Intention to Treat Analysis

Clinicians, institutions, and policy makers use results from randomized controlled trials to make decisions regarding therapeutic interventions for their patients and populations. Knowing the effect the intervention has on patients in clinical trials is critical for making both individual patient as well as population-based decisions. However, patients in clinical trials do not always adhere to the protocol. Excluding patients from the analysis who violated the research protocol (did not get their intended treatment) can have significant implications that impact the results and analysis of a study. Intention-to-treat analysis is a method for analyzing results in a prospective randomized study where all participants who are randomized are included in the statistical analysis and analyzed according to the group they were originally assigned, regardless of what treatment (if any) they received. This method allows the investigator (or consumer of the medical literature) to draw accurate (unbiased) conclusions regarding the effectiveness of an intervention. This method preserves the benefits of randomization, which cannot be assumed when using other methods of analysis. The risk of bias is increased whenever treatment groups are not analyzed according to the group to which they were originally assigned. If an intervention is truly effective (truth), an intention-to-treat analysis will provide an unbiased estimate of the efficacy of the intervention at the level of adherence in the study. This article will review the "intention-to-treat" principle and its converse, "per-protocol" analysis, and illustrate how using the wrong method of analysis can lead to a significantly biased assessment of the effectiveness of an intervention.

BackgroundMissing outcome data is a threat to the validity of treatment effect estimates in randomized controlled trials. We aimed to evaluate the extent, handling, and sensitivity analysis of missing data and intention-to-treat (ITT) analysis of randomized controlled trials (RCTs) in top tier medical journals, and compare our findings with previous reviews related to missing data and ITT in RCTs.MethodsReview of RCTs published between July and December 2013 in the BMJ, JAMA, Lancet, and New England Journal of Medicine, excluding cluster randomized trials and trials whose primary outcome was survival.ResultsOf the 77 identified eligible articles, 73 (95%) reported some missing outcome data. The median percentage of participants with a missing outcome was 9% (range 0 - 70%). The most commonly used method to handle missing data in the primary analysis was complete case analysis (33, 45%), while 20 (27%) performed simple imputation, 15 (19%) used model based methods, and 6 (8%) used multiple imputation. 27 (35%) trials with missing data reported a sensitivity analysis. However, most did not alter the assumptions of missing data from the primary analysis. Reports of ITT or modified ITT were found in 52 (85%) trials, with 21 (40%) of them including all randomized participants. A comparison to a review of trials reported in 2001 showed that missing data rates and approaches are similar, but the use of the term ITT has increased, as has the report of sensitivity analysis.ConclusionsMissing outcome data continues to be a common problem in RCTs. Definitions of the ITT approach remain inconsistent across trials. A large gap is apparent between statistical methods research related to missing data and use of these methods in application settings, including RCTs in top medical journals.

BackgroundAuthors of randomized trial reports seem to hold a variety of views regarding the relationship between missing outcome data (MOD) and intention to treat (ITT). The objectives of this study were to systematically investigate how authors of methodology articles define ITT in the presence of MOD, how they recommend handling MOD under ITT, and to make a proposal for potential improvement in the definition and use of ITT in relation to MOD.Methods and findingsWe systematically searched MEDLINE in February 2009 for methodological articles written in English that devoted at least one paragraph to ITT and two other paragraphs to either ITT or MOD. We excluded original trial reports, observational studies, and clinical systematic reviews. Working in teams of two, we independently extracted relevant information from each eligible article. Of 1007 titles and abstracts reviewed, 66 articles met eligibility criteria. Five (8%) did not provide a definition of ITT; 25 (38%) mentioned MOD but did not discuss its relationship to ITT; and 36 (55%) discussed the relationship of MOD with ITT. These 36 articles described one or more of three statements: complete follow-up is required for ITT (58%); ITT and MOD are separate issues (17%); and ITT requires a specific strategy for handling MOD (78%); 17 (47%) endorsed more than one relationship. The most frequently mentioned strategies for handling MOD within ITT were: using the last outcome carried forward (50%); sensitivity analysis (50%); and use of available data to impute missing data (46%).ConclusionWe found that there is no consensus on the definition of ITT in relation to MOD. For conceptual clarity, we suggest that both reports of randomized trials and systematic reviews separately consider and describe how they deal with participants with complete data and those with MOD.

BackgroundIndividual non-compliance with an intervention in cluster randomised trials can occur and estimating an intervention effect according to intention-to-treat ignores non-compliance and underestimates efficacy. The effect of the intervention among compliers (the complier average causal effect) provides an unbiased estimate of efficacy but inference can be complex in cluster randomised trials.MethodsWe evaluated the performance of a pragmatic bootstrapping approach accounting for clustering to obtain a 95% confidence interval (CI) for a CACE for cluster randomised trials with monotonicity and one-sided non-compliance. We investigated a variety of scenarios for correlated cluster-level prevalence of a binary outcome and non-compliance (5%, 10%, 20%, 30%, 40%). Cluster randomised trials were simulated with the minimum number of clusters to provide at least 80% and at least 90% power, to detect an ITT odds ratio (OR) of 0.5 with 100 individuals per cluster.ResultsUnder all non-compliance scenarios (5%-40%), there was negligible bias for the CACE. In the worst-case of bias, a true OR of 0.18 was estimated as 0.15 for the rarest outcome (5%) and highest non-compliance (40%). There was no under-coverage of bootstrap CIs. CIs were the correct width for an outcome prevalence of 20%-40% but too wide for a less common outcome. Loss of power for a CACE bootstrap analysis versus ITT regression analysis increased as the prevalence of the outcome decreased across all non-compliance scenarios, particularly for an outcome prevalence of less than 20%.ConclusionsOur bootstrapping approach provides an accessible and computationally simple method to evaluate efficacy in support of ITT analyses in cluster randomised trials.