.3, I2 = 0%; Analysis 5.5).

Adverse effects

One study (Bello 2011) reported the adverse effect of reactionary bleeding and reported no statistically significant difference in the number of participants with bleeding in the two arms of the study (Analysis 5.6).

Pasqualini 2005 reported wound dehiscence in 33 of the 100 participants who had primary wound closure. In an e-mail communication Dr Pasqualini reported "the parameter 'wound dehiscence' is only descriptive of the primary closure group and no comparison or inference should be done with the secondary closure group, where the mucosal dehiscence was systematically created by the surgeon. Lastly, despite the more favourable trend in pain and swelling scores, I would underline the more difficult cleaning of the wound (in the secondary closure group) in the weeks following suture removal, compared to those cases of primary closure where no dehiscence occurred, as a drawback of the secondary closure."

6. Surgical drain versus no drain (six trials)

Six included studies (Cerqueira 2004; Chukwuneke 2008; de Brabander 1988; Rakprasitkul 1997; Saglam 2003; Srinivas 2006) evaluated the use of a surgical drain following surgical extraction of mandibular third molars. Two of these studies used a parallel group design (Chukwuneke 2008; de Brabander 1988) and four used a split-mouth design in which one side of the mouth was randomly allocated a surgical drain and the other was not. One split-mouth study performed the surgery in two sessions with a two-month period between surgeries (Rakprasitkul 1997). The other three split-mouth studies allocated a drain randomly to one side of the mouth following bilateral extraction of mandibular third molars during the same session. One study was assessed at unclear risk of bias (de Brabander 1988) and the other five studies were all at high risk of bias.

None of these six studies evaluating surgical drains reported the outcomes of alveolar osteitis, infection, permanent or temporary altered tongue or chin or lower lip sensation, or adverse effects.

Pain (after 24 hours)

Three studies evaluating the use of surgical drains reported the outcome of pain (Cerqueira 2004; Chukwuneke 2008; Srinivas 2006). Two of these studies (Cerqueira 2004; Srinivas 2006) used a split-mouth design in which both mandibular third molars were extracted in the same operating session and a drain was inserted on one randomly chosen side. We considered that pain experienced by these patients could not be reliably attributed to the use of a drain or not, and so in this review we did not use pain data from split-mouth studies in which participants received both interventions during a single operative session.

Chukwuneke 2008 used a parallel group study design and found no difference in the mean pain score in the two groups (Analysis 6.1).

Swelling (after seven days)

All of the studies evaluating surgical drains reported the outcome of postoperative swelling.

One study evaluating a gauze drain used a 'u-formed calliper' to measure the thickness of the cheek (de Brabander 1988) and this study found no statistically significant difference in swelling between the two groups (MD 0.18, 95% CI -0.06 to 0.42, P value = 0.32).

The other five studies, all evaluating tube drains, took the sum of a horizontal and vertical measurement across the cheek and expressed swelling as the percentage difference compared to the preoperative measure. Meta-analysis of the data from the five studies which measured swelling in this way showed a reduction in swelling with the use of a drain (MD -0.90, 95% CI -1.62 to -0.19, P value = 0.01, I2 = 88%; Analysis 6.2). The considerable heterogeneity in this meta-analysis may be partly due to differences in the design of both the drains and the studies.

Trismus (after seven days)

The effect of the use of a surgical drain on trismus was assessed by comparing mean maximal mouth opening one week after surgery in each group. This outcome was reported by five of the included studies (Cerqueira 2004; Chukwuneke 2008; Rakprasitkul 1997; Saglam 2003; Srinivas 2006). However, it was our assessment that this outcome cannot be reliably attributed to intervention or control in split-mouth studies where both mandibular molars are extracted in the same operative session, so maximum mouth opening data from Cerqueira 2004; Saglam 2003, and Srinivas 2006 were not used in this review.

Data from studies by Chukwuneke 2008 (parallel group RCT) and Rakprasitkul 1997 (split-mouth RCT with two operative sessions two months apart) were combined in a meta-analysis and we showed that maximum mouth opening was greater in those patients who had received a surgical drain (MD 3.72 mm, 95% CI 2.84 to 4.59, P value < 0.001, I2 = 0%; Analysis 6.3).

7. Coronectomy versus complete tooth removal (two trials)

Two trials (Leung 2009; Renton 2005) included in this review set out to compare extraction of mandibular third molars with a coronectomy procedure in which only the top part of the impacted third molar was removed. Both studies were assessed at high risk of bias due to the high attrition in the coronectomy group and unit of analysis errors. Both studies included participants whose preoperative radiographs indicated proximity of the third molar root(s) to the inferior alveolar nerve. Some participants contributed two teeth to the study and some only one.

In both studies, in some of the third molars (9.4% and 38% in Leung 2009 and Renton 2005, respectively) randomly allocated to the coronectomy, the roots were inadvertently mobilised and were therefore completely removed. Also in both studies the analysis did not take account of the pairing of the data (two teeth from a single participant were incorrectly assumed to have the same independence as two teeth from two different participants). We considered that the data from these studies were not sufficiently reliable to include in this review.

However, both studies suggested that in patients where third molar roots were very close to the nerve canal, it was likely that coronectomy was associated with a reduction in nerve damage and no increase in alveolar osteitis, infection or pain in the short term.

The adverse effect of migration of the root segments occurred following coronectomy in 13% of those evaluated at 13 months in the Renton 2005 study, but extraction was not required. The proportion of participants with root migration at 12 months was 24% in Leung 2009 and two patients in the coronectomy group experienced root exposure. The mean follow-up in Leung 2009 was less than one year and in Renton 2005 follow-up was only two years, so that neither study was long enough to reliably assess whether there were any long-term adverse effects associated with coronectomy.

Discussion

Summary of main results

See Summary of findings for the main comparison; Summary of findings 2; Summary of findings 3; Summary of findings 4; Summary of findings 5; Summary of findings 6; Summary of findings 7.

The evidence included in this review is from 35 randomised controlled trials (RCTs) evaluating different aspects of the surgical removal of mandibular third molars. However, due to the number of different comparisons evaluated and the overall poor quality of the research these trials provide moderate to very low quality evidence, which means that further research may change the estimates and the confidence intervals presented. The overall results for the seven broad categories are summarised below.

  • There is moderate quality evidence that triangular flaps are associated with a reduction in alveolar osteitis at one week and pain at 24 hours compared to envelope flaps. There is low quality evidence of no difference in overall infection rates and maximum mouth opening. However, there is low quality evidence that residual swelling after one week is slightly increased in the triangular flap groups compared to envelope flap types.

  • There is low quality evidence from two studies that the use of a retractor during third molar extraction leads to more cases of temporary altered tongue, lip or chin skin sensation (up to one month). One study reported this altered sensation did not persist for more than six months in either group.

  • Due to the small number of studies, the different comparisons evaluated, the variable outcomes reported and the paucity of useful data we are not able to draw any conclusions concerning bone removal in third molar surgery.

  • There is insufficient evidence (very low quality) from single studies on irrigation method (manual versus mechanical) or irrigation volume (low or high) to determine whether there were differences or not for the outcomes of alveolar osteitis or postoperative infection.

  • There is insufficient evidence (low to very low quality) that any wound closure technique (primary versus secondary) is superior to another for the outcomes of alveolar osteitis, postoperative infection, maximum mouth opening achieved after seven days, or reactionary bleeding. There is moderate quality evidence that secondary wound closure is associated with reduced pain at 24 hours and slightly reduced swelling after one week.

  • There is moderate quality evidence that the use of a surgical drain is associated with less postoperative swelling and greater maximum mouth opening one week after surgery. There is insufficient evidence from a single study (low quality) to determine whether or not there is a difference in pain at 24 hours postoperation.

  • Although two RCTs compared coronectomy with complete extraction, flaws in the design and unit of analysis of these studies mean that there is no reliable data available for inclusion in this review. It seems likely that when the third molar root(s) are close to the nerve canal, coronectomy and complete extraction are comparable in terms of the outcomes of alveolar osteitis, infection and pain, and coronectomy is associated with fewer cases of short term nerve disturbances. However, within the relatively short-term follow-up (one to two years) root migration was observed in a substantial proportion (13% to 24%) of coronectomy patients, with root exposure in 1% of patients after 12 months.

Overall completeness and applicability of evidence

Although there are 35 trials included in this review, these describe comparisons related to seven aspects of the surgical procedures for extracting impacted mandibular third molars: type of surgical flap raised, use of retractor, techniques for bone removal, wound irrigation, wound closure, wound drainage, and complete or incomplete tooth removal. There may be only one or two trials evaluating a given aspect of the surgery or a specific comparison. Also, many trials only reported some of the primary and secondary outcomes of interest to this review. None of the trials reported the primary outcome, fracture of the mandible.

More than half of the RCTs included in this review used a split-mouth design in which one tooth was allocated to the intervention and the other to control. While this design may be efficient, if both teeth are treated in the same operative session (as was the case for most of these trials) it may be difficult to attribute outcomes such as pain or trismus to intervention or control. However, we recognise that it may also be undesirable to require participants to undergo two separate procedures, some weeks apart, in order to ascertain the effect of the intervention on these outcomes. Furthermore, the paired nature of the outcome data from split-mouth trials should be recognised in the analysis, and whether or not this was done was not always clear in the published papers.

It is unclear to what extent trials undertaken more than 25 years ago remain relevant to contemporary practice.

Quality of the evidence

The body of evidence for most of the comparisons that were included were at moderate to very low quality. This was because most of the comparisons were based on a small number of trials and participants, and the trials were at high (65%) or unclear (35%) risk of bias. Although it was possible in many trials to undertake a blinded outcome assessment this was only done in about 30%.

The follow-up period of the trial, especially for coronectomy, should be much longer in order to ensure that undertaking a quicker simpler surgical procedure is not followed by later complications such as root migration and root exposure requiring further surgery, and compromising the integrity of the individual's remaining molars.

Potential biases in the review process

We used contemporary methodology and comprehensive search strategies covering multiple databases. We made some assumptions when undertaking the analysis of the split-mouth trials, when the data were not reported, but felt this would introduce less bias than omitting the data from these trials. More up-to-date methods were used for quality assessment that specified in the protocol but we do not feel this should have caused any bias.

Agreements and disagreements with other studies or reviews

A systematic review (Carrasco-Labra 2012) on primary versus secondary wound closure has been published and the results do not support a preference for either approach, which is in disagreement with this review. Another systematic review (Long 2012) compared coronectomy versus total removal, and included four trials (two RCTs and two controlled clinical trials). The authors conclude that coronectomy is preferable to total removal of a wisdom tooth. This disagrees with the finding of this Cochrane review, which was unable to extract the same data. Lingual split with chisel versus bur for bone removal was the focus of another systematic review (Steel 2012) which included five studies. The tentative conclusions were of no difference between postoperative pain and swelling, and some evidence of less trismus, for the lingual split technique. This was similar to findings of our review although we did not find a difference for trismus. A further systematic review (Pichler 2001) looked at lingual flap retraction and prevention of lingual nerve damage associated with third molar surgery. The authors compared eight studies, seven were prospective clinical series and one was described as an RCT (Robinson 1996), which was excluded from our Cochrane review due to the high number of protocol violations resulting in unquantifiable biases. They concluded that the use of a lingual nerve retractor during third molar surgery was associated with an increased incidence of temporary nerve damage and was neither protective nor detrimental with respect to the incidence of permanent nerve damage. We agree with their findings.

Authors' conclusions

Implications for practice

Thirty-five trials were included in this review looking at a range of different surgical techniques. The comparisons in the trials related to seven broad aspects of the surgical procedures for impacted mandibular third molar removal: type of surgical flap raised, use of retractor, techniques for bone removal, wound irrigation, wound closure, wound drainage, and complete or incomplete tooth removal. The quality of the body of evidence for each of these comparisons was very low to moderate due to the small number of trials and patients, and the majority of the trials being at either high risk of bias (65%) or unclear risk of bias (35%).

Recommending changes to surgical practice must therefore be limited. However, it is useful to describe the state of the research evidence supporting practice so that surgeons can make an informed choice in adopting new techniques, or continuing with established techniques.

Implications for research

The risk of bias for the included studies was at high risk of bias (65%), or unclear. More consistent use of the CONSORT statement for reporting of randomised controlled clinical trials would increase the value of research.

  1. Detailed reporting of methods, such as generation of allocation sequence, allocation concealment, and numbers and reasons for withdrawals and exclusions.

  2. Blinding of outcome assessment if possible.

  3. Full reporting of methods used to measure facial swelling.

  4. Reporting of adverse effects of interventions.

Studies of a split-mouth design may be appropriate for comparing different surgical techniques for third molar removal but trialists need to consider which outcomes are appropriate and the design and analysis of these studies. It would be helpful to have a consensus agreement on the criteria for the measurement of facial swelling. There is a need for trials looking at coronectomy versus complete tooth removal, with a long-term follow-up so any adverse events such as root migration and root exposure requiring further surgery can be determined.

Acknowledgements

We wish to thank Anne Littlewood (Cochrane Oral Health Group) for her assistance with literature searching and Philip Riley (Cochrane Oral Health Group) for his help with the preparation of this review.
Our thanks to Niall Quigley for contributing to previous versions of the review and to Professor K Thongprasom for her assistance in locating and translating a paper published in the Thai language.

Data and analyses

Download statistical data

Comparison 1. Surgical flap type (A versus B)
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Alveolar osteitis (7 days)4 Risk Ratio (Fixed, 95% CI)Subtotals only
1.1 Triangular flap versus envelope flap3 Risk Ratio (Fixed, 95% CI)0.29 [0.11, 0.78]
1.2 Pedicle flap versus envelope flap1 Risk Ratio (Fixed, 95% CI)0.09 [0.01, 1.60]
2 Infection (7 days)4 Risk Ratio (Fixed, 95% CI)Subtotals only
2.1 Triangular flap versus envelope flap1 Risk Ratio (Fixed, 95% CI)1.55 [0.07, 35.89]
2.2 Pedicle versus envelope flap1 Risk Ratio (Fixed, 95% CI)0.20 [0.01, 4.07]
2.3 Modified triangular versus alternative single incision flap1 Risk Ratio (Fixed, 95% CI)0.33 [0.04, 3.04]
2.4 Bayonet flap versus envelope flap1 Risk Ratio (Fixed, 95% CI)5.00 [0.26, 97.99]
3 Permanently altered tongue sensation (> 6 months)1 Peto Odds Ratio (Peto, Fixed, 95% CI)Totals not selected
3.1 Triangular flap versus envelope flap1 Peto Odds Ratio (Peto, Fixed, 95% CI)0.0 [0.0, 0.0]
4 Pain at 24 hours1 Risk Ratio (M-H, Fixed, 95% CI)Totals not selected
4.1 Triangular flap versus envelope flap1 Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
5 Mean pain at 24 hours4 Mean Difference (Fixed, 95% CI)Subtotals only
5.1 Triangular flap versus envelope flap2 Mean Difference (Fixed, 95% CI)-0.21 [-0.32, -0.10]
5.2 Comma shaped incision versus modified envelope flap1 Mean Difference (Fixed, 95% CI)-1.18 [-1.37, -0.99]
5.3 Modified triangular versus alternative single incision flap1 Mean Difference (Fixed, 95% CI)-0.8 [-1.41, -0.19]
6 Swelling (after 1 week)1 Risk Ratio (M-H, Fixed, 95% CI)Totals not selected
6.1 Triangular flap versus envelope flap1 Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
7 Swelling (after 1 week)3 Mean Difference (Fixed, 95% CI)Subtotals only
7.1 Triangular flap versus envelope flap2 Mean Difference (Fixed, 95% CI)0.66 [0.26, 1.07]
7.2 Pedicle flap versus envelope flap1 Mean Difference (Fixed, 95% CI)0.21 [-0.09, 0.51]
8 Swelling (after 1 week)3 Mean Difference (Fixed, 95% CI)Subtotals only
8.1 Comma shaped incision versus modified envelope flap1 Mean Difference (Fixed, 95% CI)-2.38 [-2.81, -1.95]
8.2 Modified triangular versus alternative single incision flap1 Mean Difference (Fixed, 95% CI)0.20 [-0.22, 0.62]
8.3 Bayonet flap versus envelope flap1 Mean Difference (Fixed, 95% CI)0.71 [0.20, 1.21]
9 Trismus (after 1 week)1 Risk Ratio (M-H, Fixed, 95% CI)Totals not selected
9.1 Triangular flap versus envelope flap1 Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
10 Maximum mouth opening (after 1 week)3 Mean Difference (IV, Fixed, 95% CI)Subtotals only
10.1 Triangular flap versus envelope flap278Mean Difference (IV, Fixed, 95% CI)-0.81 [-4.94, 3.32]
10.2 Comma shaped incision versus modified envelope flap1100Mean Difference (IV, Fixed, 95% CI)-1.20 [-2.33, -0.07]
11 Adverse events- wound dehiscience (up to 30 days)2 Risk Ratio (Fixed, 95% CI)Totals not selected
11.1 Triangular flap versus envelope flap1 Risk Ratio (Fixed, 95% CI)0.0 [0.0, 0.0]
11.2 Bayonet flap versus envelope flap1 Risk Ratio (Fixed, 95% CI)0.0 [0.0, 0.0]
Analysis 1.1.

Comparison 1 Surgical flap type (A versus B), Outcome 1 Alveolar osteitis (7 days).

Analysis 1.2.

Comparison 1 Surgical flap type (A versus B), Outcome 2 Infection (7 days).

Analysis 1.3.

Comparison 1 Surgical flap type (A versus B), Outcome 3 Permanently altered tongue sensation (> 6 months).

Analysis 1.4.

Comparison 1 Surgical flap type (A versus B), Outcome 4 Pain at 24 hours.

Analysis 1.5.

Comparison 1 Surgical flap type (A versus B), Outcome 5 Mean pain at 24 hours.

Analysis 1.6.

Comparison 1 Surgical flap type (A versus B), Outcome 6 Swelling (after 1 week).

Analysis 1.7.

Comparison 1 Surgical flap type (A versus B), Outcome 7 Swelling (after 1 week).

Analysis 1.8.

Comparison 1 Surgical flap type (A versus B), Outcome 8 Swelling (after 1 week).

Analysis 1.9.

Comparison 1 Surgical flap type (A versus B), Outcome 9 Trismus (after 1 week).

Analysis 1.10.

Comparison 1 Surgical flap type (A versus B), Outcome 10 Maximum mouth opening (after 1 week).

Analysis 1.11.

Comparison 1 Surgical flap type (A versus B), Outcome 11 Adverse events- wound dehiscience (up to 30 days).

Comparison 2. Lingual nerve protection
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Temporary altered sensation (up to 1 month)2410Peto Odds Ratio (Peto, Fixed, 95% CI)5.19 [1.38, 19.49]
1.1 Subperiosteal LNP versus none1300Peto Odds Ratio (Peto, Fixed, 95% CI)3.07 [0.43, 22.06]
1.2 Free's retractor LNP versus none1110Peto Odds Ratio (Peto, Fixed, 95% CI)7.97 [1.34, 47.58]
2 Temporary altered sensation (up to 1 month)1 Peto Odds Ratio (Peto, Fixed, 95% CI)Totals not selected
2.1 Howarth's retractor LNP versus Broad retractor1 Peto Odds Ratio (Peto, Fixed, 95% CI)0.0 [0.0, 0.0]
Analysis 2.1.

Comparison 2 Lingual nerve protection, Outcome 1 Temporary altered sensation (up to 1 month).

Analysis 2.2.

Comparison 2 Lingual nerve protection, Outcome 2 Temporary altered sensation (up to 1 month).

Comparison 3. Bone removal techniques
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Infection (7 days)1 Risk Ratio (M-H, Fixed, 95% CI)Totals not selected
1.1 Lingual split chisel versus lingual split bur1 Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
2 Temporary alteration of tongue sensation (<1 month)1 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
2.1 Lingual split chisel versus lingual split bur1104Risk Ratio (M-H, Fixed, 95% CI)3.0 [0.13, 71.99]
3 Temporary alteration of chin sensation (<1 month)1 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
3.1 Lingual split chisel versus lingual split bur1104Risk Ratio (M-H, Fixed, 95% CI)0.25 [0.03, 2.16]
4 Pain at 24 hours2 Mean Difference (IV, Fixed, 95% CI)Subtotals only
4.1 Lingual split chisel versus lingual split bur160Mean Difference (IV, Fixed, 95% CI)0.5 [0.08, 0.92]
4.2 Ultrasound versus bur126Mean Difference (IV, Fixed, 95% CI)-0.20 [-1.32, 0.92]
4.3 Lingual split (bur or chisel) versus simplified bone removal190Mean Difference (IV, Fixed, 95% CI)-0.15 [-0.43, 0.13]
5 Swelling (after 7 days)2 Mean Difference (IV, Fixed, 95% CI)Subtotals only
5.1 Lingual split chisel versus lingual split bur160Mean Difference (IV, Fixed, 95% CI)-0.10 [-0.33, 0.13]
5.2 Ultrasound versus bur126Mean Difference (IV, Fixed, 95% CI)0.37 [0.20, 0.54]
5.3 Lingual split (bur or chisel) versus simplified bone removal190Mean Difference (IV, Fixed, 95% CI)0.15 [-0.01, 0.31]
6 Mouth opening (after 7 days)1 Mean Difference (IV, Fixed, 95% CI)Subtotals only
6.1 Ultrasound versus bur126Mean Difference (IV, Fixed, 95% CI)2.90 [-0.27, 6.07]
Analysis 3.1.

Comparison 3 Bone removal techniques, Outcome 1 Infection (7 days).

Analysis 3.2.

Comparison 3 Bone removal techniques, Outcome 2 Temporary alteration of tongue sensation (<1 month).

Analysis 3.3.

Comparison 3 Bone removal techniques, Outcome 3 Temporary alteration of chin sensation (<1 month).

Analysis 3.4.

Comparison 3 Bone removal techniques, Outcome 4 Pain at 24 hours.

Analysis 3.5.

Comparison 3 Bone removal techniques, Outcome 5 Swelling (after 7 days).

Analysis 3.6.

Comparison 3 Bone removal techniques, Outcome 6 Mouth opening (after 7 days).

Comparison 4. Irrigation techniques (A versus B)
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Alveolar osteitis (7 days)2 Risk Ratio (Fixed, 95% CI)Subtotals only
1.1 Mechanical versus manual surgical wound irrigation1 Risk Ratio (Fixed, 95% CI)0.33 [0.01, 8.09]
1.2 High volume versus low volume wound irrigation1 Risk Ratio (Fixed, 95% CI)0.52 [0.27, 1.02]
2 Infection (7 days)2 Risk Ratio (Fixed, 95% CI)Subtotals only
2.1 Mechanical versus manual surgical wound irrigation1 Risk Ratio (Fixed, 95% CI)0.50 [0.05, 5.43]
2.2 High volume versus low volume wound irrigation1 Risk Ratio (Fixed, 95% CI)0.17 [0.02, 1.37]
Analysis 4.1.

Comparison 4 Irrigation techniques (A versus B), Outcome 1 Alveolar osteitis (7 days).

Analysis 4.2.

Comparison 4 Irrigation techniques (A versus B), Outcome 2 Infection (7 days).

Comparison 5. Primary versus secondary wound closure
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Alveolar osteitis (7 days)3 Risk Ratio (Fixed, 95% CI)0.99 [0.41, 2.40]
2 Infection (7 days)1 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
3 Pain at 24 hours4414Mean Difference (IV, Random, 95% CI)0.79 [0.35, 1.24]
4 Swelling (after 7 days)7 Mean Difference (Random, 95% CI)0.33 [0.09, 0.57]
5 Maximum mouth opening (after 7 days)3214Std. Mean Difference (IV, Fixed, 95% CI)-0.14 [-0.41, 0.13]
6 Adverse effects - Reactionary bleeding182Risk Ratio (M-H, Fixed, 95% CI)0.41 [0.11, 1.47]
Analysis 5.1.

Comparison 5 Primary versus secondary wound closure, Outcome 1 Alveolar osteitis (7 days).

Analysis 5.2.

Comparison 5 Primary versus secondary wound closure, Outcome 2 Infection (7 days).

Analysis 5.3.

Comparison 5 Primary versus secondary wound closure, Outcome 3 Pain at 24 hours.

Analysis 5.4.

Comparison 5 Primary versus secondary wound closure, Outcome 4 Swelling (after 7 days).

Analysis 5.5.

Comparison 5 Primary versus secondary wound closure, Outcome 5 Maximum mouth opening (after 7 days).

Analysis 5.6.

Comparison 5 Primary versus secondary wound closure, Outcome 6 Adverse effects - Reactionary bleeding.

Comparison 6. Surgical drain versus no drain
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Pain at 24 hours1 Mean Difference (IV, Fixed, 95% CI)Totals not selected
2 Swelling at 7 days5 Mean Difference (Random, 95% CI)-0.90 [-1.62, -0.19]
3 Mouth opening at 7 days2 Mean Difference (Fixed, 95% CI)3.72 [2.84, 4.59]
Analysis 6.1.

Comparison 6 Surgical drain versus no drain, Outcome 1 Pain at 24 hours.

Analysis 6.2.

Comparison 6 Surgical drain versus no drain, Outcome 2 Swelling at 7 days.

Analysis 6.3.

Comparison 6 Surgical drain versus no drain, Outcome 3 Mouth opening at 7 days.

Appendices

Appendix 1. The Cochrane Oral Health Group's Trials Register search strategy

From February 2013, searches of the Oral Health Group's Trials Register were using the Cochrane Register of Studies and the search strategy below:

#1 ("third molar*" or "3rd molar*" or "mandibular molar*" or "maxillary molar*")
#2 (wisdom AND (tooth or teeth))
#3 #1 or #2
#4 (extract* or remov* or surg*)
#5 (#3 and #4) AND (INREGISTER)

Previous searches of the Oral Health Group's Trials Register for this review were undertaken using the Procite software and the search strategy below:

((molar-third OR "molar,-third" OR "molar,- third" OR "Molar, Third" OR "third molar*" OR "3rd molar*" OR (wisdom AND (tooth OR teeth)) OR "third mandibular molar*" OR "third maxillary molar*") AND (tooth-extraction* OR extract* OR remov* OR (tooth-impacted AND surg*)))

Appendix 2. Cochrane Central Register of Controlled Trials (CENTRAL) search strategy

#1 MOLAR THIRD single term (MeSH)
#2 (wisdom next tooth)
#3 (wisdom next teeth)
#4 (third near molar*)
#5 (#1 or #2 or #3 or #4)
#6 TOOTH EXTRACTION single term (MeSH)
#7 (extract* near tooth)
#8 (extract* near teeth)
#9 (extract* near (third next molar*))
#10 (remov* near tooth)
#11 (remov* near teeth)
#12 (surgical* near remov*)
#13 (surgery near remov*)
#14 (surgical* near extract*)
#15 (surgery near extract*)
#16 (#6 or #7 or #8 or #9 or #10 or #11 or #12 or #13 or #14 or #15)
#17 (#5 and #16)

Appendix 3. MEDLINE (OVID) search strategy

1. Molar, Third/
2. (third adj6 molar$).mp.
3. (wisdom adj tooth).mp.
4. (wisdom adj teeth).mp.
5. or/1-4
6. Tooth Extraction/
7. (extract$ adj6 tooth).mp.
8. (extract$ adj6 teeth).mp.
9. (extract$ adj6 (third adj molar$)).mp.
10. (extract$ adj6 (third adj3 molar$)).mp.
11. (remov$ adj6 tooth).mp.
12. (remov$ adj6 teeth).mp.
13. (surgical$ adj3 remov$).mp.
14. (surgery adj3 remov$).mp.
15. (surgical$ adj3 extract$).mp.
16. (surgery adj3 extract$).mp.
17. or/6-16
18. 5 and 17

The above subject search was linked to the Cochrane Highly Sensitive Search Strategy (CHSSS) for identifying randomised trials in MEDLINE: sensitivity maximising version (2008 revision) as referenced in Chapter 6.4.11.1 and detailed in box 6.4.c of theCochrane Handbook for Systematic Reviews of Interventions, Version 5.1.0 [updated March 2011].

1. randomized controlled trial.pt.
2. controlled clinical trial.pt.
3. randomized.ab.
4. placebo.ab.
5. drug therapy.fs.
6. randomly.ab.
7. trial.ab.
8. groups.ab.
9. or/1-8
10. exp animals/ not humans.sh.
11. 9 not 10

Appendix 4. EMBASE (OVID) search strategy

1. Molar, Third/
2. (third adj6 molar$).mp.
3. (wisdom adj tooth).mp.
4. (wisdom adj teeth).mp.
5. or/1-4
6. Tooth Extraction/
7. (extract$ adj6 tooth).mp.
8. (extract$ adj6 teeth).mp.
9. (extract$ adj6 (third adj molar$)).mp.
10. (extract$ adj6 (third adj3 molar$)).mp.
11. (remov$ adj6 tooth).mp.
12. (remov$ adj6 teeth).mp.
13. (surgical$ adj3 remov$).mp.
14. (surgery adj3 remov$).mp.
15. (surgical$ adj3 extract$).mp.
16. (surgery adj3 extract$).mp.
17. or/6-16
18. 5 and 17

The above subject search was linked to the Cochrane Oral Health Group filter for EMBASE via OVID:

1. random$.ti,ab.
2. factorial$.ti,ab.
3. (crossover$ or cross over$ or cross-over$).ti,ab.
4. placebo$.ti,ab.
5. (doubl$ adj blind$).ti,ab.
6. (singl$ adj blind$).ti,ab.
7. assign$.ti,ab.
8. allocat$.ti,ab.
9. volunteer$.ti,ab.
10. CROSSOVER PROCEDURE.sh.
11. DOUBLE-BLIND PROCEDURE.sh.
12. RANDOMIZED CONTROLLED TRIAL.sh.
13. SINGLE BLIND PROCEDURE.sh.
14. or/1-13
15. (exp animal/ or animal.hw. or nonhuman/) not (exp human/ or human cell/ or (human or humans).ti.)
16. 14 NOT 15

Appendix 5. US National Institutes of Health Trials Register (ClinicalTrials.gov) search strategy

(surgery and ("third molar" or "3rd molar" or "wisdom tooth" or "wisdom teeth"))

Appendix 6. WHO International Clinical Trials Registry Platform search strategy

Advanced search:

Title: "third molar"
Intervention: surgery

History

Protocol first published: Issue 3, 2003
Review first published: Issue 7, 2014

DateEventDescription
12 May 2009AmendedConverted to new review format.

Contributions of authors

Conceiving, designing and co-ordination of the review (Paul Coulthard (PC).
Undertaking the searches (PC, Marco Esposito (ME)).
Screening search results and retrieval of papers against inclusion criteria (PC, ME, Edmund Bailey (EB)).
Risk of bias assessment and extracting data from papers (PC, ME, Tara Renton (TR), EB, Susan Furness (SF)).
Writing to authors for additional information (PC, EB).
Data management for the review and entering data into RevMan (Helen Worthington (HW), SF, EB).
Analysis and interpretation of data (HW, PC, EB, SF).
Writing the review (PC, EB, SF).
Providing general advice on the review (ME, TR).

Declarations of interest

Tara Renton is also the author of a study included in this review; however, she was not involved in its quality assessment.
Paul Coulthard, Edmund Bailey, Marco Esposito, Susan Furness and Helen V Worthington: no interests to declare.

Sources of support

Internal sources

  • School of Dentistry, The University of Manchester, UK.

  • The Sahlgrenska Academy at Goteborg University, Sweden.

  • GKTDI, King's College, London, UK.

  • MAHSC, UK.

    The Cochrane Oral Health Group is supported by the Manchester Academic Health Sciences Centre (MAHSC) and the NIHR Manchester Biomedical Research Centre.

External sources

  • Swedish Medical Research Council (9495), Sweden.

  • Cochrane Oral Health Group Global Alliance, UK.

    All reviews in the Cochrane Oral Health Group are supported by Global Alliance member organisations (British Association of Oral Surgeons, UK; British Orthodontic Society, UK; British Society of Paediatric Dentistry, UK; British Society of Periodontology, UK; Canadian Dental Hygienists Association, Canada; Mayo Clinic, USA; National Center for Dental Hygiene Research & Practice, USA; New York University College of Dentistry, USA; and Royal College of Surgeons of Edinburgh, UK) providing funding for the editorial process (http://ohg.cochrane.org/).

  • National Institute for Health Research (NIHR), UK.

    CRG funding acknowledgement:
    The NIHR is the largest single funder of the Cochrane Oral Health Group.

    Disclaimer:
    The views and opinions expressed therein are those of the authors and do not necessarily reflect those of the NIHR, NHS or the Department of Health.

Differences between protocol and review

During the preparation of this review it was decided to exclude studies of germectomy as this procedure is fundamentally different from the extraction of impacted mandibular third molars. We also excluded studies looking at 'periodontal outcomes relating to the second permanent molar from the review', although this was not specified in the protocol.

There have been some changes to the prespecified outcomes and prioritisation of outcomes.

The protocol was published in 2003 and new methods have been applied for: quality assessment, inclusion of Summary of Findings tables and GRADE.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Absi 1993

Methods

Study design: RCT (split-mouth)

Conducted in: Department of Oral Surgery, University Dental School, Cardiff, Wales, UK

Participants

Inclusion criteria: 52 consecutive healthy patients scheduled for surgery entered the trial after assessment with dental panoramic radiograph. All had similarly impacted bilateral lower third molars

Exclusion criteria: Patients were excluded if they had pericoronitis in the 6 weeks before surgery, or if they were allergic to any of the drugs in the standard regimen

Age: Mean 22 years

Number of patients randomised: 52

Number of patients evaluated: 52

Interventions

Lingual split with chisel versus bur for bone removal under general anaesthesia

Group A (n = 52 teeth): Lingual split with chisel for bone removal

Group B (n = 52 teeth): Lingual split with bur for bone removal

Follow-up: 4 weeks

All procedures were carried out under general anaesthetic. 43/52 patients had maxillary third molars extracted in same session

OutcomesQuestionnaire assessment of lingual and inferior alveolar nerve function, swelling and pain were measured by a 4-point scale at 6 hrs, 24 hrs, 48 hrs and 7 days after the procedure. Patients also asked to indicate which side they felt was more swollen at these intervals. Infection was assessed by presence of dry sockets or purulence or both
Notes

Sample size calculation: Not reported

Email sent to author (12/2/2003). Unpublished data supplied

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

Quote: "...the method for a particular side and the third molar removed first were selected randomly"

After contacting author, method of randomisation disclosed as "nurse blindly selecting a piece of paper on which was written either 'left' or 'right' from a bag in which were placed equal numbers of pieces of paper with 'left' and 'right' written on them"

Allocation concealment (selection bias)Low riskThe operator was blinded to the above randomisation procedure. Comment: Allocation concealed
Blinding (performance bias and detection bias)
patient
Low risk

Quote: "...the trial was single-blind to the patient as far as the surgical method was concerned"

Comment: As the procedures were carried out under general anaesthetic, then it should be considered that the patients were blinded to which side received which intervention

Blinding (performance bias and detection bias)
assessor
Low risk

For pain, swelling, and sensory disturbances the outcome assessors (the patients) were blinded

"on day 7... the wounds were examined by an independent observer"

Incomplete outcome data (attrition bias)
All outcomes
Low riskNo drop-outs
Selective reporting (reporting bias)Low riskAlthough author was unable to specify which group (and indeed which socket - upper or lower) received an acute abscess, despite direct correspondence on the matter
Other biasUnclear risk

In 43 of the 52 patients, maxillary 3rd molars were also extracted, however after contacting the author he assured us that none of these were surgical

Author notes "higher complication rates might have been found if this information had been supplemented by clinical assessments", noting the limitations of subjective assessment

Baqain 2012

Methods

Study design: Split-mouth RCT

Conducted in: Jordan University Hospital, Amman, Jordan

Participants

Inclusion criteria: symmetrically impacted mandibular third molars, with comparable positioning and angulation, no acute local inflammation or pathology

Exclusion criteria: systemic diseases, pregnancy, lactation, smokers, medications that would influence the surgical procedure or wound healing

Age: Mean 21.4 ± 2.3, 18-26 years

Number of patients randomised: 20

Number of patients evaluated: 19

Interventions

Buccal envelope flap versus triangular flap

Group A (n = 19 teeth): Sulcular incision from first to second mandibular molar with distal incision along mandibular ramus

Group B (n = 19 teeth): Incision commenced distally from the mandibular ramus to the disto-buccal aspect of the second molar then a sulcular incision near mesio-buccal edge of M2 was made extending to its distal surface, finally a relieving incision from disto-buccal aspect of M2 curving forward into mandibular vestibule

Follow-up: 14 days, time interval between 2 sides 2 weeks

All procedures were carried out by the same surgeon, using same instruments, rotary and irrigation devices and materials, under sedation with IV midazolam and local anaesthetic

OutcomesPain (VAS 1-10), swelling, trismus, periodontal examination of adjacent M2, alveolar osteitis, wound infection at 2, 7 and 14 days
NotesSample size calculation: Not reported
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskQuote: "...randomly assigned using electronic randomization tables with patients numbered according to the order in which they presented for surgery"
Allocation concealment (selection bias)Unclear riskUnclear who performed the allocation and whether it was concealed from operator
Blinding (performance bias and detection bias)
patient
Unclear riskNot mentioned
Blinding (performance bias and detection bias)
assessor
High riskNot mentioned and assessment probably performed by operator
Incomplete outcome data (attrition bias)
All outcomes
Low riskOne randomised patient did not have second procedure. Unlikely to have introduced a bias
Selective reporting (reporting bias)Low riskPlanned outcomes reported in full
Other biasLow riskNo other sources of bias identified

Barone 2010

Methods

Study design: RCT (parallel group)

Conducted in: Versilia Hospital, Lido di Camaiore, Italy

Participants

Inclusion criteria: Patients referred for lower third molar extraction at Versilia Hospital who were systemically healthy

Exclusion criteria: Patients with a history of systemic diseases that would contraindicate surgery, pregnant and lactating women, patients in whom there was no need to raise the mucoperiosteal flap to remove the third molar, and patients who smoked more than 10 cigarettes per day

Age: Mean 31.2 years

Number of patients randomised: 26

Number of patients evaluated: 26

Interventions

Ultrasound versus rotary instruments for bone removal

Group A (n = 13): Surgical removal of lower 3rd molar using ultrasonic bone surgery under local anaesthesia

Group B (n = 13): Surgical removal of lower 3rd molar using traditional rotary instruments under local anaesthesia

(rotary instruments were used for sectioning of teeth where necessary)

All procedures performed under local anaesthetic

Follow-up: At days 1, 3, 5 and 7

Outcomes

Surgical time (start of first incision to last suture)

Pain (VAS) at days 1, 3, 5 and 7

Trismus (inter-incisal distance measured using callipers) at days 1, 3, 5 and 7

Cheek swelling (measured with a standard calliper from the lingual aspect of the midportion of the crown of the first mandibular molar to the tangent of the cheek's skin) at days 1, 3, 5 and 7

Notes

Sample size calculation: Not reported

At baseline (parallel groups only) groups were comparable with regard to type of impaction

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskQuote: "An independent evaluator allocated the patients into the test and control groups according to a computer-generated randomisation list"
Allocation concealment (selection bias)Low riskAllocation was performed by an independent evaluator
Blinding (performance bias and detection bias)
patient
Unclear riskAll surgery performed on patients under local anaesthetic only. Therefore, blinding not possible
Blinding (performance bias and detection bias)
assessor
Unclear riskNot clear as to who carried out the postoperative assessments
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll randomised patients evaluated
Selective reporting (reporting bias)Low riskPain, swelling, trismus, analgesic consumption planned and recorded
Other biasLow riskGroups appeared similar at baseline

Bello 2011

Methods

Study design: RCT parallel group

Conducted in: Department of Oral and Maxillofacial Surgery, National Hospital, Abuja, Nigeria

Number of centres: 1

Recruitment period: Not stated

Participants

Inclusion criteria: Patients referred for extraction of 1 or 2 impacted mandibular third molars

Exclusion criteria: Patients with acute pericoronal infection, systemic diseases, or bleeding disorder, patients receiving steroid therapy or contraceptives, and smokers were excluded.  Patients whose extraction procedure took more than 35 minutes were also excluded

Number randomised: Unclear

Number evaluated: 82

Interventions

Partial versus complete wound closure

Group  A (n = 40): Partial wound closure was achieved using 4 interrupted sutures leaving a window communicating with the oral cavity

Group  B (n = 42): Complete wound closure was achieved using 5 interrupted sutures which sealed off communication with the oral cavity

All procedures done under local anaesthetic by the same surgeon. All patients received pre-emptive antibiotics (amoxicillin and metronidazole) for 5 days and 3 days of diclofenac for pain and inflammation

OutcomesPain (VAS) reported daily for 7 days. Maximal interincisal distance (as % of baseline value), and swelling (difference from baseline) were evaluated on days 2, 5 and 7. Numbers of postoperative complications (dry socket, infection and secondary haemorrhage) were also noted
Notes

Sample size calculation: Not reported

E-mail sent to sabello2004@yahoo.com 12/3/2012 requesting further information about the methods used. Reply received 21/3/2012 with unpublished data

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk

Quote: "randomly allocated"

Comment: Allocated by drawing lots to either partial or total (e-mail correspondence)

Allocation concealment (selection bias)Low riskA paper was drawn by the assistant and shown to the surgeon prior to suturing (e-mail correspondence)
Blinding (performance bias and detection bias)
patient
Unclear riskNot mentioned
Blinding (performance bias and detection bias)
assessor
Unclear riskAssessment done using a 'clean proforma' but it is possible that the assessor knew which group the patient was in
Incomplete outcome data (attrition bias)
All outcomes
Low risk90 patients were screened for inclusion but unclear how many were randomised. Patients were excluded when the procedure took more than 35 minutes (n = 3). 8 patients did not return for follow-up assessment, 5 and 3 in each group
Selective reporting (reporting bias)Low riskPlanned outcomes of pain, trismus, swelling and complications reported
Other biasLow riskNo other sources of bias identified

Briguglio 2011

Methods

Study design: Parallel group RCT

Conducted in: Brazil (but researchers based in Italy)

Number of centres: 1

Participants

Inclusion criteria: Patients between the ages of 18-41 requiring extraction of mandibular third molars. Those with moderate impaction mesio-angularly with a tilt degree more than 25 degrees in relation to the second molar. Only impacted third molars with distil periodontal defects at the second molar with PPD ≥ 7 mm and CAL ≥ 6 mm were selected

Exclusion criteria: Systemic disease, pregnancy, smoking and medication (unspecified)

Age: 18-45 years

Number randomised: 45

Number evaluated: 45

Interventions

Laskin triangular flap versus Thibault and Parant modified envelope flap versus Laskin envelope flap

Group A (n = 15): Laskin triangular flap

Group B (n = 15): Thibault and Parant modified envelope flap

Group C (n = 15): Laskin envelope flap

All participants had a preoperative dental hygiene check and procedures were performed under local anaesthetic. All participants given 1 g amoxicillin+ sulbactam preoperation and all used CHX mouthwash pre and postoperation

OutcomesShort-term complications (pain swelling and infection) and PPD and CAL at 3, 6, 12 and 24 months
Notes

Sample size calculation: Not reported

Gaetano Isola provided additional data on short-term complications by e-mail 7/3/2012

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskQuote: "randomly assigned" using a prepared randomisation list
Allocation concealment (selection bias)Low riskAllocation done by third person not involved in the trial procedures
Blinding (performance bias and detection bias)
patient
Low riskDouble blind
Blinding (performance bias and detection bias)
assessor
Low riskDouble blind – examiner evaluating periodontal outcomes did not know treatment allocation till end of study. Unclear who evaluated short-term complications
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll randomised patients included in outcomes up to 24 months
Selective reporting (reporting bias)Unclear riskPaper reports only periodontal outcomes but data on short-term complications supplied by author
Other biasLow risk No other sources of bias identified

Butler 1977

Methods

Study design: RCT (split-mouth)

Conducted in: Bethesda, Maryland, USA

Participants

Inclusion criteria: Patients with bilaterally symmetrical impactions with regard to depth and angulation. Partial or complete impactions were accepted also

Exclusion criteria: Patients with evidence of acute infection or severe pericoronitis around the wisdom teeth were excluded from the study

Number of patients randomised: 211

Number of patients evaluated: 211, 422 teeth

Interventions

High volume irrigation versus low volume irrigation

Group A (n = 211): Postextraction irrigation with 175 ml sterile saline under IV sedation

Group B (n = 211): Postextraction irrigation with 25 ml sterile saline under IV sedation

Follow-up: At 4, 5 or 6 days

All procedures performed under intravenous sedation

OutcomesPresence of alveolar osteitis at recall 4-6 days later
Notes

Sample size calculation: Not reported

In 32 cases a mechanical irrigation device was used for irrigation in the higher volume site

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

Quote: "a random selection technique was used"

Comment: Method of sequence generation not described

Allocation concealment (selection bias)Unclear riskNot mentioned
Blinding (performance bias and detection bias)
patient
Unclear riskAs the patients were sedated they were highly unlikely to be aware of which volume of irrigant they received and where. This however, may be confounded by the 32 cases in which a mechanical irrigating device was used (see below)
Blinding (performance bias and detection bias)
assessor
High riskNo mention of assessor blinding
Incomplete outcome data (attrition bias)
All outcomes
Low riskNo withdrawals
Selective reporting (reporting bias)Unclear riskThe authors state that "patients were examined on the fourth, fifth, or sixth postoperative day for evidence of localised osteitis". They then, however, go on to mention that most 'infections' occurred between the seventh and tenth postoperative days. The outcome of infection was not mentioned at the outset, regardless of the fact of a non-significant result
Other biasHigh risk

Quote: "..most patients also had one or two maxillary third molars removed, but these operations were not included in the investigation"

Comment: Multiple extraction sites may influence the presence of alveolar osteitis and therefore should be considered as a confounding factor

In 32 cases a mechanical irrigating device was used to provide the higher volume of irrigant, thus also confounding the results

Cerqueira 2004

Methods

Study design: RCT split-mouth

Conducted in: Pernambuco, Brazil

Number of centres: 1

Recruitment period: Not stated

Participants

Inclusion criteria: Patients aged 14-30 years, with bilateral impacted third molars in similar positions on each side of the mouth

Exclusion criteria: Patients using medications that could interfere with healing or those with systemic diseases

Number invited: 5 patients underwent surgery "with the purpose of calibration" and further 12 were excluded because they "proved to be unsuitable"

Number randomised: 53

Number evaluated: 53

Interventions

Drain versus no drain

Group A (n = 53): 1 side of the mouth, chosen at random, had a silicon tube drain inserted into the buccal fold. Drain in situ for 4 days

Group B (n = 53): On the opposing side the wound was sutured with no drain

All patients receive preoperative antibiotic prophylaxis (amoxicillin) and postoperative cetoprophen for 4 days

All procedures performed under local anaesthesia

OutcomesPain (VAS), maximal mouth opening, swelling (% of preoperative) on postoperative days 1, 3, 7 and 15
Notes

Sample size calculation: Not reported

E-mail sent to Dr Vasconcelos belmiroc@terra.com.br on 6/3/2012 requesting further information. No reply received

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskThe choice of treatment "was made randomly"
Allocation concealment (selection bias)Unclear riskNot mentioned
Blinding (performance bias and detection bias)
patient
Unclear riskNot mentioned and probably not possible
Blinding (performance bias and detection bias)
assessor
High riskNot mentioned
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll 53 patients were evaluated
Selective reporting (reporting bias)Low riskPlanned outcomes of pain, swelling and trismus reported
Other biasLow riskNo other sources of bias identified

Chukwuneke 2008

Methods

Study design: RCT parallel group single blind trial

Conducted in: Oral surgery department of University of Nigeria Teaching Hospital, Enugu, Nigeria

Participants

Inclusion criteria: patients who were willing to come for their follow-up appointments, who were free from pain or any other inflammatory symptoms (swelling, hyperaemia, TMD), had impacted lower wisdom teeth, were not on medication that could interfere with healing and did not smoke or have any systemic disease

Exclusion criteria: Pregnant or lactating females were excluded from the study

Number of patients randomised: 100

Number of patients evaluated: 100

Interventions

Rubber tube (Penrose) drain versus no drain

Group A (n = 50): Sutures plus Penrose rubber drain placement for 72 hrs

Group B (n = 50): Sutures only postoperatively

All patients received 2 g amoxicillin preoperatively and procedures were performed under local anaesthesia

Follow-up: 24 hrs, 72 hrs and 5 days

OutcomesPain (VAS), swelling (horizontal and vertical guide with tape and reference points) and trismus (inter-incisal callipers) Evaluated at 24 hrs, 72 hrs and 5 days postoperatively
NotesSample size calculation: Not reported
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

Quote: "...a prospective, randomised, single-blind experimental study was undertaken"

Comment: Method of sequence generation not described

Allocation concealment (selection bias)Unclear riskNot mentioned
Blinding (performance bias and detection bias)
patient
High riskPatients were unable to be blinded from whichever treatment group they were in, as they would be aware of the presence of the Penrose rubber drain in their mouths. This is evidenced in the photographs
Blinding (performance bias and detection bias)
assessor
Unclear riskThere is no mention of assessor blinding other than in the sentence "a prospective, randomised, single-blind experimental study was undertaken." However it is not clear to whom this single blinding refers
Incomplete outcome data (attrition bias)
All outcomes
Low riskNo withdrawals
Selective reporting (reporting bias)Low riskAll planned outcomes reported on
Other biasUnclear risk

The study accepted all impactions, but there was no report of measures to account for possible differing degrees of difficulty with the extractions

While appropriate randomisation should result in groups with similar proportions of high risk patients no baseline characteristics were reported in this trial. It is unclear whether these biases were avoided

Danda 2010

Methods

Study design: Split-mouth RCT

Conducted in: India

Number of centres: 1

Recruitment period: May 2005 to March 2008

Participants

Inclusion criteria: Patients requiring removal of bilateral impacted third molars, for prophylactic or therapeutic reasons. Partial or complete bony impaction

Exclusion criteria: Patients with medical problems which would contraindicate oral surgery, bone pathology, immunocompromised patients and those with soft tissue impaction of mandibular third molars

Number randomised: 93

Number evaluated: 93

Interventions

Primary versus secondary closure

Group A (n = 93): Primary closure (2 sutures on distal arm and 1 on mesial arm of incision)

Group B (n = 93): Secondary closure (wedge of mucosa removed distil to 2nd molar then 1 suture on mesial and another on distil arm of the incision

All procedures performed under local anaesthesia

OutcomesPain and swelling measure on 5-point VAS daily for 7 days. Alveolar osteitis and nerve damage also reported
Notes

Sample size calculation: Not reported

E-mail send to Dr Danda (anilomfs@gmail.com) on 6/3/2012 requesting further information. No reply received

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNo information at all regarding the method by which side of mouth was selected for each procedure
Allocation concealment (selection bias)Unclear riskNot mentioned
Blinding (performance bias and detection bias)
patient
Unclear riskNot mentioned
Blinding (performance bias and detection bias)
assessor
High riskPatients assessed the outcomes
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNo mention of drop-outs but unlikely that all randomised patients included in outcomes
Selective reporting (reporting bias)Low riskAll planned outcomes reported
Other biasLow riskNo other sources of bias identified

de Brabander 1988

Methods

Study design: RCT (parallel)

Conducted in: Eastman Dental Center, New York, USA

Participants

Inclusion criteria: Unilateral wisdom tooth needing extraction

Exclusion criteria: Patients with clinical signs of pericoronitis, or those whose surgery took longer than 20 minutes from the first incision

Number of patients randomised: 21

Number of patients evaluated: 21

Interventions

Gauze drain versus no drain

Group A (n = 11): Postextraction placement of a vaseline coated gauze drain partially submerged into the socket, sutured in place

Group B (n = 10): Postextraction removal of a wedge of tissue distal to the second molar before closure

Surgery was performed under local anaesthetic

Follow-up: 2 and 7 days

OutcomesPain (VAS), swelling (examiner VAS, and comparisons with preoperative calliper measurements), trismus (inter-incisal distance), dry socket
Notes

Sample size calculation: Not reported

Characteristics of the groups at baseline not reported

Letter sent to author (May 2003). Reply that no additional data are available

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskQuote: "...made on a randomised basis by an individual other than the surgeon performing the procedure"; "the operator did not know whether a drain would be inserted until the time of closure"
Allocation concealment (selection bias)Low risk

Quote: "The decision on whether a drain should be inserted was made on a randomised basis by an individual other than the surgeon performing the procedure"

Comment: Unclear if was this concealed from the surgeon?

Blinding (performance bias and detection bias)
patient
Unclear riskQuote: "..the patients were not informed of the potential advantages or disadvantages of the drain", though patients still likely to detect presence of drain in operation site, therefore unable to blind patients
Blinding (performance bias and detection bias)
assessor
Unclear riskSubsequently, pain assessments by the patients using a VAS were not blinded either: "...the patients were routinely examined by an examiner who was not the surgeon and not aware of the treatment given" therefore assessor was blinded for swelling, trismus and dry socket. Unclear when drain was removed and whether examiner could have determined whether a drain had been placed
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNo withdrawals mentioned, but numbers evaluated not stated. Only means per group were reported
Selective reporting (reporting bias)Low riskAlthough the authors only mention dry socket in the results section. It should be stated that the investigators were looking for dry socket at the outset, regardless of the fact that there was no incidence of this outcome in either group
Other biasUnclear risk1 of the exclusion criteria precludes any cases that took longer than 20 minutes from the first incision to removal of the tooth. The authors do not mention whether or not more subjects were included in the study beforehand but subsequently had to withdraw because of lengthy surgery, these subjects were not accounted for in the text, if indeed this applied to any

Erdogan 2011

Methods

Study design: Split-mouth RCT

Conducted in: Adana, Turkey

Number of centres: 1

Recruitment period: January 2008 to June 2009

Participants

Inclusion criteria: Patients were aged 20-32 years with bilateral symmetrically impacted mandibular third molars. Participants were free of systemic disease, had no history of pericoronal infection or recent anti-inflammatory drug use. Included teeth were all class I or II and position A or B according to Pell and Gregory classification

Exclusion criteria: Deeply impacted cases were not included in the study

Age: Mean 23.9 ± 4.3, 20-32 years

Number randomised: 20

Number evaluated: 20

Interventions

Envelope flap versus triangular flap

Group A (n = 20): Sulcular incision extending from the lateral border of the  mandibular ramus to the second premolar with no releasing vertical incision

Group B (n = 20): Buccal releasing incision positioned on the mesial aspect of the second molar

All patients had preoperative single dose of oral penicillin and rinsed with CHX

All surgical procedures were performed by the same surgeon under local anaesthetic, and incisions were closed with secondary wound closure

Second extraction was performed after 3 weeks

OutcomesOperating time, mouth opening, VAS pain (resting and chewing), analgesic consumption
Notes

No sample size calculation reported. Probably underpowered. E-mail sent 25/3/2013 and reply

from authors provided additional information

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk

Quotes: "..selected randomly to have their first operations on the left side….For the first extraction the flap design was chosen randomly"

"A coin toss was used to determine the flap type and surgical site at the day of the first surgery" (e-mail communication)

Allocation concealment (selection bias)Unclear riskQuote: "Coin toss was conducted by junior surgeon, and an experienced surgeon performed the surgery" (e-mail communication)
Blinding (performance bias and detection bias)
patient
Low riskDouble blind
Blinding (performance bias and detection bias)
assessor
Low riskQuote: "A blinded surgeon who was not aware of the flap design, conducted the measurements"
Incomplete outcome data (attrition bias)
All outcomes
Low riskOutcome data for all 40 procedures on all 20 participants reported
Selective reporting (reporting bias)Low riskPlanned outcomes reported
Other biasLow riskNo other sources of bias identified

Gargallo-Albiol 2000

Methods

Study design: RCT (parallel group)

Conducted in: Department of Oral Surgery, Odontology, University of Barcelona, Spain

Participants

Inclusion criteria: 300 consecutive patients who needed 1 lower impacted wisdom tooth extracted

Exclusion criteria: If the tooth did not need to be sectioned during the procedure, then it was excluded from the study

Age: Mean 27.4 years

Number of patients randomised: 300

Number of patients evaluated: 300

Interventions

Lingual nerve protection (subperiosteal retractor) versus none

Group A (n = 142): Lower third molar removed with subperiosteal insertion of retractor for lingual nerve protection

Group B (n = 158): Lower third molar removed without lingual nerve protection

All molars removed under local anaesthetic

Follow-up: Day 7, day 21 and day 60

OutcomesVerbal self assessment and mechanosensory testing of lingual nerve function
Notes

Sample size calculation: Not reported

The author notes that the low incidence of sensory disturbance in this study may be related to the fact that the procedures were performed under local anaesthetic. Previous studies may indicate that procedures performed under general anaesthetic are associated with higher levels of sensory disturbances, but as the author rightly points out: "the choice of general anaesthesia [...] may also be related to the degree of difficulty when removing the third molar". Any other issues

Baseline comparability: Information about the comparability of the groups at baseline not reported

Letter sent to author. Reply that no additional data are available

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

Quote: "...patients were strictly randomised"

Comment: Method of sequence generation not stated

Allocation concealment (selection bias)Unclear riskNot mentioned
Blinding (performance bias and detection bias)
patient
Unclear riskPatient blinding is not mentioned, but likely that patients were aware of whether a retractor was used
Blinding (performance bias and detection bias)
assessor
High riskQuote: "...lingual nerve function was tested at one week [...] and was carried out by the same surgeon who performed the procedure"
Incomplete outcome data (attrition bias)
All outcomes
Low riskNo withdrawals mentioned, results reported as percentage and appear to include all randomised participants. Confirmed by correspondence with author
Selective reporting (reporting bias)Low riskAll pre-specified outcomes reported on
Other biasLow riskNo other sources of bias identified

Goldsmith 2012

Methods

Study design: Split-mouth RCT

Conducted in: University of Otago, New Zealand

Participants

Inclusion criteria: Patients aged 16-40, ASA I or II, with bilateral symmetrically impacted partially erupted mandibular third molars, no associated pathology, no medical conditions that might alter wound healing potential

Exclusion criteria: History of abuse of midazolam, allergy to any of the medications to be used, pregnancy, present or previous radiotherapy to third molar region of lower jaw, long term steroid or bisphosphonate use, bone disorder or fibrous dysplasia

Number of patients randomised: 57

Number of patients evaluated: 52 (42 for pain outcome)

Interventions

Envelope flap versus pedicle flap

Group A (n = 52 teeth): Incision placed in the buccal gingival sulcus from the mesio-buccal line angle of the first molar to the most distal visible aspect of the third molar. The relieving incision then extended up the external oblique ridge

Group B (n = 52 teeth): Pedicle flap design involved the same initial incision, in the buccal gingival sulcus, but distil to the third molar the incision was extended approximately 1 cm and then curved towards the buccal sulcus allowing for rotation of the flap and primary closure over sound bone

Follow-up: 7 days

3 weeks between procedures. All procedures were carried out by the same surgeon under sedation with midazolam and local anaesthetic. All patients received standard pain relief medication regimen (ibuprofen/paracetamol plus codeine phosphate if required) and 0.2% CHX mouthrinse to be used 3 times daily for 5 days

OutcomesAlveolar osteitis, wound infection, pain, swelling,trismus, wound dehiscence on days 2 and 7 (envelope flap only)
Notes

Sample size calculation: Stated that sample size was determined by a power calculation using data previously collected

Funding: New Zealand Dental Research Foundation and University of Otago Fuller Scholarship

E-mail from authors 22/8/2013 provided additional information

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk

Quote: "randomly allocated into one of two groups using the Logan envelope technique.... In one group the first procedure was carried out using the envelope flap design ..... both flap designs were randomly allocated to left or right side of each participants lower jaw again using Logan envelope technique"

E-mail from author: "..a larger envelope contained smaller pieces of paper that had the type of flap to be assigned to a patient, on the day of the procedure the surgeon's assistant would "blindly" pick out a piece of paper and the surgeon would then use that type of flap. The same was done for assigning which side of the mouth would be operated on"

Allocation concealment (selection bias)Low riskAllocation concealed from operator
Blinding (performance bias and detection bias)
patient
Low riskE-mail from author: "Patients were not aware of what type of flap they received nor were the clinicians who did the follow-up clinical outcomes"
Blinding (performance bias and detection bias)
assessor
Low riskE-mail from author: "Patients were not aware of what type of flap they received nor were the clinicians who did the follow up clinical outcomes"
Incomplete outcome data (attrition bias)
All outcomes
Low risk10 patients excluded from pain evaluation due to missing data, but unlikely to result in bias in split-mouth study
Selective reporting (reporting bias)Unclear riskTrismus outcome data not reported
Other biasLow riskNo other sources of bias identified

Gomes 2005

Methods

Study design: RCT (split-mouth)

Conducted in: Department of Oral and Maxillofacial Surgery, University of Pernambuco, Camaragibe, Brazil

Participants

Inclusion criteria: Patients with bilateral mandibular impacted third molars, and all procedures had to be performed by the same operator

Exclusion criteria: Patients with medical problems that could contraindicate the procedure were excluded, as were any procedures in which complete fractures of the lingual cortex were likely

Age: Not stated

Number of patients randomised: 55

Number of patients evaluated: 55

Interventions

Lingual nerve protection (Free's retractor) versus none

Group A (n = 55 teeth): Lingual flap with Free's retractor

Group B (n = 55 teeth): Without lingual flap

Procedures under local anaesthesia or general anaesthesia with local anaesthesia. 1 surgeon

Follow-up: 3 months

OutcomesPin-prick test to confirm nerve injury at 1 and 7 days postoperatively
Notes

Sample size calculation: Not reported

In this study, the authors note in the discussion that the lingual bone plate was preserved in all cases, and that this was responsible for a higher degree of difficulty during the procedure, "especially in deeper impactions". This may affect the homogeneity of the study

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

Quote: "..all patients were randomly allotted"

Comment: Method of sequence generation not reported

Allocation concealment (selection bias)Unclear riskNot mentioned
Blinding (performance bias and detection bias)
patient
Unclear riskBlinding not mentioned. Patients who underwent general anaesthetic are likely to be unaware of the procedure they received but those with only local anaesthetic may have been aware of the procedure used
Blinding (performance bias and detection bias)
assessor
Low riskAssessor was blinded. Quote: "An oral and maxillofacial surgeon who knew the proposal of the study but did not know which side was an experimental or a control group performed this evaluation"
Incomplete outcome data (attrition bias)
All outcomes
Low riskNo withdrawals
Selective reporting (reporting bias)Unclear riskSingle outcome of nerve injury reported
Other biasUnclear riskThe authors note in the discussion that the lingual bone plate was preserved in all cases, and that this was responsible for a higher degree of difficulty during the procedure, "especially in deeper impactions"

Greenwood 1994

Methods

Study design: RCT (split-mouth)

Conducted in: Department of Oral and Maxillofacial Surgery, The University of Manchester, UK

Participants

Inclusion criteria: "150 patients undergoing third molar removal under general anaesthesia were entered into the study. Cases were selected so that the left and right sides were close to identical for tooth position and degree of difficulty"

Exclusion criteria: None described

Age: Not stated

Number of patients randomised: 150

Number of patients evaluated: 150

Interventions

Howarth's elevator versus broad retractor for lingual nerve protection

Group A (n = 150): Lingual flap retraction with Howarth's elevator

Group B (n = 150): Lingual flap retraction using broad retractor

All procedures performed under general anaesthesia, all required bone removal with either drill or chisel. Operators had varying experience from house officers to consultant. Both extractions for each patient were completed by the same operator

Follow-up: 1 month

OutcomesVerbal self assessment of lingual nerve function, immediately, at 10 days and 30 days postoperatively
Notes

Sample size calculation: Not reported

E-mail sent 15/1/2003 and reply received 20/1/2003 with unpublished data

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk

Quote: "Left and right sides were then allocated at random..."

Comment: After contacting the author, the following further information was garnered: "the randomisation code was computer generated"

Allocation concealment (selection bias)Low riskIn the same letter, the author notes that "the allocation was concealed until surgery"
Blinding (performance bias and detection bias)
patient
Low riskAll procedures were performed under general anaesthetic, so it may be assumed that the patients were blinded as to which side received the broader retractor, however, this is not specified in the paper
Blinding (performance bias and detection bias)
assessor
Low riskQuote: "..the nurse and surgeon assessors were blinded" (from private correspondence)
Incomplete outcome data (attrition bias)
All outcomes
Low riskNo withdrawals
Selective reporting (reporting bias)Low riskAll pre-specified outcomes reported on
Other biasHigh risk

Method of bone removal was not standardised: "the tooth was then removed employing either drill or chisel for bone removal, according to the operator's personal preference." Bone removal technique may possibly confound the results, and it was not recorded how many in each group had bone removal by each technique

The Howarth's elevator was used to raise the initial flap for both sides, and then the broad retractor was introduced to 1 side

Haraji 2010

Methods

Study design: Split-mouth RCT

Conducted in: Iran

Number of centres: 1

Recruitment period: Not stated

Participants

Inclusion criteria: Patients with bilaterally impacted third molars with similar difficulty index

Exclusion criteria: Pre-existing medical conditions, oral contraceptive use, systemic or neurological conditions, pregnancy, pericoronitis or pathological conditions associated with third molars

Age: Mean 19.94 ±1.5 years

Number randomised: 17

Number evaluated: Unclear

Interventions

Buccal envelope versus modified triangular flap

Group  A (n = 17): Buccal envelope flap

Group  B (n = 17): Modified triangular flap

All patients received local anaesthetic, oral cephalexin 500 mg 6 hourly for 5 days and 500 mg acetaminophen codeine postoperatively

OutcomesAlveolar osteitis, and 'healing scores'
Notes

Sample size calculation: Not reported

E-mail sent to Dr.a.Haraji@Dentaliau.ir on 12 March 2012 requesting further information. No reply received

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

Quote: "allocated randomly"

Comment: Method of sequence generation not described

Allocation concealment (selection bias)Unclear riskNot mentioned
Blinding (performance bias and detection bias)
patient
Low riskDouble blinded
Blinding (performance bias and detection bias)
assessor
Low riskDouble blinded. Paper does not state who conducted outcome assessment
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskOutcomes reported as percentage only with no indication of how many patients were evaluated
Selective reporting (reporting bias)High riskAlveolar osteitis reported as percentage in each group and P value, no indication if paired data taken into account in analysis. Suggestion that a number of patients had bilateral osteitis. Pain and healing not reported
Other biasUnclear riskControl group rate of alveolar osteitis 41.7% is very high and it seems unlikely that a different flap design is solely responsible for the lower rate in the intervention group. ? Co-intervention or aseptic conditions for surgery

Hashemi 2012

Methods

Study design: Split-mouth RCT

Conducted in: Tehran, Iran

Number of Centres: 1 (Tehran University Hospital)

Recruitment period: September 2008 to January 2010

Participants

Inclusion criteria: Bilateral bony mandibular third molars that were fairly similar in terms of angulation, degree of impaction, and estimated difficulty of removal

Exclusion criteria: Presence of any medical problem that would contraindicate extraction, pathological lesion near teeth to be extracted

Number of patients randomised: 30

Number of patients evaluated: 30

Interventions

No sutures versus multiple sutures for wound closure

Group A (n = 30 teeth)

Group B (n =30 teeth)

Follow-up: 7 days

All procedures were carried out by a single surgeon under local anaesthetic

OutcomesPain (6-point VAS) and swelling on days 1, 3 and 7
NotesSample size calculation: Not reported
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

Quote: "..one of the two impacted mandibular third molars in each patient was randomly allocated"

Comment: Method of sequence generation not described

Allocation concealment (selection bias)Unclear riskNo mention of who conducted the randomisation and whether it was concealed from the surgeon
Blinding (performance bias and detection bias)
patient
Unclear riskNot possible
Blinding (performance bias and detection bias)
assessor
High riskNot mentioned
Incomplete outcome data (attrition bias)
All outcomes
Low riskIt appears that all the randomised participants were included in the outcome valuation
Selective reporting (reporting bias)High riskPlanned outcomes reported but unclear if paired nature of data was taken into account
Other biasLow riskNo other sources of bias identified

Kirk 2007

Methods

Study design: RCT (split-mouth)

Conducted in: New Zealand Defence Force, Taranaki Base Hospital

Participants

Inclusion criteria: Patients with bilateral, symmetrically impacted lower wisdom teeth

Exclusion criteria: Patients were excluded if they had a pre-existing medical condition, or were taking medication that would influence the ability to undergo surgery or alter wound healing. Patients were also excluded if they had any discernible active pathology associated with the third molars, or if the impactions were such that surgical time and trauma would be excessive and mask the possible influence of flap design

Age: Mean 24.2 years

Number of patients randomised: 35

Number of patients evaluated: 32

Interventions

Modified triangular flap versus envelope flap

Group A (n = 32): Modified triangular flap (on randomly selected side of mouth)

Group B (n = 32): Envelope flap (on the other side)

Both lower 8s removed at same visit and all procedures performed under intravenous sedation, by same surgical operator and dental assistant

Follow-up: Day 2 and day 7

OutcomesPain (VAS), alveolar osteitis, Infection, trismus, swelling (measured by evaluation of laser scans of the subjects' cheeks)
NotesSample size calculation: Not reported
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

Quote: "..the flap design and side of mouth were randomly assigned for each patient"

Comment: Method of sequence generation not described

Allocation concealment (selection bias)Unclear riskNot mentioned
Blinding (performance bias and detection bias)
patient
Unclear riskPatients likely to be unaware of which flap they received, but blinding was not specified
Blinding (performance bias and detection bias)
assessor
High riskNot mentioned whether or not assessors were blinded
Incomplete outcome data (attrition bias)
All outcomes
Low risk3 withdrawals, 1 prior to first surgery and 2 before the second surgery. These 2 were excluded from the outcome data but unlikely to have introduced bias
Selective reporting (reporting bias)Low riskAll pre-specified outcomes reported on
Other biasLow riskNo other sources of bias identified

Leung 2009

Methods

Study design: RCT (parallel group)

Conducted in: Discipline of Oral and Maxillofacial Surgery, Faculty of Dentistry, The University of Hong Kong, China

Participants

Inclusion criteria: The wisdom tooth root touched or overlapped with the superior cortical line of the IDN on radiographs. Radiographic signs were used to assess a close relationship with the nerve

Exclusion criteria: Wisdom tooth roots did not touch the IDN cortical lines, or if wisdom teeth were associated with apical pathology or cystic or neoplastic lesions. Patients were also excluded if they had any of the following:

  • systemic conditions predisposing to local infection, such as diabetes mellitus or AIDS, or concurrent cancer chemotherapy;

  • local factors predisposing to infection, such as fibrous dysplasia or a history of radiotherapy on mandible;

  • craniofacial syndromes with preexisting IDN deficit;

  • any plans for orthognathic surgery

Number of patients randomised: 231

Number of patients evaluated: 231

Interventions

Coronectomy versus complete tooth removal

Group A (n = 171 teeth): Underwent coronectomy

Group B (n = 178 teeth): Underwent conventional extraction

Failed coronectomy (n = 16 teeth)

Surgical residents undertook treatment under general anaesthesia in 50.3% of test patients and 48.3% of patients in the control group, intravenous sedation with local anaesthesia in 3.5% of test patients and 5.6% in the control group. Local anaesthesia was used in 46.2% of the test patients and 46.1% of the control patients

Follow-up: Postoperatively, patients were assessed at 1 week and at 1, 3, 6, 12, and 24 months. Mean length of follow-up for all groups 10.6 months

Outcomes

The primary outcome of the study was the presence of IDN deficit 1 week postoperatively. The secondary outcomes were:

  • the presence of lingual nerve deficit

  • recovery from IDN and lingual nerve deficit

  • pain

  • infection

  • dry socket

  • root exposure

  • root migration

  • and the need for re-operation

Notes

Sample size calculation: The sample size calculation was based on assuming the incidence of IDN deficit in the control group (conventional extraction) and the study group (coronectomy) would be 5% and 0%, respectively. If these assumptions were correct, 152 patients per group would be sufficient to detect a statistical difference, with a 2-sided type 1 error of 5% and a power of 80%

Baseline comparability: "There were no statistical differences between the 2 groups in terms of age and sex of the patients; eruption status, pattern and depth of impaction, and root shape of the wisdom teeth; the type of anaesthesia used; or the presence and type of radiographic signs"

Any other issues: Withdrawals clearly stated

Unit of analysis problem as patients were randomised but data are presented at the tooth level. There were 231 patients and 349 teeth. This means the confidence intervals will be narrower than they should be as the teeth are clustered within patients

E-mail sent to author (30/9/2011). Unpublished data supplied

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskQuote: "A house officer not participating in the study generated a randomisation table using a computer program. Patients were assigned to the 1 or 2 groups according to the randomisation table"
Allocation concealment (selection bias)Low riskThe allocation sequence was kept by an assigned nurse and concealed from both the operator and patient until the patient was assigned
Blinding (performance bias and detection bias)
patient
Low riskComment: All procedures performed under general anaesthetic and patients will look similar after operation regardless of intervention. Patients unlikely to be aware of which procedure was performed but only coronectomy group received orthopantomograms at 1 week
Blinding (performance bias and detection bias)
assessor
High riskNot mentioned
Incomplete outcome data (attrition bias)
All outcomes
Unclear risk16/171 wisdom teeth considered as failed coronectomy were excluded from analysis reported in the paper. However data on these patients were supplied by the authors
Selective reporting (reporting bias)High riskOutcomes: Neurosensory deficit, infection rate, pain and root migration planned and reported, but not by randomised person
Other biasUnclear risk231 patients contributed 349 teeth to the study. Seems that analysis did not take account of this

Mocan 1996

Methods

Study design: RCT (parallel)

Conducted in: Department of Oral and Maxillofacial Surgery, University of Ankara, Turkey

Participants

Inclusion criteria: Patients with third molars requiring extraction. A criterion for inclusion in the study was that the third molars should be either partially or fully covered by bone, and only unilateral cases were included

Exclusion criteria: Patients with complicating systemic disorders were accepted (ASA I and II)

Age: Mean 21.5 years

Number of patients randomised: 20

Number of patients evaluated: 20

Interventions

Chisel versus bur for bone removal

Group A (n = ?): Lingual split with chisel for bone removal

Group B (n = ?): Buccal approach with bur for bone removal

All procedures performed under local anaesthesia

Follow-up: Day 7

OutcomesAnalytical stereometric photogrammetrical assessment of swelling, calliper measure of mouth opening, and VAS self assessment of postoperative pain
Notes

Sample size calculation: Not reported

Baseline comparability: The lingual split group had 4 mesioangular impactions, 2 distoangular impactions and 4 vertical impactions whereas the buccal approach group had 3 mesioangular impactions, 0 distoangular impactions and 7 vertical impactions

E-mail correspondence in 2003 - no unpublished data available

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

Quote: "..the patients were divided randomly into groups"

Comment: Method of sequence generation not described

Allocation concealment (selection bias)Unclear riskNot mentioned
Blinding (performance bias and detection bias)
patient
Unclear riskThe patients were not blinded in this study
Blinding (performance bias and detection bias)
assessor
High riskNo mention of assessor blinding
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNo withdrawals mentioned but numbers evaluated unclear
Selective reporting (reporting bias)High risk

Quote: "...no one experienced sensory impairment of the inferior alveolar or lingual nerves"

Comment: It was not mentioned at the outset that sensory assessments were being made, and no method of assessment was described. Raw data and standard deviations not reported for primary outcomes, and not supplied by authors

Other biasHigh risk

Very small sample size with only 10 subjects in each intervention group

Different distribution of impactions in the 2 intervention groups at entry

Unable to include in meta-analysis as raw data not available in the paper, or after contacting author

Nageshwar 2002

Methods

Study design: Parallel group RCT

Conducted in: India

Number of centres: 1

Recruitment period: Not stated

Participants

Inclusion criteria: Patients scheduled to undergo surgical removal of impacted mandibular third molars

Exclusion criteria: Not explicit

Age: Mean 25.66 ± 4.45 years

Number randomised: 100

Number evaluated: 100 (e-mail from author)

Interventions

Comma incision versus modified envelope incision

Group A (n = 50): New comma incision

Group B (n = 50): Conventional modified envelope incision

All patients had local anaesthesia, conventional methods of bone removal and tooth sectioning as required. All had prescribed antibiotics and analgesics as indicated and CHX mouthwash until suture removal

OutcomesPain (VAS 0-10), swelling, trismus (compared to baseline) and periodontal sequelae measured on days 1, 3, 7 and 14
Notes

Sample size calculation: Not reported

E-mail sent 12 March 2012 requesting further information on randomisation and variance of outcome estimates. Reply with unpublished information received from Dr N Iyer 15/3/2012

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskQuotes: "..divided at random"; "Table a random numbers was used in sequence" (e-mail communication)
Allocation concealment (selection bias)Unclear riskSurgeon allocated each surgical site to 1 of the 2 groups just prior to the procedure (e-mail communication)
Blinding (performance bias and detection bias)
patient
Unclear riskNot mentioned
Blinding (performance bias and detection bias)
assessor
Low riskQuote: "The person who assessed all the study parameters did not know which kind of incision each patient received" (e-mail communication)
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll randomised patients included in the analysis (e-mail communication)
Selective reporting (reporting bias)Unclear riskPlanned outcomes pain swelling and trismus reported but no estimates of variance provided so data cannot be used in meta-analysis. Data supplied by author via e-mail
Other biasLow riskNo other sources of bias identified

Osunde 2011

Methods

Study design: RCT parallel group

Conducted in: Department of Dental and Maxillofacial Surgery, Aminu Kano Teaching Hospital, Kano, Nigeria

Number of centres: 1

Recruitment period: January to December 2007

Participants

Inclusion criteria: Patients referred for extraction of impacted lower third molars

Exclusion criteria: Patients with a perceptible level of pain at time of surgery were excluded

Number randomised: Unclear

Number evaluated: 50

Interventions

Partial versus complete wound closure

Group A (n = 25): A single 3-0 silk suture for closing the socket was placed at the distal relieving incision

Group B (n = 25): Multiple sutures for closing the socket; the sutures were placed at the interdental papilla between the second and third molars and at the distal relieving incision

All procedures performed under local anaesthetic

Both treatment groups received oral antibiotics (amoxicillin and metronidazole for 5 days), analgesics (ibuprofen for 3 days) and instructions to use a warm saline mouthrinse

OutcomesPatients assessed at days 1, 2, 3, 5, and 7 postoperatively to evaluate the degree of pain, swelling, and trismus
Notes

Sample size calculation: Not reported

E-mail sent to otdany@yahoo.co.uk on 12 March 2012 requesting additional information on randomisation and participants. No reply received

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

Quote: "..randomized into two groups"

Comment: Method of sequence generation not described

Allocation concealment (selection bias)Unclear riskNot mentioned
Blinding (performance bias and detection bias)
patient
Low riskQuote: "Double blind"
Blinding (performance bias and detection bias)
assessor
Low risk

Quote: "Double blind"

Comment: Outcomes were assessed by an independent evaluator who was unaware of the treatment group to which the patients belonged

Incomplete outcome data (attrition bias)
All outcomes
Unclear riskUnclear how many patients were originally randomised to treatment. Patients with perceptible pain at baseline were excluded from the study
Selective reporting (reporting bias)Unclear riskPain, swelling and trismus reported but not postoperative complications
Other biasLow riskNo other sources of bias identified

Osunde 2012

Methods

Study design: Parallel group RCT

Conducted in: Benin City, Nigeria

Number of Centres: 1 (Aminu Kano Teaching Hospital)

Recruitment Period: Not stated

Participants

Inclusion criteria: Patients aged 18-38 years with mesioangular, distoangular, horizontal and vertical impactions with a difficulty index of 3-8 according to Peterson's criteria. No symptoms of pain, facial swellin or trismus in 10 days preceding surgery, non-smokers, no concomitant medications or systemic diseases that could interfere with healing

Exclusion criteria: Pregnant or lactating females, patients with more than 1 third molar requiring treatment

Number of patients randomised: 80

Number of patients evaluated: 80

Interventions

No sutures versus multiple sutures for wound closure

Group A (n = 40): No sutures

Group B (n = 40): Multiple sutures using 3/0 silk, placed at the interdental papilla immediately distil to the second molar, the buccal relieving incision and the at the distil relieving incision

Follow-up: 7 days

All procedures were carried out by the same surgeon and assistant under local anaesthetic

OutcomesPain (10 cm VAS), trismus, swelling, on days 1, 2, and 7
Notes

Sample size calculation: Not reported

E-mail sent to otdany@yahoo.co.uk on 12 March 2012 requesting additional information on randomisation and participants. No reply received

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

Quote: "Consecutively randomised into two treatment groups"

Comment: Method of sequence generation not described

Allocation concealment (selection bias)Unclear riskNo mention of who conducted the random allocation and whether it was concealed from the surgeon
Blinding (performance bias and detection bias)
patient
Unclear riskNot possible
Blinding (performance bias and detection bias)
assessor
Low riskQuote: "..the patients were evaluated in a blinded manner by the same independent observer"
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll randomised participants are included in the outcome assessment
Selective reporting (reporting bias)Low riskPlanned outcomes reported in full
Other biasLow riskNo other sources of bias identified

Pasqualini 2005

Methods

Study design: Parallel group RCT

Conducted in: Italy

Number of centres: 1

Recruitment period: Not started

Participants

Inclusion criteria: Totally or partially bone-impacted mandibular third molar with mesial inclination between 25 7 30 degrees

Exclusion criteria: No systemic disease, good general health, age less than 30 years, non-smoker, no inflammation of the oral cavity, co-operation with the study and with postoperative follow-up, and no contraindication to anaesthetics or study drugs

Number of patients randomised: 200

Number evaluated: 200

Interventions

Primary versus secondary wound closure

Group A (n = 100): Primary wound closure - "flap repositioned and sutured hermetically"

Group B (n = 100): Secondary wound closure - "a sedge of mucosa 5-6 mm was removed from second molar and flap was repositioned and sutured"

All procedures performed under local anaesthesia

All patients also received antibiotics (amoxicillin 2 g/day for 5 days and nimesulide 200 mg/day for 3 days)

OutcomesPain and swelling on 5-point VAS daily for 7 days
Notes

Sample size calculation: Not reported

E-mail sent to Dr Pasqualini (damianox@mac.com) who provided additional information on the methods of this trial 6/3/2012

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskQuotes: "..randomly divided"; "Randomization was performed using a computer generated random numbers sequence created by an independent research office" (e-mail from author)
Allocation concealment (selection bias)Low riskQuote: "Allocations were kept in sealed serially numbered opaque envelopes which were opened in sequence and showed to the surgeon at the moment of surgical closure, with at least one independent witness present (generally a nurse)" (e-mail from author)
Blinding (performance bias and detection bias)
patient
Low riskThe patients were not aware of the type of closure (e-mail from author)
Blinding (performance bias and detection bias)
assessor
Unclear riskQuote: "The examiner who assessed the postoperative outcome at 7 and 30 days was not aware of the allocation. However they could have presumed it by looking at the residual wound" (e-mail from author)
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll randomised participants included in the outcome evaluation
Selective reporting (reporting bias)Low riskPain, swelling and infection reported as planned
Other biasLow riskAuthor confirmed that no wound dressings were used in either group

Praveen 2007

Methods

Study design: Parallel group RCT

Conducted in: India

Number of centres: 1

Recruitment period: Not stated

Participants

Inclusion criteria: Healthy patients with symptomatic impacted mandibular third molars

Exclusion criteria: Not explicitly stated

Number randomised: 90

Number evaluated: Unclear

Interventions

Lingual split with chisel versus surgical bur versus simplified split bone technique

Group A (n = 30): Lingual split, bone removed with a 5 mm mono bevelled chisel

Group B (n = 30): Bone removal with 702 bur at 15,000 rpm

Group C (n = 30): "Simplified split bone technique" using chisel from buccal aspect

"The lingual nerve was protected by a Howarth's periosteal elevator in all cases." All procedures performed under local anaesthetic

OutcomesPain, swelling and sensory disturbances recorded at 6, 24 and 48 hours and on day 7
Notes

Sample Size calculation: Not reported

E-mail sent to Dr Rajesh (rajeshomfs@gmail.com) seeking clarifications on 28/2/2012. No reply received

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

Quote: "The methods for a particular patient were selected randomly"

Comment: Method of sequence generation not described

Allocation concealment (selection bias)Unclear riskNot mentioned
Blinding (performance bias and detection bias)
patient
Unclear riskNot mentioned
Blinding (performance bias and detection bias)
assessor
High riskNot mentioned
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskUnclear how many extractions are included in the outcomes reported
Selective reporting (reporting bias)Unclear riskThere appear to be omissions and errors in the reported data, where different aspects are contradictory
Other biasLow riskNo other sources of bias identified

Rakprasitkul 1997

Methods

Study design: RCT (split-mouth)

Conducted in: Oral and Maxillofacial Surgery Department, Faculty of Dentistry, Mahidol University, Bangkok, Thailand

Participants

Inclusion criteria: Healthy patients requiring bilaterally impacted 3rd molars, who would co-operate with the study and postoperative follow-up. All teeth were fully covered by mucosa, and partially or completely covered by bone

Exclusion criteria: Patients with significant medical diseases or a history of bleeding problems were excluded, as were pregnant women. In addition, patients with any sign of pericoronitis were excluded from the study

Number of patients randomised: 23

Number of patients evaluated: 23

Interventions

Tube drain versus no drain

Group A (n = 23): Surgical drain placement for 3 days

Group B (n = 23): Simple primary wound closure with no surgical drain placement

Surgery performed by the same surgeon on 2 occasions 2 months apart, under local anaesthetic

Follow-up: Day 7

OutcomesPain (VAS), swelling (measured by distance of 2 transecting lines across cheek, and by patient grading), mouth opening (inter-incisal distance)
Notes

Sample size calculation: Not reported

E-mail sent to author (March 2003). Reply that no additional data are available

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

Quote: "...the patients were assigned to test and control groups by random selection"

Comment: Method of sequence generation not described

Allocation concealment (selection bias)Unclear riskNot mentioned
Blinding (performance bias and detection bias)
patient
Unclear riskAlthough not specified, patients would be aware of the presence of a surgical drain in their mouths
Blinding (performance bias and detection bias)
assessor
High riskQuote: "..the patients were examined by the same person [surgeon] immediately preoperatively, and on the third and seventh postoperative days"
Incomplete outcome data (attrition bias)
All outcomes
Low riskNo withdrawals
Selective reporting (reporting bias)Low riskAll pre-specified outcomes reported on
Other biasUnclear riskBy having a 2-month gap between the 2 extractions (1 side then the other) patients' perception of pain may be altered by that previous experience

Refo'a 2011

Methods

Study design: Parallel group RCT

Conducted in: Tehran, Iran

Number of centres: 1

Recruitment period: Not stated

Participants

Inclusion criteria: Patients aged 20–25 years with wholly bone-impacted mandibular third molar with mesioangular inclination willing to participate in study

Exclusion criteria: Systemic medical conditions, smoking, inflammation in the oral cavity, history of drug use

Number randomised: 32

Number evaluated: Unclear

Interventions

Primary versus secondary wound closure

Group A (n = 16): Triangular flap was raised, teeth were extracted and following saline irrigation flaps were repositioned and sutured completely using 0.5 inch round cutting needle with 3.0 silk suture

Group B (n = 16): Triangular flap was raised, teeth were extracted and following saline irrigation flaps were repositioned and 5-6 mm of distil extension to second molars was kept open, while other parts of the flap were repositioned and sutured

All surgical procedures were performed by the same surgeon under local anaesthetic. All patients received amoxicillin and ibuprofen and used CHX mouthrinse twice daily postoperatively

OutcomesPain (6-point VAS), swelling and trismus after 3 days
NotesNo sample size calculation reported
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskQuotes: "..randomly divided into two quantitatively equal groups using a computer generated random number table…. The groups were equalised regarding gender"
Allocation concealment (selection bias)Low riskThe surgeon was unaware of the type of closure until suturing
Blinding (performance bias and detection bias)
patient
Unclear riskNot specifically mentioned
Blinding (performance bias and detection bias)
assessor
High riskBlinding of outcome assessment not mentioned. Patients self assessed pain and it is unclear if lack of blinding would have introduced a risk of bias
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNumbers of participants included in the outcome evaluation is not stated
Selective reporting (reporting bias)High riskMethods section planned to report pain swelling and trismus. Swelling and trismus data not reported, only graph without estimates of variance and P values for difference
Other biasLow riskNo other sources of bias identified

Renton 2005

Methods

Study design: RCT (parallel)

Conducted in: Department of Oral and Maxillofacial Surgery, Guy's Dental Hospital London, UK

Participants

Inclusion criteria: Patients who required removal of third molars and were judged to be at high risk of injury to the inferior alveolar nerve based on radiographic features

Exclusion criteria: Patients who were predisposed to local infection, or who had systemic infections, and those with previous or existing defects of the inferior alveolar nerve. Patients with neuromuscular disorders or non-vital third molars were also excluded

Number of patients randomised: 128 patients, 196 teeth

Number of patients evaluated: Unclear

Interventions

Coronectomy versus complete surgical removal

Group A (n = 94 teeth): Coronectomy - sectioning 3-4 mm below the crown, reducing roots with bur and leaving in situ. No treatment to the pulp

Group B (n = 102 teeth): Complete surgical removal of teeth

60% of teeth were treated under general anaesthesia, 30% under local anaesthesia and 10% under sedation + local anaesthesia. 3 surgeons performed procedures

Follow-up: 2 years

OutcomesVerbal assessment and mechanosensory testing of inferior alveolar nerve, dry socket infection or soft tissue infection assessed immediately postoperation, on day 3 and after 1-2 weeks
Notes

Sample size calculation: Not reported

Unit of randomisation is teeth. Patients having non-surgical extraction were excluded. In order to overcome the problems due to the study being a mixture of split-mouth and parallel group designs, one site per patient was randomly selected

Additional information supplied by author

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskQuote: "..the teeth to be removed were randomised (using a table of random numbers that was concealed from the surgeon)"
Allocation concealment (selection bias)Low riskQuote: "..the teeth to be removed were randomised (using a table of random numbers that was concealed from the surgeon)"
Blinding (performance bias and detection bias)
patient
Unclear riskBlinding of patients not mentioned. Surgeons would have an ethical duty to inform their patients that some of their tooth was remaining in situ
Blinding (performance bias and detection bias)
assessor
High riskNot mentioned whether or not outcome assessors were blinded. Method of pain assessment not discussed
Incomplete outcome data (attrition bias)
All outcomes
High risk

128 patients were included but unclear how many patients in each group. Denominators for outcomes are teeth not people, and participants not analysed in the groups into which they were originally randomised: presence of the failed coronectomy subgroup confounds the intention-to-treat analysis

22 of the 196 teeth were simple elevation extractions, and there does not appear to be any statistical accounting for this. "Of the 58 patients who had coronectomy 47 (81%) attended the department for review within the first 6 months." No reasons given for these failures of follow-up

No mention of follow-up in extraction group

Selective reporting (reporting bias)High riskAll pre-specified outcomes reported on, but some outcomes not reported for each randomised group. Some data was supplied by the authors, but unclear when and how pain was assessed. Pain is reported per tooth, but patients with 2 teeth in the study would be expected to have greater pain
Other biasUnclear riskNo mention of how pain was assessed or if any statistical tests were done on it. 196 teeth from 128 patients and often unclear which numbers were used in analysis

Roode 2010

Methods

Study design: RCT (split-mouth)

Conducted in: Department of Maxillofacial and Oral Surgery, Faculty of Health Sciences, University of Pretoria, South Africa

Number of centres: 1

Recruitment period: Not stated

Participants

Inclusion criteria: No pre-existing medical conditions or medication use which would influence their ability to undergo surgery. Symmetrical, bilateral impacted lower third molars fully covered by mucosa with no discernable active pathology associated

Mean age: 19 years, 26 female, 10 male

Number of patients randomised: 36 patients, 72 teeth

Number of patients evaluated: 33

Interventions

Reverse L-flap versus straight line incision

Group A (n = 33): Reverse L-flap method of raising surgical flap for access to impacted tooth

Group B (n = 33): Alternative surgical flap method, a straight line incision

All patients had both types of flap in a single procedure. The side of mouth was randomly allocated

All patients treated under general anaesthesia

Follow-up: Clinical assessment at day 3, questionnaires collected with 7 days of postoperative data compiled by the patient

OutcomesOutcomes: Duration of procedure, infection incidence reported. Pain and swelling using a VAS assessed daily (every morning) from the day after surgery to day 7
Notes

Sample size calculation: Not mentioned

All procedures performed by the same surgeon

E-mail sent to authors requesting additional information 2/7/2012. Unpublished data supplied

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskThe side for the intervention and the control were selected by the "cast of a die"
Allocation concealment (selection bias)Unclear riskNot mentioned
Blinding (performance bias and detection bias)
patient
Low riskAll procedures carried out under a general anaesthetic
Blinding (performance bias and detection bias)
assessor
High riskNot mentioned whether or not clinical outcome assessors were blinded. Some outcomes patient reported
Incomplete outcome data (attrition bias)
All outcomes
Low risk36 patients were included in the trial but questionnaires were only returned by 33. Unlikely to introduce a bias in this split-mouth trial
Selective reporting (reporting bias)Low riskAll pre-specified outcomes reported, and estimates of variance for outcomes supplied by e-mail from authors
Other biasLow riskNo other sources of bias identified

Rullo 2013

Methods

Study design: RCT (split-mouth), procedures 30 days apart

Conducted in: Naples, Italy

Number of centres: 1

Recruitment period: Not stated

Participants

Inclusion criteria: (a) The presence in each subject of bilateral and symmetrically oriented impacted lower third molars to be extracted for prophylactic reasons; (b) forceps extractions not requiring osteotomy, were excluded; (c) no systemic diseases; (d) age> 18 years; (e) non-smoker; (f) not pregnant; and (g) no allergy to penicillin or other drugs used in the standardised postoperative therapy

Exclusion criteria: Patients who were taking antibiotics for current infection or who had acute pericoronitis or severe periodontal disease at the time of operation

Mean Age: 26.2 years, range 18-54 years. Male 20, female 32

Number of patients randomised: 52

Number of patients evaluated: 52

Interventions

Piezoelectric bone removal versus bur

Group A (n = 52): Piezoelectric handpiece operating with modulated ultrasound with a functional frequency of 25e29 kHz and a digital modulation of 30 kHz. The inserts moved with a linear vibration of between 60 and 210 mm

Group B (n = 52): Osteotomies using a conventional rotating drill were carried out with a Stryker tungsten carbide bur mounted on a surgical high-speed handpiece

Procedures subgrouped into 'simple extractions' and complex extractions. All procedures performed under local anaesthetic, and drain inserted, all patients received amoxicillin (500 mg 3 times daily for 7 days starting day before surgery), ibuprofen 600 mg 3 times daily for 4 days, and CHX mouthwash

Follow-up: VAS for pain completed daily for 6 days

OutcomesOutcomes: Duration of procedure, pain (100-point VAS), surgical difficulty (Parant scale) histological analysis of bone biopsy samples
Notes

Sample size calculation: Not reported

All procedures performed by the same surgeon

E-mail sent to authors requesting additional information

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskQuotes: "..the instruments were randomly selected using a coin toss", "instrument sequence was random"
Allocation concealment (selection bias)Unclear riskUnclear who performed the coin toss, and exactly how the first extraction side was chosen
Blinding (performance bias and detection bias)
patient
Unclear riskPatients and clinicians could not be blinded to allocated treatments
Blinding (performance bias and detection bias)
assessor
High riskNo blinding of outcome assessment undertaken
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskUnclear how many procedures were included in the simple and complex subgroups, for each procedure
Selective reporting (reporting bias)High riskPain measured and reported. It seems likely that attempts were made to measure swelling and trismus but these outcomes were not reported because the measures were "not reproducible"
Other biasHigh riskThe outcomes in the graphs and in the tables are contradictory

Saglam 2003

Methods

Study design: RCT (split-mouth)

Conducted in: Department of Oral and Maxillofacial Surgery, School of Dentistry at Suleyman Demirel University, Isparta, Turkey

Participants

Inclusion criteria: Healthy, co-operative patients aged 15-39 years who had bilateral fully impacted mandibular third molars, partly or completely covered by bone

Exclusion criteria: Patients with significant medical diseases or history of bleeding problems. Pregnant women and patients with signs of pericoronitis were also excluded from the study

Number of patients randomised: 13

Number of patients evaluated: Unclear - no mention of withdrawals but numbers evaluated not stated

Interventions

Tube drain versus no drain

Group A (n = 13 teeth): Small surgical tube drain applied via a stab incision in buccal fold between first and second molars; drain was removed 3 days postoperation

Group B (n = 13 teeth): No drain used; flap approximated without tension

All procedures performed by 1 surgeon and all patients received same antimicrobial and analgesic drugs. Seems likely that procedures performed on 2 separate visits, but timing unclear

All procedures performed under local anaesthetic

Follow-up: 7 days

OutcomesSwelling by measuring distance from commissures to ear lobe and distance from outer canthus of eye to angulus mandibulae. Maximum mouth opening measured between edges of maxillary and mandibular central incisors
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

Quote: "..the teeth were assigned [...] by random selection and in a crossover pattern"

Comment: Method of sequence generation not described

Allocation concealment (selection bias)Unclear riskUnclear
Blinding (performance bias and detection bias)
patient
Unclear riskNot mentioned and not possible to blind patients to the presence of an intraoral surgical drain
Blinding (performance bias and detection bias)
assessor
High risk

Quote: "..the patients were examined by the same surgeon immediately preoperatively, and on the first, second, third and seventh post-operative days."

Comment: Unclear whether the outcome assessor was the surgeon who performed the procedure. Probably obvious whether drain was used

Incomplete outcome data (attrition bias)
All outcomes
Low riskNo mention of withdrawals. Likely that all 26 were included in evaluation
Selective reporting (reporting bias)Low riskAll pre-specified outcomes reported on
Other biasLow riskNo other sources of bias identified

Sandhu 2010

Methods

Study design: Split-mouth cross-over RCT

Conducted in: India

Number of centres: 1

Recruitment period: Not stated

Participants

Inclusion criteria: Patients requiring extraction of bilateral impacted third molars, with no history of medical illness or medication use that could influence wound healing, healthy dental and periodontal status at the time of surgery. Attempt made to include those with teeth of comparable position and expected difficulty during extraction

Exclusion criteria: Not explicitly stated

Age: Mean 25 years

Number randomised: 20 (40 teeth)

Number evaluated: 20

Interventions

Bayonet flap versus envelope flap

Group A (n = 20): Bayonet flap raised

Group B (n = 20): Envelope flap raised

Minimum of 1 month between procedures

All procedures performed under local anaesthetic. All patients given prophylactic intravenous amoxicillin/clavulanic acid, ibuprofen tablet and CHX mouthrinse prior to surgery and ibuprofen and CHX mouthrinses in the postoperative period

OutcomesPain, facial swelling, trismus, wound dehiscence evaluated on days 1, 3, 7, 14 and 30
Notes

Sample size calculation: Not reported

E-mail sent 26/3/2012 requesting further information. Reply received 31/3/2012 with unpublished data

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskQuotes: "...randomized by systematic allocation"; "Both the type of flap used and the side operated were randomized by tossing a coin which was carried out by the surgeon, and communicated to the evaluator after the surgical procedure for recording" (e-mail communication)
Allocation concealment (selection bias)High riskCoin toss done by operating surgeon (e-mail communication)
Blinding (performance bias and detection bias)
patient
Low riskQuote: "..both patients and evaluator were blinded to the flap groups"
Blinding (performance bias and detection bias)
assessor
Low riskQuote: "..both patients and evaluator were blinded to the flap groups"
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll randomised participants included in the outcome evaluations
Selective reporting (reporting bias)Low riskAll planned outcomes reported
Other biasLow riskNo other sources of bias identified

Srinivas 2006

Methods

Study design: Split-mouth RCT

Conducted in: Bangalore, India

Number of centres: 1

Recruitment period: Not stated

Participants

Inclusion criteria: Patients aged 15-39 years willing to participate in the study. No significant medical history, non-smokers, non-alcoholics with bilateral and symmetrically positioned impacted lower third molars, which are completely covered by mucosa/partially or completely covered by bone

Exclusion criteria: None stated

Number randomised: 14

Number evaluated: 14

Interventions

Tube drain versus no drain

Group A (n = 14): Mucoperiosteal flap raised following envelope incision, flap was reflected and bone removed with a bur. Tooth was removed and socket was irrigated with saline. Small surgical drain was placed via stab incision in buccal fold between 1st and 2nd molar and closed. Tube was removed on postoperative day 3

Group B (n = 14): Mucoperiosteal flap raised following envelope incision, flap was reflected and bone removed with a bur. Tooth was removed and socket was irrigated with saline. Flap was approximated, closed with interrupted 3-0 silk sutures

All surgical procedures were performed by the same surgeon under local anaesthetic

Second extraction was performed after 2 months

OutcomesPain (present/absent), swelling (vertical/horizontal measurements), trismus (MMO)
NotesNo sample size calculation reported
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

Quote: "randomly chosen"

Comment: Method of sequence generation not described

Allocation concealment (selection bias)Unclear riskNot mentioned
Blinding (performance bias and detection bias)
patient
Unclear riskNot mentioned
Blinding (performance bias and detection bias)
assessor
High riskNot mentioned. Likely that the same operator performed the procedures and assessed the outcomes
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll randomised participants included in the outcome
Selective reporting (reporting bias)Low riskAll planned outcomes reported
Other biasLow riskNo other sources of bias identified

Sweet 1976

Methods

Study design: RCT (split-mouth)

Conducted in: United States Public Health Service Hospital, New York, USA

Participants

Inclusion criteria: Male patients from 17 to 27 years of age, who were in good health, and who required bilateral, similarly impacted wisdom teeth extracted. Medical health was ascertained by a "complete physical examination by a physician, normal hospital screening tests, a resident's admission examination, and a complete medical history." In addition, "only patients with soft-tissue or osseous-tissue impactions which were asymptomatic were accepted for the study"

Exclusion criteria: "..patients with a preoperative infection or pericoronitis were eliminated from the study"

Number of patients randomised: 103 men, 206 teeth

Number of patients evaluated: 99 - no withdrawals but 4 patients with infection excluded from other outcome assessments

Interventions

Mechanical irrigation versus manual irrigation

Group A (n = 103 teeth): Postextraction mechanical lavage (350 ml sterile saline)

Group B (n = 103 teeth): Conventional manual syringe lavage (350 ml sterile saline)

Procedures performed under general anaesthetic, both teeth extracted in same session by same surgeon

Follow-up: Days 3 and 5

OutcomesAlveolar osteitis, infection, pain (4-point scale), swelling (4-point scale)
Notes

Sample size calculation: Not reported

4 patients who presented with alveolar osteitis or infection were excluded from other outcome assessments

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskQuote: "The type of irrigation [...] was predetermined by random selection technique, with the use of random sampling numbers, before the study was begun"
Allocation concealment (selection bias)Unclear riskNot mentioned
Blinding (performance bias and detection bias)
patient
Low riskBlinding not mentioned but likely that patients were unaware of lavage volume as they were sedated with pentobarbital
Blinding (performance bias and detection bias)
assessor
Low risk

Quotes: "These examinations were made by a dental surgeon who was not involved with the operation"; "the surgical sites were observed by a dental surgeon who was not involved with the operation, and who was unaware of the irrigation methods used"

Comment: Assessor blinding successful

Incomplete outcome data (attrition bias)
All outcomes
Low riskNo withdrawals, but 4 patients who had infections excluded from wound healing. However in a split-mouth study this is unlikely to introduce bias. Quote: "once a patient was treated, he was then counted in the 'treated group', and was not evaluated for any healing results at the 3- or 5-day levels"
Selective reporting (reporting bias)Low riskAll planned outcomes reported
Other biasLow riskNo other sources of bias identified

Xavier 2008

  1. a

    CAL = clinical attachment level; CHX = chlorhexidine; IDN = inferior dental nerve; MMO = maximum mouth opening; PPD = probing pocket depth; RCT = randomised controlled trial; TMD = temporomandibular disorder; VAS = visual analogue scale

Methods

Study design: RCT (split-mouth)

Conducted in: Recife, Brazil

Recruitment period: May to September 2004

Participants

Inclusion criteria: Patients consecutively enrolled between May and September 2004 for surgical extraction of bilateral impacted lower third molars. Both wisdom teeth had to be in similar position according to Pell and Gregory classification

Exclusion criteria: History of significant systemic pathology, or use of any medication that could interfere with the repair process

Number of patients randomised: 40 patients

Number of patients evaluated: 40 patients

Interventions

Partial wound closure versus complete wound closure

Group A (n = 20 teeth): Sutures on attached gum only

Group B (n = 20 teeth): Complete suture was performed on free and attached gums

Procedures performed under local anaesthetic, both teeth extracted in same session by same surgeon

Follow-up: Days 3, 7 and 15 and 3 months

OutcomesPain, swelling, trismus at 7 days, probing depth 3 months postoperation
Notes

Sample size calculation: Not reported

E-mail sent 29/3/2012 requesting further information. No reply received

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

Quote: "2 groups were established on randomised basis (by allotment)"

Comment: Method of sequence generation not described

Allocation concealment (selection bias)Unclear riskNot mentioned
Blinding (performance bias and detection bias)
patient
Low riskDouble blinded
Blinding (performance bias and detection bias)
assessor
Low riskDouble blinded - assumed that both patients and clinical outcome assessor blinded to treatment
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll randomised participants included in outcome evaluation
Selective reporting (reporting bias)Unclear riskAll planned outcomes reported
Other biasUnclear riskNo other sources of bias identified

Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion
  1. a

    RCT = randomised controlled trial

Abu-Serriah 2004Most study participants had both maxillary and mandibular third molars extracted in the same procedure. Also bur group were irrigated with saline and the laser group were irrigated with water. This was thought to be a confounding factor
Akota 1998Surgical drain used in 1 group was coated with chlortetracycline ointment. It was thought that the antibiotic confounded the effect of the drain
Ayad 1995Unclear study design - both parallel group and split-mouth study. Also wisdom teeth in the maxilla were included
Cetinkaya 2009Primary outcome measures of this review not reported. Trial looks at effects on periodontal disease
Chossegros 2002Study describes germectomy not extraction of third molars
Clauser 1994Study not eligible as not all patients had incision and flap raised. Some third molars were removed using an elevator
Ding 2000No mention of incision or flap being raised
Dubois 1982Split-mouth trial but allocation of 1 side of face to treatment was not randomised
Finne 1981Study describes germectomy not extraction of third molars
Genu 2008Unclear whether study was truly randomised, appeared to have 2 interventions and there was confounding of the effects. Unable to contact authors to obtain further information
Gonzalez 2001Abstract only - insufficient information to include. No subsequent publication identified
Goyal 2012Allocation by alternation
Holland 1984Comparison is between 2 types of wound closure but 1 incision also had a dressing used, which acted as a confounder to the interventions studied in this review
Jakse 2002Not RCT
Kerdvongbundit 1989Unable to locate a copy of this paper. There is no abstract, and based on title it is unclear whether this was a randomised trial
Quee 1985Trial of the effect of flap design on subsequent periodontal health. Not relevant to this review
Robinson 1996The study compared removal of wisdom teeth with or without lingual flap retraction. Prior to operation the operator was allocated the random group procedure. The authors found significant differences for the grades of surgical difficulty between the 2 groups. On investigation some of the operators had deviated from the protocol. In an unknown number of easier surgical cases the lingual flap was not raised when it should have been. We are unable to quantify the bias caused by this protocol violation
Rosa 2002Not RCT
Shevel 2001This study included both mandibular and maxillary molar teeth. The removal of the maxillary teeth may affect pain and swelling so we could not use the data for the mandibular teeth
Sivolella 2011Study describes germectomy not extraction of third molars
Sortino 2008Not RCT
Strukmeier 1980Probably not RCT after translation from German
Suarez-Cunqueiro 2003This study is confounded by the surgical removal of unerupted maxillary third molars
Suddhasthira 1991Paper in Thai language. Sent to translator. Not an RCT
Sweet 1978Contradiction on state of teeth. They are described as all being impacted and then some are described as being erupted. We are also unsure whether the same surgical technique was used bilaterally
Torres-Lagares 2006Intervention is postsurgical use of chlorhexidine gel to prevent infection. Not a surgical intervention
Tuffin 1990Design fault. This study compared irrigation of the socket at the end of surgery with bupivacaine versus no irrigation. Apart from the planned interventions the patients were also treated by chisel or drill technique, with 1 (10%) of patient in the control having the drill compared with 15 (65%) of patient in the treatment group. We felt this could be a confounding factor
Zhang 1997No mention of incision or flap being raised

Characteristics of studies awaiting assessment [ordered by study ID]

Kumar 2013

MethodsRCT cross-over design
Participants20 participants
InterventionsStandard incision versus comma shaped incision and its influence on postoperative complications in surgical removal of impacted third molar
OutcomesPostoperative complications
NotesCategorical data - need to dichotomise and ask authors for paired data

Ozveri Koyuncu 2013

  1. a

    RCT = randomised controlled trial

MethodsRCT parallel group
Participants36 patients
Interventions3-cornered flap versus modified triangular flap
OutcomesDehiscence, pain, swelling, mouth opening
Notes 

Characteristics of ongoing studies [ordered by study ID]

ACTRN 2010

  1. a

    RCT = randomised controlled trial

Trial name or titleRCT split-mouth design
Methods 
ParticipantsUnclear
InterventionsPedicle flap design versus control flap design
OutcomesUnclear
Starting dateUnclear
Contact information www.anzctr.org.au/ACTRN12610000500055.aspx
Notes 

Ancillary