Types of studies

We will include both individually randomised controlled trials (RCTs), that is where the participant is the unit of randomisation, and cluster‐RCTs. In addition, we will include the following types of studies for the primary adverse event outcome 'infection‐related adverse events, defined as reconsultation rates with the same symptoms (sore throat) or progressive/new symptoms, and/or complications (e.g. peritonsillar abscess) within one month': quasi‐RCT (Q‐RCT), non‐RCT (NRCT), prospective cohort study, and retrospective cohort study. We aim to include these types of observational studies for our primary adverse event outcome since RCTs are usually underpowered to identify potentially important differences in the occurrence of rare complications between treatment groups; this approach might therefore allow us to more carefully weigh benefits and harms of NSAIDs use in patients with acute sore throat.

We will include studies irrespective of publication status, date of publication, or language.

Types of participants

We will include both children and adults presenting in a primary healthcare setting (i.e. patients who are not referred by a physician because of the current sore throat episode, or patients presenting through self‐referral in ambulatory care or an emergency department) with sore throat. We will include the following conditions:

1. clinical signs and symptoms suggestive of acute tonsillitis ‐ inflammation of the tonsils;

2. clinical signs and symptoms suggestive of pharyngitis ‐ inflammation of the oropharynx; and

3. clinical syndrome of sore throat (painful throat, odynophagia).

We will exclude studies of referred patients, hospitalised patients, as well as those involving highly specialised populations (e.g. those with glandular fever, immunodeficiency, peritonsillar abscess, or sore throat following tonsillectomy).

Types of interventions

We will include studies irrespective of the dose, duration, or method of administration of NSAIDs.

We will include trials comparing:

1. NSAIDs with placebo;

2. NSAIDs with no treatment;

3. NSAIDs with paracetamol (acetaminophen);

4. NSAID and paracetamol (acetaminophen) with placebo and paracetamol (acetaminophen);

5. NSAIDs and paracetamol (acetaminophen) with paracetamol (acetaminophen) alone;

6. NSAIDs with other pharmacological treatments such as lozenges (not containing NSAIDs) and herbal medicinal products; and

7. NSAIDs with non‐pharmacological treatments such as physical cooling (e.g. ice cubes, wet cloths) and topical treatments (e.g. gargling with povidone‐iodine or salt).

We will include trials reporting combined interventions if they allow a direct comparison between the NSAID and control groups (e.g. co‐administration of antibiotics in both groups). We will address the following comparison pairs separately:

1. NSAIDs and antibiotics versus placebo and antibiotics; and

2. NSAIDs and antibiotics versus antibiotics alone.

Types of outcome measures

We will analyse the following outcomes in the review, but will not use them as a basis for including or excluding studies.

Electronic searches

We will search the following databases from inception to present, using the search terms and strategy described in Appendix 1:

1. Cochrane Central Register of Controlled Trials (CENTRAL), published in the Cochrane Library;

2. MEDLINE (Ovid);

3. Embase (Ovid); and

4. CINAHL (Cumulative Index to Nursing and Allied Health Literature) (EBSCOHost).

We will also search the World Health Organization (WHO) International Clinical Trials Registry Platform (www.who.int/ictrp/en/) and US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (clinicaltrials.gov) for trials in progress. We will also run searches of Google Scholar using the same terms to retrieve grey literature and other trials. We will report all search details in the final version of the review.

Searching other resources

We may contact original authors for clarification and further data if trial reports are unclear. We will arrange translations of papers where necessary.

We will also check the reference lists of all included studies and those of relevant review articles identified by the searches.

Selection of studies

Two review authors (RPV, RTvU) will independently screen titles and abstracts of the records obtained from the searches. We will first identify and exclude duplicates, and then screen the titles and abstracts of the unique records to identify potentially relevant studies. We will retrieve the full‐text study reports/publications of all potentially eligible studies. The same two review authors will independently review the full texts against the predefined inclusion and exclusion criteria for eligibility. Any disagreements will be resolved through discussion or by consulting a third review author (PL) if required.

We will collate multiple reports of the same study so that each study, rather than each report, is the unit of interest in the review. We will record the selection process in sufficient detail to complete a PRISMA flow diagram and 'Characteristics of excluded studies' table (Moher 2009).

Data extraction and management

We will use a data collection form for the extraction of study characteristics and outcome data that has been piloted on at least one study in the review. Two review authors (RPV, RTvU) will independently extract the following study characteristics from the included studies:

1. methods: study design, total duration of study, details of any 'run‐in' period, number of study centres and location, study setting, withdrawals, and date of study;

2. participants: number, mean age, age range, gender, severity of condition, diagnostic criteria, inclusion criteria, and exclusion criteria;

3. interventions: intervention (including instruction about NSAIDs use and NSAID type), comparison, concomitant medications, and excluded medications;

4. outcomes: primary and secondary outcomes specified and collected, and time points reported; and

5. notes: funding for trial, and notable conflicts of interest of trial authors.

Two review authors (RPV, RTvU) will independently extract outcome data from the included studies. We will note in the 'Characteristics of included studies' table (see Table 1 for template) if outcome data are not reported in a useable way. Any disagreements will be resolved by consensus or by involving a third review author (PL). One review author (RPV) will transfer data into the Review Manager file using RevMan Web (RevMan Web 2019). We will double‐check that data are entered correctly by comparing the data presented in the systematic review with the study reports. A second review author (RTvU) will spot‐check study characteristics for accuracy against the trial report. 

Assessment of risk of bias in included studies

Two review authors (RPV, RTvU) will independently assess risk of bias for the included RCTs using the Cochrane 'Risk of bias' tool for randomised trials (RoB 2) (Sterne 2019), according to the guidance in the latest version of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2019). Any disagreements will be resolved by discussion or by involving another review author (PL). We will assess risk of bias as 'low', 'high', or 'some concerns' for each of the following five domains:

1. bias arising from the randomisation process;

2. bias due to deviations from the intended interventions;

3. bias due to missing outcome data;

4. bias in measurement of outcome; and

5. bias in selection of the reported results.

For cluster‐RCTs, we will assess risk of bias using these five domains in addition to a sixth domain described in the latest version of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2019), namely "Bias arising from the timing of identification and recruitment of participants". 

We will provide a quote from the study report together with a justification for our judgement in the 'Risk of bias' table. We will summarise the 'Risk of bias' judgements across different studies for each of the domains listed. Where information on risk of bias relates to unpublished data or correspondence with an individual trial author, we will note this in the 'Risk of bias' table.

We will manage the risk of bias through the Microsoft Excel tool available at www.riskofbias.info (Sterne 2019). For assessment of the overall risk of bias, we will use signalling questions and algorithm tools (Higgins 2019). We will assess risk of bias of the RCTs using the RoB 2 tool based on the following outcomes:

1. complete resolution of pain at two to three days (48 to 72 hours; when symptoms are usually worst);

2. infection‐related adverse events, defined as reconsultation with the same symptoms (sore throat) or progressive/new symptoms, and/or complications (e.g. peritonsillar abscess) within one month;

3. complete resolution of pain at various other time points (24 hours, four to seven days);

4. mean time to resolution of fever;

5. mean time to complete resolution of total symptoms (recovery);

6. mean time to return to normal diet; and

7. adverse effects likely to be related to the use of NSAIDs.

We will store the consensus decisions for the signalling questions as supplemental data.

The effect of interest will be ‘effect of assignment’ (intention‐to‐treat (ITT) principle).

We will assess risk of bias for relevant non‐randomised studies using the ROBINS‐I tool (Sterne 2016). We will use this tool to assess the risk of bias of the primary adverse event outcome 'infection‐related adverse events, defined as reconsultation rates with the same symptoms (sore throat) or progressive/new symptoms, and/or complications (e.g. peritonsillar abscess) within one month'. We will assess the following confounders: age, comorbidity, aetiology, fever, symptom severity, prior duration of symptoms, exudate, severe inflammation of the pharynx, and co‐treatment with antibiotics.

Measures of treatment effect

We will enter the outcome data for each study into the data tables in RevMan Web to calculate the treatment effects (RevMan Web 2019). We will use risk ratios (RR) with 95% confidence intervals (CIs) for dichotomous outcomes and mean differences (MD) with standard deviations (SD) or as standardised mean differences (SMD) for continuous outcomes. Where SMD has been computed, we will re‐express the outcomes in a single natural scale (to be determined based on the scales from the observed studies).

For each of the key outcomes listed in the 'Summary of findings' table, we will present the results for the absolute numbers (in addition to the relative numbers) based on the pooled results and compared to the assumed risk. Where applicable, we will also calculate the number needed to treat for an additional beneficial outcome (NNTB) and number needed to treat for an additional harmful outcome (NNTH) using the pooled results.

Unit of analysis issues

For cluster‐RCTs, we will consider potential differences between the intervention effects being estimated and check whether clustering was taken into account in the analysis of the individual trials. Additionally, as per the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2019), we will report on whether or not the sample sizes were estimated through the intracluster/intraclass correlation coefficient (ICC) and whether the analyses were carried out at the cluster level. Where trials take multiple measurements or observations of a single outcome in the same participants (repeated measurements), we will only extract and analyse the datapoints consistent with the times of follow‐up defined in the primary and secondary outcomes in this protocol.

Dealing with missing data

We will contact individual trial authors or study sponsors to verify key study characteristics and to obtain missing numerical outcome data where possible (e.g. when a study is identified as abstract only). Where this is not possible, and the missing data are thought to introduce serious bias, we will explore the impact of including such studies in the overall assessment of results by a sensitivity analysis, provided sufficient data are available. 

If SDs are missing and they cannot be obtained from the trial authors, we will approximate them using other available statistics such as P values or alternative methods according to the methods described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2019).

Apart from imputations for missing SDs, we will not conduct any other imputations. We will extract and analyse all data using the available‐case analysis method based on the ITT principle, that is by analysing all participants in the groups to which they had originally been randomised.

Assessment of heterogeneity

We will assess clinical heterogeneity by reviewing the included trials for potential differences between trials in study populations, interventions or comparisons used, and outcomes measured.

We will assess statistical heterogeneity for each outcome by visually inspecting the forest plots and by using the I² statistic, with investigation of I² values over the ranges described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2019).

Assessment of reporting biases

To assess outcome reporting bias, we will search the internet and ClinicalTrials.gov (clinicaltrials.gov) for available study protocols to determine whether the outcomes reported in the included trials were predefined and whether all outcomes listed in the study protocol are reported in the publication. We propose that if there are sufficient trials, we will assess reporting bias by using funnel plots. If we are able to pool more than 10 trials, we will create and examine a funnel plot to explore possible small‐study and publication biases.

Data synthesis

We will analyse the data according to the ITT principle. Where data are missing or there was loss to follow‐up, we will conduct an available‐case analysis.

We will pool data from studies we judge to be clinically homogeneous using RevMan Web (RevMan Web 2019). If more than one study provides useable data in any single comparison, we will perform a meta‐analysis using a random‐effects (DerSimonian and Laird) model to provide a more conservative estimate. We will include all eligible studies in the primary analyses, but we will perform separate meta‐analyses for RCTs and non‐RCTs (for the primary adverse event outcome).

For non‐RCTs, our primary analyses will include eligible studies where the data have been adjusted (with common criteria). Secondary analyses will consist of combining eligible studies where the data have not been adjusted. We will discuss the limitations of these analyses in both cases. 

Subgroup analysis and investigation of heterogeneity

We will perform subgroup analyses planned a priori for the following characteristics if there are at least 10 studies available for each subgroup stratum:

1. age (children versus adults);

2. Centor criteria (0 to 2 versus 3 to 4);

3. prior duration (three days or less versus over three days);

4. fever (body temperature > 38 °C present versus absent);

5. NSAID type (sprays versus lozenges versus tablets/capsules/suppositories/injections); and

6. instruction about NSAIDs use (at regular time intervals versus as needed).

We will use the Chi² test to test for subgroup interactions in Review Manager (RevMan Web 2019).

Sensitivity analysis

We plan to carry out sensitivity analyses to determine the robustness of our 'Risk of bias' findings for the included studies by excluding studies with high risk of overall bias for the results as assessed using the Cochrane RoB 2 tool (Higgins 2019).