Non-steroidal anti-inflammatory drugs for the common cold

  • Review
  • Intervention

Authors


Abstract

Background

Non-steroidal anti-inflammatory drugs (NSAIDs) have been widely used for the treatment of pain and fever associated with the common cold. However, there is no systematic review to assess the effects of NSAIDs in treating the common cold.

Objectives

To determine the effects of NSAIDs versus placebo (and other treatments) on signs and symptoms of the common cold, and to determine any adverse effects of NSAIDs in people with the common cold.

Search methods

We searched CENTRAL (The Cochrane Library 2013, Issue 1), MEDLINE (January 1966 to April week 4, 2013), EMBASE (January 1980 to April 2013), CINAHL (January 1982 to April 2013) and ProQuest Digital Dissertations (January 1938 to April 2013).

Selection criteria

Randomised controlled trials (RCTs) of NSAIDS in adults or children with the common cold.

Data collection and analysis

Four review authors extracted data. We subdivided trials into placebo-controlled RCTs and head-to-head comparisons of NSAIDs. We extracted and summarised data on global efficacies of analgesic effects (such as reduction of headache and myalgia), non-analgesic effects (such as reduction of nasal symptoms, cough, sputum and sneezing) and side effects. We expressed dichotomous outcomes as risk ratios (RR) with 95% confidence intervals (CI) and continuous data as mean differences (MD) or standardised mean differences (SMD). We pooled data using the fixed- and random-effects models.

Main results

We included nine RCTs with 1069 participants, describing 37 comparisons: six were NSAIDs versus placebo and three were NSAIDs versus NSAIDs. The overall risk of bias in the included studies was mixed. In a pooled analysis, NSAIDs did not significantly reduce the total symptom score (SMD -0.40, 95% CI -1.03 to 0.24, three studies, random-effects model), or duration of colds (MD -0.23, 95% CI -1.75 to 1.29, two studies, random-effects model). For respiratory symptoms, cough did not improve (SMD -0.05, 95% CI -0.66 to 0.56, two studies, random-effects model) but the sneezing score significantly improved (SMD -0.44, 95% CI -0.75 to -0.12, two studies, random-effects model). For outcomes related to the analgesic effects of NSAIDs (headache, ear pain, and muscle and joint pain) the treatment produced significant benefits. The risk of adverse effects was not high with NSAIDs (RR 2.94, 95% CI 0.51 to 17.03, two studies, random-effects model) and it is difficult to conclude that such drugs are not different from placebo.

Authors' conclusions

NSAIDs are somewhat effective in relieving discomfort caused by a cold but there is no clear evidence of their effect in easing respiratory symptoms. The balance of benefit and harms needs to be considered when using NSAIDs for colds.

Resumen

Antecedentes

Agentes antiinflamatorios no esteroides para el resfriado común

Aunque los agentes antiinflamatorios no esteroides (AINE) han sido ampliamente utilizados para el tratamiento del dolor y la fiebre asociados con el resfriado común, no hay revisiones sistemáticas para evaluar los efectos de los AINE en los pacientes con resfriado común.

Objetivos

Determinar los efectos de los AINE versus placebo y otros tratamientos en los signos y síntomas del resfriado común. Determinar cualquier efecto adverso de los AINE en los pacientes tratados con AINE para el resfriado común.

Estrategia de búsqueda

Se realizaron búsquedas en el Registro Cochrane Central de Ensayos Controlados (Cochrane Central Register of Controlled Trials, CENTRAL) (The Cochrane Library 2009, número 1) que incluyen el Registro Especializado de Ensayos Coontrolados del Grupo de Infecciones Respiratorias Infecciosas (Acute Respiratory Infections Group's Specialized Register); MEDLINE (enero 1966 hasta marzo 2009); EMBASE (enero 1980 hasta marzo 2009); CINAHL (enero 1982 hasta marzo 2009); ProQuest Digital Dissertations (enero 1938 hasta marzo 2009); KoreaMed (enero 1958 hasta marzo 2009) and KMbase (enero 1949 hasta marzo 2009).

Criterios de selección

Ensayos controlados aleatorios (ECA) que estudian el tratamiento del resfriado común con AINE en adultos o niños.

Obtención y análisis de los datos

Cuatro autores de la revisión extrajeron los datos (SYK, YSM, YJC, YWH). Se subdividieron los ensayos en ECA controlados con placebo y ECA de AINE versus AINE. Se extrajeron y resumieron los datos sobre eficacia general: los efectos analgésicos como reducción de la cefalea y la mialgia; los efectos no analgésicos como la reducción de los síntomas nasales, la tos, el esputo y los estornudo; y efectos secundarios.

Resultados principales

Esta revisión incluye nueve ECA, que describen 37 comparaciones: seis eran AINE versus placebo, y tres eran AINE versus AINE. Se incluyó un total de 1064 pacientes con resfriado común. En un análisis agrupado, los AINE no redujeron significativamente la puntuación total de síntomas ni la duración de los resfriados.

Sin embargo, para las medidas de resultado relacionadas con los efectos analgésicos de los AINE (cefalea, dolor de oído y dolor muscular y de las articulaciones), los AINE produjeron beneficios significativos, y el malestar general mostró un beneficio dudoso, aunque no se mejoró la irritación de la garganta. Los escalofríos mostraron resultados contradictorios. En lo que respecta a los síntomas respiratorios, no se mejoraron las puntuaciones de tos y de secreción nasal, pero la puntuación de estornudos mejoró significativamente. El riesgo de efectos adversos no fue alto y es difícil concluir que no son diferentes de los del placebo.

Conclusiones de los autores

Los AINE tienen cierta eficacia en el alivio del malestar causado por el resfriado, pero no existen pruebas claras de su efecto para aliviar los síntomas respiratorios. Debe considerarse la relación entre los efectos beneficiosos y perjudiciales cuando se utilizan los AINE para el tratamiento del resfriado.

Traducción

Traducción realizada por el Centro Cochrane Iberoamericano

摘要

背景

非類固醇類抗發炎藥物(Nonsteroidal antiinflammatory drugs)治療感冒

雖然非類固醇類抗發炎藥物(NSAIDs)已被廣泛用於治療感冒相關的疼痛和發燒,但並沒有系統性回顧(systematic review)來評估非類固醇類抗發炎藥物在感冒患者的效果。

目標

確定非類固醇類抗發炎藥物對照安慰劑和其他治療對感冒症狀與症候的效果。確定任何非類固醇類抗發炎藥物在治療感冒病患時的不良反應。

搜尋策略

我們檢索了Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2009, issue 1)資料庫;其中包含了急性呼吸道感染(acute Respiratory Infections;ARI)群組的特別登錄資料。同時也搜尋MEDLINE(1966年1月至2009年3月)、EMBASE(1980年1月至2009年3月)、CINAHL(1982年1月至2009年3月)、ProQuest Digital Dissertations(1938年1月至2009年3月)、KoreaMed(1958年1月至2009年3月)和KMbase(1949年1月至2009年3月)等資料庫。

選擇標準

研究以非類固醇類抗發炎藥物治療成人或兒童感冒的隨機對照試驗(randomised controlled trials;RCTs)。

資料收集與分析

四位審查作者負責選取數據(SYK, YSM, YJC, YWH)。我們再細分試驗為安慰劑控制的隨機對照試驗及非類固醇類抗發炎藥物與非類固醇類抗發炎藥物的隨機對照試驗。我們選取和歸納全體功效的資料:鎮痛效果,如減少頭痛和肌肉疼痛;非鎮痛效果,如減少鼻腔症狀、咳嗽、咳痰及打噴嚏;以及副作用。

主要結論

此回顧包括9個隨機對照試驗,共描述37個比較項目:其中6個是非類固醇類抗發炎藥物與安慰劑對照,另外3個則是非類固醇類抗發炎藥物與非類固醇類抗發炎藥物對照。全部共包含1064位感冒患者。匯總分析發現,非類固醇類抗發炎藥物並未顯著減少總症候評分或感冒持續時間。然而,非類固醇類抗發炎藥物對鎮痛作用相關(頭痛,耳朵痛,肌肉和關節疼痛)的效果有顯著益處;對疲倦則有邊緣性益處;但是對喉嚨不適沒有改善;對改善寒顫則好壞參半。在呼吸道症狀部分,對咳嗽和鼻腔分泌物的評分沒有改善,但打噴嚏則顯著改善。我們在非類固醇類抗發炎藥物治療組並未發現增加副作用頻率的證據。

作者結論

作者建議使用非類固醇類抗發炎藥物以緩解感冒造成的疼痛或不適。對於使用非類固醇類抗發炎藥物來緩解如咳嗽和鼻腔分泌物等呼吸道症狀,則需要進一步研究。

翻譯人

本摘要由臺灣大學附設醫院耿立達翻譯。

此翻譯計畫由臺灣國家衛生研究院(National Health Research Institutes, Taiwan)統籌。

總結

感冒是人類最常見與最廣泛的疾病。非類固醇類抗發炎藥物是具鎮痛(減少疼痛)、解熱(退燒),並在高劑量下具有抗發炎作用。非類固醇類抗發炎藥物被廣泛用於治療感冒相關的疼痛與發燒長達一個多世紀之久。此回顧發現,有9項臨床研究(含1064位病患)比較不同種非類固醇類抗發炎藥物和安慰劑與其他非類固醇類抗發炎藥物。我們的研究結果認為非類固醇類抗發炎藥物最能改善感冒引起的鎮痛相關症狀;但沒有明確的證據顯示,非類固醇類抗發炎藥物可有效地改善感冒引起的咳嗽或流鼻涕。

Résumé scientifique

Non-steroidal anti-inflammatory drugs for the common cold

Contexte

Bien que les anti-inflammatoires non stéroïdiens (AINS) aient été largement utilisés pour le traitement de la douleur et de la fièvre associés rhinopharyngite/rhume, aucune revue systématique n’a évalué les effets des AINS chez les personnes souffrant de la rhinopharyngite/rhume

Objectifs

Déterminer les effets des AINS par comparaison avec un placebo et d’autres traitements sur les signes et les symptômes de la rhinopharyngite/rhume.
Déterminer les effets indésirables des AINS chez les patients traités avec des AINS pour la rhinopharyngite/rhume.

Stratégie de recherche documentaire

Nous avons effectué une recherche dans le registre Cochrane des essais contrôlés (CENTRAL) (The Cochrane Library 2011, numéro 1) qui contient la base de données spécialisée du groupe Cochrane sur les infections respiratoires aiguës, MEDLINE (janvier 1996 à la 4e semaine du mois de mars 2011), EMBASE (janvier 1980 à avril 2011), CINAHL (janvier 1982 à avril 2011), ProQuest Digital Dissertations (janvier 1938 à avril 2011), KoreaMed (janvier 1958 à avril 2011) et KMbase (janvier 1949 à avril 2011).

Critères de sélection

Essais contrôlés randomisés (ECR) étudiant le traitement de la rhinopharyngite/rhume par des AINS chez les adultes ou les enfants

Recueil et analyse des données

Quatre auteurs (SYK, YSM, YJC, YWH) ont extrait les données. Nous avons sous-divisé les essais en ECR contrôlés versus placebo et ECR comparant des AINS en face à face. Nous avons extrait et résumé les données sur l'efficacité globale des effets analgésiques (comme la diminution des maux de tête et des myalgies), des effets non analgésiques (comme la diminution des symptômes nasaux, de la toux, des expectorations et des éternuements) et des effets secondaires. Nous avons exprimé les résultats dichotomiques sous la forme de rapports de risques (RR) avec des intervalles de confiance (IC) à 95 % et les données continues en tant que différence moyenne (DM) ou différences moyennes standardisées (DMS). Les données ont été combinées à l’aide d’un modèle à effets fixes et à effets aléatoires.

Résultats principaux

Nous avons inclus neuf ECR comprenant 1 069 participants et décrivant 37 comparaisons : six d’entre eux étudiaient des AINS versus placebo et trois d'entre eux comparaient un AINS à un autre AINS. Le risque de biais global dans les études incluses variait. Dans une analyse combinée, les AINS n’ont pas permis de réduire significativement le score de symptômes total (DMS -0,40 ; IC à 95 % -1,03 à 0,24, trois études, modèle à effets aléatoires) ni la durée des rhumes (DM -0,23 ; IC à 95 % -1,75 à 1,29, deux études, modèle à effets aléatoires). Un bénéfice limite était observé pour les sensations de malaise (DMS -0,29 ; IC à 95 % -0,6 à 0,03, deux études, modèle à effets aléatoires), l’irritation de la gorge ne s’est pas améliorée (DMS -0,01 ; IC à 95 % -0,33 à 0,30, deux études, modèle à effets aléatoires) et les résultats sur les frissons étaient variables (DMS -0,02 ; IC à 95 % -0,34 à 0,30, deux études, modèle à effets aléatoires). Pour ce qui est des symptômes respiratoires, les scores de toux (DMS -0,05 ; IC à 95 % -0,66 à 0,56, deux études, modèle à effets aléatoires) et d'écoulement nasal ne se sont pas du tout améliorés, mais le score d’éternuement (DMS -0,44 ; IC à 95 % -0,75 à -0,12, deux études, modèle à effets aléatoires) s’est sensiblement amélioré. Toutefois, pour ce qui est des résultats liés aux effets analgésiques des AINS (maux de tête, douleurs auriculaires et douleurs musculaires et articulaires), ce type de traitement a permis d’obtenir des bénéfices significatifs. Le risque d’effets indésirables n’était pas élevé avec les AINS (RR 2,94, IC à 95 % 0,51 à 17,03, deux études, modèle à effets aléatoires) et il est difficile de conclure que ces médicaments ne sont pas différents du placebo.

Conclusions des auteurs

Les AINS se révèlent relativement efficaces pour soulager les douleurs causées par un rhume, mais aucune preuve manifeste de leur effet sur le soulagement des symptômes respiratoires n’a été apportée. L’équilibre entre le bénéfice et les préjudices doit être pris en considération en cas d’utilisation des AINS pour traiter les rhumes.

Plain language summary

Non-steroidal anti-inflammatory drugs for the common cold

The common cold is the most common and widespread illness known to humans. Non-steroidal anti-inflammatory drugs (NSAIDs) for example, aspirin, ibuprofen and naproxen, have analgesic (pain reducing), antipyretic (fever reducing) and, in higher doses, anti-inflammatory effects. NSAIDs have been widely used for over a century for the treatment of pain and fever associated with the common cold and so it important to assess the efficacy of using NSAIDs for treating pain or fever associated with the common cold.

This review found nine studies (including 1069 participants) that compared various NSAIDs either with each other or with an inactive substance that has no treatment value (placebo). Our findings suggest that NSAIDs may improve most analgesia-related symptoms caused by the common cold, but there is no clear evidence that NSAIDs are effective in improving coughs and runny noses caused by the common cold.

Some of the included trials reported gastrointestinal complaints, rash and oedema in the NSAIDs group. Major limitations of this review were that the results of the studies are quite diverse and the number of studies for one result is quite small. For this reason, it is difficult to draw any strong conclusions.

Résumé simplifié

Anti-inflammatoires non stéroïdiens pour le traitement du rhume

Le rhume constitue la maladie la plus courante et la plus répandue chez l'homme. Les anti-inflammatoires non stéroïdiens (AINS) sont des médicaments qui ont des effets analgésiques (diminution de la douleur), antipyrétiques (baisse de la fièvre) et, à des doses plus élevées, anti-inflammatoires. Les AINS sont largement utilisés depuis plus d’un siècle pour le traitement de la douleur et de la fièvre associées au rhume. Cette revue a trouvé neuf études (incluant 1 069 participants) qui ont comparé divers AINS soit entre eux, soit avec une substance inactive n'ayant aucune valeur thérapeutique (placebo). Nos résultats suggèrent que les AINS sont susceptibles d’améliorer la plupart des symptômes liés à l’analgésie qui sont causés par le coryza, mais aucune preuve claire attestant de l’efficacité des AINS dans l’amélioration de la toux et des écoulements nasaux provoqués par le rhume n’a été apportée.

Notes de traduction

Traduit par: French Cochrane Centre 1st November, 2011
Traduction financée par: Ministère du Travail, de l'Emploi et de la Santé Français

Background

Description of the condition

The common cold is an acute respiratory tract infection (ARTI) and is the most common and widespread illness known to humans, affecting all age groups. Young children suffer an average of six to eight colds a year, while adults experience approximately two to four colds a year. Although the common cold is usually mild, with symptoms lasting one to two weeks, it is a leading cause of medical visits and days missed from school and work (Heikkinen 2003). Nasal congestion, rhinorrhoea, sneezing and coughing accompanied by general malaise are typical symptoms of the common cold. Over 200 serologically different viral types are responsible for common colds, with the rhinovirus being the most common cause (Eccles 2005).

Description of the intervention

Despite ongoing research into antiviral drugs, there are no effective therapies for the prevention or treatment of the common cold. Therefore, treatment of colds is normally aimed at relieving the symptoms of the illness. Several classes of drugs are currently available, including decongestants, anticholinergics, antihistamines and antitussives. These are effective, to a greater or lesser extent, in treating symptoms of the common cold (AlBalawi 2011; De Sutter 2012; Li 2010; Ostberg 1997; Saraswat 2011; Smith 2012).

How the intervention might work

NSAIDs have been widely used for over a century for the treatment of pain and fever associated with the common cold. Despite their widespread present day use and the long medical history of the use of NSAIDs in relieving pain associated with the common cold, there is a lack of clinical data to support the efficacy of NSAIDs treating this condition. There is some evidence that cold symptoms might be the result of inflammatory mediators such as kinins and prostaglandins, which can be blocked by NSAIDs, rather than the result of the direct cytopathic effects of viruses (Eccles 2005; Gwaltney 2002).

Why it is important to do this review

Several studies have proposed that NSAIDs could be effective in alleviating common cold symptoms, including sneezing and coughing (Sperber 1989; Sperber 1992; Winther 2001). However, no consensus has been reached on this issue. This systematic review is an update of a Cochrane review first published in 2009 (Kim 2009).

Objectives

To determine the effects of NSAIDs versus placebo (and other treatments) on signs and symptoms of the common cold, and to determine any adverse effects of NSAIDs in people with the common cold.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised controlled trials (RCTs) comparing NSAIDs used either alone or in combination with other medications versus placebo and other therapies for the treatment of signs and symptoms of the common cold in adults and children.

Types of participants

We included adults and children with the common cold, who had no other acute illness or severe, chronic conditions. The case definition of the common cold used was: recent onset of symptoms of runny or stuffy nose (or both), and sneezing, with or without symptoms of headache and cough. We excluded participants if they suffered from allergic rhinitis, had a concurrent lower or chronic respiratory infection or another chronic disease, atopic eczema, asthma, fever (> 38°C), sinusitis or exudative pharyngitis.

Source populations were volunteers from the community, hospital or community outpatient departments, and primary care settings. We accessed additional evidence from studies of healthy volunteers exposed to rhinovirus in experimental conditions.

Types of interventions

NSAIDs versus placebo as a treatment for symptoms of the common cold. We considered variable doses and routes of administration of the NSAID treatments. We included trials that allowed concurrent use of other medications if they permitted equal access for patients in both the NSAIDs and placebo groups (Ta'i 2012).

Types of outcome measures

We did not consider objective assessments such as rhinometry and rhinoscopy.

Primary outcomes
  1. Global evaluation of efficacy in the treatment of common cold symptoms.

  2. Decrease in the number or duration of individual common cold symptoms. These symptoms were assessed by severity scale.

Secondary outcomes
  1. Any reported side effects.

Search methods for identification of studies

Electronic searches

In the previous review we searched the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2011, Issue 1) which contains the Cochrane Acute Respiratory Infections Group's Specialised Register, MEDLINE (January 1966 to March week 4, 2011), EMBASE (January 1980 to April 2011), CINAHL (January 1982 to April 2011) and ProQuest Digital Dissertations (January 1938 to April 2011).

For this 2013 update we searched the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2013, Issue 1) which contains the Cochrane Acute Respiratory Infections Group's Specialised Register, MEDLINE (January 2011 to March week 4, 2013), EMBASE (January 2011 to April 2013), CINAHL (January 2011 to April 2013) and ProQuest Digital Dissertations (January 2011 to April 2013).

See Appendix 1 and Appendix 2 for the EMBASE and CINAHL search strategies and Appendix 3 for the search strategy used for MEDLINE and CENTRAL. We combined the MEDLINE search terms with the highly sensitive search strategy designed by The Cochrane Collaboration for identifying RCTs (Lefebvre 2011). We adapted these search terms to search EMBASE.

We imposed no language or publication restrictions.

Searching other resources

We assessed non-English language papers and, if necessary and possible, translated them, with the assistance of native language speakers. We searched reference lists of review articles and of all included studies to find other potentially eligible studies. We contacted authors of the included trials to request unpublished studies.

Data collection and analysis

Selection of studies

We used the search strategy detailed above to obtain titles and abstracts of studies that might be relevant to the review. Three review authors (YSM, YJC, YWH) independently screened titles and abstracts and one review author (SYK) collated the results. All review authors participated in resolving discrepancies until a consensus was reached.

Data extraction and management

The same review authors (YSM, YJC, YWH) independently carried out data extraction using standard data extraction forms. We translated studies reported in non-English language journals before assessment. Where more than one publication of one trial existed, we included only the publication with the most complete data. We resolved disagreements by discussion.

Assessment of risk of bias in included studies

Three review authors (YSM, YJC, YWH) independently assessed the methodological quality of included studies using The Cochrane Collaboration's 'Risk of bias' tool (Higgins 2011). One review author (SYK) collated the results. All review authors participated in resolving discrepancies until a consensus was reached.

Measures of treatment effect

The effect of NSAIDs on common cold signs and symptoms was our primary measure of interest. We expressed results as risk ratios (RRs) with 95% confidence intervals (CIs) for dichotomous outcomes.

We used the standardised mean difference (SMD) where continuous scales of measurement were used to assess the effects of treatment (for example, mean severity scores and time to symptom relief), because different scales were used in most of the trials.

We summarised adverse effects when reported. We calculated the RR with 95% CI for each adverse effect, where possible, either compared to no treatment or to another treatment.

Unit of analysis issues

We split trials including more than two comparisons and analysed them as individual pair-wise comparisons. By dividing the placebo case, we ensured that we did not count cases in the placebo group more than once when conducting a meta-analysis. We had no special issues in the analysis of studies with non-standard designs.

Dealing with missing data

We attempted to contact the trial authors for additional information if data from the trial reports were unclear or missing. We have excluded data from the meta-analysis and clearly stated the reason if we judged missing data to render the result uninterpretable.

Assessment of heterogeneity

We assessed heterogeneity amongst trials by using the Chi2 test for heterogeneity (with a 10% level of statistical significance) and the I2 statistic.

We considered other sources of heterogeneity, apart from differences in interventions, namely clinical diversity (children/adults, different classes of NSAIDs and different dosages) and study quality. Heterogeneity in treatments could be related to prior agent(s) used, and the agent, dose and duration of the therapy.

Assessment of reporting biases

There were insufficient trials for us to assess the likelihood of publication bias by examining the funnel plot for asymmetry.

Data synthesis

We pooled data using a fixed-effect model. We also used the random-effects model to ensure the robustness of the chosen mode and susceptibility to outliers.

Subgroup analysis and investigation of heterogeneity

We intended to conduct subgroup analyses where data were available, for example, by age (adult, child), NSAID class and whether the common cold was artificial or natural.

Sensitivity analysis

We pooled data using the fixed-effect model but we also analysed the random-effects model to ensure robustness of the mode chosen and susceptibility to outliers.

Results

Description of studies

In the vast majority of studies, the clinical symptoms of the common cold, requirements for inclusion, type and dose of NSAIDs, outcomes of trials and duration of therapy were quite diverse, which caused difficulties in quantitative analysis.

Results of the search

In the previous search, we identified 57 trials; of these, nine met the inclusion criteria. In this 2013 update search, we identified three more potential trials and all three were excluded. All included studies were double-blinded RCTs. Four of the six trials of community-acquired colds were multi-centre trials.

Included studies

The nine included studies involved 1069 participants suffering from colds. In six studies, 891 participants had community-acquired colds, and in three studies, 178 participants were experimentally infected with cold viruses. For experimentally infected colds, inoculated populations were analysed. Only 72.5% of experimentally infected participants had cold symptoms. Therefore, we included non-symptomatic infected participants in this analysis.

Three studies were performed in the USA, four in Japan, and one each in Belgium and Denmark. Trials took place in a total of 154 settings. Most were participants from hospitals, clinics and outpatient departments. One trial involved medical students and university staff. Three trials of experimentally infected colds did not report the trial setting. One trial involved mainly students and two trials reported participants only as volunteers.

Five trials compared NSAIDs with a placebo, three trials compared one NSAID with another, and one trial compared two NSAIDs with a placebo.

Five studies used ibuprofen, two used aspirin and two studies used loxoprofen. Ketoprofen, fenoprofen, fentiazac and naproxen were used in one study. Seven trials used visually identical capsules, one trial used a double-dummy method and one trial used coded vials. The duration of treatment varied from a single dose to two daily doses for seven days.

Three studies used a general symptom score and five studies used a symptom severity score.

The Characteristics of included studies table includes a summary of the randomisation process, cold acquisition route, inclusion criteria, population, interventions and comparisons, outcome measures, adverse events and methodological qualities.

Excluded studies

We excluded 51 trials: four studies were not randomised or the randomisation allocation was unclear; one study included febrile participants; 46 studies included participants with diagnoses other than common colds (see Characteristics of excluded studies table).

Risk of bias in included studies

The overall risk of bias in the included studies was mixed, largely due to missing information regarding randomisation procedures. We assessed two studies (Goto 2007; Ryan 1987) as being of high quality.

Allocation

Out of the nine included studies (Goto 2007; Graham 1990; Itoh 1980; Katsu 1993; Nagaoka 1980; Ryan 1987; Sperber 1989; Sperber 1992; Winther 2001), two studies (Goto 2007; Ryan 1987) used a computer-generated random numbers table to generate the allocation sequence. The remaining studies contained insufficient information about the sequence generation process.

In four studies (Goto 2007; Itoh 1980; Nagaoka 1980; Ryan 1987) the allocation method was adequately concealed. In two Japanese studies (Itoh 1980; Nagaoka 1980) the randomisation process was carried out by two controllers who retained the key codes. In the remaining two studies, treatment was allocated by a third party (Goto 2007), or considered adequately concealed because the single oral dose was administered using a double-blind method (Ryan 1987).

Blinding

All studies were described as 'double-blind' and considered 'adequate'; either the active drug and placebo were identical, or an 'identical capsule double-dummy' method was used.

Incomplete outcome data

Among the included studies, eight (Goto 2007; Graham 1990; Itoh 1980; Nagaoka 1980; Ryan 1987; Sperber 1989; Sperber 1992; Winther 2001) adequately addressed incomplete outcome data. Three experimental rhinovirus cold trials (Graham 1990; Sperber 1989; Sperber 1992) excluded participants who were not infected, in which case the reason for exclusion may be justifiable. In six studies (Itoh 1980; Nagaoka 1980; Ryan 1987; Sperber 1989; Sperber 1992; Winther 2001), the number of withdrawals was zero or very small. One study had insufficient information to permit judgement of 'low risk' or 'high risk' of bias (Katsu 1993).

Selective reporting

All studies were considered as 'unclear' risk of bias as all trials failed to include the study protocol. They had insufficient information to permit a judgement of either 'low risk' or 'high risk' of bias.

Other potential sources of bias

Amongst the included studies, none were stopped early or reported claims of fraudulence against it. One study (Winther 2001) did not contain data to assess the baseline balance. The overall quality of studies was mixed, largely due to missing information regarding randomisation procedures (Figure 1; 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

Effects of interventions

In total, we identified 37 outcomes. Eight studies assessed effectiveness and five studies assessed adverse effects. Twenty-one (56.7%) of the 37 outcomes were assessed only by a single trial. Among the 16 outcomes assessed by two trials or more, six outcomes had an I2 statistic of ≥ 50% (overall symptom score, duration of colds, cough score, headache score, chills score and overall side effects).

Outcomes included in the meta-analyses

One trial (Sperber 1992) reported the daily symptom scores during six post-challenge days and a six-day cumulative symptom score. Because other trials reported cumulative symptom scores, we included the cumulative symptom score in the meta-analysis for comparison.

One trial (Sperber 1989) reported cumulative symptom scores for individual symptoms, such as rhinorrhoea and nasal obstruction, as well as cumulative symptom scores for individual areas (that is, nasal symptom score). To prevent double counting and to compare data, we included only cumulative symptom scores of individual symptoms in the meta-analysis.

Graham 1990 used aspirin (4 g/day) and ibuprofen (1.2 g/day). Because the dose of ibuprofen was the usual prescribed dose for the common cold and that of aspirin was not, we chose to use the ibuprofen group in the meta-analysis.

1. Global evaluation of efficacy in the treatment of common cold symptoms

i. Non-steroidal anti-inflammatory drugs (NSAIDs) versus placebo

Three trials (Goto 2007; Sperber 1989; Sperber 1992) assessed the total symptom score improvement of NSAIDs on the course of the common cold. The first trial (Sperber 1989) included 40 young adults and compared the effect of ibuprofen at a dose of 200 mg/four times a day for five days with that of a placebo. During six post-challenge days, the daily total symptom score was not significantly different between the two groups. The second trial (Sperber 1992) included 79 young adults and compared naproxen at a dose ranging from 3.0 to 5.0 g for five days with placebo. The total five-day symptom score judged by the modified Jackson criteria was reduced by 29% (95% CI 16% to 42%) in the naproxen group compared with the placebo group. The third trial (Goto 2007) included 174 adults and compared the effects of loxoprofen at a dose of 60 mg/twice a day for seven days with placebo. Duration of illness, number of days with limited daily activities and total symptom score were not significantly different between the two groups. We conducted a meta-analysis of data from the three trials. The results of the pooled analysis were not significant (standardised mean difference (SMD) -0.40, 95% confidence interval (CI) -1.03 to 0.24, random-effects model) (Analysis 1.2) and there was heterogeneity (I2 statistic = 83%).

Two trials assessed the duration of colds (Goto 2007; Sperber 1989). The results of the pooled analysis were not significant (mean difference (MD) -0.23, 95% CI -1.75 to 1.29, random-effects model) (Analysis 1.5) and there was heterogeneity (I2 statistic = 80%).

One trial assessed the proportion of patients with symptoms of moderate to marked severity (Sperber 1989); no significant effect was detected.

ii. Head-to-head comparisons

Three trials involving participants with natural colds assessed the effect of one NSAID compared to other NSAID and ranked the severity of global symptoms on a five- to seven-point scale; all three trials were performed in Japan (Itoh 1980; Katsu 1993; Nagaoka 1980).

Nagaoka 1980, which involved 222 participants, compared fentiazac (300 mg/day) with ibuprofen (600 mg/day). Katsu 1993 involved 167 participants and compared loxoprofen (80 mg/day) with ibuprofen (600 mg/day). Itoh 1980 enrolled 184 participants with upper respiratory tract infections and compared aspirin with ketoprofen. Itoh 1980 reported that there was no statistically significant difference between the groups in a subgroup analysis for the population with common colds, but the estimates and the number of participants included in the study population were not reported. Therefore, we could not use this result in a pooled analysis of efficacy.

Marked improvement and moderate to marked improvement (in a global improvement rating) were significant in only one study (Nagaoka 1980).

2. Decrease in the number or duration of individual common cold symptoms

i. NSAIDs versus placebo: analgesic effects

Two trials (Sperber 1992; Winther 2001) measured nine outcomes evaluating the analgesic effects of NSAIDs. The types of NSAIDs and the scale of outcomes differed between these studies.

As mentioned above, Sperber 1992 assessed the effect of naproxen in participants with an experimental cold and reported daily symptom scores and total (five-day) symptom scores. Winther 2001 enrolled 80 participants with natural colds. The effect of ibuprofen at a dose of 400 mg/three times a day for three days was studied and the severity of symptoms was then ranked on a four-point scale (not present, mild, moderate, severe) and a three-day cumulative symptom score was reported.

Firstly, the cumulative throat irritation score was used in two trials (Sperber 1992; Winther 2001). In Sperber 1992, total (five-day) and daily throat scores were not significantly different between the treatment groups. In Winther 2001, the total throat irritation/pain score was not significantly different between the treatment groups. As expected, the results of the pooled analysis were not significant (SMD -0.01, 95% CI -0.33 to 0.30, fixed-effect and random-effects models) (Analysis 2.2) and there was no heterogeneity.

Secondly, cumulative headache scores were reported in the same two trials (Sperber 1992; Winther 2001). All trials reported that headache scores were significantly lower in the NSAIDs groups than in the placebo groups. In a pooled analysis, NSAIDs significantly reduced headache scores (SMD -0.65, 95% CI -0.97 to -0.32) (Analysis 2.3). The results of fixed-effect and random-effects models were the same; there was marginal heterogeneity.

Thirdly, cumulative pain scores in the muscles and joints were also reported in these two trials (Sperber 1992; Winther 2001). In Winther 2001, the pain score in muscles and joints did not differ significantly between the treatment groups. In Sperber 1992, the myalgia score was significantly reduced in the naproxen group. In a pooled analysis, NSAIDs significantly reduced the score for pain in muscles and joints (SMD -0.40, 95% CI -0.77 to -0.03) in fixed-effect models; there was no heterogeneity.

Fourthly, the two studies assessed a cumulative malaise score (Sperber 1992; Winther 2001). All trials reported that the malaise score was not significantly different between the two treatment groups. However, in a pooled analysis there was a trend towards reduction of malaise (SMD -0.29, 95% CI -0.6 to 0.03, fixed-effect and random-effects models) (Analysis 2.7, Analysis 2.8).

Fifthly, the two studies assessed a cumulative chills score; the results were mixed. One trial reported a significant reduction (Sperber 1992) and another reported a significant increase (Winther 2001). In a pooled analysis, the statistical significance of the difference disappeared and heterogeneity was detected (SMD -0.02, 95% CI -0.34 to 0.30, P = 0.89, P = 0.0006; I2 statistic = 91.5%) (Analysis 2.9).

The cumulative earache score was significantly reduced in the ibuprofen group compared to the placebo group (Winther 2001).

ii. NSAIDs versus placebo: non-analgesic effects

Four trials (Graham 1990; Sperber 1989; Sperber 1992; Winther 2001) measured 15 outcomes irrelevant to the analgesic effect. The scales of outcomes were quite diverse. Three trials (Graham 1990; Sperber 1989; Winther 2001) tested ibuprofen and one trial (Sperber 1992) tested naproxen.

Firstly, two trials reported a cumulative cough score (Sperber 1992; Winther 2001). In Sperber 1992, the cumulative cough score was not significant (0.8 and 1.6, naproxen and placebo, respectively), but the daily score was significantly reduced at four days (P < 0.01). Winther 2001 evaluated the cumulative cough score, but there was no difference between the groups. The results of a pooled analysis for cumulative cough score were not significant.

Secondly, two trials evaluated a cumulative sneezing score (Sperber 1992; Winther 2001). In Sperber 1992, the cumulative sneezing score was not significant (1.5 and 2.2, naproxen and placebo, respectively) but daily scores were reduced in the naproxen group at one and four days. The statistically insignificant differences between scores were at two and three days. In Winther 2001, the cumulative sneezing score was significantly reduced in the ibuprofen group, and the result of a pooled analysis supported this effect (SMD -0.44, 95% CI -0.75 to -0.12, the P value of the heterogeneity test was 0.44; the results of fixed-effect and random-effects models were the same) (Analysis 3.4). Winther also examined the total number of sneezes and the result was significant.

Three trials studied a cumulative rhinorrhoea score and a cumulative nasal obstruction score, and found no differences between the groups (Sperber 1989; Sperber 1992; Winther 2001).

The proportion of nasal obstruction scores greater than five points (Graham 1990), total mucus weight, total tissue count (Sperber 1989), total number of nose blows, cumulative nasal dryness score, cumulative score for reduced sense of smell, cumulative hoarseness score, cumulative fatigue score and cumulative malaise score were quantified in a single study (Winther 2001) and the results were not significantly different between the treatment groups.

The cumulative nose irritation score, cumulative pain on swallowing score and cumulative eye itching score were also not significantly different between the treatment groups (Winther 2001).

3. Any reported side effects

i. NSAIDs versus placebo: adverse effects

Five trials reported adverse effects. One study (Goto 2007) reported that adverse effects were more frequent in the loxoprofen group (9.5% versus 1.1%, P < 0.05). Otherwise we could not find any evidence of increasing frequency of adverse effects in the active treatment groups. These outcomes included overall side effects, gastrointestinal complaints and other problems such as rash and oedema.

Two trials assessed the overall side effects of NSAIDs and there was moderate heterogeneity (Goto 2007; Sperber 1989). The results of a pooled analysis for overall side effects was significant with a fixed-effect model (risk ratio (RR) 2.88, 95% CI 1.11 to 7.45, P = 0.03) (Analysis 4.1) but not with a random-effects model (RR 2.94, 95% CI 0.51 to 17.03) (Analysis 4.2). Three trials reported gastrointestinal adverse effects and found no differences between the groups (Ryan 1987; Sperber 1989; Sperber 1992). Lethargy/drowsiness, feeling hyperactive, feeling more awake, flushed face, difficulty sleeping, light-headedness and dry mouth were reported in one or two trials and the results were not significantly different between the treatment groups.

Discussion

In summary, if non-steroidal anti-inflammatory drugs (NSAIDs) are administered to community-infected or experimentally infected cold patients, their analgesic effect against pain and irritation induced by the cold is relatively effective, but reports on whether they are helpful in relieving respiratory symptoms, such as coughing and sneezing, are not consistent and the evidence is insufficient.

Despite a comprehensive search, only nine studies met the inclusion criteria, six of which were placebo-controlled RCTs and three of which were head-to-head RCTs. When we evaluated the methodological quality of the included studies using The Cochrane Collaboration's tool for assessing risk of bias (Higgins 2011), the overall quality of studies was mixed, largely due to missing information regarding randomisation procedures. We assessed two studies (Goto 2007; Ryan 1987) as being of high quality. Our outcomes were mainly subjective and blinding of participants may be critical. All nine studies were described as 'double-blind' and considered 'adequate'.

Among the results used to examine the effect of NSAIDs on the common cold, the ones looking at the analgesic effect evaluated headache, throat irritation, muscle and joint pain, ear pain, malaise and chills. Among them, headache, ear pain and muscle and joint pain showed significant results and malaise showed borderline significance. However, throat irritation was not improved, and chills showed mixed results. For some cases where symptoms did not improve, the reasons were uncertain. Whether the cold was community-acquired or experimentally infected, the trial quality and dose of NSAIDs could not explain the differences. In the case of throat irritation, if the cold was an infection with a rhinovirus, there was the possibility that the treatment was not effective because throat pain disappeared naturally over a short period of time (Heikkinen 2003). There is also the possibility that the mechanism of throat pain may be different from that of headache and muscle pain. In the case of chills, NSAIDs were obviously effective in one trial, but worsened the symptoms in the other trial. Because chills are known to happen mainly when the fever has lowered, the measure of improvement may be different from the other symptoms and depends on whether there was a fever before the administration of treatment or not. However, because there was no information on the body temperature before starting the treatment in the two trials, we cannot draw a conclusion on this matter. Apart from these two symptoms, NSAIDs improved most of the analgesia-related symptoms caused by a cold. Therefore, we recommend the use of NSAIDs for these symptoms.

Three trials (Goto 2007; Sperber 1989; Sperber 1992) studied whether NSAIDs had a comprehensive effect on various symptoms caused by the common cold. Two of them (Sperber 1989; Sperber 1992) were conducted with participants whose cold was experimentally infected by a rhinovirus. One of those (Sperber 1992) showed a statistically significant difference in the effect of NSAIDs and when the results of the two studies were merged the results were significant. However, one recently published trial (Goto 2007) reported that the total symptom score showed no significant difference between the two groups. The results of the pooled analysis were not significant and there was heterogeneity, but the reason for this was unclear.

Among the studies two trials examined whether NSAIDs reduced the duration of a cold (Goto 2007; Sperber 1989). The results of the pooled analysis were not significant and there was heterogeneity. NSAIDs did not have any effect on the severity or duration of a cold. Because there were only two trials and the number of subjects in the studies was small, it is hard to draw a definite conclusion about the effects of NSAIDs on the duration of a cold.

One of the current issues related to the administration of NSAIDs for the common cold is whether NSAIDs are helpful in relieving respiratory symptoms such as cough. Many of the studies on the common cold recommend the administration of NSAIDs to ease coughing caused by a cold (Heikkinen 2003; Irwin 2000). The recently published American College of Chest Physicians (ACCP) guidelines recommend the combined administration of first-generation antihistamine and nasal decongestant or the administration of naproxen for cough caused by a cold (Pratter 2006). Respiratory symptoms examined in this review were cough, nasal discharge and sneezing. The medication was not effective for cough in two trials (Sperber 1992; Winther 2001) and pooled results did not show a significant improvement. None of the three trials (Sperber 1989; Sperber 1992; Winther 2001) showed a significant result for nasal discharge, and pooled results were not significant. However, in the case of sneezing, one trial (Winther 2001) showed a significant improvement and pooled results showed a moderate effect. Considering these results, which differ from existing guidelines, there is no clear evidence that NSAIDs are effective for coughs caused by a cold, or should be recommended in order to ease cough caused by a cold.

NSAIDs draw attention due to their adverse effects. For some NSAIDs, their long-term use increases the risk of cardiovascular disease (Matchaba 2004) and may cause gastrointestinal side effects (Ofman 2002). The frequency of gastrointestinal side effects increases in proportion to the dose and period of NSAID medication but the risk of gastrointestinal side effects cannot be excluded in short-term use (Hernández-Díaz 2000). In trials included in this review, the risk of side effects was not high but it is difficult to conclude that they are not different from placebo in terms of side effects.

In this review, three trials (Itoh 1980; Katsu 1993; Nagaoka 1980) studied which specific NSAIDs were more effective in treating a cold. One study (Nagaoka 1980) found that fentiazac was more effective than ibuprofen. However, this is probably because the dose of ibuprofen used in the trial was 600 mg/day, lower than that used in other trials.

The absence of epithelial destruction during rhinovirus infections has led to the idea that the clinical symptoms of the common cold may not be caused by a direct cytopathic effect of the viruses but instead are primarily caused by the inflammatory response of the host by media such as kinins, leukotrienes and histamines (Heikkinen 2003). Accordingly, NSAIDs are believed to ease not only fever and irritation but also respiratory symptoms such as coughing. However, this was not proven in the review. Further research is needed to examine their effects.

For analgesic effects on a cold, acetaminophen was also frequently used along with NSAIDs. However, in this review, we did not examine which of the medications was superior in terms of effect and safety. Further research is needed to evaluate this.

Major limitations of this review are that the results of the research are quite diverse and the number of studies for one result is quite small. For this reason, it is somewhat difficult to draw clear conclusions.

In conclusion, NSAIDs are recommended for relieving irritation or pain caused by a cold but the notion that NSAIDs are effective in relieving respiratory symptoms such as cough and nasal discharge needs more solid evidence.

Summary of main results

If NSAIDs are administered to community-infected or experimentally infected cold patients, their analgesic effect against pain and irritation induced by the cold is somewhat effective but reports on whether they are helpful in relieving respiratory symptoms such as coughing and sneezing are not consistent and the evidence is insufficient.

Overall completeness and applicability of evidence

The trials included in the analyses involved mainly young adults of both sex. Therefore the results of these trials may not be applicable to children and older people.

Quality of the evidence

We did not assess the quality of evidence using GRADE methodology but the quality of evidence may be estimated at 'low' to 'moderate' because of imprecision and risk of bias of the evidence.

Potential biases in the review process

Among the analgesic effect outcomes of NSAIDs, headache, pain in muscles and joints, and earache were statistically significant. However, these were mainly based on only two trials.

Agreements and disagreements with other studies or reviews

Some studies (Heikkinen 2003; Irwin 2000) and ACCP guidelines (Pratter 2006) recommend administration of NSAIDs for coughs caused by a cold. However, this review concluded that there is no clear evidence that NSAIDs are effective for coughs caused by a cold.

Authors' conclusions

Implications for practice

Non-steroidal anti-inflammatory drugs (NSAIDs) are somewhat effective in relieving discomfort caused by a cold but there is no clear evidence of their effect in easing respiratory symptoms. The balance of benefit and harms needs to be considered when using NSAIDs for colds.

Implications for research

We were unable to support the theory that NSAIDs are effective in reducing cough based upon the data included in this review. A large trial to study the effect of NSAIDs on colds may make this relationship clearer. For analgesic effects on the common cold, acetaminophen is also frequently used along with NSAIDs. However, in this review, we did not examine which of these treatments was superior in terms of effect or safety. For this evaluation, we consider another review necessary.

Acknowledgements

We would like to acknowledge the helpful comments of the panel of experts who refereed our review. We are grateful to Liz Dooley and Hayley Edmonds, Cochrane ARI Group Managing Editor and former Assistant Managing Editor. We also wish to thank the following people for commenting on the draft review: Tracey Lloyd, Owen Hendley, Rick Shoemaker and Bruce Arroll.

Data and analyses

Download statistical data

Comparison 1. NSAIDs versus placebo, global effect
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Sum of overall symptom score (fixed-effect model)3293Std. Mean Difference (IV, Fixed, 95% CI)-0.26 [-0.49, -0.03]
2 Sum of overall symptom score (random-effects model)3293Std. Mean Difference (IV, Random, 95% CI)-0.40 [-1.03, 0.24]
3 Moderate to marked severity140Risk Ratio (M-H, Fixed, 95% CI)0.61 [0.18, 2.11]
4 Duration of colds (fixed-effect model)2214Mean Difference (IV, Fixed, 95% CI)-0.24 [-0.93, 0.44]
5 Duration of colds (random-effects model)2214Mean Difference (IV, Random, 95% CI)-0.23 [-1.75, 1.29]
6 Duration of restriction of daily activities1174Mean Difference (IV, Fixed, 95% CI)-0.56 [-1.24, 0.12]
Analysis 1.1.

Comparison 1 NSAIDs versus placebo, global effect, Outcome 1 Sum of overall symptom score (fixed-effect model).

Analysis 1.2.

Comparison 1 NSAIDs versus placebo, global effect, Outcome 2 Sum of overall symptom score (random-effects model).

Analysis 1.3.

Comparison 1 NSAIDs versus placebo, global effect, Outcome 3 Moderate to marked severity.

Analysis 1.4.

Comparison 1 NSAIDs versus placebo, global effect, Outcome 4 Duration of colds (fixed-effect model).

Analysis 1.5.

Comparison 1 NSAIDs versus placebo, global effect, Outcome 5 Duration of colds (random-effects model).

Analysis 1.6.

Comparison 1 NSAIDs versus placebo, global effect, Outcome 6 Duration of restriction of daily activities.

Comparison 2. NSAIDs versus placebo, analgesic effect
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Throat irritation score (fixed-effect model)2159Std. Mean Difference (IV, Fixed, 95% CI)-0.01 [-0.33, 0.30]
2 Throat irritation score (random-effects model)2159Std. Mean Difference (IV, Random, 95% CI)-0.01 [-0.33, 0.30]
3 Headache score (fixed-effect model)2159Std. Mean Difference (IV, Fixed, 95% CI)-0.65 [-0.97, -0.32]
4 Headache score (random-effects model)2159Std. Mean Difference (IV, Random, 95% CI)-0.65 [-1.11, -0.19]
5 Score of pain in muscle/joints score (fixed-effect model)2114Std. Mean Difference (IV, Fixed, 95% CI)-0.40 [-0.77, -0.03]
6 Score of pain in muscle/joints (random-effects model)2114Std. Mean Difference (IV, Random, 95% CI)-0.42 [-0.86, 0.01]
7 Malaise score (fixed-effect model)2159Std. Mean Difference (IV, Fixed, 95% CI)-0.29 [-0.60, 0.03]
8 Malaise score (random-effects model)2159Std. Mean Difference (IV, Random, 95% CI)-0.29 [-0.60, 0.03]
9 Chilliness score (fixed-effect model)2159Std. Mean Difference (IV, Fixed, 95% CI)-0.02 [-0.34, 0.30]
10 Chilliness score (random-effects model)2159Std. Mean Difference (IV, Random, 95% CI)-0.03 [-1.12, 1.06]
11 Nose irritation score180Std. Mean Difference (IV, Fixed, 95% CI)-0.04 [-0.48, 0.40]
12 Score of pain on swallowing180Std. Mean Difference (IV, Fixed, 95% CI)-0.07 [-0.51, 0.37]
13 Eye itching score180Std. Mean Difference (IV, Fixed, 95% CI)-0.14 [-0.58, 0.30]
14 Earache score180Std. Mean Difference (IV, Fixed, 95% CI)-0.59 [-1.04, -0.14]
Analysis 2.1.

Comparison 2 NSAIDs versus placebo, analgesic effect, Outcome 1 Throat irritation score (fixed-effect model).

Analysis 2.2.

Comparison 2 NSAIDs versus placebo, analgesic effect, Outcome 2 Throat irritation score (random-effects model).

Analysis 2.3.

Comparison 2 NSAIDs versus placebo, analgesic effect, Outcome 3 Headache score (fixed-effect model).

Analysis 2.4.

Comparison 2 NSAIDs versus placebo, analgesic effect, Outcome 4 Headache score (random-effects model).

Analysis 2.5.

Comparison 2 NSAIDs versus placebo, analgesic effect, Outcome 5 Score of pain in muscle/joints score (fixed-effect model).

Analysis 2.6.

Comparison 2 NSAIDs versus placebo, analgesic effect, Outcome 6 Score of pain in muscle/joints (random-effects model).

Analysis 2.7.

Comparison 2 NSAIDs versus placebo, analgesic effect, Outcome 7 Malaise score (fixed-effect model).

Analysis 2.8.

Comparison 2 NSAIDs versus placebo, analgesic effect, Outcome 8 Malaise score (random-effects model).

Analysis 2.9.

Comparison 2 NSAIDs versus placebo, analgesic effect, Outcome 9 Chilliness score (fixed-effect model).

Analysis 2.10.

Comparison 2 NSAIDs versus placebo, analgesic effect, Outcome 10 Chilliness score (random-effects model).

Analysis 2.11.

Comparison 2 NSAIDs versus placebo, analgesic effect, Outcome 11 Nose irritation score.

Analysis 2.12.

Comparison 2 NSAIDs versus placebo, analgesic effect, Outcome 12 Score of pain on swallowing.

Analysis 2.13.

Comparison 2 NSAIDs versus placebo, analgesic effect, Outcome 13 Eye itching score.

Analysis 2.14.

Comparison 2 NSAIDs versus placebo, analgesic effect, Outcome 14 Earache score.

Comparison 3. NSAIDs versus placebo, non-analgesic effect
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Cough score (fixed-effect model)2159Std. Mean Difference (IV, Fixed, 95% CI)-0.05 [-0.36, 0.27]
2 Cough score (random-effects model)2159Std. Mean Difference (IV, Random, 95% CI)-0.05 [-0.66, 0.56]
3 Sneezing score (fixed-effect model)2159Std. Mean Difference (IV, Fixed, 95% CI)-0.44 [-0.75, -0.12]
4 Sneezing score (random-effects model)2159Std. Mean Difference (IV, Random, 95% CI)-0.44 [-0.75, -0.12]
5 Total number of sneezes180Std. Mean Difference (IV, Fixed, 95% CI)-0.51 [-0.95, -0.06]
6 Rhinorrhoea score (fixed-effect model)3199Std. Mean Difference (IV, Fixed, 95% CI)0.03 [-0.25, 0.30]
7 Rhinorrhoea score (random-effects model)3199Std. Mean Difference (IV, Random, 95% CI)0.03 [-0.25, 0.30]
8 Nasal obstruction score (fixed-effect model)3199Std. Mean Difference (IV, Fixed, 95% CI)-0.15 [-0.43, 0.13]
9 Nasal obstruction score (random-effects model)3199Std. Mean Difference (IV, Random, 95% CI)-0.16 [-0.46, 0.14]
10 Nasal obstruction score > 5127Risk Ratio (M-H, Fixed, 95% CI)5.36 [0.28, 102.12]
11 Total number of nose blows180Std. Mean Difference (IV, Fixed, 95% CI)0.17 [-0.27, 0.61]
12 Total mucus weight140Std. Mean Difference (IV, Fixed, 95% CI)0.13 [-0.49, 0.76]
13 Total tissue number count140Std. Mean Difference (IV, Fixed, 95% CI)-0.20 [-0.83, 0.42]
14 Score of dryness in the nose180Std. Mean Difference (IV, Fixed, 95% CI)0.04 [-0.40, 0.48]
15 Score of reduced sense of smell180Std. Mean Difference (IV, Fixed, 95% CI)0.08 [-0.36, 0.51]
16 Hoarseness score180Std. Mean Difference (IV, Fixed, 95% CI)0.32 [-0.12, 0.76]
17 Fatigue score180Std. Mean Difference (IV, Fixed, 95% CI)0.18 [-0.26, 0.62]
Analysis 3.1.

Comparison 3 NSAIDs versus placebo, non-analgesic effect, Outcome 1 Cough score (fixed-effect model).

Analysis 3.2.

Comparison 3 NSAIDs versus placebo, non-analgesic effect, Outcome 2 Cough score (random-effects model).

Analysis 3.3.

Comparison 3 NSAIDs versus placebo, non-analgesic effect, Outcome 3 Sneezing score (fixed-effect model).

Analysis 3.4.

Comparison 3 NSAIDs versus placebo, non-analgesic effect, Outcome 4 Sneezing score (random-effects model).

Analysis 3.5.

Comparison 3 NSAIDs versus placebo, non-analgesic effect, Outcome 5 Total number of sneezes.

Analysis 3.6.

Comparison 3 NSAIDs versus placebo, non-analgesic effect, Outcome 6 Rhinorrhoea score (fixed-effect model).

Analysis 3.7.

Comparison 3 NSAIDs versus placebo, non-analgesic effect, Outcome 7 Rhinorrhoea score (random-effects model).

Analysis 3.8.

Comparison 3 NSAIDs versus placebo, non-analgesic effect, Outcome 8 Nasal obstruction score (fixed-effect model).

Analysis 3.9.

Comparison 3 NSAIDs versus placebo, non-analgesic effect, Outcome 9 Nasal obstruction score (random-effects model).

Analysis 3.10.

Comparison 3 NSAIDs versus placebo, non-analgesic effect, Outcome 10 Nasal obstruction score > 5.

Analysis 3.11.

Comparison 3 NSAIDs versus placebo, non-analgesic effect, Outcome 11 Total number of nose blows.

Analysis 3.12.

Comparison 3 NSAIDs versus placebo, non-analgesic effect, Outcome 12 Total mucus weight.

Analysis 3.13.

Comparison 3 NSAIDs versus placebo, non-analgesic effect, Outcome 13 Total tissue number count.

Analysis 3.14.

Comparison 3 NSAIDs versus placebo, non-analgesic effect, Outcome 14 Score of dryness in the nose.

Analysis 3.15.

Comparison 3 NSAIDs versus placebo, non-analgesic effect, Outcome 15 Score of reduced sense of smell.

Analysis 3.16.

Comparison 3 NSAIDs versus placebo, non-analgesic effect, Outcome 16 Hoarseness score.

Analysis 3.17.

Comparison 3 NSAIDs versus placebo, non-analgesic effect, Outcome 17 Fatigue score.

Comparison 4. NSAIDs versus placebo, adverse effects
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Overall side effects (fixed-effect model)2220Risk Ratio (M-H, Fixed, 95% CI)2.88 [1.11, 7.45]
2 Overall side effects (random-effects model)2220Risk Ratio (M-H, Random, 95% CI)2.94 [0.51, 17.03]
3 GI complaint (fixed-effect model)3189Risk Ratio (M-H, Fixed, 95% CI)0.76 [0.17, 3.32]
4 GI complaint (random-effects model)3189Risk Ratio (M-H, Random, 95% CI)0.73 [0.14, 3.68]
5 Lethargy/drowsiness (fixed-effect model)2110Risk Ratio (M-H, Fixed, 95% CI)1.0 [0.14, 6.91]
6 Lethargy/drowsiness (random-effects model)2110Risk Ratio (M-H, Random, 95% CI)1.00 [0.11, 9.28]
7 Feeling hyperactive146Risk Ratio (M-H, Fixed, 95% CI)3.0 [0.13, 70.02]
8 Feeling more awake146Risk Ratio (M-H, Fixed, 95% CI)3.0 [0.13, 70.02]
9 Flushed face146Risk Ratio (M-H, Fixed, 95% CI)3.0 [0.13, 70.02]
10 Difficulty sleeping146Risk Ratio (M-H, Fixed, 95% CI)0.33 [0.01, 7.78]
11 Light-headedness146Risk Ratio (M-H, Fixed, 95% CI)1.0 [0.15, 6.51]
12 Dry mouth146Risk Ratio (M-H, Fixed, 95% CI)3.0 [0.13, 70.02]
Analysis 4.1.

Comparison 4 NSAIDs versus placebo, adverse effects, Outcome 1 Overall side effects (fixed-effect model).

Analysis 4.2.

Comparison 4 NSAIDs versus placebo, adverse effects, Outcome 2 Overall side effects (random-effects model).

Analysis 4.3.

Comparison 4 NSAIDs versus placebo, adverse effects, Outcome 3 GI complaint (fixed-effect model).

Analysis 4.4.

Comparison 4 NSAIDs versus placebo, adverse effects, Outcome 4 GI complaint (random-effects model).

Analysis 4.5.

Comparison 4 NSAIDs versus placebo, adverse effects, Outcome 5 Lethargy/drowsiness (fixed-effect model).

Analysis 4.6.

Comparison 4 NSAIDs versus placebo, adverse effects, Outcome 6 Lethargy/drowsiness (random-effects model).

Analysis 4.7.

Comparison 4 NSAIDs versus placebo, adverse effects, Outcome 7 Feeling hyperactive.

Analysis 4.8.

Comparison 4 NSAIDs versus placebo, adverse effects, Outcome 8 Feeling more awake.

Analysis 4.9.

Comparison 4 NSAIDs versus placebo, adverse effects, Outcome 9 Flushed face.

Analysis 4.10.

Comparison 4 NSAIDs versus placebo, adverse effects, Outcome 10 Difficulty sleeping.

Analysis 4.11.

Comparison 4 NSAIDs versus placebo, adverse effects, Outcome 11 Light-headedness.

Analysis 4.12.

Comparison 4 NSAIDs versus placebo, adverse effects, Outcome 12 Dry mouth.

Comparison 5. Head to head comparison, global effect
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Global improvement rating, marked improvement (fixed-effect model)2365Risk Ratio (M-H, Fixed, 95% CI)1.52 [0.99, 2.34]
2 Global improvement rating, marked improvement (random-effects model)2365Risk Ratio (M-H, Random, 95% CI)1.53 [1.00, 2.35]
3 Global improvement rating, moderate to marked improvement (fixed-effect model)2365Risk Ratio (M-H, Fixed, 95% CI)1.20 [1.02, 1.41]
4 Global improvement rating, moderate to marked improvement (random-effects model)2365Risk Ratio (M-H, Random, 95% CI)1.20 [0.96, 1.50]
Analysis 5.1.

Comparison 5 Head to head comparison, global effect, Outcome 1 Global improvement rating, marked improvement (fixed-effect model).

Analysis 5.2.

Comparison 5 Head to head comparison, global effect, Outcome 2 Global improvement rating, marked improvement (random-effects model).

Analysis 5.3.

Comparison 5 Head to head comparison, global effect, Outcome 3 Global improvement rating, moderate to marked improvement (fixed-effect model).

Analysis 5.4.

Comparison 5 Head to head comparison, global effect, Outcome 4 Global improvement rating, moderate to marked improvement (random-effects model).

Appendices

Appendix 1. EMBASE search strategy

/* COMMON COLD */

#1 'common cold'/exp OR (common cold*):ti,ab

#2 coryza:ti,ab

#3 ('upper respiratory infection'):ti,ab OR ('upper respiratory infections'):ti,ab

#4 ('upper respiratory tract infection'):ti,ab OR ('upper respiratory tract infections'):ti,ab

#5 urti:ti,ab

#6 ((respiratory tract infection:ti,ab) OR (respiratory tract infections:ti,ab)) AND upper:ti,ab

#7 'rhinitis'/exp OR rhinit*:ti,ab

#8 'pharyngitis'/exp OR pharyngit*:ti,ab

#9 'sore throat':ti,ab OR 'sore throats':ti,ab

#10 'rhinopharyngitis'/exp OR nasopharyngit*:ti,ab

#11 'laryngitis'/exp OR laryngit*:ti,ab

#12 'coughing'/exp OR cough*:ti,ab

#13 'nose obstruction'/exp OR 'nasal obstruction':ti,ab

#14 'sneezing'/exp OR sneez*:ti,ab

#15 'rhinovirus'/exp OR rhinovirus:ti,ab

#16 OR/#1-#15

/* NSAIDS */

#17 'nonsteroid antiinflammatory agent'/exp OR nsaid*:ti,ab OR (((non-steroid OR nonsteroid OR 'non steroid' OR 'non steroids') AND (anti-inflammatory OR antiinflammatory OR 'anti inflammatory')):ti,ab)

#18 'azapropazone'/exp OR apazone:ti,ab

#19 'acetylsalicylic acid'/exp OR aspirin:ti,ab

#20 'celecoxib'/exp OR celecoxib:ti,ab

#21 'diclofenac'/exp OR diclofenac:ti,ab

#22 'diflunisal'/exp OR diflunisal:ti,ab

#23 'etodolac'/exp OR etodolac:ti,ab

#24 'fenoprofen'/exp OR fenoprofen:ti,ab

#25 'flurbiprofen'/exp OR flurbiprofen:ti,ab

#26 'ibuprofen'/exp OR ibuprofen:ti,ab

#27 'indometacin'/exp OR indomethacin:ti,ab

#28 'ketoprofen'/exp OR ketoprofen:ti,ab

#29 'ketorolac'/exp OR ketorolac:ti,ab

#30 'meclofenamic acid'/exp OR meclofenamate:ti,ab

#31 'meloxicam'/exp OR meloxicam:ti,ab

#32 'salicylic acid methyl ester'/exp OR methylsalicylate:ti,ab OR 'methyl salicylate':ti,ab

#33 'nabumetone'/exp OR nabumetone:ti,ab

#34 'naproxen'/exp OR naproxen:ti,ab

#35 'nimesulide'/exp OR nimesulide:ti,ab

#36 'oxaprozin'/exp OR oxaprozin:ti,ab

#37 'phenylbutazone'/exp OR phenylbutazone:ti,ab

#38 'piroxicam'/exp OR piroxicam:ti,ab

#39 'salicylic acid'/exp OR salicylate:ti,ab

#40 'sulindac'/exp OR sulindac:ti,ab

#41 'tenoxicam'/exp OR tenoxicam:ti,ab

#42 'tolmetin'/exp OR tolmetin:ti,ab

#43 OR/#17-#42 

/* RCT */

#44 'clinical trial'/exp OR 'clinical trial':ti,ab

#45 'randomized controlled trial'/exp OR 'randomized controlled trial':ti,ab

#46 'randomisation'/exp OR random*:ti,ab

#47 'single blind procedure'/exp OR (singl*:ti,ab AND (mask*:ti,ab OR blind*:ti,ab))

#48 'double blind procedure'/exp OR (doubl*:ti,ab AND (mask*:ti,ab OR blind*:ti,ab))

#49 'triple blind procedure'/exp OR (trip*:ti,ab AND (mask*:ti,ab OR blind*:ti,ab))

#50 'placebo'/exp OR placebo:ti,ab

#51 OR #44-#50

/* Combine & Limit */

#52 #16 AND #43 AND #51

#53 #16 AND #43 AND [randomized controlled trial]/lim

#54 (#52 OR #53) AND [human]/lim

#55 #54 AND [2009-2011]/py

Appendix 2. CINAHL search strategy

/* COMMON COLD */

S1 (MH "Common Cold") OR (TX "common cold*")

S2 TX coryza

S3 (MH "Respiratory Tract Infections") or TX "upper respiratory infection*"

S4 TX "upper respiratory tract infection*"

S5 TX URTI

S6 (TX "respiratory tract infection*") AND (TX upper)

S7 (MH "Rhinitis") OR (TX rhinit*)

S8 (MH "Pharyngitis") OR (TX pharyngit*)

S9 TX "sore throat*"

S10 (MH "Nasopharynx") OR (TX nasopharyngit*)

S11 (MH "Laryngitis") OR (TX laryngit*)

S12 (MH "Cough") OR (TX cough*)

S13 (MH "Nasal Obstruction") OR (TX nasal obstruction*)

S14 (MH "Sneezing") OR (TX sneez*)

S15 TX rhinovirus

S16 OR/S1-S15

 

/* NSAIDS */

S17 (MH "Antiinflammatory Agents, Non-Steroidal") OR (TX nsaid*) OR (TX (non-steroid* OR nonsteroid* OR "non steroid*") AND TX (anti-inflammator* OR antiinflammator* OR "anti inflammator*"))

S18 TX azapropazone

S19 MH "Aspirin" OR aspirin

S20 MH "Cox-2 Inhibitors" OR TX celecoxib

S21 MH "Diclofenac" OR TX diclofenac

S22 TX diflunisal

S23 MH "Etodolac" OR TX etodolac

S24 TX fenoprofen

S25 MH "Flurbiprofen" OR TX flurbiprofen

S26 MH "Ibuprofen" OR TX ibuprofen

S27 MH "Indomethacin" OR TX indomethacin

S28 TX ketoprofen

S29 MH "Ketorolac" OR TX ketorolac

S30 TX meclofenamate

S31 TX meloxicam

S32 TX (methylsalicylate OR "methyl salicylate")

S33 TX nabumetone

S34 MH "Naproxen" OR TX naproxen

S35 TX nimesulide

S36 TX oxaprozin

S37 MH "Phenylbutazone" OR TX phenylbutazone

S38 MH "Piroxicam" OR TX piroxicam

S39 MH "Salicylic Acids" OR TX salicylate

S40 MH "Sulindac" OR TX sulindac

S41 TX tenoxicam

S42 MH "Tolmetin" OR TX tolmetin

S43 OR/S11-S42

 

/* RCT */

S44 MH "Clinical trial" OR TX "clinical trial"

S45 MH "Randomized Controlled Trials" OR TX "randomized controlled trial"

S46 MH "Random Sample" OR TX random*

S47 MH "Single-Blind Studies" OR TX (singl* AND (mask* OR blind*))

S48 MH "Double-Blind Studies" OR TX (doubl* AND (mask* OR blind*))

S49 MH "Triple-Blind Studies" OR TX (trilp AND (mask* OR blind*))

S50 MH "Placebos" OR TX placebo

S51 OR S44-S50

 

/* Combine & Limit */

S52 S16 AND S43 AND S51

S53 S16 AND S43 AND [crinical trial]/lim

S54 S52 OR S53

S55 S54 AND [2009-2011]/py

Appendix 3. MEDLINE and CENTRAL search strategy

MEDLINE (Ovid)

1 Common Cold/
2 common cold*.tw.
3 coryza.tw.
4 upper respiratory infection*.tw.
5 upper respiratory tract infections*.tw.
6 urti.tw.
7 respiratory tract infections.sh. and upper.tw.
8 Rhinitis/
9 rhinit*.tw.
10 exp Pharyngitis/
11 pharyngit*.tw.
12 sore throat*.tw.
13 exp Nasopharyngitis/
14 nasopharyngit*.tw.
15 exp Laryngitis/
16 laryngit*.tw.
17 Cough/
18 cough*.tw.
19 Nasal Obstruction/
20 nasal obstruction*.tw.
21 Sneezing/
22 sneez*.tw.
23 Rhinovirus/
24 rhinovirus*.tw.
25 or/1-24
26 exp Anti-Inflammatory Agents, Non-Steroidal/
27 nsaid*.tw.
28 ((non-steroid* or nonsteroid* or non steroid*) and (anti-inflammator* or antiinflammator* or anti inflammator*)).tw.
29 Apazone.sh. or apazone.tw.
30 Aspirin.sh. or aspirin.tw.
31 celecoxib.nm. or celecoxib.tw.
32 diclofenac.sh. or diclofenac.tw.
33 diflunisal.sh. or diflunisal.tw.
34 etodolac.sh. or etodolac.tw.
35 fenoprofen.sh. or fenoprofen.tw.
36 flurbiprofen.sh. or flurbiprofen.tw.
37 ibuprofen.sh. or ibuprofen.tw.
38 indomethacin.sh. or indomethacin.tw.
39 ketoprofen.sh. or ketoprofen.tw.
40 ketorolac.sh. or ketorolac.tw.
41 Meclofenamic Acid/
42 meclofenamate.tw. or meloxicam.nm. or meloxicam.tw.
43 methyl salicylate.nm. or methylsalicylate.tw. or methyl salicylate.tw.
44 nabumetone.nm. or nabumetone.tw.
45 naproxen.sh. or naproxen.tw.
46 nimesulide.nm. or nimesulide.tw.
47 oxaprozin.nm. or oxaprozin.tw.
48 phenylbutazone.sh. or phenylbutazone.tw.
49 piroxicam.sh. or piroxicam.tw.
50 salicylate.mp.
51 sulindac.sh. or sulindac.tw.
52 tenoxicam.nm. or tenoxicam.tw.
53 tolmetin.sh. or tolmetin.tw.
54 or/26-53
55 25 and 54

Feedback

Non-steroidal anti-inflammatory drugs for the common cold, 8 December 2009

Summary

In their Cochrane Review on non-steroidal anti-inflammatory drugs for the common cold, Kim et al. (1) "recommend NSAIDs for relieving discomfort or pain caused by the common cold” without any reservations. However, the common cold is a rather harmless condition, whereas NSAIDs can have serious and even lethal adverse effects (2-4). The review also has methodological shortcomings.

One problem is the excessive number of outcomes; the review authors report on no less than 26 primary outcomes. Four of these, sneezing, headache, pain in muscles/joints and earache, were statistically significant, but the first 3 outcomes were based on only 2 trials, including 159 participants, and the last outcome on only 1 trial.

One of these 2 trials was an experimental study (5) of 87 healthy volunteers that were inoculated with rhinovirus. The trial had unclear sequence generation, unclear concealment of allocation and was not analysed using intention to treat, as 8 people were excluded from the analysis. The volunteers were treated with very high doses of naproxen, up to 1500 mg daily, which is higher than what has been approved for treatment of acute pain conditions (6), and as the risk of harms increases linearly with the dose (7),  this is particularly problematic. This trial is also included in the analysis for global effects where it had the largest effect of the 3 included trials and contributed to substantial heterogeneity, which suggests bias or problems with generalisability. Further, as it can be problematic to generalise findings from experimental settings to patients (8), it is questionable to pool this trial with trials from clinical settings.

The second trial (9) included 80 patients with natural colds that received 1200 mg ibuprofen daily. This trial also had unclear sequence generation and unclear allocation concealment. Additionally, for analysis 2.9 and 2.10 of Chilliness score, the authors have erroneously extracted the results from the placebo arm of this trial as though they belonged to another trial (5) and vice versa. This raises the question whether there were other data extraction errors. Data extraction errors are frequent in meta-analyses using SMD (10).

Adverse effects are not mentioned in the Discussion and only briefly in Results. According to the authors, 5 trials assessed adverse effects but they only reported data from 4 trials. The omitted trial (9) reported adverse effects (e.g. pain in abdomen, ear buzzing) as continuous outcomes, and not as binary (5). While it is reasonable not to pool trials with binary and continuous outcomes, we are puzzled as to why the authors omitted reporting any adverse effects data from this trial in their Cochrane Review. We wonder whether adverse effects from other trials were similarly ignored.

The 4 trials where the review authors reported adverse effects assessed 9 outcomes and for all outcomes, the confidence intervals were wide (e.g. for overall adverse effects, RR 2.94 [0.51, 17.03]). Based on this uncertainty, adverse effects of NSAIDs cannot be dismissed and it is therefore surprising that the authors did not refer to additional evidence, as recommended in The Cochrane Handbook (11). NSAIDs are known to cause serious harms (2-4).

Additionally, in Methods the authors state “We assessed heterogeneity amongst trials by using the Chi2 test for heterogeneity with a 10% level of statistical significance and I2 test.” In their protocol the I2 is not mentioned at all. While there was substantial heterogeneity for overall side effects (I2 = 58%) the Chi2 test for heterogeneity was not statistically significant (P = 0.12). So, based on their own criteria the authors should have analysed the data using a fixed-effect model, which would have shown a significant increase in overall side effects, relative risk 2.88 [1.11, 7.45] (P = 0.03).

Additionally, there are some discrepancies between what was reported in the protocol and what was done in the review. Kim et al. originally stated in their protocol (1) that they would search databases for unpublished trials, contact authors for missing data and examine publication bias, but apparently did not do any of this. The identified trials were all very small. It is therefore likely that the identified sample of published trials is biased (12), as small trials with non-statistical findings are often not published.

In their abstract, the authors recommend NSAIDs for “reliving discomfort or pain”. This statement is highly misleading, as it indicates that NSAIDs have other clinical effects than their analgesic effect. The authors do not use the word “discomfort” anywhere else in the review, but we assume it refers to either global outcomes or non-analgesic outcomes. However, the authors found no effect on global outcomes and the effect on “sneezing” is likely spurious, as it occurred for only one out of 13 non-analgesic outcomes, and was based on the 2 problematic trials already described.

Based on these methodological problems, and the serious adverse effects of the drugs, we believe there is no sound basis for recommending NSAIDs for the common cold and urge the authors to present a more balanced view.

References

1) Kim SY, Chang YJ, Cho HM, Hwang YW, Moon YS. Non-steroidal anti-inflammatory drugs for the common cold. Cochrane Database of Systematic Reviews 2009, Issue 3. Art. No.: CD006362. DOI: 10.1002/14651858.CD006362.pub2.
2) Hernández-Díaz S, Rodríguez LA. Association between nonsteroidal anti-inflammatory drugs and upper gastrointestinal tract bleeding/perforation: an overview of epidemiologic studies published in the 1990s. Arch Intern Med. 2000;160:2093-9.
3) Fosbøl EL, Gislason GH, Jacobsen S, Folke F, Hansen ML, Schramm TK, Sørensen R, Rasmussen JN, Andersen SS, Abildstrom SZ, Traerup J, Poulsen HE, Rasmussen S, Køber L, Torp-Pedersen C. Risk of myocardial infarction and death associated with the use of nonsteroidal anti-inflammatory drugs (NSAIDs) among healthy individuals: a nationwide cohort study. Clin Pharmacol Ther. 2009;85:190-7.
4) Gislason GH, Rasmussen JN, Abildstrom SZ, Schramm TK, Hansen ML, Fosbøl EL, Sørensen R, Folke F, Buch P, Gadsbøll N, Rasmussen S, Poulsen HE, Køber L, Madsen M, Torp-Pedersen C. Increased mortality and cardiovascular morbidity associated with use of nonsteroidal anti-inflammatory drugs in chronic heart failure. Arch Intern Med. 2009;169:141-9. 
5) Sperber SJ, Hendley JO, Hayden FG, Riker DK, Sorrentino JV, Gwaltney JM Jr. Effects of naproxen on experimental rhinovirus colds. A randomised, double-blind, controlled trial Annals of Internal Medicine. 1992;117:3741.
6) Naproxen. DrugDex ® Evaluations.Thomson Micromedex. Modified: 12 July 2009. [http://www.micromedex.dk/hcs/librarian/ND_T/HCS/ND_PR/Main/CS/726463/DUPLICATIONSHIELDSYNC/2A5F1A/ND_PG/PRIH/ND_B/HCS/SBK/4/ND_P/Main/PFActionId/hcs.common.RetrieveDocumentCommon/DocId/0004/ContentSetId/31#TopOfPage] (Accessed 19 November 2009)
7) Gøtzsche PC. NSAIDs. In: Young C, ed. Clinical Evidence Handbook. London: BMJ Publishing Group Limited, June 2009:384-5.
8) Rothwell PM. External validity of randomized controlled trials: "to whom do the results of this trial apply?". Lancet. 2005;365:82-93.
9) Winther B, Mygind N. The therapeutic effectiveness of ibuprofen on the symptoms of naturally acquired common colds. American Journal of Rhinology. 2001;15:23942.
10) Gøtzsche PC, Hróbjartsson A, Maric K, Tendal B. Data extraction errors in meta-analyses that use standardized mean differences. JAMA. 2007;298:430-7. Erratum in: JAMA. 2007;298:2264.
11) Higgins JPT, Green S (editors). Cochrane Handbook for Systematic Reviews of Interventions Version 5.0.2 [updated September 2009]. The Cochrane Collaboration, 2008. [www.cochrane-handbook.org] (Accessed 19 November 2009)
12) Hopewell S, Loudon K, Clarke MJ, Oxman AD, Dickersin K. Publication bias in clinical trials due to statistical significance or direction of trial results. Cochrane Database of Systematic Reviews 2009, Issue 1. Art. No.: MR000006. DOI: 10.1002/14651858.MR000006.pub3.

Reply

Thank you for your feedback.
I think that the comments fall into four main areas.

1. Adverse effects.
2. Discrepancy between the protocol and review.
3. The methodological issues of weak studies, and multiple outcomes.
4. The heterogeneity tests used, and the choice of random-effects or fixed-effect model.

We will discuss the feedback according to the four main areas.

1. Adverse effect issues
Safety-related issues of NSAIDs, in particular, the issues of cardiovascular disease and gastrointestinal disease have been reviewed in many studies but no clear conclusion has been drawn on what problems there can be in short-term uses such as the use for a common cold. Of course, the risk of gastrointestinal side effects may increase even in short-term usage.
We agree with the commentators that there have not been many safety-related discussions in the review and the power is not high enough to conclude on the safety of NSAIDs based on trials in this review. We also agree that review of other systematic reviews related to safety issues is necessary.
As for the trial on which the commentators stated that it omitted safety-related results, the trial author mentioned that there was no abnormal adverse events in the trial and the outcomes mentioned by the commentators classified it as effectiveness outcomes.

2. Discrepancy between the protocol and review - search of unpublished trials and publication bias
In the methods, we did make some efforts to search for unpublished trials.
“We searched reference lists of review articles and of all included studies to find other potentially eligible studies. We contacted authors of the included trials to request unpublished studies”. However, we did not find any additional trials.

We did examine publication bias by funnel plot analysis. We omitted them because there were too many funnel plots in our review.

3. The methodological issues of weak studies, and multiple (26) outcomes
As mentioned by the commentators, the number of results may be too large. This problem is mainly because outcomes of trials and duration or dose of therapy were quite diverse, so it was inevitable (in this sense).
The effect of NSAIDs may not be different according to whether a cold is induced experimentally or happens naturally.
A calculation error that the commentators pointed out was corrected.
We added the following to the Discussion:
“Major limitations of this review is that the results of the research are quite diverse and the number of studies for one result is quite small. For this reason, it is somewhat difficult to draw clear conclusions."

4. The heterogeneity tests used, and the choice of random-effects or fixed-effect model
The reason for changing the protocol and review methodology in connection to heterogeneity is because The Cochrane Handbook was upgraded from 4.2 to 5.0 during the review and the 5.0 version recommends the use of I2 statistic and so we added it. In the Chi2 test, some heterogeneity was observed as I2 statistic = 58%, although not statistically significant, so in the actual analysis we presented both the fixed-effect model and the random-effects model.
For the above reason, we are going to add new text to the Results, Discussion and Conclusions sections.

Results
Two trials assessed the overall side effects of NSAIDs, and there was moderate heterogeneity. The results of a pooled analysis for overall side effects was significant in the fixed-effect model (risk ratio (RR) 2.88 (95% CI 1.11 to 7.45), (P = 0.03), but not in random-effects model (RR 2.94, 95% CI 0.51 to 17.03).
Three trials reported gastrointestinal adverse effects and found no differences between the groups.
Lethargy/drowsiness, feeling hyperactive, feeling more awake, flushed face, difficulty sleeping, light-headedness and dry mouth were reported in one to two trials and the results were not significantly different between the treatment groups.

Discussion
NSAIDs are drawing attention for their side effects. For some NSAIDs, their long-term use increases the risk of cardiovascular disease and may cause gastrointestinal side effects. The frequency of gastrointestinal side effects increases in proportion to the dose and period of medication with NSAIDs but the risk of gastrointestinal side effects cannot be excluded in short-term use. In trials included in this review, the risk of side effects was not evidently high; it is hard to conclude that they are not different from placebo in terms of side effects.

Conclusion
NSAIDs are somewhat effective in relieving discomfort caused by a cold, but there is no clear evidence of their effect in easing respiratory symptoms. The use of NSAIDs for a cold should be decided in consideration of side effects.

Contributors

Andreas Lundh, Britta Tendal. The Nordic Cochrane Centre, Rigshospitalet, Dept. 3343, Blegdamsvej 9, 2100 Copenhagen Ø, Denmark 

What's new

DateEventDescription
17 April 2013New citation required but conclusions have not changedOur conclusions remain unchanged.
17 April 2013New search has been performedSearches updated. Three new trials were identified and excluded (Chachtel 2011; Ruperto 2011; Russo 2013).

History

Protocol first published: Issue 1, 2007
Review first published: Issue 3, 2009

DateEventDescription
6 April 2011New search has been performedSearches updated. Two new trials were identified and excluded (Azuma 2010; Azuma 2011). Our conclusions remain unchanged.

Contributions of authors

Soo young Kim (SYK), Yoon-Jung Chang (YJC), Ye-won Hwang (YWH) and Yoo Sun Moon (YSM) were responsible for study selection, methodological quality assessment, data extraction and analyses, and writing the review.
Hye Min Cho (HMC) was responsible for the literature search and writing the review.

Declarations of interest

None known.

Characteristics of studies

Characteristics of included studies [author-defined order]

Graham 1990

MethodsDouble-blind, placebo-controlled, experimental colds
Participants59 inoculated; 42 colds. Mean age 20.1 years, 43.3% women, university students
Interventions2 groups: aspirin 4 g/day and ibuprofen 1.2 g/day for 7 days
OutcomesThe proportion of nasal obstruction score > 5 in the aspirin group (6/15) significantly differed from that of the placebo group (0/14, P < 0.05)
Mean mucus weight, mean tissue count, mean overall symptom score and mean overall side effect score were reported but any other statistical parameters such as SD, SE, 95% CI, P value of each group or difference between these groups were not reported
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk"a randomised double-blind, placebo-controlled clinical trial"
Allocation concealment (selection bias)Unclear riskInsufficient information to permit judgement
Blinding (performance bias and detection bias)
All outcomes
Low risk

"identical capsules containing aspirin (500 mg), ibuprofen (200 mg) or placebo"

Comment: probably done

Incomplete outcome data (attrition bias)
All outcomes
Low risk

"4 volunteers who were considered uninfected and were excluded from further analyses"

Comment: probably done

Selective reporting (reporting bias)Unclear riskNo protocol, no convincing text
Other biasLow riskThe study appears to be free of other sources of bias

Goto 2007

MethodsDouble-blind, placebo-controlled, natural colds
Participants174 adults, age 18 to 65 years, 35% women, 23 outpatients facilities, URTI onset 2 days or less
InterventionsLoxoprofen 60 mg 2 times for 7 days
OutcomesDuration of illness, the number of days with limited daily activities were not significantly different between groups
NotesThe primary outcome was duration of illness in days
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk

"Randomisation was based on simple computer-generated random digits"

Comment: probably done

Allocation concealment (selection bias)Low risk

"self-drawing a sealed opaque envelope in the physician’s sight....the correspondence between the digits and the group assignment was held in the central, secured location by a third party independent of the investigators until data collection was completed. Thus, allocation was concealed and masked from both patients and physicians"

Comment: probably done

Blinding (performance bias and detection bias)
All outcomes
Low risk

"A double-blind, randomised, placebo-controlled trial"; "those in the control group were to take a placebo which was quite similar to active loxoprofen in shape and taste"

Comment: probably done

Incomplete outcome data (attrition bias)
All outcomes
Low risk

"six (two in loxoprofen group and four in placebo group) withdrew from the study, because two patients (one in loxoprofen and another in placebo) did not complete the diary; three patients (one in loxoprofen and the others in placebo) did not return the diary; and one patient (placebo) decided not to continue the study after the allocation. We excluded nine more participants (two in loxoprofen and seven in placebo) from analyses"

Comments: probably done (missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups)

Selective reporting (reporting bias)Unclear riskNo protocol, no convincing text
Other biasLow riskThe study appears to be free of other sources of bias

Itoh 1980

MethodsDouble-blind, head-to-head comparison, natural colds
Participants184 adults, mean age, sex not reported for the subgroup of colds, 29 centres, outpatient departments of hospitals and clinics, URTI onset ≤ 3 days
Interventions2 groups: ketoprofen 50 mg 3 times and aspirin 500 mg 3 times for 3 days
OutcomesNo significant difference in FGIR between 2 groups
NotesNo available data on adverse effects for the subgroup of common colds
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskComments: insufficient information about the sequence generation process
Allocation concealment (selection bias)Low risk

"randomisation process was done by two controllers and key codes were kept by controllers (in Japanese)"

Comment: probably done

Blinding (performance bias and detection bias)
All outcomes
Low risk

"double-blind method...active drug capsule and aspirin capsule were quite similar in shape (in Japanese)"

Comment: probably done

Incomplete outcome data (attrition bias)
All outcomes
Low risk

"91/93 cases in ketoprofen group and 89/91 cases in aspirin group were finally analysed"

Number of withdrawals was too small to make any important difference to the estimated intervention effect

Comment: probably done

Selective reporting (reporting bias)Unclear riskNo protocol, no convincing text
Other biasLow riskThe study appears to be free of other sources of bias

Katsu 1993

MethodsDouble-blind, double-dummy, head-to-head comparison, natural colds
Participants167 adults, mean age, sex not reported for the subgroup of colds, 32 centres, outpatient departments of hospitals and clinics, moderate to severe URTI, not requiring antibiotics
Interventions2 groups: loxoprofen 180 mg/day and ibuprofen 600 mg for 3 days
OutcomesNo significant difference in FGIR between 2 groups
NotesNo available data on adverse effects for the subgroup of common colds
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

"were randomly assigned to receive"

Comments: insufficient information about the sequence generation process

Allocation concealment (selection bias)Unclear riskComments: insufficient information about the allocation concealment
Blinding (performance bias and detection bias)
All outcomes
Low risk

"double-blind, double-dummy method...active drug and placebo were quite similar in shape (in Japanese)"

Comment: probably done

Incomplete outcome data (attrition bias)
All outcomes
Unclear risk

"112/130 of the CS-600 group and 113/132 group were evaluated in the assessment improvement ratings"

There are no reasons for missing participants. Insufficient reporting of attrition/exclusions to permit judgement

Selective reporting (reporting bias)Unclear riskNo protocol, no convincing text
Other biasLow riskThe study appears to be free of other sources of bias

Nagaoka 1980

MethodsDouble-blind, head-to-head comparison, natural colds
Participants222 adults, sex not reported for the subgroup of colds, 51 centres, outpatient departments of hospitals and clinics, URTI onset ≤ 2 days and fever ≤ 39°C
Interventions2 groups: fentiazac 300 mg/day and ibuprofen 600 mg/day for 3 days
OutcomesModerate to marked improvement of FGIR was more frequent in the fenoprofen group than the placebo (P < 0.05)
NotesNo available data on adverse effects for the subgroup of common colds
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskComments: insufficient information about the sequence generation process
Allocation concealment (selection bias)Low risk"randomisation process was done by two controllers and key codes were kept by controllers (in Japanese)"
Blinding (performance bias and detection bias)
All outcomes
Low risk

"double-blind, double-dummy method...active drug and placebo were quite similar in shape (in Japanese)"

Comment: probably done

Incomplete outcome data (attrition bias)
All outcomes
Low risk

"243 out of 244 patients were analysed after the elimination of 1 drop-out case"

Number of withdrawals was too small to make any important difference to the estimated intervention effect

Comment: probably done

Selective reporting (reporting bias)Unclear riskNo protocol, no convincing text
Other biasLow riskThe study appears to be free of other sources of bias

Ryan 1987

MethodsDouble-blind, placebo-controlled, natural colds
Participants64 adults, age range 18 to 60 years, 75% women, single family centre, fever ≤ 37.8°C with moderate pain due to malaise/aches
InterventionsFenoprofen 200 mg single dose
OutcomesNo available data on efficacy
NotesOnly 2 adverse effects (1 stomach discomfort and 1 drowsiness), both in the fenoprofen group
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk"assigned to one of three treatment groups via a computer-generated random table"
Allocation concealment (selection bias)Low risk

"Each dose of medication was dispensed in identically appearing capsules"

Single oral dose was given

Comment: probably done

Blinding (performance bias and detection bias)
All outcomes
Low risk

"Each dose of medication was dispensed in identically appearing capsules in double-blind method"

Single oral dose was given

Comment: probably done

Incomplete outcome data (attrition bias)
All outcomes
Low riskAll participants who entered the study completed treatment and were included in the assessment of effectiveness and side effects
Selective reporting (reporting bias)Unclear riskNo protocol, no convincing text
Other biasLow riskThe study appears to be free of other sources of bias

Sperber 1989

MethodsDouble-blind, placebo-controlled, experimental colds
Participants40 inoculated, 31 colds, mean age 21 years, 39.1% women, setting not reported, fever ≤ 37.7°C
InterventionsIbuprofen 200 mg, 2 doses for the first day and 4 doses for the subsequent 4 days
Outcomes4-point scale. Moderate to marked severity (2- to 3-point) was reduced in the ibuprofen group (18% versus 29%) but statistical significance was not reported
NotesAdverse effects were slightly more frequent in the ibuprofen group (6/23) than in the control group (4/23)
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

"were randomly assigned to receive"

Comments: insufficient information about the sequence generation process

Allocation concealment (selection bias)Unclear riskComments: insufficient information about the allocation concealment
Blinding (performance bias and detection bias)
All outcomes
Low risk

"two identically appearing capsules"

Comment: probably done

Incomplete outcome data (attrition bias)
All outcomes
Low riskAmong 58 inoculated subjects, 8 were excluded (7 not infected, 1 infected with wild type virus), 1 was withdrawn
Selective reporting (reporting bias)Unclear riskNo protocol, no convincing text
Other biasLow riskThe study appears to be free of other sources of bias

Sperber 1992

MethodsDouble-blind, placebo-controlled, experimental colds
Participants79 inoculated (first cohort 34, second cohort 24 and third cohort 21); 56 colds. Mean age 21.4 years, 52% women. Setting not reported
InterventionsFor first cohort, naproxen loading dose of 400 mg followed by 200 mg 3 times daily, and for second and third cohort, naproxen loading dose of 500 mg followed by 500 mg 3 times daily for 5 days
Outcomes5-point symptom score. Total cumulative 5-day score for headache was lower in the naproxen group (0.5 versus 2.5, P < 0.001)
Notes1 in the naproxen group and 2 in the placebo group experienced gastrointestinal complaints
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

"Participants were randomly assigned to receive..."

Comments: insufficient information about the sequence generation process

Allocation concealment (selection bias)Unclear riskComments: insufficient information about the allocation concealment
Blinding (performance bias and detection bias)
All outcomes
Low risk

"The study drug and placebo were supplied in identically appearing capsules"

Comment: probably done

Incomplete outcome data (attrition bias)
All outcomes
Low risk

"Among 87 volunteers completed, 79 were considered evaluable"

The reason for exclusion (infected with wild type rhinovirus, not infected, missed dose of study drug) is unlikely to be related to the outcome of the trial (symptomatic improvement of common cold symptoms)

Comment: probably done

Selective reporting (reporting bias)Unclear riskNo protocol, no convincing text
Other biasLow riskThe study appears to be free of other sources of bias

Winther 2001

  1. a

    CI: confidence interval
    FGIR: final global improvement rating
    SD: standard deviation
    SE: standard error
    URTI: upper respiratory tract infection

MethodsDouble-blind, placebo-controlled, natural colds
Participants80 adults, mean age 30.1 years, 60% women, single centre, medical students and members of the staff at the university
InterventionsIbuprofen 400 mg 3 times for 3 days
Outcomes4-point symptom score by patients. Sneezing, earache, headache, and pain in muscles and joints were significantly reduced in the ibuprofen group compared with placebo group. Number of sneezing episodes was also reduced (21.33 ± 3.3 and 12.44 ± 1.5, P = 0.02)
NotesNo adverse effects in either group
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk

"randomised study of two parallel groups"

Comments: insufficient information about the sequence generation process

Allocation concealment (selection bias)Unclear riskComments: insufficient information about the allocation concealment
Blinding (performance bias and detection bias)
All outcomes
Low risk

"Coded vials with ibuprofen and placebo tablets were provided by Benzon Pharma"

Comment: probably done

Incomplete outcome data (attrition bias)
All outcomes
Low risk

"All patients who entered the study completed treatment and were included in the assessment of effectiveness and side effects"

Comments: probably done

Selective reporting (reporting bias)Unclear riskNo protocol, no convincing text
Other biasUnclear riskNo data on baseline imbalance

Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion
  1. a

    URTI: upper respiratory tract infection

Aggarwal 1997Not common cold
Azuma 2010Not common cold
Azuma 2011Not common cold
Bachert 2005Febrile URTI
Banchini 1993Not common cold
Batista 1985Not common cold
Bellussi 1993Not common cold
Bellussi 1996Not common cold
Benrimoj 2001Not common cold
Bernstein 1974Not common cold
Blagden 2002Not common cold
Bonifaci 1977Not common cold
Cappella 1993Not common cold
Chachtel 2011Not common cold
Ebel 1985Not common cold
Eccles 2003Not common cold
Fujimori 1982Not randomised
Fujimori 1983Not common cold
Gehanno 2003Not common cold
Gruber 1977Not common cold
Kandoth 1984Not common cold
Katsu 1977Not common cold
Katsu 1978Not common cold
Katsu 1982Randomisation is not clear
Katsu 1983Not common cold
Kierszenbaum 1991Not common cold
Lopes 1991Not common cold
Martinez Gallardo 1994Randomisation is not clear
Matsumoto 1984Not common cold
Moore 2002Not common cold
Nagaoka 1985Not common cold
Nagaoka 1986aNot common cold
Nagaoka 1986bNot common cold
Nouri 1984Not common cold
Nouri 1993Not common cold
Pagella 2001Not common cold
Passali 1989Not common cold
Passali 1997Not common cold
Reiner 1983Not common cold
Ruperto 2011Not common cold
Russo 2013Not common cold
Salmon 1993Not common cold
Salzberg 1993Not common cold
Sanchez 1999Not common cold
Schachtel 1993Not common cold
Schachtel 2002Not common cold
Stanley 1975Randomisation is not clear
Tamura 1984Not common cold
Ulukol 1999Not common cold
Vauzelle 1996Not common cold
Watson 2000Not common cold

Ancillary