Background/Purpose: Selective outcome reporting may bias treatment effect estimates of clinical trials. Registration of clinical trials was established to improve transparency in their conduct and reporting. We studied the discrepancies between the registered and published primary outcomes (PO) of rheumatoid arthritis (RA) randomized controlled trials (RCTs).

Methods: RA RCTs that were registered at ClinicalTrials.gov (CTG) with completion date before January 1, 2010 and ≥ 1 publication in a peer-reviewed journal were studied. Registered and published POs were extracted, and two authors independently categorized presence and type of PO discrepancies using standardized method (see Table) with resolution of differences by consensus. Associations of presence of PO discrepancies with RCT characteristics were assessed by Chi-square or Fisher’s Exact test.

Results: Among 95 eligible RCTs, 56 (58.9%) had identical registered and published POs. PO discrepancies could not be assessed for 10 RCTs (3 had missing PO in CTG records; while ambiguous/unclear PO were recorded in 4 CTG records and 3 manuscripts). 29 (30.5%) had explicit discrepancies between the registered and published POs (Table). Discrepant or ambiguous PO reporting was associated with funding source [Industry funding (21/65, 32.3%) vs non-profit source funding (18/30, 60%), p = 0.011). No association was found with study phase; year of study registration; and number of study centers. Among 39 RCTs with ambiguous/explicit PO changes, discrepancies were considered to be clinically relevant in 8 RCTs [7 with shorter published PO assessment time (5 non-profit & 2 industry funded); 1 with statistically significant published PO but insignificant CTG PO (industry funded)]; not clinically relevant in 12 RCTs (both CTG and published POs had identical statistical significance); and were unable to assess clinical relevance in 19 RCTs (10 with missing or ambiguous PO, 7 where only difference was published PO time assessment specification, and 2 with statistically significant published POs not specified in CTG).

Conclusion: More than quarter of RA RCTs had ambiguous or explicit PO discrepancies, and 8 (5.6%) were clearly considered to be of potentially clinical relevance. Industry funding was associated with less likelihood of ambiguous/discrepant PO reporting. Improvement in reporting of registered and published POs is needed to improve utility of trial registries.

Table. Types of discrepancies between CTG registered and published POs.

*5 RCTs had > 1 PO discrepancies, hence total (%) RCTs > 95 (100%)

Ref: Arthritis Rheumatol. 2014;66:2664-74

« Back to 2015 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/discrepancies-between-registered-and-published-primary-outcomes-in-randomized-controlled-trials-of-rheumatoid-arthritis/