Review ArticleModified intention-to-treat analysis did not bias trial results
Introduction
The intention-to-treat (ITT) principle is recommended for analysis as it protects against attrition bias by providing an unbiased estimate of treatment effect when reporting randomized controlled trials [1], [2], [3]. The ITT principle requires all patients in a randomized trial to be analyzed according to their original randomized allocation and requires all patients to be included in the analysis [4]. This conserves randomization and avoids attrition bias when evaluating a treatment assignment [5], [6]. Frequently, a less strict ITT approach is used [7], [8], [9], [10], [11]—commonly referred to as modified ITT (mITT). Generally mITT is based on exclusion of patients—for different reason(s)—postrandomization [12], which may introduce bias [13], [14].
Both ITT and mITT analyses are affected by missing data which also can bias estimates of treatment effect [15], [16], [17]. Methods to handle missing data in randomized trials are as follows: last observation carried forward [10], [11], [18], [19], [20], [21], nonresponder imputation, and multiple imputation [11], [22]. Last observation carried forward is widely criticized and advised against [16], [18], [19], [20], [23], [24], [25]. Nonresponder imputation attributes withdrawal to lack of efficacy—and missing data are imputed as “failure” [18], [26]—and represents a more conservative method than last observation carried forward for dichotomous or categorical outcomes. Preferably, multiple imputation should be used to handle missing data [18], [23].
Estimates of treatment effect can be biased by the use of mITT and common different imputations methods, but the impact and direction of bias has not been assessed previously within rheumatology. To examine whether ITT treatment effect is different from mITT treatment effect and to examine the impact of different imputation methods, we carried out a meta-epidemiological study focusing on trials assessing US Food and Drug Administration or European Medicines Agency–approved biological or targeted interventions (see Methods in the following for definition) in patients with rheumatoid arthritis. These interventions are relatively recent (first approved in 1998) and yield a more modern approaches to trial design, that are believed to have low risk of bias [27]. They therefore represent a good case for examining the nuances of attrition bias.
Section snippets
Methods
Study selection, assessment of eligibility criteria, data extraction, and statistical analysis plan were conducted based on a predefined protocol (PROSPERO CRD42013006702) according to the Methodological Expectations of Cochrane Intervention Reviews (MECIR) [28], [29] and reported according to the “Preferred Reporting Items for Systematic reviews and Meta-Analyses” (PRISMA) [30].
Characteristics of included trials
Of 5,237 identified trials, 72 were eligible (Fig. 1). All approved biologics and targeted agents were represented. Table 1 summarizes the characteristics of each analysis population and the associated trials. Detailed characteristics are available in Appendix Table B/Appendix B at www.jclinepi.com. The total number of randomized patients was 24,160, and the total number of analyzed patients was 23,842 (99%). Of the 72 included trials, 30 analyzed the ITT population (9,057 patients), and 37
Discussion
In this meta-epidemiological study of 72 randomized trials, we observed no variation in the treatment effect of biologics for rheumatoid arthritis when mITT was compared to ITT. Subsequently, we did an mITT subgroup analysis and found no evidence supporting the notion that estimates of the net benefit—depending on the type and number of modifications applied—had any impact on the effect of a treatment. Our confidence in these estimates corresponds to moderate-quality evidence within the field
Conclusion
In conclusion, randomized trials using the mITT population for analysis have been suspected to be inferior to randomized trials using the ITT population for analysis. Our study of biologics and targeted interventions for rheumatoid arthritis does not support this statement, as we find a comparable treatment effect when ITT and mITT analyses are applied.
References (57)
- et al.
A systematic review finds variable use of the intention-to-treat principle in musculoskeletal randomized controlled trials with missing data
J Clin Epidemiol
(2015) - et al.
Addition of infliximab compared with addition of sulfasalazine and hydroxychloroquine to methotrexate in patients with early rheumatoid arthritis (Swefot trial): 1-year results of a randomised trial
Lancet
(2009) - et al.
Handling missing data issues in clinical trials for rheumatic diseases
Contemp Clin Trials
(2011) - et al.
Publication bias in clinical research
Lancet
(1991) - et al.
CONSORT 2010 explanation and elaboration: updated guidelines for reporting parallel group randomised trials
Int J Surg
(2012) - et al.
CONSORT 2010 Explanation and Elaboration: updated guidelines for reporting parallel group randomised trials
J Clin Epidemiol
(2010) - et al.
The Cochrane Collaboration's tool for assessing risk of bias in randomised trials
BMJ
(2011) - et al.
Chapter 8: assessing risk of bias in included studies
- et al.
Chapter 16: special topics in statistics
- et al.
Biologics for rheumatoid arthritis: an overview of Cochrane reviews
Cochrane Database Syst Rev
(2009)
Beyond intention to treat: what is the right question?
Clin Trials
The intention-to-treat approach in randomized controlled trials: are authors saying what they do and doing what they say?
Clin Trials
Intention-to-treat analysis: who is in? Who is out?
J Fam Pract
What is meant by intention to treat analysis? Survey of published randomised controlled trials
BMJ
Violation of the intent-to-treat principle and rate of missing data in superiority trials assessing structural outcomes in rheumatic diseases
Arthritis Rheum
Modified versus standard intention-to-treat reporting: are there differences in methodological quality, sponsorship, and findings in randomized trials? A cross-sectional study
Trials
Investigating patient exclusion bias in meta-analysis
Int J Epidemiol
The effects of excluding patients from the analysis in randomised controlled trials: meta-epidemiological study
BMJ
Missing data
BMJ
Missing data in trials: do we have to keep carrying the last observation forward?
Arthritis Rheum
Potential impact on estimated treatment effects of information lost to follow-up in randomised controlled trials (LOST-IT): systematic review
BMJ
Guideline on missing data in confirmatory clinical trials
Does analysis using “last observation carried forward” introduce bias in dementia research?
CMAJ
Missing outcomes in randomized trials: addressing the dilemma
Open Med
Interpreting incomplete data in studies of diet and weight loss
N Engl J Med
Reporting of radiographic methods in randomised controlled trials assessing structural outcomes in rheumatoid arthritis
Ann Rheum Dis
Missing data and the trouble with LOCF
Evid Baed Ment Health
Cited by (41)
Randomized controlled trial of stress reduction with meditation and health education in black men and women with high normal and normal blood pressure
2021, American Journal of Preventive CardiologyComparative effectiveness of endoscopic versus microscopic transsphenoidal surgery for patients with growth hormone secreting pituitary adenoma: An emulated trial
2021, Clinical Neurology and NeurosurgeryCitation Excerpt :We performed several sensitivity analyses to assess the robustness of our results. First, we used logistic regression on only complete cases with follow-up to estimate the treatment effect [20]. Second, we used multiple imputation for the unmeasured follow-up data [21].
Allocation Concealment and Intention-To-Treat Analysis Do Not Influence the Treatment Effects of Physical Therapy Interventions in Low Back Pain Trials: a Meta-epidemiologic Study
2019, Archives of Physical Medicine and RehabilitationCitation Excerpt :In contrast, among the meta-epidemiologic studies that evaluated the influence of allocation concealment on categorical outcomes, some of them7,13,31-33 found association between the lack of allocation concealment and overestimation of treatment effects (ratio of treatment effect ranging from 0.63 to 0.94), whereas other previous studies reported no association.18,34-36 Regarding the influence of ITT analysis on treatment effects of categorical outcomes, most of them found no influence of this methodological characteristic.11,13,37,38 Only Abraha et al,39 using a different definition for deviation from ITT analysis without taking into account the occurrence of post-randomization exclusions, found that the treatment effect of trials that performed modified ITT analysis was overestimated by 17% compared with trials that performed ITT analysis.
Methodological progress note: Choosing analytic methods for randomized trials
2024, Journal of Hospital Medicine
Funding: This research received grants from the Michaelsen Foundation. No sponsor was involved in study design, and no sponsor had authority in the collection, management, analysis, and interpretation of data. Writing of the report and the decision to submit the results for publication were made strictly by the authors. Musculoskeletal Statistics Unit, The Parker Institute, is supported by unrestricted grants from the Oak Foundation. The Copenhagen Trial Unit is funded by the Danish state.
All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare no financial relationships with any organizations that might have an interest in the submitted work in the previous 3 years; no other relationships or activities that could appear to have influenced the submitted work.