Criteria for considering studies for this review
Types of studies
Randomized controlled trials (RCTs) will be considered for inclusion.
Types of participants
Participants of any age with active Crohn’s disease defined by a combination of clinical (e.g. Crohn's disease activity index, CDAI > 150), radiographic, endoscopic, and histological criteria will be considered for inclusion.
Types of interventions
RCTs where low dose naltrexone was given by any route versus placebo or an active comparator will be considered for inclusion.
Types of outcome measures
The primary outcome will be the proportion of patients achieving remission (e.g. CDAI < 150) as defined by the included studies, and expressed as a percentage of the patients randomized (intention to treat analysis).
Secondary outcome measures will include the proportion of patients with:
A. Clinical response (as defined by the included studies);
B. Endoscopic response or remission (as defined by the included studies);
C. Improvement in quality of life (as defined by the included studies);
D. Adverse events;
E. Withdrawal because of adverse events; and
F. Serious adverse events.
Search methods for identification of studies
We will search the following databases for relevant trials:
MEDLINE (from 1966 to February 2013)
PubMed (from 1966 to the February 2013);
EMBASE OVID (from 1974 to the February 2013);
The Cochrane Library (CENTRAL, February 2013); and
The Cochrane Inflammatory Bowel Disease and Functional Bowel Bowel Disorders Review Group Specialised Register (SR-IBD).
The search strategies are listed in Appendix 1.
Searching other resources
We will perform a manual review of conference proceedings including Digestive Disease Week (DDW) and United European Gastroenterology Week (UEGW). Reference lists from retrieved articles will be scanned to identify additional citations that may have been overlooked by the electronic searches.
Data collection and analysis
Selection of studies
All studies identified by the search strategy will be independently assessed by two authors (DS and JKM) for inclusion, according to pre-specified criteria listed above. Studies in abstract form will only be included if the authors can be contacted for further information. Any disagreements will be resolved by consensus. If necessary a third party (NC) will be engaged to resolve any ongoing disagreements.
Data extraction and management
A standardized form will be used to extract relevant data from the included studies. The form will be based on the Cochrane checklist of items to consider for data extraction (Higgins 2011a). Two authors (DS and JKM) will independently extract and record the data. Any disagreements will be resolved by consensus with possible input from a third author (NC). The following data will be extracted:
A. Methods (methods of randomization, allocation, concealment, and blinding; inclusion and exclusion criteria, definition of outcomes);
B. Participants (number, age, gender, disease activity, co-medication);
C. Interventions (formulation, dose, frequency, duration); and
D. Outcomes (clinical remission and response rates, endoscopic remission and response rates, quality of life, adverse events, withdrawals due to adverse events, and serious adverse events).
Assessment of risk of bias in included studies
Two authors (DS and JKM) will independently assess the methodological quality of the included studies using the Cochrane risk of bias tool (Higgins 2011b). Any disagreements will be resolved by consensus with possible input from a third author (NC). Factors to be assessed will include:
A. Sequence generation;
B. Allocation concealment;
C. Blinding of participants and personnel;
D. Blinding of outcome assessment;
E. Completeness of outcome data;
F. Selective reporting; and
G. Other sources of bias.
Studies will be assigned a low risk of bias, high risk of bias, or unclear risk of bias for each category.
The quality of the total body of evidence for the primary outcomes of interest will be assessed using the GRADE criteria (Guyatt 2008; Schünemann 2011). Randomized trials are considered high quality and will be downgraded if the following weaknesses are present: within-study risk of bias, indirectness of evidence, heterogeneity, imprecision of effect estimates, and risk of publication bias. The quality of the body of evidence for a particular outcome will then be graded as:
A. High: Further research is very unlikely to change our confidence in estimate of effect;
B. Moderate: Further research is likely to have an important impact on our confidence in estimate of effect and may change estimate;
C. Low: Further research is very likely to have an important impact on our confidence in estimate of effect and likely may change estimate; and
D. Very low: Any estimate of effect is very uncertain.
Measures of treatment effect
Data will be analyzed using Review Manager (RevMan 5.2). All data will be analyzed on an intention-to-treat basis. For dichotomous outcomes, we will calculate the relative risk and 95% confidence intervals (95% CI). For continuous outcomes with uniform units, we will calculate a mean difference (MD) and the corresponding 95% CI. The standardized mean difference (SMD) and 95% CI will be calculated when different scales are used for the same outcome.
Unit of analysis issues
The primary analysis will combine data from the trial defined primary observation time-point. For studies that report outcomes at fixed intervals, outcomes will be combined at those time points as well. For a parallel group design a single measurement for each outcome per participant will be collected. For crossover trials, data from the first part of the trial (prior to crossover) will be extracted. For cluster-randomized trials, data will be extracted at the level of the individual. When more than one efficacy or safety event is reported per subject, we will use the proportion of subjects with at least one event as the outcome. If studies allocate subjects to different active treatment arms they will be combined for the primary analysis. In this case, subgroup analysis will be attempted to assess outcomes among different doses of naltrexone. Separate analyses will be conducted for LDN versus placebo, and LDN versus active comparator.
Dealing with missing data
As the primary analysis will be carried out as intention-to-treat, missing data will be assumed to be negative (treatment failure). Whenever possible, study authors will be contacted to request missing data, or to ascertain the reason for data loss. A sensitivity analysis will be used to assess the impact on the results of the assumption of treatment failure when data are missing.
Assessment of heterogeneity
Heterogeneity in included studies will be evaluated using the Chi2 test (a P value of 0.10 will be considered statistically significant) and the I2 statistic. Sensitivity analysis will be used as indicated to investigate sources of heterogeneity.
Assessment of reporting biases
If there are a sufficient number of studies publication bias will be assessed by funnel plots (Egger 1997).
Studies will be pooled for meta-analyses when studies are similar in terms of treatments, comparators, and outcome definitions (determined by consensus). Data will not be pooled for analysis if there is a high degree of heterogeneity (I2 > 75%). A fixed-effect model will be used in the absence of statistically significant heterogeneity; while a random-effects model will be used if there is statistically significant heterogeneity (P < 0.10).
Subgroup analysis and investigation of heterogeneity
Subgroup analysis will be undertaken to investigate treatment effects for the following if possible:
A. LDN dose;
B. Disease activity;
C. Duration of treatment;
D. Age; and
Sensitivity analyses will be done by repeating the meta-analysis with new data sets that:
A. Exclude lower quality studies; and
B. Incorporate specific assumptions for missing data.