Background/Purpose: Randomized controlled trials (RCTs) are considered the best research design for assessing healthcare intervention. Concerns have been raised about the increased likelihood of positive RCT outcome (bias) with trial characteristics such as financial conflicts of interests (FCOIs) and inadequate methodological quality parameters reporting. Our objective was to assess whether reported trial characteristics are associated with outcome of RCTs of rheumatoid arthritis (RA) drug therapy.

Methods: We identified original, non-phase 1, parallel-group, drug therapy RCTs of RA published in English in the years 2002-3, 2006-7, and 2010-11 by searching Medline and CENTRAL databases. RCT efficacy was assessed for primary outcome as positive (statistically significant result favoring experimental intervention) or negative. RCT characteristics [experimental intervention (traditional anti-rheumatic drugs, biologics, small molecules, others), study phase (phase 2, non-phase 2), funding source (industry, non-profit), FCOIs of authors related to industry sponsor (honoraria/consultation fee, employment, research grant, stock ownership), number of subjects enrolled, study center (single, multiple), study duration, and placebo use], and reported methodological quality measures [adequate description of random sequence generation, allocation concealment, blinding, subject follow-up, intent-to-treat analysis] were assessed independently by two investigators. Univariable associations of trial characteristics and methodological quality measures with positive study outcome were assessed using Chi-square, Fisher’s exact test, likelihood ratio, or t-test. Multivariable logistic regression (MLR) was performed by including funding source and all variables associated with positive outcome with P≤0.1

Results: 146 eligible RCTs were identified. Efficacy outcome could be assessed for 125 (85.6%) RCTs. Studies were excluded for the following reasons:  primary outcome safety (11), no intervention declared experimental a priori(10). Positive outcome was noted in 86 (68.8%) RCTs. Non-phase 2, higher number of enrolled patients and author FCOI (honoraria/consultation fee) increased likelihood, while adequate random sequence generation description decreased this likelihood of positive outcome on univariable analysis. Author FCOI (honoraria/consultation fee), number of enrolled patients, and adequate description of randomization were independently associated with positive outcomes in MLR analysis (Table 1).

Conclusion: Certain trial characteristics (number of enrolled subjects, author’s receipt of honoraria/consultation fee) and reported trial quality measure (random sequence generation description) were independently associated with positive outcomes of RA drug therapy RCTs.

Table 1. Univariable and multivariable analysis of clinical trial characteristics and study outcome.

Variable	Univariable analysis			Multivariable analysis
	Referent	OR (95% CI)*	P	OR (95% CI)*	P
Funding source, industry**	Non-profit	0.89 (0.40-1.97)	0.769	0.53 (0.18-1.53)	0.241
Honoraria/consulting fee receipt by author, yes	No	2.41 (0.99-5.88)	0.043	3.24 (1.06-9.88)	0.039
Non-phase 2	Phase 2	3.45 (1.41-8.45)	0.005	2.68 (0.88-8.11)	0.082
Number of patient enrolled***	D 1	1.59 (1.09-2.32)	0.013	1.73 (1.11-2.69)	0.016
Duration of study, months***	D 1	1.49 (0.95-2.32)	0.075	1.28 (0.78-2.10)	0.321
Adequate reporting of random sequence generation, yes	No	0.40 (0.18-0.89)	0.022	0.32 (0.13-0.78)	0.013
*OR: odds ratio; CI: confidence interval *RCTs with partial or complete funding by industry were considered as industry funded, while those with explicit non-profit source or unspecified funding source were considered non-profit. **log transformed

---

« Back to 2012 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/association-of-clinical-trial-characteristics-with-positive-study-outcome-reporting-in-randomized-controlled-trials-of-rheumatoid-arthritis-therapy/