Delayed antibiotics for respiratory infections

  • Review
  • Intervention

Authors


Abstract

Background

Concerns exist regarding antibiotic prescribing for acute respiratory tract infections (ARTIs) owing to adverse reactions, cost and antibacterial resistance. One strategy to reduce antibiotic prescribing is to provide prescriptions but to advise delay in the hope symptoms will resolve first. This is an update of a Cochrane Review originally published in 2007 and updated in 2010.

Objectives

To evaluate the use of delayed antibiotics compared to immediate or no antibiotics as a prescribing strategy for ARTIs. We evaluated clinical outcomes including duration and severity measures for pain, malaise, fever, cough and rhinorrhoea in sore throat, acute otitis media, bronchitis (cough) and the common cold. We also evaluated the outcomes of antibiotic use, patient satisfaction, antibiotic resistance and re-consultation rates and use of alternative therapies.

Search methods

We searched CENTRAL (The Cochrane Library 2013, Issue 2), which includes the Acute Respiratory Infection Group's Specialised Register; Ovid MEDLINE (January 1966 to February Week 3 2013); Ovid MEDLINE In-Process & Other Non-Indexed Citations (28 February 2013); EMBASE (1990 to 2013 Week 08); Science Citation Index - Web of Science (2007 to May 2012) and EBSCO CINAHL (1982 to 28 February 2013).

Selection criteria

Randomised controlled trials (RCTs) involving participants of all ages defined as having an ARTI, where delayed antibiotics were compared to antibiotics used immediately or no antibiotics.

Data collection and analysis

Three review authors independently extracted and collected data. Important adverse effects, including adverse effects of antibiotics and complications of disease, were included as secondary outcomes. We assessed the risk of bias of all included trials. We contacted trial authors to obtain missing information where available.

Main results

Ten studies, with a total of 3157 participants, were included in this review. Heterogeneity of the 10 included studies and their results generally precluded meta-analysis with patient satisfaction being an exception.

There was no difference between delayed, immediate and no prescribed antibiotics for the clinical outcomes evaluated in cough and common cold. In patients with acute otitis media (AOM) and sore throat immediate antibiotics were more effective than delayed for fever, pain and malaise in some studies. There were only minor differences in adverse effects with no significant difference in complication rates.

Delayed antibiotics resulted in a significant reduction in antibiotic use compared to immediate antibiotics. A strategy of no antibiotics resulted in least antibiotic use.

Patient satisfaction favoured immediate antibiotics over delayed (odds ratio (OR) 0.52; 95% confidence interval (CI) 0.35 to 0.76). Delayed and no antibiotics had similar satisfaction rates with both strategies achieving over 80% satisfaction (OR 1.44; 95% CI 0.99 to 2.10).

There was no difference in re-consultation rates for immediate and delayed groups.

None of the included studies evaluated antibiotic resistance.

Authors' conclusions

Most clinical outcomes show no difference between strategies. Delay slightly reduces patient satisfaction compared to immediate antibiotics (87% versus 92%) but not compared to none (87% versus 83%). In patients with respiratory infections where clinicians feel it is safe not to prescribe antibiotics immediately, no antibiotics with advice to return if symptoms do not resolve is likely to result in the least antibiotic use, while maintaining similar patient satisfaction and clinical outcomes to delayed antibiotics.

Résumé scientifique

Antibiothérapie différée pour les infections respiratoires

Contexte

Des préoccupations existent concernant la prescription d'antibiotiques pour les infections respiratoires aiguës (IRA) en raison des réactions indésirables, du coût et de la résistance antibactérienne. Une stratégie pour réduire la prescription d'antibiotiques est de fournir des prescriptions mais conseiller de les différer dans l'espoir que les symptômes se résolvent d'eux-mêmes. Ceci est une mise à jour d'une revue systématique Cochrane publiée initialement en 2007 et mise à jour en 2010.

Objectifs

Évaluer l'utilisation de différée des antibiotiques par rapport à leur utilisation immédiate ou leur absence d'utilisation en tant que stratégie de prescription pour l'IRA. Nous avons évalué les résultats cliniques, notamment la durée et la gravité des mesures pour la douleur, les malaises, la fièvre, la toux et la rhinorrhée pour le mal de gorge, l'otite moyenne aiguë, la bronchite (la toux) et le rhume banal. Nous avons également évalué les résultats de l'utilisation d'antibiotiques, la satisfaction du patient, la résistance aux antibiotiques et les taux de nouvelle consultation et l'utilisation de traitements alternatifs.

Stratégie de recherche documentaire

Nous avons effectué des recherches dans CENTRAL ( La Bibliothèque Cochrane 2013, numéro 2), qui inclut le registre spécialisé du groupe Cochrane sur les infections respiratoires aiguës, Ovid MEDLINE (de janvier 1966 à la 3ème semaine de février 2013) ; Ovid MEDLINE In-Process & Other Non-Indexed Citations (28 février 2013) ; EMBASE (de 1990 à 2013 la semaine 8) ; Science Citation Index - Web of Science (de 2007 à mai 2012) et EBSCO CINAHL (de 1982 jusqu'au 28 février 2013).

Critères de sélection

Essais contrôlés randomisés (ECR) portant sur des participants de tous âges définis comme présentant une IRA, où l'antibiothérapie différée était comparée à l'antibiothérapie immédiate ou absente

Recueil et analyse des données

Trois auteurs de la revue ont extrait et recueilli les données de manière indépendante. Les effets indésirables importants, y compris les effets indésirables des antibiotiques et les complications de la maladie, ont été inclus comme critères de jugement secondaires. Nous avons évalué le risque de biais de tous les essais inclus. Nous avons contacté les auteurs des essais pour obtenir des informations manquantes lorsqu' elles étaient disponibles.

Résultats principaux

Dix études, avec un total de 3157 participants, ont été inclues dans cette revue. L'hétérogénéité des 10 études incluses et de leurs résultats généralement empêchait une méta-analyse, la satisfaction des patients étant une exception.

Il n'y avait aucune différence entre l'antibiothérapie différée, immédiate ou absente pour les résultats cliniques évalués dans la toux et le rhume banal. Chez les patients souffrant d'une otite moyenne aiguë (OMA) ou de mal de gorge l'antibiothérapie immédiate était plus efficace que la différée pour la fièvre, la douleur et les malaises dans certaines études. Il n'y avait que des différences mineures concernant les effets indésirables avec aucune différence significative dans les taux de complication.

L'antibiothérapie différée entraînait une réduction significative de l'utilisation d'antibiotiques par rapport à l'antibiothérapie immédiate. Une stratégie d'absence d'antibiothérapie entraînait la moindre utilisation d'antibiotiques.

La satisfaction des patients étaient en faveur de l'antibiothérapie immédiate par rapport à l'antibiothérapie différée (rapport des cotes (RC) 0,52 ; intervalle de confiance (IC) à 95 % 0,35 à 0,76). Les stratégies d'antibiothérapie immédiate et d'absence d'antibiothérapie présentaient des taux de satisfaction similaires obtenant plus de 80 % (RC 1,44 ; IC à 95 % 0,99 à 2,10).

Il n'y avait aucune différence en termes de taux de nouvelle consultation entre les groupes d'antibiothérapie immédiate et d'antibiothérapie différée

Aucune des études incluses n'évaluaient la résistance aux antibiotiques.

Conclusions des auteurs

La plupart des critères de jugement cliniques n'ont montré aucune différence entre les stratégies. L'antibiothérapie différée réduit légèrement la satisfaction du patient par rapport à l'antibiothérapie immédiate (87 % versus 92 %) mais pas par rapport à l'absence d'antibiothérapie (87 % contre 83 %). Chez les patients souffrant d'infections respiratoires lorsque les cliniciens estiment qu'il est sans danger de ne pas prescrire d'antibiotiques immédiatement , tout en conseillant au patient de revenir si les symptômes ne disparaissent pas, est susceptible d'entraîner la moindre utilisation d'antibiotiques, en maintenant une satisfaction et des résultats cliniques pour le patient similaires à l'antibiothérapie différée.

Plain language summary

Delayed antibiotics for symptoms and complications of acute respiratory tract infections

Previous reviews indicate that antibiotics have, at best, only modest benefit for acute respiratory tract infections (ARTIs). These benefits need to be balanced against adverse effects, costs and the risk of bacteria becoming resistant to antibiotics. One way for doctors to reduce their use is to prescribe delayed antibiotics (meaning providing the prescription but advising the patient/carer to delay their use in the hope that symptoms resolve first). Delayed prescribing resulted in 32% of patients using antibiotics compared to 93% of patients in the immediate prescription group. However, not prescribing antibiotics at all results in the least antibiotic prescribing (14% of patients used antibiotics).

This review found 10 studies, involving 3157 participants, looking at prescribing strategies for respiratory infections. It was generally not possible to combine results from different studies because of incomplete information from some studies and the different types of patients in each study. There were only three trials comparing the strategies of delayed and no antibiotics.

For most symptoms like fever, pain and malaise, there was no difference between immediate, delayed and no antibiotics. The only differences were small and favoured immediate antibiotics for relieving pain and fever for sore throat and pain and malaise for middle ear infections. There was little difference in adverse effects of antibiotics for the three prescribing strategies and no significant difference in complication rates.

Patient satisfaction was slightly reduced in the delayed antibiotic group (87% satisfied) compared to the immediate antibiotic group (92% satisfied). Satisfaction rates were similar between delayed and no antibiotic groups (83% satisfied).

No included studies evaluated antibiotic resistance.

When doctors feel it is safe not to prescribe antibiotics immediately, prescribing none with advice to return if symptoms do not resolve rather than delaying them will result in lower subsequent antibiotic use, while maintaining similar patient satisfaction and symptom outcomes.

Laički sažetak

Davanje antibiotika s odgodom za simptome i komplikacije akutnih upala dišnog sustava

Prethodno objavljeni sustavni pregledi su pokazali da antibiotici u najmanju ruku imaju umjeren učinak kod akutnih infekcija dišnog sustava. Ove koristi trebaju se razmotriti u svjetlu mogućih nuspojava, cijene i rizika od razvoja otpornosti bakterija na antibiotike. Jedan način na koji liječnici mogu smanjiti uporabu antibiotika jest da propišu ODGOĐENU primjenu antibiotika – što znači da pacijentima daju recept za antibiotike, ali s uputom da ih ne uzmu odmah nego da ih uzmu nakon određenog vremena ako se simptomi upale ne počnu povlačiti. Odgođeno propisivanje antibiotika značajno smanjuje uporabu antibiotika među pacijentima – ako se antibiotici propišu sa savjetom za odgodu uzme ih 32% pacijenata, a ako se propišu sa savjetom da se počnu uzimati odmah, onda ih uzme 93% pacijenata. Međutim, ako se pacijentima ne propišu antibiotici, to dovodi do najmanje uporabe antibiotika jer ih u toj skupini na koncu ipak uzme 14% pacijenata.

U Cochrane sustavni pregled uključeno je 10 kliničkih ispitivanja s ukupno 3157 ispitanika, u kojima su istražene različite strategije propisivanja antibiotika za dišne upale. Općenito nije bilo moguće kombinirati rezultate iz različitih studija zbog nedovoljne količine informacija u nekim studijama i različite vrste pacijenata koji su uključivani u pojedine studije. Samo su 3 istraživanja usporedila odgođenu primjenu antibiotika i uskraćivanje antibiotika.

Za većinu simptoma, kao što su vrućica, bol i umor, nije bilo razlike između trenutne primjene, odgođene primjene i uskraćivanja antibiotika. Jedina razlika, koja je bila malena i u korist trenutnog propisivanja antibiotika, uočena je za ublažavanje boli i vrućice kod grlobolje i za ublažavanje boli i umora kod upale srednjeg uha. Između ove tri strategije propisivanja antibiotika uočene su male razlike u nuspojavama, i nije bilo značajne razlike u učestalosti komplikacija.

Zadovoljstvo pacijenata bilo je neznatno umanjeno u skupini koja je dobila recept za odgođeno propisivanje lijekova (87% zadovoljnih) u usporedbi s pacijentima kojima je antibiotik trenutno propisan (92% zadovoljnih). Zadovoljstvo pacijenata bilo je slično između skupine s odgođenim propisivanjem i pacijenata kojima je antiobiotik uskraćen (83% zadovoljnih).

Niti jedna od analiziranih studija nije analizirala otpornost bakterija na antibiotike.

Kad liječnici smatraju da je sigurno ne propisati antibiotike odmah, bolje je ne propisati ih uopće nego savjetovati pacijentima da se vrate ako im simptomi ne popuste, jer će se na taj način smanjiti uporaba antibiotika, a zadovoljstvo pacijenata i ishodi simptoma će ostati slični.

Bilješke prijevoda

Prevoditelj:: Croatian Branch of the Italian Cochrane Centre

Résumé simplifié

Antibiothérapie différée pour les symptômes et les complications des infections aiguës des voies respiratoires

De précédentes Revues indiquent que les antibiotiques ont, au mieux, un bénéfice modeste pour les infections respiratoires aiguës (IRA). Ces effets bénéfiques doivent être mis en balance avec les effets indésirables, les coûts et le risque d'antibiorésistance. Un moyen pour les médecins de réduire leur utilisation est la prescription différée (c'est à dire en fournir l'ordonnance mais conseiller au patient/soignant de différer leur utilisation dans l'espoir que les symptômes se résolvent d'eux-mêmes). La prescription différée a entraîné 32 % de patients à utiliser des antibiotiques contre 93 % des patients dans le groupe à prescription immédiate. Cependant, l'absence totale d'une prescription d'antibiotiques a comme résultat la prescription la moins élevée d'antibiotiques (14 % des patients ont utilisé des antibiotiques).

Cette revue a identifié 10 études, impliquant 3157 participants, portant sur des stratégies de prescription pour les infections respiratoires. Il n'a généralement pas été possible de combiner les résultats des différentes études en raison d'informations incomplètes et de différents types de patients dans chaque étude. Seuls trois essais comparaient les stratégies d'antibiothérapie différée par rapport à l'absence d'antibiothérapie

Pour la plupart des symptômes tels que la fièvre, la douleur et les malaises, il n'y avait aucune différence entre antibiothérapie différée et absente. Les seules différences étaient petites et favorables à l'antibiothérapie immédiate pour soulager la douleur et la fièvre pour le mal de gorge et la douleur et les malaises pour les infections de l'oreille moyenne. Il y avait peu de différence concernant les effets indésirables des antibiotiques pour les trois stratégies de prescription, et aucune différence significative dans les taux de complication.

La satisfaction des patients était légèrement réduite dans le Groupe sous antibiotiques différés (87 % de patients satisfaits) par rapport au Groupe sous antibiotiques immédiats (92 % de patients satisfaits). Le taux de satisfaction était similaire entre les Groupes d'antibiothérapie différée et absente (83 % de patients satisfaits).

Aucune des études incluses n'évaluait la résistance aux antibiotiques.

Dans le cas où les médecins estiment qu'il est sans dangerde ne pas prescrire d'antibiotiques immédiatement , alors ne pas en prescrire du tout en conseillant au patient de revenir si les symptômes ne disparaissent pas, plutôt que leur prescription différée permet de réduire l'utilisation ultérieure d'antibiotiques, avec des résultats similaires sur la satisfaction du patient et les symptômes.

Notes de traduction

Traduit par: French Cochrane Centre 1st December, 2013
Traduction financée par: Instituts de Recherche en Sant� du Canada, Minist�re de la Sant� et des Services Sociaux du Qu�bec, Fonds de recherche du Qu�bec-Sant� et Institut National d'Excellence en Sant� et en Services Sociaux

Background

Description of the condition

The use of antibiotics for acute respiratory tract infections (ARTIs) is controversial. Empirical evidence suggests that antibiotics have only a modest benefit in acute otitis media (AOM) (Venekamp 2013), pharyngitis (Spinks 2011) and acute bronchitis (Smith 2011) and no effect in the common cold (Arroll 2010). Any benefits have to be weighed up against common adverse reactions (including rash, abdominal pain, diarrhoea and vomiting) and cost (Berman 1997; Niemela 1999). Over-prescribing may also contribute to community bacterial resistance to antibiotics (Arason 1996; Brook 1998; Verkatesum 1995).

Description of the intervention

There has been interest in strategies to reduce antibiotic prescribing for ARTIs. One of these strategies is to advise patients to 'delay' filling their script and only to fill it if their symptoms persist or deteriorate. Delayed antibiotics are advocated as a means of demonstrating to patients that antibiotics are not always necessary, without making them feel under-serviced (Arroll 2002b). Two ways of using this strategy have been deployed: giving the patient the antibiotic (with instructions not to use unless there is deterioration); and making the prescription available at the clinic reception (to be picked up in the event of deterioration).

How the intervention might work

Delaying antibiotics may provide a feeling of safety for both patient and clinician should an illness deteriorate. This intervention then provides the safety of having a prescription of antibiotics available, yet an educational way of experiencing whether the illness resolves spontaneously without their use.

A systematic review showed that using delayed antibiotics in ARTIs significantly reduces antibiotic prescribing (Arroll 2003a). The reduction ranges from a risk ratio (RR) of 0.77 (95% confidence interval (CI) 0.73 to 0.81) (Dowell 2001) to RR 0.25 (95% CI 0.19 to 0.34) (Little 1997).

Why it is important to do this review

The delayed antibiotic strategy has also been advocated more recently as a safety net for avoiding rare but important complications of initially uncomplicated ARTIs (Little 2005b). The same authors also advocated delayed antibiotics for reducing antibiotic use, allowing adequate control of symptoms, while providing high levels of patient satisfaction (Little 2005b).

This review asks specifically what effect delayed antibiotics have on clinical outcomes of ARTIs compared to immediate antibiotics and no antibiotics. This review also evaluates the available data on antibiotic use, patient satisfaction and antibiotic resistance for the three prescribing strategies of delayed antibiotics, immediate antibiotics and no antibiotics. This is an update of a Cochrane Review originally published in 2007 (Spurling 2007), with an updated version published in 2010 (Spurling 2010).

Objectives

To evaluate the use of delayed antibiotics compared to immediate or no antibiotics as a prescribing strategy for ARTIs. We aimed to evaluate clinical outcomes including duration and severity measures for pain, malaise, fever, cough and rhinorrhoea in sore throat, AOM, bronchitis (cough) and the common cold. We also aimed to evaluate the outcomes of antibiotic use, patient satisfaction, antibiotic resistance and re-consultation rates and use of alternative therapies.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials (RCTs) studying the treatment of ARTIs with delayed antibiotics versus immediate or no antibiotics. Open randomised trials were accepted.

Types of participants

Patients of all ages defined as having ARTIs.

Types of interventions

  1. 'Delayed antibiotic use' was defined as a strategy involving the use of or advice to use antibiotics more than 48 hours after the initial consultation.

  2. 'Immediate antibiotic use' was defined as the immediate use of a prescription of oral antibiotics given at the initial consultation.

  3. 'No antibiotic use' was defined as no prescription of antibiotics at the initial consultation.

Types of outcome measures

Primary outcomes

We compared delayed antibiotics with immediate antibiotics and delayed antibiotics with no antibiotics where data were available.

  1. Clinical outcomes for sore throat, AOM, bronchitis (cough) and common cold (we included duration and severity measures for the following symptoms: pain, malaise, fever, cough and rhinorrhoea)

  2. Antibiotic use

  3. Patient satisfaction (where patient satisfaction is measured on a four to six-point Likert scale; we defined satisfaction as including both satisfied and very satisfied)

  4. Antibiotic resistance

Secondary outcomes
  1. Adverse effects of antibiotics

  2. Complications of disease

  3. Re-consultation

  4. Use of alternative therapies

Search methods for identification of studies

Electronic searches

For this updated review we searched the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2013, Issue 2), which includes the Acute Respiratory Infection Group's Specialised Register; Ovid MEDLINE (January 1966 to February Week 3 2013); Ovid MEDLINE In-Process & Other Non-Indexed Citations (28 February 2013); EMBASE (1990 to 2013 Week 08); Science Citation Index - Web of Science (2007 to May 2012) and EBSCO CINAHL (1982 to 28 February 2013).

In the original version of this review MEDLINE was searched using the following keywords and MeSH terms in conjunction with the highly sensitive search strategy designed by The Cochrane Collaboration for identifying randomised controlled trials (Dickersin 1994). For this update we applied no trial filters. We used the MEDLINE search strategy to search CENTRAL (Appendix 1) and adapted this to search EMBASE (Appendix 2) and CINAHL (Appendix 3).

Ovid MEDLINE

1     exp Respiratory Tract Infections/ (114895)
2     (upper respiratory tract infection$ or urti).mp. (2482)
3     exp Otitis Media/ (8289)
4     otitis media.mp. (10100)
5     exp Pharyngitis/ (4870)
6     pharyngitis.mp. (3733)
7     exp Tonsillitis/ (2065)
8     tonsillitis.mp. (2423)
9     exp Common Cold/ (1492)
10     common cold.mp. (2207)
11     exp Bronchitis/ (8275)
12     bronchitis.mp. (8027)
13     exp Sinusitis/ (8071)
14     sinusitis.mp. (10465)
15     sore throat$.mp. (2080)
16     or/1-15 (133707)
17     exp Anti-Bacterial Agents/ (215537)
18     antibiotic$.mp. (127408)
19     or/17-18 (278179)
20     (delay$ adj15 prescri$).mp. (474)
21     and/16,19-20 (55)

There were no language or date of publication restrictions in any of the electronic database searches.

Searching other resources

We scanned abstracts from the search results to identify trials that loosely met the inclusion criteria. We checked references of all relevant retrieved trials to identify any other articles.

Data collection and analysis

Selection of studies

In the original publication of this review, we scanned abstracts from the initial search results to identify trials that loosely met the inclusion criteria. We checked references of all relevant retrieved trials to identify any other articles. Three review authors (RFo, LD, CDM) independently reviewed the full-text articles of the retrieved trials.

In the 2010 update, one further study was found to meet the inclusion criteria (Chao 2008) and two review authors (LD, CDM) independently assessed the methodological quality of the new included study that met the inclusion criteria at that time (Chao 2008).

Similarly, in this updated review (2013), three authors (RFo, GS, RFa) scanned abstracts from the updated searches to identify trials that met the inclusion criteria, checking the references of all retrieved trials to identify other articles. Three review authors (LD, CDM, RFa) independently reviewed the full-text articles of the retrieved trials and applied the inclusion criteria.

We identified two papers, Little 2006 and Moore 2009, as reporting longer-term outcomes from previously included studies (Little 2001; Little 2005a).

Data extraction and management

In the initial publication of this review, three review authors (RFo, LD and CDM) independently extracted data for each study trial to be included. We extracted data in a blinded manner (that is, without the knowledge of the study results, the names of the authors, institutions or journal of publication). We extracted additional data from graphs of the published articles of El-Daher 1991 and Pichichero 1987 on fever severity and symptom scores.

In this most recent update (2013), two review authors (LD, CDM) independently extracted data from the two new included papers. We contacted the authors of Little 2006 to obtain original data for the outcomes of earache at three months and one year that had been reported as odds ratios (ORs) in the published trial. The complete data were unavailable and there was some inconsistency between what was provided and the published numbers. These results have been included in the text of this review, in the form of the published ORs.

Assessment of risk of bias in included studies

In the first publication of this review three review authors (RFo, LD, CDM) independently assessed the quality of each of the study trials that met the inclusion criteria. We resolved disagreements by consensus. Assessment was blinded (that is, without the knowledge of the study results, the names of the authors, institutions or journal of publication).

We rated the quality of each eligible RCT according to the 'Risk of bias' tool available in RevMan 5.2 and criteria set out in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We assessed methodological quality under the headings of allocation, blinding, incomplete outcome data, selective reporting and other potential sources of bias.

Two review authors (LD, CDM) independently assessed the methodological quality of the trial included in the 2010 update.

We resolved disagreements by discussion.

Measures of treatment effect

We analysed data using RevMan 5.2. We expressed continuous data comparisons using mean differences (MD) where there was one study or standardised MD where more than one study used different measurement scales. We expressed dichotomous data using odds ratios (OR). We pooled data into clinical outcomes where multiple trial results for the same clinical presentation existed and there was no heterogeneity.

Unit of analysis issues

The units of analysis for each outcome are the individual research participants.

Dealing with missing data

Six studies included an intention-to-treat (ITT) analysis. Three other studies described their minimal drop-out rates. One study (El-Daher 1991) did not discuss the drop-out rate, though it was small.

Assessment of heterogeneity

We did not undertake a meta-analysis for most clinical outcomes owing to multiple analyses with only one or two study results. We pooled results where satisfactorily low I2 statistic and non-significant Chi2 test results were found. We did not undertake a meta-analysis for antibiotic use owing to the heterogeneity of the included study results, likely owing to different antibiotic indications for different clinical presentations.

Assessment of reporting biases

Two studies collected data on clinical outcomes yet did not report them in detail (Dowell 2001; Gerber 1990). In both cases, the studies reported that there was no difference between control and intervention groups.

Data synthesis

Most of the data in this review are reported as a narrative synthesis describing outcome measures. As indicated previously, we pooled results where satisfactorily low I2 statistic and non-significant Chi2 test results were found. We undertook a meta-analysis for the outcomes of fever for sore throat and patient satisfaction.

Subgroup analysis and investigation of heterogeneity

Subgroup analyses were considered for all outcomes and included year of publication, clinical presentation, differences in the intervention and risk of bias.

We describe in the results section the two subgroup analyses that showed differences in outcomes. We explored heterogeneity of antibiotic use in delayed antibiotic arms further with analysis of different methods of the delay strategy. We explored heterogeneity of patients satisfaction further with respect to blinding of outcome assessor and patient.

Results

Description of studies

Results of the search

Searches conducted for this review have resulted in 244 articles being identified by electronic searching; 28 were retrieved for more detailed evaluation and 17 studies have been formally evaluated.

Five studies were excluded and are described in the Excluded studies section. Two studies identified in this 2013 update reported longer-term outcomes from previously included studies (Little 2006; Moore 2009) and while their data have been added to this review, they are considered part of the original included studies.

Ten trials were eligible for inclusion. They included 1159 participants in their delayed antibiotic arm, with 1067 participants in the immediate antibiotic arm of nine trials and 465 participants in the no antibiotic arm of three trials.

In this most recent update (2013), following removal of duplicated studies, searches resulted in the identification of 77 articles (out of the 244 previously mentioned). Five articles were retrieved for further evaluation (out of 28). Three studies were excluded (out of a total of five) because they were not randomised. The remaining two reported longer-term outcomes from previously included studies (Little 2006; Moore 2009) and while their data have been added to this review, they are considered part of the original included studies. Therefore, there are no more included studies as a result of this 2013 update.

Included studies

Nine trials compared immediate antibiotics with delayed antibiotics. Four of these trials investigated acute pharyngitis/sore throat; two with AOM; two with cough and one dealt with the common cold. Early studies of sore throat (El-Daher 1991; Gerber 1990; Pichichero 1987) were designed as efficacy trials to identify the rate of relapse of group A beta-haemolytic streptococcus (GABHS) throat in immediate versus delayed antibiotic groups. Subsequent trials (Arroll 2002a; Dowell 2001; Little 1997; Little 2001; Spiro 2006) comparing delayed antibiotics and immediate antibiotics were conducted with a view to evaluate the use of delayed antibiotics to reduce the use of antibiotics for upper respiratory tract infections (RTIs).

Three studies compared the prescribing strategy of no antibiotics with delayed antibiotics (Chao 2008; Little 1997; Little 2005a). These three trials investigated the presentations of sore throat (Little 1997), cough (Little 2005a) and AOM (Chao 2008). This last trial (Chao 2008) also asked patients in the no antibiotic arm to return if their symptoms had not resolved.

Excluded studies

Since the first publication of this review, five trials have been excluded. One because it used a before-and-after study design (Cates 1999) and four because they were not randomised.

Risk of bias in included studies

Summaries of the bias in included studies are provided in Figure 1 and Figure 2.

Figure 1.

Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.

Figure 2.

Methodological quality summary: review authors' judgements about each methodological quality item for each included study.

Allocation

Eight included studies were adequately randomised using random number tables or computer-generated randomisation. In two studies the method of randomisation was not described (El-Daher 1991; Little 1997). Only four trials described adequate allocation concealment using opaque envelopes (Arroll 2002a; Little 2001; Little 2005a; Spiro 2006)

Blinding

Three studies attempted to blind the patient and the doctor without mentioning the outcome assessor (Arroll 2002a; El-Daher 1991; Pichichero 1987). In one study patients were told only that they would be given one of two sets of instructions about taking antibiotics for their colds. Participants read an information sheet and then completed a consent form. Thus, patients were blinded to what the other group would take (Arroll 2002a). Two studies used placebo tablets to blind patients (El-Daher 1991; Pichichero 1987). Seven studies attempted to blind some or all aspects of the study; that is, the patients, the doctor and the outcome assessor. For four studies (Chao 2008; Dowell 2001; Little 2005a; Spiro 2006), the outcomes assessor was blinded but not the patient or the care giver. For the remaining three studies no blinding was undertaken (Gerber 1990; Little 1997; Little 2001).

Incomplete outcome data

Only one trial (El-Daher 1991) had incomplete outcome data and did not adequately address it.

Selective reporting

Only one trial (Gerber 1990) reported collecting important information (in this case related to clinical outcomes) without fully reporting it.

Other potential sources of bias

No other sources of bias were identified.

Effects of interventions

For most outcomes meta-analyses were not possible: some studies did not describe their data in sufficient detail and others were too heterogeneous to safely allow meta-analysis. Therefore, few forest plots have more than one study. Table 1 summarises the statistical outcomes available for each study. However, for patient satisfaction, data were available and homogenous, so pooled results using a random-effects model are presented. For sore throat, two trials with minimal heterogeneity have been pooled for the outcome of fever severity on day three.

Table 1. Summary of outcomes
  1. AOM: acute otitis media
    CI: confidence interval
    IRR: incident rate ratio
    LRTI: lower respiratory tract infection
    MD: mean difference
    OR: odds ratio
    SMD: standardised mean difference

StudyOutcomeFavoursResult (with 95% CI)Notes
 Sore throat   
 Outcomes in this table are the result of a comparison between delayed and immediate antibiotics unless otherwise specified   
Pichichero 1987Fever severity on day 3 SMD 0.40 (0.05 to 0.75) 
 Malaise severity on day 3No differenceMD 0.20 (-0.11 to 0.51) 
 Pain severity on day 3No differenceMD 0.30 (-0.15 to 0.75) 
 ComplianceNo difference100% in both groups 
Gerber 1990Recurrence rateNo difference  
 ComplianceDelayed antibiotics88% in immediate group and 93% in the delayed group 
El Daher 1991VomitingImmediate antibioticsOR 25.00 (8.65 to 72.25) 
 Pain on day 3Immediate antibioticsOR 14.51 (7.14 to 29.50) 
 Malaise on day 3Immediate antibioticsOR 16.49 (5.68 to 47.83) 
 Fever severity on day 3Immediate antibioticsSMD 0.58 (0.31 to 0.84) 
 Compliance   
Little 1997VomitingNo differenceOR 1.00 (0.49 to 2.05) 
 DiarrhoeaNo differenceOR 1.23 (0.67 to 2.28) 
 RashNo differenceOR 0.93 (0.41 to 2.11) 
 Stomach acheNo differenceOR 0.82 (0.53 to 1.27) 
 Fever (> 37.0 ºC)Immediate antibiotics  
 Sore throatNo difference  
 CoughNo difference  
 MalaiseNo difference  
 Analgesic useNo difference  
 Time off workNo difference  
 AOM   
Little 2001DiarrhoeaDelayed antibioticsOR 0.45 (0.22 to 0.91) 
 RashNo differenceOR 1.21 (0.41 to 2.58) 
 Patients with pain on day 3No differenceOR 1.93 (0.96 to 3.88) 
 Patients with pain on day 7No differenceOR 6.55 (0.33 to 128.35) 
 Patients with malaise on day 3Immediate antibioticsOR 2.62 (1.44 to 4.76) 
 Malaise severity day 3Immediate antibioticsMD 0.43 (0.11 to 0.75) 
 Malaise severity on day 7No differenceMD 0.01 (-0.11 to 0.13) 
 Pain severity on day 3Immediate antibioticsMD 0.75 (0.26 to 1.24) 
 Pain severity on day 7No differenceMD 0.12 (-0.04 to 0.28) 
 Paracetamol consumptionImmediate antibioticsMD 0.59 (0.25 to 0.93) 
 Last day of cryingImmediate antibioticsMD 0.69 (0.31 to 1.07) 
Little 2001 (published in Little 2006)Episodes of earache in the 3 months since randomisationNo differenceOR 0.89 (0.48 to 1.65) 
 Episodes of earache over 1 yearNo differenceOR 1.03 (0.6 to 1.78) 
Spiro 2006Fever day 4 to 6No differenceOR 0.88 (0.53 to 1.47) 
 VomitingNo differenceOR 1.01 (0.47 to 2.16) 
 DiarrhoeaDelayed antibioticsOR 0.27 (0.13 to 0.58) 
Chao 2008Fever day 3No differenceOR 1.45 (0.50 to 4.24) 
 Pain day 3No differenceOR 0.64 (0.29 to 1.38) 
 Cough   
Dowell 2001Clinical outcomesNo difference  
Little 2005aAll clinical outcomesNo difference  
 Common cold   
Arroll 2002Patients with fever on day 3No differenceOR 0.75 (0.22 to 2.6) 
 Patients with fever on day 7No differenceOR 0.68 (0.15 to 3.17) 
 Patients with diarrhoeaNo differenceOR 0.79 (0.53 to 1.19) 
 Patients with pain on day 3No differenceOR 1.47 (0.58 to 3.77) 
 Patients with pain on day 7No differenceOR 0.31 (0.03 to 3.03) 
 Patients with cough on day 3No differenceOR 0.90 (0.37 to 2.18) 
 Patients with cough on day 7No differenceOR 0.72 (0.32 to 1.58) 
 Fever severity day 3No differenceMD -0.24 (-0.48 to 0.00) 
 Fever severity on day 7Delayed antibioticsMD -0.32 (-0.57 to -0.07)Mean temperature for both < 37 ºC
 Antibiotic use   
 Sore throat   
Little 1997Antibiotic use (none versus delayed)No antibiotics (least antibiotic use)OR 3.18 (1.85 to 5.46) 
 Antibiotic use (delayed versus immediate)Delayed antibiotics (less than immediate)OR 0.00 (0.00 to 0.02) 
 AOM   
Little 2001Antibiotic useDelayed antibioticsOR 0.05 (0.02 to 0.08) 
Spiro 2006Antibiotic useDelayed antibioticsOR 0.09 (0.05 to 0.17) 
Chao 2008Antibiotic useNo antibioticsOR 4.06 (2.01 to 8.19) 
 Cough   
Dowell 2001Antibiotic useDelayed antibioticsOR 0.00 (0.00 to 0.07) 
Little 2005Antibiotic use (none versus delayed)No differenceOR 1.30 (0.77 to 2.21) 
Little 2005Antibiotic use (delayed versus immediate)Delayed antibioticsOR 0.01 (0.00 to 0.02) 
 Common cold   
Arroll 2002Antibiotic useDelayed antibioticsOR 0.20 (0.09 to 0.44) 
 Patient satisfaction   
 Sore throat   
Little 1997Patient satisfaction (none versus delayed)No differenceOR 1.49 (0.70 to 3.19) 
 Patient satisfaction (delayed versus immediate)No differenceOR 0.61 (0.25 to 1.49) 
 AOM   
Little 2001Patient satisfaction (immediate versus delayed)Immediate antibioticsOR 0.32 (0.16 to 0.65) 
Chao 2008Patient satisfaction (delayed versus none)No differenceOR 2.00 (0.65 to 6.18) 
 Cough   
Dowell 2001Patient satisfactionImmediate antibioticsOR 0.19 (0.01 to 4.01) 
Little 2005Patient satisfaction (none versus delayed)No differenceOR 1.34 (0.84 to 2.14) 
Little 2005Patient satisfaction (delayed versus immediate)Immediate antibioticsOR 0.58 (0.34 to 0.97) 
 Common cold   
Arroll 2002Patient satisfactionNo differenceOR 1.47 (0.32 to 6.85) 
 Secondary outcomes   
 Sore throat   
Pichichero 1987Re-consultation rateNo differenceOR 0.83 (0.30 to 2.29) 
 AOM   
Spiro 2006Re-consultation rateNo differenceOR 1.21 (0.52 to 2.81) 
 LRTI   
Little 2005a (published in Moore 2009)Re-consultation in the year following the index consultation (excluding the first month after consultation)No differenceIRR 0.81 (0.51 to 1.28) 

Results are outlined under the headings of clinical outcomes, antibiotic use and patient satisfaction in order to reflect the important clinical considerations relevant to the strategy of prescribing delayed antibiotics. The strategy of delayed antibiotics is compared to the strategies of immediate antibiotics and no antibiotics, depending on the available data. For each illness category there is at least one RCT (for example, common cold) with a maximum of four (sore throat). Given the low numbers of trials for each illness category, conclusions for illness categories need to be treated with caution. The multiplicity of comparisons for the clinical outcomes stratified by illness, makes a type I error more likely. However, clinical outcomes are stratified by illness owing to known differences in the effect of antibiotics on different types of respiratory infections. Antibiotic use and patient satisfaction data have been presented without this stratification as they are less likely to be affected by illness type and to show more clearly the effect of prescribing strategies.

Clinical outcomes

See Table 1.

Sore throat

Four included studies examined sore throat (El-Daher 1991; Gerber 1990; Little 1997; Pichichero 1987).

Delayed antibiotics versus immediate antibiotics

Pain was reduced on day three in the immediate antibiotic group compared to delayed antibiotics in one study (Analysis 1.1). Pain was not significantly different between delayed and immediate antibiotic groups in three studies (Gerber 1990; Little 1997; Pichichero 1987).

Malaise was reduced on day three in the immediate antibiotic group compared to delayed antibiotics in one study (Analysis 2.1) and no difference was found in the other study measuring this outcome (Analysis 2.2).

Fever severity on day three was reduced with immediate antibiotics compared to delayed antibiotics in two studies (pooled results odds ratio (OR) 0.53; 95% confidence interval (CI) 0.31 to 0.74) (Analysis 3.1). The number of days with fever was reduced in the immediate antibiotic group of Little 1997 and there was no difference found in the fourth study (Gerber 1990).

Delayed antibiotics versus no antibiotics

One study examining sore throat compared the prescribing strategy of delayed antibiotics with no antibiotics (Little 1997). This study found no difference in any clinical outcome between these two prescribing strategies.

Complications

Data on complications of sore throat such as rheumatic fever, post-streptococcal glomerulonephritis and peri-tonsillar abscess were not reported in any of the four studies looking at sore throat for the three prescribing strategies of immediate, delayed and no antibiotics.

Acute otitis media (AOM)

Three included trials examined AOM (Chao 2008; Little 2001; Spiro 2006).

Delayed antibiotics versus immediate antibiotics

Pain and malaise were greater using delayed antibiotics compared to immediate antibiotics in one study measuring these outcomes on day three (Analysis 4.1). One study examined clinical outcomes on days four to six and found no difference (Analysis 5.1).

Other proxies for malaise outcomes reported by Little 2001 included last day of crying, which favoured the immediate antibiotic group by approximately 16 hours in children with AOM (0.69 days; 95% CI 0.31 to 0.07). In the same study, just over half a spoon of paracetamol a day less was used in the immediate antibiotic group (0.59; 95% CI 0.25 to 0.93). On day one there were no significant differences between immediate and delayed antibiotic groups in symptom outcome measures and by day seven there was no difference between immediate and delayed antibiotic groups (Little 2001).

Further analysis of earache from one trial (Little 2001) found the delayed prescribing strategy did not significantly increase risk of earache at three months (OR 0.89; 95% CI 0.48 to 1.65) or one year (OR 1.03; 95% CI 0.60 to 1.78) (Little 2006).

Delayed antibiotics versus no antibiotics

Only one study compared delayed antibiotics with no antibiotics with no significant difference for pain or fever on day three (Analysis 8.1; Analysis 9.1). This trial also advised participants in the no antibiotic arm to re-present in two to three days if symptoms did not resolve.

Complications

Data on complications of AOM such as mastoiditis, rheumatic fever and post-streptococcal glomerulonephritis were not reported in any of the three studies looking at AOM for the prescribing strategies of immediate and delayed antibiotics. However, Spiro 2006 and Chao 2008 noted that there were no serious adverse events for participants in the study.

Bronchitis (cough)

Two studies examined the prescribing strategies of immediate versus delayed antibiotics for the clinical presentation of cough (Dowell 2001; Little 2005a) and neither found any difference in clinical outcomes, including fever and cough.

Complications

Little 2005a also looked at delayed antibiotics versus no antibiotics and found no difference in clinical outcomes between the two prescribing strategies. One patient in the no antibiotic group (out of 273) of this study developed pneumonia and recovered with antibiotics in hospital.

Dowell 2001 did not report on complications in the immediate and delayed antibiotic groups.

Common cold

One study looked at immediate antibiotics versus delayed antibiotics (Arroll 2002a) and found no difference between the two prescribing strategies for the clinical outcomes of fever, cough, pain and malaise (Analysis 10.1; Analysis 11.4; Analysis 12.1).

Antibiotic use

See Table 1.

Delayed antibiotics

The three studies included in this systematic review published prior to 1992 examined the concern that immediate antibiotics for streptococcal pharyngitis might impair the body's immune response and predispose the patient to a relapse of pharyngitis. Compliance in both immediate and delayed antibiotic groups was close to 100%. Six of the included studies published after 1992 were conducted to evaluate the role of delayed antibiotics as a way of reducing antibiotic use for respiratory infections compared to immediate antibiotics. All six studies found that antibiotic use was significantly reduced in the delayed antibiotic group compared to the immediate antibiotic group. There were significant differences in the way antibiotics were delayed which may have resulted in the marked heterogeneity of this result. Of the seven studies published after 1991, four had the delayed script kept at reception to be picked up (Dowell 2001; Little 1997; Little 2001; Little 2005a) and in three, the script was issued to patients with instructions to delay (Arroll 2002a; Chao 2008; Spiro 2006). For the delayed arms of the four studies where the script was left at reception, antibiotics were used in 28% of cases (173/618) compared with antibiotics being used in 40% of cases (122/305) where antibiotics were issued to patients with instructions to delay.

Overall, the seven trials post 1992 providing a delayed antibiotic arm found 295 prescriptions filled out of 923 participants (32.0%).

Immediate antibiotics

Six trials published post 1992 provided immediate antibiotic arms examining this outcome resulting in 790 participants filling prescriptions out of 847 participants (93.3%) (Analysis 13.1).

No antibiotics

Three studies compared delayed antibiotics with no antibiotics. Little 1997 found that there was less antibiotic use with the no antibiotic strategy compared to delayed antibiotics. Little 2005a found no differences. Chao 2008 is the most recent and only study conducted comparing delayed antibiotics only with no antibiotics and also found that fewer antibiotics were prescribed in the no antibiotic group (Analysis 15.1).

Overall, 65 patients filled scripts out of 466 participants (13.9%).

Patient satisfaction

See Table 1.

Delayed antibiotics versus immediate antibiotics

(Analysis 16.1)

Patient satisfaction has been measured in five out of seven studies evaluating the prescribing strategy of delayed antibiotics since 1992 (Arroll 2002a; Dowell 2001; Little 1997; Little 2001; Little 2005a). Two of these studies indicated that study participants were more satisfied with the strategy of immediate antibiotics than delayed antibiotics (Little 2001; Little 2005a). There was no difference found in the other three studies (Arroll 2002a; Dowell 2001; Little 1997). The pooled result for this outcome with these five studies was an odds ratio (OR) of 0.52 (95% CI 0.35 to 0.76) favouring immediate antibiotics. Fixed- and random-effects analyses gave similar results. A breakdown of the trials by blinding gave two trials (Dowell 2001; Little 2005a) which blinded the outcome assessor and one blinded the patient and the doctor (Arroll 2002a) to give an odds ratio for all three studies of 0.62 (95% CI 0.38 to 1.01). The two completely unblinded trials (Little 1997; Little 2001) give an OR of 0.42 (95% CI 0.22 to 0.78). Overall 92% of the participants in the immediate antibiotics arms were satisfied versus 87% in the delayed arms.

Delayed antibiotics versus no antibiotics

(Analysis 17.1)

Three studies examined patient satisfaction comparing the prescribing strategies of delayed antibiotics and no antibiotics (Chao 2008; Little 1997; Little 2005a). While there was no difference in patient satisfaction for any of these studies, the pooled result for these three studies was an odds ratio of 1.44 (95% CI 0.99 to 2.10) showing no statistically significant difference. Fixed- and random-effects analyses gave similar results. A breakdown of the trials by blinding gave two trials (Chao 2008; Little 2005a) which blinded the outcome assessor to give an odds ratio for these two trials of 1.42 (95% CI 0.92 to 2.19). The one completely unblinded trial (Little 1997) gave an odds ratio of 1.49 (95% CI 0.70 to 3.19). In the delayed antibiotic arm 413 of the participants were satisfied or very satisfied out of 473 participants (87.3%) compared to 387 out of 465 participants in the no antibiotics group (83.2%).

Adverse effects of antibiotics

Adverse effects are considered under different clinical headings owing to differences in antibiotic prescribing recommendations for each condition. This is likely to have contributed to the heterogeneity evident in the forest plots for these outcomes preventing pooling of results. Adverse results are presented graphically for delayed versus immediate antibiotics (Analysis 17.1; Analysis 17.2; Analysis 17.3; Analysis 18.4) and delayed versus no antibiotics (Analysis 18.1; Analysis 18.2; Analysis 18.3; Analysis 18.4).

Sore throat
Delayed antibiotics versus immediate antibiotics

One study (Little 1997) found no difference for diarrhoea, vomiting, rash and stomach ache. El-Daher 1991 found more vomiting in the delayed group compared to the immediate antibiotics.

Delayed antibiotics versus no antibiotics

One study (Little 1997) found no difference for diarrhoea, vomiting, rash and stomach ache.

AOM
Delayed antibiotics versus immediate antibiotics

Little 2001 and Spiro 2006 found reduced diarrhoea in the delayed antibiotic group. Spiro 2006 did not find any difference between delayed and immediate antibiotics for vomiting and Little 2001 found no difference for the outcome of rash.

Delayed antibiotics versus no antibiotics

There were no adverse events in either group reported by Chao 2008.

Bronchitis (cough)
Delayed antibiotics versus immediate antibiotics

Little 2005a found no difference for adverse effects.

Delayed antibiotics versus no antibiotics

Little 2005a found no difference for adverse effects.

Common cold
Delayed antibiotics versus immediate antibiotics

There was no significant difference between the groups for diarrhoea, a potential adverse effect of antibiotics (Arroll 2002a).

Re-consultation rates

Re-consultation rates were the same between delayed and immediate antibiotic groups in two studies (Analysis 19.1). Subsequent consultation rates in the 12 months (excluding the first month) were also the same between delayed and immediate antibiotic groups in one study (Little 2001). Participants with sore throat in one study were more likely to intend to consult again if they received immediate antibiotics compared to delayed antibiotics (Little 1997).

Discussion

Summary of main results

Small differences were found between prescribing strategies for clinical outcomes with immediate antibiotics most likely to show benefit over delayed antibiotics in participants with sore throat and acute otitis media (AOM). All strategies appear to have similar safety with no advantage found for delayed antibiotics over no antibiotics for disease complications. Delay and no antibiotic strategies dramatically reduce the use of antibiotics for acute respiratory tract infections (ARTIs) compared to immediate antibiotics. The least antibiotic use was in the no antibiotic group followed by delay and then immediate. The number needed to treat to prevent one antibiotic prescription using the delay strategy is 1.6 compared to immediate antibiotics. The number needed to treat to prevent one antibiotic prescription using a no antibiotic strategy compared to delay is 5.6. Patient satisfaction was highest in the immediate antibiotic group with 92.2% being satisfied or very satisfied with the consultation. The delay and no groups had similar quite high satisfaction rates at 87.3% and 83.2%, respectively. These high satisfaction results may reflect patient involvement in studies where their treating physicians are more thorough in their explanations than usual (Hawthorne effect) (French 1950; Levitt 2011). Results for satisfaction may not be as high in routine general practice.

Overall completeness and applicability of evidence

Studies comparing delayed and immediate antibiotics have been performed for two different motives. The studies of Pichichero 1987, Gerber 1990 and El-Daher 1991 were concerned that immediate antibiotics for streptococcal pharyngitis might impair the body's immune response and predispose the patient to a relapse of pharyngitis. These studies are useful for determining the effect of delayed versus immediate antibiotics on the clinical course of suspected streptococcal pharyngitis. Six of the remaining studies were conducted to determine if the strategy of delayed antibiotics reduces the number of prescriptions filled for upper ARTIs (Arroll 2002a; Dowell 2001; Little 1997; Little 2001) while maintaining patient safety and satisfaction. The most recent study may indicate evolution in prescribing habits as it was the first to drop the immediate antibiotic arm (Chao 2008).

Useful data were collected for many symptom outcomes in all studies but were not always reported in a way that could be analysed. This problem was partially overcome by obtaining raw data from some trial authors. The seven studies conducted after 1992 all reported useful data on antibiotic use and six on patient satisfaction.

There are only three trials comparing delayed antibiotics with no antibiotics.

Quality of the evidence

All but one trial (El-Daher 1991) were adequately randomised and accounted for incomplete data. El-Daher 1991 did find large differences for clinical outcomes for sore throat in favour of immediate antibiotics compared to delayed antibiotics.

This intervention does not lend itself to blinding. However, three trials attempted to blind patients and doctors (Arroll 2002a; El-Daher 1991; Pichichero 1987). For four studies (Chao 2008; Dowell 2001; Little 2005a; Spiro 2006), the outcomes assessor was blinded but not the patient nor the care giver.

Otherwise, studies were well reported and appeared to be high quality.

Potential biases in the review process

Heterogeneity of randomised controlled trials (RCTs) is one limitation of this review. Heterogeneity may have resulted from variable clinical presentations, differences in delay method, differences in antibiotic use and quality of included studies. Potential for type I error is another limitation of this review given the large number of reported outcome results. For example, multiple outcome measures are reported for the clinical outcomes comparing delayed and immediate antibiotic groups.

Agreements and disagreements with other studies or reviews

Findings for certain clinical outcomes in our review might have been anticipated. Systematic reviews on antibiotics for sore throat and AOM found that their time of greatest benefit for symptoms is apparent at days three or four after treatment has started (Spinks 2011; Venekamp 2013). Thus delaying antibiotics by 48 hours or more would overshoot this zenith. Nor is it surprising that we found more adverse reactions to antibiotics from immediate antibiotics in line with known adverse events from comparison RCTs with no antibiotics.

The greatest difference in clinical outcomes was found in the only trial of delayed antibiotics conducted in a low socio-economic environment, favouring immediate antibiotics over delay (El-Daher 1991). This trial was also the least methodologically sound but it highlighted that concerns expressed about delayed antibiotics for children, the elderly (Datta 2008) and those with language or cultural difficulties (Johnson 2007) may also need to be extended to low socio-economic populations.

A parallel RCT of patients with acute infective conjunctivitis similarly reported shortest symptom duration with immediate, followed by delayed and then no antibiotics (the last resulting in least antibiotic use). There was no difference between the groups for patient satisfaction (Everitt 2006).

A recent randomised controlled trial published in 2010 (Worrall 2010) comparing delayed prescriptions dated either the day of the office visit or two days later, but not comparing with either immediate or no antibiotics, demonstrated no significant difference between the two groups in terms of antibiotic use.

RCTs comparing delayed with no antibiotics (concluding that they were both equally acceptable alternatives to immediate antibiotics as a means of reducing antibiotic prescriptions) (Little 2001; Little 2005a) led to recommending delayed instead of no antibiotics to address concerns about risks of complications (Little 2005b). Doctors worried about the risk of serious infective complications consequent to adopting a no antibiotic rather than delayed strategy might take comfort from a UK observational study showing that reduced prescribing resulted in no increase in admissions to hospital for peri-tonsillar abscess or rheumatic fever (Sharland 2005), although mastoiditis might be a risk at the rate of 2500 children needing to be treated with antibiotics to prevent one case (Van Zuijlen 2001). Thirty-five per cent of parents in the AOM trials (Chao 2008; Little 2001; Spiro 2006) used their delayed script suggesting that the number of delayed scripts required to prevent one case of mastoiditis would be significantly higher than 2500. Doctors often find it difficult to identify patients at risk of serious complications from respiratory infections (Kumar 2003). Patients probably perform even less well, despite their self confidence in making this decision if given a delayed antibiotic prescription. This concern is supported by empirical data: respiratory disease severity does not correlate with patients' immediate preference for an antibiotic prescription (Macfarlane 1997). This review did not find any significant difference for complication rates between prescribing strategies.

There is little controversy within published guidelines that immediate antibiotics are recommended for patients who appear to be seriously unwell, fit multiple criteria indicating bacterial tonsillitis, are under six months of age with AOM, have bilateral AOM or have AOM with otorrhoea (Tan 2008). American guidelines also recommend immediate antibiotics for children under two with definite AOM (OMTG 2004). It seems then that for the majority of respiratory infections that do not meet these criteria, clinicians have the option of delayed or no antibiotics. It seems clear that no antibiotics will result in least antibiotic use and therefore less antibiotic resistance. Concerns about patient and doctor satisfaction with no antibiotics appear to be driving the use of a delayed strategy. Some doctors use the delay strategy to reduce antibiotic use, empower patients and save the patient time and money without jeopardising the doctor-patient relationship (Arroll 2002b). A qualitative study conducted in 2002 (Arroll 2002b) found that while some patients appreciated the option of controlling the decision as to whether and when to take antibiotics, others expected "the physician to decide". Concern was expressed by one physician that patients might view delayed prescribing as physician incompetence, substantiated by comments from some patients. Shared decision-making (Butler 2001; Legare 2007) and education campaigns for doctors (Sung 2006) have been proposed as ways of helping doctors and patients avoid unnecessary antibiotic use. One suggestion is that delayed antibiotics may in time become redundant as doctors and their patients gain more reassurance in the safety of not using antibiotics (Arroll 2003b).

Authors' conclusions

Implications for practice

A strategy of immediate antibiotics is more likely to confer the modest benefits of antibiotics on clinical outcomes such as symptoms for acute otitis media and sore throat than delayed antibiotics. There were no differences in complication rates between immediate and delayed antibiotics nor between delayed and no antibiotics. Immediate antibiotics had slightly higher levels of patient satisfaction than delayed antibiotics, which reached statistical significance but is of marginal clinical significance (92% versus 87%). Patient satisfaction was similarly high in the delayed and no antibiotic groups with a trend towards delayed antibiotics that was neither statistically nor clinically significant (87% versus 83%). Delayed antibiotic prescribing strategies achieved lower rates of antibiotic use compared to immediate antibiotics (32% versus 93%). No antibiotics achieved lower rates of antibiotic use compared to delayed antibiotics (13% versus 32%).

Delayed antibiotics for respiratory infections is a strategy which reduces antibiotic use compared to immediate antibiotics but has not been shown by this review to be different to no antibiotics in terms of symptom control and disease complications. In patients with respiratory infections where clinicians feel it is safe not to prescribe antibiotics immediately, no antibiotics with advice to return if symptoms do not resolve is likely to result in the least antibiotic use, while maintaining similar patient satisfaction and clinical outcomes to delayed antibiotics.

Implications for research

Further research into antibiotic prescribing strategies for respiratory infections may best be focused on identifying patient groups at high risk of disease complications, enhancing doctors' communication with patients to maintain satisfaction and ways of reducing doctors' anxieties about not prescribing antibiotics for respiratory infections. Future randomised controlled trials of delaying antibiotics as an intervention should fully report symptoms, patient satisfaction, doctor satisfaction and disease complications as well as changes in prescription rates. They should also include a no antibiotic arm.

Acknowledgements

The review authors wish to thank the following people for commenting on the 2010 updated review: Alison Thomas, Bill Hueston, Max Bulsara and Tom Fahey. We also thank the following people for commenting on this 2013 updated review: Amanda Roberts, Sandra Arnold, Bill Hueston, Sree Nair and Tom Fahey.

Data and analyses

Download statistical data

Comparison 1. Sore throat - pain; delayed versus immediate antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Pain on day 31 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
2 Pain severity on day 31 Mean Difference (IV, Fixed, 95% CI)Totals not selected
Analysis 1.1.

Comparison 1 Sore throat - pain; delayed versus immediate antibiotics, Outcome 1 Pain on day 3.

Analysis 1.2.

Comparison 1 Sore throat - pain; delayed versus immediate antibiotics, Outcome 2 Pain severity on day 3.

Comparison 2. Sore throat - malaise; delayed versus immediate antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Malaise on day 31 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
2 Malaise severity1 Mean Difference (IV, Fixed, 95% CI)Totals not selected
Analysis 2.1.

Comparison 2 Sore throat - malaise; delayed versus immediate antibiotics, Outcome 1 Malaise on day 3.

Analysis 2.2.

Comparison 2 Sore throat - malaise; delayed versus immediate antibiotics, Outcome 2 Malaise severity.

Comparison 3. Sore throat - fever; delayed versus immediate antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Fever severity on day 32343Std. Mean Difference (IV, Fixed, 95% CI)0.53 [0.31, 0.74]
2 Fever severity on day 12343Std. Mean Difference (IV, Fixed, 95% CI)-0.07 [-0.29, 0.14]
Analysis 3.1.

Comparison 3 Sore throat - fever; delayed versus immediate antibiotics, Outcome 1 Fever severity on day 3.

Analysis 3.2.

Comparison 3 Sore throat - fever; delayed versus immediate antibiotics, Outcome 2 Fever severity on day 1.

Comparison 4. AOM - pain; delayed versus immediate antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Pain on day 31 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
2 Pain on days 4 to 61 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
3 Pain on day 71 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
4 Pain severity on day 31 Mean Difference (IV, Fixed, 95% CI)Totals not selected
5 Pain severity on day 71 Mean Difference (IV, Fixed, 95% CI)Totals not selected
Analysis 4.1.

Comparison 4 AOM - pain; delayed versus immediate antibiotics, Outcome 1 Pain on day 3.

Analysis 4.2.

Comparison 4 AOM - pain; delayed versus immediate antibiotics, Outcome 2 Pain on days 4 to 6.

Analysis 4.3.

Comparison 4 AOM - pain; delayed versus immediate antibiotics, Outcome 3 Pain on day 7.

Analysis 4.4.

Comparison 4 AOM - pain; delayed versus immediate antibiotics, Outcome 4 Pain severity on day 3.

Analysis 4.5.

Comparison 4 AOM - pain; delayed versus immediate antibiotics, Outcome 5 Pain severity on day 7.

Comparison 5. AOM - malaise; delayed versus immediate antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Malaise on day 31 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
2 Malaise severity on day 31 Mean Difference (IV, Fixed, 95% CI)Totals not selected
3 Malaise severity on day 71 Mean Difference (IV, Fixed, 95% CI)Totals not selected
Analysis 5.1.

Comparison 5 AOM - malaise; delayed versus immediate antibiotics, Outcome 1 Malaise on day 3.

Analysis 5.2.

Comparison 5 AOM - malaise; delayed versus immediate antibiotics, Outcome 2 Malaise severity on day 3.

Analysis 5.3.

Comparison 5 AOM - malaise; delayed versus immediate antibiotics, Outcome 3 Malaise severity on day 7.

Comparison 6. Supplementary medicine consumption; delayed versus immediate antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Spoons of paracetamol/day1 Mean Difference (IV, Fixed, 95% CI)Totals not selected
2 Use of paracetamol and ibuprofen1 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
Analysis 6.1.

Comparison 6 Supplementary medicine consumption; delayed versus immediate antibiotics, Outcome 1 Spoons of paracetamol/day.

Analysis 6.2.

Comparison 6 Supplementary medicine consumption; delayed versus immediate antibiotics, Outcome 2 Use of paracetamol and ibuprofen.

Comparison 7. AOM - fever; delayed versus immediate antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Fever Days 4 to 61 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
Analysis 7.1.

Comparison 7 AOM - fever; delayed versus immediate antibiotics, Outcome 1 Fever Days 4 to 6.

Comparison 8. AOM - pain; delayed versus no antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Otitis media pain on Day 3 delayed versus none1 Odds Ratio (M-H, Random, 95% CI)Totals not selected
Analysis 8.1.

Comparison 8 AOM - pain; delayed versus no antibiotics, Outcome 1 Otitis media pain on Day 3 delayed versus none.

Comparison 9. AOM - fever; delayed versus no antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Otitis media number of patients with fever on day 3 delayed versus none1 Odds Ratio (M-H, Random, 95% CI)Totals not selected
Analysis 9.1.

Comparison 9 AOM - fever; delayed versus no antibiotics, Outcome 1 Otitis media number of patients with fever on day 3 delayed versus none.

Comparison 10. Common cold - pain; delayed versus immediate antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Pain on day 31 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
2 Pain on day 71 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
Analysis 10.1.

Comparison 10 Common cold - pain; delayed versus immediate antibiotics, Outcome 1 Pain on day 3.

Analysis 10.2.

Comparison 10 Common cold - pain; delayed versus immediate antibiotics, Outcome 2 Pain on day 7.

Comparison 11. Common cold - fever; delayed versus immediate antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Fever on day 31 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
2 Fever on day 71 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
3 Fever severity on day 11 Mean Difference (IV, Fixed, 95% CI)Totals not selected
4 Fever severity on day 31 Mean Difference (IV, Fixed, 95% CI)Totals not selected
5 Fever severity on day 71 Mean Difference (IV, Fixed, 95% CI)Totals not selected
Analysis 11.1.

Comparison 11 Common cold - fever; delayed versus immediate antibiotics, Outcome 1 Fever on day 3.

Analysis 11.2.

Comparison 11 Common cold - fever; delayed versus immediate antibiotics, Outcome 2 Fever on day 7.

Analysis 11.3.

Comparison 11 Common cold - fever; delayed versus immediate antibiotics, Outcome 3 Fever severity on day 1.

Analysis 11.4.

Comparison 11 Common cold - fever; delayed versus immediate antibiotics, Outcome 4 Fever severity on day 3.

Analysis 11.5.

Comparison 11 Common cold - fever; delayed versus immediate antibiotics, Outcome 5 Fever severity on day 7.

Comparison 12. Common cold - cough; delayed versus immediate antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Cough on day 31 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
2 Cough on day 71 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
Analysis 12.1.

Comparison 12 Common cold - cough; delayed versus immediate antibiotics, Outcome 1 Cough on day 3.

Analysis 12.2.

Comparison 12 Common cold - cough; delayed versus immediate antibiotics, Outcome 2 Cough on day 7.

Comparison 13. Antibiotic use: delayed versus immediate antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Antibiotic use: delayed versus immediate antibiotics6 Odds Ratio (M-H, Random, 95% CI)Totals not selected
1.1 Antibiotic use: delayed (prescription at time of visit) versus immediate antibiotics2 Odds Ratio (M-H, Random, 95% CI)0.0 [0.0, 0.0]
1.2 Antibiotic use: delayed (return for prescription) versus immediate antibiotics4 Odds Ratio (M-H, Random, 95% CI)0.0 [0.0, 0.0]
Analysis 13.1.

Comparison 13 Antibiotic use: delayed versus immediate antibiotics, Outcome 1 Antibiotic use: delayed versus immediate antibiotics.

Comparison 14. Antibiotic use: delayed versus no antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Antibiotic use: delayed versus no antibiotics3 Odds Ratio (M-H, Random, 95% CI)Totals not selected
1.1 Antibiotic use: delayed (prescription at time of visit) versus no antibiotics1 Odds Ratio (M-H, Random, 95% CI)0.0 [0.0, 0.0]
1.2 Antibiotic use: delayed (return for prescription) versus no antibiotics2 Odds Ratio (M-H, Random, 95% CI)0.0 [0.0, 0.0]
Analysis 14.1.

Comparison 14 Antibiotic use: delayed versus no antibiotics, Outcome 1 Antibiotic use: delayed versus no antibiotics.

Comparison 15. Patient satisfaction: delayed versus immediate antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Patient satisfaction: delayed versus immediate antibiotics51334Odds Ratio (M-H, Random, 95% CI)0.52 [0.35, 0.76]
1.1 Patient satisfaction: delayed (prescription at time of consult) versus immediate antibiotics1129Odds Ratio (M-H, Random, 95% CI)1.47 [0.32, 6.85]
1.2 Patient satisfaction: delayed (return for prescription) versus immediate antibiotics41205Odds Ratio (M-H, Random, 95% CI)0.48 [0.33, 0.71]
Analysis 15.1.

Comparison 15 Patient satisfaction: delayed versus immediate antibiotics, Outcome 1 Patient satisfaction: delayed versus immediate antibiotics.

Comparison 16. Patient satisfaction: delayed versus no antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Patient satisfaction: delayed versus no antibiotics3938Odds Ratio (M-H, Random, 95% CI)1.44 [0.99, 2.10]
1.1 Patient satisfaction: delayed (prescription provided at visit) versus no antibiotics1206Odds Ratio (M-H, Random, 95% CI)2.00 [0.65, 6.18]
1.2 Patient satisfaction: delayed (return for prescription) versus no antibiotics2732Odds Ratio (M-H, Random, 95% CI)1.38 [0.93, 2.06]
Analysis 16.1.

Comparison 16 Patient satisfaction: delayed versus no antibiotics, Outcome 1 Patient satisfaction: delayed versus no antibiotics.

Comparison 17. Adverse events: delayed versus immediate antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Vomiting3 Odds Ratio (M-H, Random, 95% CI)Totals not selected
2 Diarrhoea4 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
3 Rash2 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
4 Stomach ache1 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
Analysis 17.1.

Comparison 17 Adverse events: delayed versus immediate antibiotics, Outcome 1 Vomiting.

Analysis 17.2.

Comparison 17 Adverse events: delayed versus immediate antibiotics, Outcome 2 Diarrhoea.

Analysis 17.3.

Comparison 17 Adverse events: delayed versus immediate antibiotics, Outcome 3 Rash.

Analysis 17.4.

Comparison 17 Adverse events: delayed versus immediate antibiotics, Outcome 4 Stomach ache.

Comparison 18. Adverse events: delayed versus no antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Vomiting1 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
2 Diarrhoea2 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
3 Rash1 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
4 Stomach ache1 Odds Ratio (M-H, Fixed, 95% CI)Totals not selected
Analysis 18.1.

Comparison 18 Adverse events: delayed versus no antibiotics, Outcome 1 Vomiting.

Analysis 18.2.

Comparison 18 Adverse events: delayed versus no antibiotics, Outcome 2 Diarrhoea.

Analysis 18.3.

Comparison 18 Adverse events: delayed versus no antibiotics, Outcome 3 Rash.

Analysis 18.4.

Comparison 18 Adverse events: delayed versus no antibiotics, Outcome 4 Stomach ache.

Comparison 19. Re-consultation rate; delayed versus immediate antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Re-consultation rate2379Odds Ratio (M-H, Fixed, 95% CI)1.04 [0.55, 1.98]
Analysis 19.1.

Comparison 19 Re-consultation rate; delayed versus immediate antibiotics, Outcome 1 Re-consultation rate.

Comparison 20. Subsequent consultation rates in the 12 months following the index consultation (excluding first month following consultation); delayed versus immediate antibiotics
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Re-consultation in the 12 months following the index consultation (excluding the first month following the index consultation)1 Rate Ratio (Fixed, 95% CI)Totals not selected
Analysis 20.1.

Comparison 20 Subsequent consultation rates in the 12 months following the index consultation (excluding first month following consultation); delayed versus immediate antibiotics, Outcome 1 Re-consultation in the 12 months following the index consultation (excluding the first month following the index consultation).

Appendices

Appendix 1. Ovid EMBASE search strategy

1     exp Respiratory Tract Infection/ (172448)
2     exp Upper Respiratory Tract Infection/ (22007)
3     (upper respiratory tract infection$ or urti).mp. (14226)
4     exp Otitis Media/ (15047)
5     otitis media.mp. (16846)
6     exp Pharyngitis/ (13679)
7     pharyngitis.mp. (9017)
8     exp Tonsillitis/ (5085)
9     tonsillitis.mp. (4596)
10     exp Common Cold/ (4421)
11     common cold.mp. (5401)
12     exp Bronchitis/ (24102)
13     bronchitis.mp. (17391)
14     exp Sinusitis/ (19381)
15     sinusitis.mp. (18397)
16     sore throat$.mp. (8421)
17     or/1-16 (234854)
18     exp Antibiotic Agent/ (544500)
19     antibiotic$.mp. (328859)
20     or/18-19 (628363)
21     (delay$ adj15 prescri$).mp. (841)
22     17 and 20 and 21 (102)

Appendix 2. CENTRAL search strategy

#1        MeSH descriptor: [Respiratory Tract Infections] explode all trees       9072
#2        (upper next respiratory next tract infection*) or URTI             1061
#3        MeSH descriptor: [Otitis Media] explode all trees        1009
#4        otitis next media          1926
#5        MeSH descriptor: [Pharyngitis] explode all trees          841
#6        pharyngitis       1237
#7        MeSH descriptor: [Tonsillitis] explode all trees            322
#8        tonsillitis          651
#9        MeSH descriptor: [Common Cold] explode all trees    375
#10      common next cold*     729
#11      MeSH descriptor: [Bronchitis] explode all trees           1416
#12      bronchitis         2754
#13      MeSH descriptor: [Sinusitis] explode all trees  626
#14      sinusitis            1362
#15      sore next throat*          826
#16      #1 or #2 or #3 or #4 or #5 or #6 or #7 or #8 or #9 or #10 or #11 or #12 or #13 or #14 or #15           14213
#17      MeSH descriptor: [Anti-Bacterial Agents] explode all trees     8199
#18      antibiotic*        15634
#19      #17 or #18       19843
#20      delay* near/15 prescri*           87
#21      #16 and #19 and #20   28

Appendix 3. EBSCO CINAHL search strategy

S15 S10 and S13 and S14
S14 TI delay* N15 prescri* or AB delay* N15 prescri*
S13 S11 or S12
S12 TI antibiotic* or AB antibiotic*
S11 (MH "Antibiotics+")
S10 S1 or S2 or S3 or S4 or S5 or S6 or S7 or S8 or S9
S9 TI ( otitis media or pharyngitis or tonsillitis or common cold* or bronchitis or sinusitis or sore throat* ) or AB ( otitis media or pharyngitis or tonsillitis or common cold* or bronchitis or sinusitis or sore throat* )
S8 (MH "Sinusitis+")
S7 (MH "Bronchitis+")
S6 (MH "Common Cold")
S5 (MH "Tonsillitis+")
S4 (MH "Pharyngitis")
S3 (MH "Otitis Media+")
S2 TI ( upper respiratory tract infection* or urti ) or AB ( upper respiratory tract infection* or urti )
S1 (MH "Respiratory Tract Infections+")

Appendix 4. ISI Current Contents Connect search strategy

#14 #13 AND #12 AND #9
#13 TS=antibiotic*
#12 #11 OR #10
#11 TS=immediate*
#10 TS=delay*
#9 #8 OR #7 OR #6 OR #5 OR #4 OR #3 OR #2 OR #1
#8 TS=sore throat
#7 TS=sinusitis
#6 TS=bronchitis
#5 TS=common cold*
#4 TS=tonsillitis
#3 TS=pharyngitis
#2 TS=otitis media
#1 TS=respiratory tract infection*

Feedback

Feedback: Analysis 15.01 Comparison 15 may have some errors, 9 June 2008

Summary

Feedback: Analysis 15.01 Comparison 15 Patient satisfaction immediate versus delayed antibiotics, Outcome 01 Patient satisfaction: immediate versus delayed antibiotics may have some errors.

We think that the extracted data has been entered under the wrong headings, i.e. for Little 1997, it reports that 165/177 were satisfied with delayed antibiotics but the RevMan forest plot has 165/177 under the immediate antibiotics.

Data extracted from one article (Dowell 2001) may have been entered incorrectly, i.e. the percentage has been entered into RevMan directly rather than as the actual number. In other words, for Dowell 2001, the paper reports 100% (73% very satisfied and 27% moderately satisfied), whereas the forest plot has reported the 73% as 73/75. This is a double query ? see below for issue of inconsistent grouping of satisfaction scores.

Suggest that the data extracted for Dowell 2001 should be consistent with the logic used for Arroll 2002 in their results for the same outcome.
We think that possibly the forest plot analysis should be conducted with the figures below. We have looked at all the original papers.

Arroll 2002a   
 64/67* Delayed Antibiotics
 58/62* Immediate Antibiotics
Dowell 2001
 71/73# Delayed Antibiotics
 75/75# Immediate Antibiotics
Little 1997    
 165/177 Delayed Antibiotics   
 202/211 Immediate Antibiotics
Little 2001
 115/150 Delayed Antibiotics
 123/135 Immediate Antibiotics
Little 2005a   
 147/190 Delayed Antibiotics
 166/194 Immediate Antibiotics

Arroll et al noted that  for these results, groups responding 1 and 2 have been combined and groups 3 and 4 have been combined where: 1= very satisfied; 2= moderately satisfied; 3 = slightly satisfied; 4 = not at all satisfied.

Using similar logic as Arroll et al, results for groups responding ?very satisfied? and ?moderately satisfied? have been combined, as have ?not very satisfied? and ?not at all satisfied? to get the figures in the table above for Dowell 2001. (Note: in the review table, the figures were extracted directly from the ?very satisfied? column only, where they were presented as a percentage without then recalculating them as a whole figure).

We don't think these possible errors effect the overall conclusions made by the authors in the review.

Submitter agrees with default conflict of interest statement:
I certify that I have no affiliations with or involvement in any organisation or entity with a financial interest in the subject matter of my feedback.

Reply

We thank those who have given feedback on this review. We greatly appreciate the work you have done to uncover these errors and the opportunity you have given us to correct them. We agree with all the feedback you have submitted and have made corrections to analysis 15 comparison 15.1, analysis 16 comparison 16.1, analysis 13 comparison 13.1 (antibiotic use delayed versus immediate), analysis 14 comparison 14.1 (antibiotic use delayed versus none) and analysis 3 comparison 3.1 (fever severity on day 3). We have also added an analysis 17: adverse events delayed versus no antibiotics.

Theses changes have not fundamentally changed the results of the review. However the text and outcome tables have been amended to reflect changes made.

Geoff Spurling, Chris Del Mar, Liz Dooley
Feedback reply added 25 June 2008

Contributors

Dianne Lowe, Rebecca Ryan
Feedback comment added 16 June 2008

It would be interesting to explore the comparative evidence base for the most effective method of delayed prescription, 18 March 2009

Summary

Feedback: It would be interesting to explore the comparative evidence base for the most effective method of "delayed prescription" e.g.:
1. Script dated today given to patient
2. Script dated 2-3 days from now - given to patient
3. Script held at practice

Submitter agrees with default conflict of interest statement:
I certify that I have no affiliations with or involvement in any organization or entity with a financial interest in the subject matter of my feedback.

Reply

We thank you for your feedback on this review. We agree that it would be interesting to explore the comparative evidence base for the most effective method of delayed prescription. Subgroups highlighting the method of delayed prescribing have been added for the outcomes antibiotic use and patient satisfaction. Unfortunately, there was great heterogeneity in the methods of delayed prescribing that makes combining studies difficult. Methods of delayed prescribing ranged from issuing a prescription at the time of the initial consults with instruction to delay, to holding the delayed prescription at reception to be picked up if symptoms hadn't improved after a specified period of time. The recommended periods of delay ranged from three to fourteen days.

The three studies included in this systematic review published prior to 1992 examined the concern that immediate antibiotics for streptococcal pharyngitis might impair the body's immune response and predispose the patient to a relapse of pharyngitis. Six of the included studies published after 1992 were conducted to evaluate the role of delayed antibiotics as a way of reducing antibiotic use for respiratory infections compared to immediate antibiotics. While all six studies found that antibiotic use was significantly reduced in the delayed antibiotic group compared to the immediate antibiotic group. There were significant differences in the way antibiotics were delayed which may have contributed to the marked heterogeneity of this result. Of the seven studies published after 1991, four had the delayed script kept at reception to be picked up (Dowell 2001; Little 1997; Little 2001; Little 2005a) and in three, the script was issued to patients with instructions to delay (Arroll 2002a; Chao 2008; Spiro 2006). For the delayed arms of the four studies where the script was left at reception, antibiotics were used in 28% of cases (173/618) compared with antibiotics being used in 40% of cases (122/305) where antibiotics were issued to patients with instructions to delay.

None of the included studies specifically addressed whether or not prescriptions had been post-dated. However, a recent randomised controlled trial published in 2010, (Worrall 2010) comparing delayed prescriptions dated either the day of the office visit or 2 days later, but not comparing with either immediate or no antibiotics, demonstrated no significant difference between the two groups in terms of antibiotic use.

Geoff Spurling, Chris Del Mar, Liz Dooley, Rebecca Farley
Feedback reply added 25 March 2012

Contributors

Jas Janjuha, Occupation Pharmacist

What's new

Last assessed as up-to-date: 28 February 2013.

DateEventDescription
28 February 2013New search has been performedThe searches have been updated. Two new papers (Little 2006; Moore 2009) were included. They reported longer-term outcomes of two previously included studies (Little 2001; Little 2005a) including impact of delayed antibiotic prescribing on earache recurrence and subsequent consultation rates in the 12 months following the initial consultation. Three new trials were excluded (Fischer 2009; Newson 2009; Vouloumanou 2009). Our conclusions remain unchanged.
28 February 2013New citation required but conclusions have not changedA new author joined the team to update the review.

History

Protocol first published: Issue 4, 2003
Review first published: Issue 4, 2004

DateEventDescription
5 August 2010AmendedContact details updated.
27 March 2009New search has been performedSearches conducted. This 2009 update contains one new study (Chao 2008) and Feedback on a comment submitted via The Cochrane Library.
16 June 2008Feedback has been incorporatedFeedback comment added.
16 June 2008AmendedConverted to new review format.
21 January 2007New search has been performedSearches conducted.
9 January 2004New search has been performedSearches conducted.

Contributions of authors

Chris Del Mar (CDM) conceived the review.
Geoff Spurling (GS) and CDM designed the review.
Ruth Foxlee (RF) and GS performed the literature searches.
RF, Liz Dooley (LD) and CDM appraised articles and extracted data from these articles.
GS and Rebecca Farley (RFa) entered data into RevMan with contributions from LD, RF and CDM.
GS secured funding for the review with the assistance of CDM.

Declarations of interest

No known conflict of interest.

Sources of support

Internal sources

  • Bond University, Gold Coast, Australia.

  • The Discipline of General Practice at the University of Queensland, Australia.

    For providing the infrastructure which allowed the first publication of this review to be conducted.

External sources

  • General Practice Education and Training, Australia.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Arroll 2002a

MethodsRandomised controlled trial
ParticipantsAdults and children with the common cold
InterventionsDelayed antibiotics (patients given script and instructed to fill within 72 hours) versus immediate antibiotics
OutcomesFever, duration of fever, cough, duration of cough, pain, absence from school/work, diarrhoea, antibiotic use
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandomisation using Excel
Allocation concealment (selection bias)Low riskYes - opaque envelopes
Blinding (performance bias and detection bias)
All outcomes
Low riskPatient and care provider were blinded but unsure regarding outcome assessor
Incomplete outcome data (attrition bias)
All outcomes
Low riskITT analysis used
Selective reporting (reporting bias)Low riskPre-specified outcomes were reported
Other biasLow riskNo information

Chao 2008

MethodsRandomised controlled trial
ParticipantsChildren with AOM
InterventionsNo antibiotics (observation) versus delayed antibiotics (observation plus prescription) - patients given script and instructed to fill the script if required
OutcomesFever, pain, antibiotic use, patient satisfaction, adverse events
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandom number table
Allocation concealment (selection bias)High riskNot described
Blinding (performance bias and detection bias)
All outcomes
High riskOutcome assessor blinded. Patient and care provider not blinded
Incomplete outcome data (attrition bias)
All outcomes
Low riskMissing data were described and ITT analysis applied
Selective reporting (reporting bias)Low riskPre-specified outcomes were reported
Other biasLow riskNo Information

Dowell 2001

MethodsRandomised controlled trial
ParticipantsAdults and children with cough
InterventionsDelayed antibiotics (script left at reception and patients instructed to pick up the script after 1 week of delay) versus immediate antibiotics (antibiotic of GP's choice)
OutcomesDuration of cough, antibiotic use
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandom number table
Allocation concealment (selection bias)Unclear riskNumbered envelopes (opacity not mentioned)
Blinding (performance bias and detection bias)
All outcomes
High riskOutcome assessor blinded but not patient nor care provider
Incomplete outcome data (attrition bias)
All outcomes
Low riskDrop-out numbers were described and intention-to-treat analysis used
Selective reporting (reporting bias)Low riskPre-specified clinical outcomes were not published but authors provided this information
Other biasLow riskNo Information

El-Daher 1991

MethodsRandomised controlled trial
ParticipantsChildren with sore throat (suspected group A beta haemolytic Streptococcus)
InterventionsDelayed antibiotics (48 hours) versus immediate antibiotics (penicillin V 50,000 IU/kg/day)
OutcomesPain, malaise, vomiting, temperature
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskMethod not described
Allocation concealment (selection bias)High riskNot described
Blinding (performance bias and detection bias)
All outcomes
Low riskBlinding of patient and care provider but unsure about outcome assessor
Incomplete outcome data (attrition bias)
All outcomes
High riskDrop-outs not described
Selective reporting (reporting bias)Low riskPre-specified outcomes reported
Other biasLow riskNo Information

Gerber 1990

MethodsRandomised controlled trial
ParticipantsAdults and children with sore throat (suspected group A beta haemolytic Streptococcus)
InterventionsDelayed antibiotics (48 hours) versus immediate antibiotics (penicillin V, 250 mg qds for 10 days)
OutcomesMalaise
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandom number table
Allocation concealment (selection bias)High riskNo Information
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding
Incomplete outcome data (attrition bias)
All outcomes
Low riskDrop outs described
Selective reporting (reporting bias)Unclear riskClinical outcomes reported as one outcome
Other biasLow riskNo Information

Little 1997

MethodsOpen randomised controlled trial
ParticipantsAdults and children with sore throat
InterventionsDelayed antibiotics (script left at reception and patients instructed to pick it up 72 hours later if required) versus immediate antibiotics versus no antibiotics (penicillin V 250 mg qds in both groups)
OutcomesFever, cough, duration of pain, duration of malaise, absence from school, diarrhoea and rash
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskRandomisation method not described
Allocation concealment (selection bias)Unclear risk"Sealed envelopes"; no mention of opacity
Blinding (performance bias and detection bias)
All outcomes
High riskNo Information
Incomplete outcome data (attrition bias)
All outcomes
Low riskITT analysis undertaken
Selective reporting (reporting bias)Low riskNo Information
Other biasLow riskNo Information

Little 2001

MethodsPragmatic randomised controlled trial
ParticipantsChildren aged 6 months to 10 years with AOM
InterventionsDelayed antibiotics (72 hours, parents were advised to use antibiotics if their child had significant otalgia or fever after 72 hours, or if discharge lasted for 10 days or more) versus immediate antibiotics (amoxicillin 250 mg tds for 1 week)
OutcomesFever, severity of pain, duration of malaise, absence from school, use of paracetamol, antibiotic use, further earache at 3 and 12 months
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskQuote: "patients were randomised to a group"
Allocation concealment (selection bias)Low riskQuote: "doctor opened a sealed numbered opaque envelope"
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding undertaken
Incomplete outcome data (attrition bias)
All outcomes
Low riskA comparison of responders versus non-responders was undertaken
Selective reporting (reporting bias)Low riskPre-specified outcomes have been reported
Other biasLow riskNo Information

Little 2005a

MethodsRandomised controlled trial
ParticipantsAdults and children aged 3 years and over with cough and at least 1 symptom or sign localising to the lower respiratory tract
InterventionsDelayed antibiotics (script left at reception and patients instructed to pick up the script after 14 days if required) versus immediate antibiotics versus no antibiotics
OutcomesFever, cough, duration of cough, severity of cough, malaise, duration of malaise, complications of disease, hospital admissions, diarrhoea, antibiotic use, re-consultation in the 12 months following the index consultation, excluding the first month after the index consultation
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputer-generated random number tables and block randomisation (block size 6)
Allocation concealment (selection bias)Low riskOpaque sealed envelopes
Blinding (performance bias and detection bias)
All outcomes
High riskOutcome assessor was blinded. Patient and care provider were not blinded
Incomplete outcome data (attrition bias)
All outcomes
Low riskMissing data were described and intention-to-treat analysis used
Selective reporting (reporting bias)Low riskPre-specified outcomes were reported
Other biasLow riskNo Information

Pichichero 1987

MethodsOpen randomised controlled trial
ParticipantsChildren with sore throat (suspected group A beta haemolytic Streptococcus)
InterventionsDelayed antibiotics (48 hours) versus immediate antibiotics (penicillin 250 mg tds for 10 days)
OutcomesFever, duration of fever, malaise, re-consultation rates, vomiting
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskTable of random numbers
Allocation concealment (selection bias)High riskNot used
Blinding (performance bias and detection bias)
All outcomes
Low riskPatient and doctor blinded but unsure about outcome assessor
Incomplete outcome data (attrition bias)
All outcomes
Low riskNo drop-outs
Selective reporting (reporting bias)Low riskPre-specified outcomes were reported
Other biasLow riskNo Information

Spiro 2006

  1. a

    AOM: acute otitis media
    ITT: intention-to-treat
    IU: international units
    qds: four times a day
    tds: three times a day

MethodsPlacebo and randomised controlled trial
ParticipantsChildren aged 6 months to 12 years
InterventionsDelayed antibiotics (patients given a script which was to expire after 72 hours) versus immediate antibiotics
OutcomesFever, duration of fever, pain, duration of pain, vomiting, diarrhoea, rash, antibiotic use
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputer-assisted randomisation
Allocation concealment (selection bias)Low riskSealed opaque envelopes
Blinding (performance bias and detection bias)
All outcomes
High riskStudy participants were not blinded but outcome assessors were blinded
Incomplete outcome data (attrition bias)
All outcomes
Low riskMore people in the wait and see prescription group stayed in the trial, however this was acknowledged and addressed
Selective reporting (reporting bias)Low riskPre-specified outcomes were reported
Other biasLow riskNo Information

Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion
Cates 1999Not a randomised controlled trial
Fischer 2009Not a randomised controlled trial
Newson 2009Not a randomised controlled trial
Siegel 2003Not a randomised controlled trial
Vouloumanou 2009Not a randomised controlled trial