Many randomized trials in a large systematic review were not registered and had evidence of selective outcome reporting: a metaepidemiological study

Abstract

OBJECTIVES: The primary objectives were to describe characteristics of trial registration in the chronic low back pain (CLBP) field and assess the association of trial registration status (registered vs unregistered, prospectively registered vs retrospectively registered) with risk of bias, sufficient sample size, quality of reporting, and treatment effect estimates. Secondary objectives were to describe trial registration consistency with the final report and assess its association with risk of bias, sufficient sample size, and treatment effect estimates. STUDY DESIGN AND SETTING: A cross-sectional metaepidemiological study of trials included in a large Cochrane review on exercise treatments for CLBP. We extracted relevant trial and registration information and assessed trials' risk of bias using the Cochrane Risk of Bias 1 tool. We performed descriptive analyses, logistic regressions, and subgroup meta-analyses. RESULTS: We included 361 trials, of which 23.3% were prospectively registered. Registered trials had lower risk of bias (odds ratio [OR] 0.6; 95% confidence interval [CI] 0.5, 0.7) and higher reporting quality (OR 1.6; 95% CI 1.4, 1.8) than unregistered trials. Prospectively registered trials were more likely to have low risk of reporting bias (OR 2.7; 95% CI 1.2, 6.5) and higher quality of reporting (OR 1.3; 95% CI 1.1, 1.6) than retrospectively registered trials. Trial registration status was not associated with effect estimates. Among prospectively registered trials, 64.3% clearly defined primary outcome(s) in their registration, 58.3% had consistent sample sizes, and 22.6% had no evidence of selective outcome reporting. Trials that clearly defined primary outcome(s) were more likely to report larger effect estimates for pain intensity (mean difference -15.8; 95% CI -22.7, -8.9 vs -6.0; 95% CI -10.6, -1.5; Q = 6.7, P = .01), although the difference was small, the 95% CIs overlapped, and no difference was found for functional limitations. CONCLUSION: A small proportion of trials in the CLBP field were registered prospectively and many presented registration inconsistencies. Registered trials tend to have lower risk of bias and higher quality of reporting. Policies are needed to improve prospective registration and registration consistency in the field. PLAIN LANGUAGE SUMMARY: Prospective trial registration is the practice of documenting the planned methods of a randomized controlled trial on a publicly available online platform (ie, website) before enrolling participants. Medical journals require trialists to prospectively register their trials to encourage the conduct of high-quality research and reduce the chance of trialists changing their research plan to report only positive or significant results (known as selective outcome reporting). We investigated whether trialists within the chronic low back pain field were registering their trials, and whether they followed their registered research plan.We used data from a large systematic review of 456 trials that tested the effectiveness of exercise as a treatment for chronic low back pain. We assessed each trials' registration status and whether prospectively registered trials had inconsistencies between their registered research plan and their research conduct (eg, evidence of selective outcome reporting). We also looked at the association among trial registration with trials' quality of reporting (ie, a marker of research transparency), risk of bias (ie, a marker of research quality), and the amount of low back pain improvement reported by the trials (ie, effect estimates).We found that less than 25% of trials were prospectively registered, and many had inconsistencies between their registered research plan and their research conduct. Overall, registered trials had lower risk of bias and higher quality of reporting. However, trial registration status and selective outcome reporting were not associated with effect estimates (the amount of back pain improvement reported by trials). Our findings highlight the need for trialists and journals to better follow trial registration guidelines and policies in the chronic low back pain field. Knowledge users should be cautious when consuming information from unregistered trials as they appear to be more likely to have quality concerns.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 6 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.