Types of studies

We will include all randomised or quasi‐randomised controlled clinical trials (RCTs or CCTs) comparing allopurinol with another therapy (placebo or other urate‐lowering treatment) in adults with chronic gout.

Only trials that are published as full articles or available as a full trial report will be included.

Types of participants

Trials that include adults (age > 18 years) with a diagnosis of chronic gout will be included.

The American College of Rheumatology (ACR) preliminary criteria for the classification of acute arthritis of primary gout remain the most frequently used criteria for chronic gout diagnosis in clinical trials (Wallace 1977). According to these criteria, a patient can be classified as having gout if MSU crystals are identified in a synovial fluid sample or tophus aspirate, or any of six out of 12 following clinical, radiographic and laboratory criteria are fulfilled.

1. More than one attack of acute arthritis

2. Maximum inflammation developed within one day

3. Monoarthritis attack

4. Redness observed over joints

5. First metatarsophalangeal joint painful or swollen

6. Unilateral first metatarsophalangeal joint attack

7. Unilateral tarsal joint attack

8. Tophus (proved or suspected)

9. Hyperuricaemia

10. Asymmetric swelling within a joint on x‐ray film

11. Subcortical cysts without erosions on x‐ray film

12. Joint culture negative for organism during attack

Trials will be included where adults have either chronic gout based upon the ACR criteria, or author‐defined gout.

Types of interventions

All trials that evaluate allopurinol in any dose or dosing interval will be included.

Comparators could be any of the following.

1. Placebo

2. No treatment

3. Another urate lowering therapy including febuxostat, probenecid, benzbromarone, sulfinpyrazone, pegloticase or rasburicase

4. One regimen of allopurinol versus another

5. A combination of urate‐lowering therapies

Types of outcome measures

OMERACT (Outcome Measures in Rheumatology) is an international network interested in outcome measures across the spectrum of rheumatology intervention studies. At the OMERACT 9 conference, core and discretionary domains for outcome measures in clinical studies of acute and chronic gout were defined (Schumacher 2009). We will include the core domains nominated for chronic gout which include acute gout attacks, serum urate, pain, health‐related quality of life (HRQoL), and function.

Electronic searches

The following electronic databases will be searched from inception.

1. OVID MEDLINE 1948 to present (Appendix 1)

2. EMBASE 1980 to present (Appendix 2)

3. The Cochrane Central Register of Controlled Trials (CENTRAL) (Appendix 3)

No language restrictions will be applied.

Searching other resources

Abstracts from the two major international rheumatology scientific meetings ‐ the American College of Rheumatology (ACR) and the European League Against Rheumatism (EULAR) will be searched for the years 2011 and 2012. We will also search the reference lists of included articles for additional trials.

Selection of studies

The results of the search will be independently reviewed by two authors (RS and AK) to identify trials that fulfil the inclusion criteria. We will review titles and abstracts and if more information is required to determine whether a trial meets the inclusion criteria, we will obtain the full text. A record for the reasons for excluding studies will be kept and any disagreement will be resolved by discussion. If no agreement can be met, a third review author (CE) will act as arbiter. Articles in a language other than English will be translated.

Data extraction and management

Data will be independently extracted from the included trials by the two review authors (RS and AK) using the pre‐set forms. The collected data will be entered into Revman 5.1 (RevMan 2011).

Two independent reviewers (RS and AK) will extract relevant information from the included trials including study design, characteristics of study population, treatment regimen and duration, outcomes and timing of outcome assessment, using predetermined forms. Differences in data extraction will be resolved by referring back to the original articles and establishing consensus. A third reviewer (CE) will act as arbiter to help resolve differences if necessary.

The raw data (means and standard deviations for continuous outcomes and the number of events for dichotomous outcomes) will be extracted for outcomes of interest.

Assessment of risk of bias in included studies

Two review authors (RS and AK) will independently assess the risk of bias of all included studies. The Cochrane Collaboration's tool for assessing risk of bias will be used (Higgins 2011a). This includes assessing the bias in each of the following domains: random sequence generation, allocation concealment, blinding of participants, personnel and outcome assessors, completeness of outcome data, selective reporting and other sources of bias.

Each of these criteria will be graded as 'high risk' of bias, 'low risk' of bias or 'unclear risk' of bias. Disagreements will be resolved by consensus; if a consensus cannot be reached, a third review author (CE) will act as arbiter.

Measures of treatment effect

The data will be summarized in a meta‐analysis only if there is sufficient clinical and statistical homogeneity.

For dichotomous data, the results will be presented as risk ratios (RR) with corresponding 95% confidence intervals (95% CI). A RR > 1.0 will indicate a beneficial effect of allopurinol.

For continuous data, the results will be presented as mean differences (MD) between the intervention and comparator groups with the corresponding 95% CIs.

When different scales are used to measure the same conceptual domain, standardised mean differences (SMD) with corresponding 95% CIs will be calculated instead. For the calculation of SMD, MD is divided by the standard deviation, resulting in a unit‐less measure of treatment effect. SMDs larger than zero indicate a beneficial effect of allopurinol. An SMD of 0.2 indicates a small beneficial effect, 0.5 a medium effect and 0.8 a large effect in favour of allopurinol. SMD will be re‐expressed as MD by multiplying the SMD by a typical among‐person standard deviation using a familiar scale in order to facilitate appraisal by clinicians (Schünemann 2011b).

Unit of analysis issues

We will assess whether each study evaluates the number of people with acute flares or the number of acute flares as a unit of analysis, and we plan to evaluate number of people with acute flares as the preferred outcome.

We will avoid a potential unit of analysis issue by making multiple pair‐wise comparisons between all possible pairs of intervention groups for trials with multiple treatment groups, or alternatively, by including only the pair with accepted drug dosages (Higgins 2011c).

Dealing with missing data

If data are missing or incomplete, we will seek further information from the study authors.

In cases where individual data are missing from the reported results and no further information is available from the study authors, we will assume the missing values to have a poor outcome. For dichotomous variables that measure AEs, the withdrawal rate will be calculated using the number of patients who received the treatment as the denominator (worst‐case analysis). For dichotomous outcomes that measure benefits, the worst case analysis will be calculated using the number of randomised individuals as the denominator. For continuous variables, we will calculate the MD or the SMD based on the number of patients analysed at the time point. If the number of patients analysed is not available, the number of randomised participants in each group at baseline will be used.

Where possible, missing standard deviations will be calculated from other statistics such as standard errors, CIs or P values, according to methods recommended in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b). If standard deviations cannot be calculated, they will be imputed from other studies in the meta‐analysis (Higgins 2011c).

Assessment of heterogeneity

We will assess clinical and statistical heterogeneity between studies.

We will initially assess studies for clinical homogeneity with respect to study participants, intervention groups, outcome measures and timing of outcome.

For studies judged as clinically similar, statistical heterogeneity will be assessed using the I² statistic (Deeks 2011). The following thresholds for the interpretation of I² will be used: 0% to 40% heterogeneity might not be important, 30% to 60% represents moderate heterogeneity, 50% to 90% represents substantial heterogeneity and > 75% represents considerable heterogeneity. In cases of considerable heterogeneity, we will explore the data further, including subgroup analyses, in an attempt to explain heterogeneity.

Assessment of reporting biases

To assess the potential for reporting bias, we will determine whether the protocol of the trial was published before recruitment of patients began. For trials published after 1 July 2005, we will screen the Clinical Trials Register at the International Clinical Trials Registry Platform of the World Health Organization (http://apps.who.int/trialsearch/). We will evaluate if selective reporting of outcomes is present.

We will compare the fixed‐effect estimate against random‐effects model to assess the possible presence of small sample bias in the published literature (i.e. in which the intervention effect is more beneficial in smaller studies). In the presence of small sample bias, the random‐effects estimate of the intervention is more beneficial than the fixed‐effect estimate (Sterne 2011).

Reporting bias will be explored by a funnel plot if 10 or more studies are available (Sterne 2011).

Data synthesis

If studies are considered to be sufficiently homogenous, we will pool data in a meta‐analysis using a random‐effects model, irrespective of the I² results.

Subgroup analysis and investigation of heterogeneity

We suspect that responses to treatment may differ according to the participant's age and gender. Elderly patients can present with more associated conditions and possibly a greater chance of adverse effects (Busquets 2011) while reports indicate that gout in women may have different epidemiological and clinical characteristics compared to gout in men (Harrold 2006).

Therefore, we plan the following subgroup analyses if sufficient data are available.

1. Patient's age (> 65 years or < 65 years)

2. Gender (men versus women).

Ideally, we would extract the outcome (acute gout attacks) separately for men and women, and the outcome, withdrawals due to adverse events separately by age subgroups from within each trial. We plan to informally compare the magnitudes of effect to assess possible differences in response to treatment by considering the overlap of the confidence intervals of the summary estimates in the two subgroups ‐ non‐overlap of the confidence intervals indicates statistical significance. However, we anticipate that the outcomes may not be reported by subgroups within the trials, precluding the planned analyses.

Sensitivity analysis

Where sufficient studies exist, sensitivity analyses are also planned to explore the impact of any bias attributable to inadequate or unclear allocation concealment and outcome assessor blinding.

We will assess the presence of small study bias (i.e., intervention effect is more beneficial in smaller studies) in the meta‐analysis by comparing the fixed‐effect estimate and the random‐effects estimate.

We will investigate the effect of any missing or imputed data by sensitivity analysis.