Determinants of Citation impaCt for Large CLin- iCaL triaLs in CritiCaL Care

introduction: Clinical research is the foundation of quality patient care. Randomized controlled clinical trials (RCTs) provide the most reliable evidence of efficacy or harm, but are costly and time-consuming to conduct. As a result, in many fields, the majority of RCTs are initiated and run by the pharmaceutical or medical device industry. Concerns have been voiced that the resulting knowledge base is unduly influenced by commercial considerations. At the same time, influential trials have been increasingly led by investigator-led trials groups from around the world. We sought to understand the influence of funding model on the impact of large ICU-based clinical trials. methods: We conducted a systematic review of RCTs published in five high impact general interest medical journals, five critical care journals, and one pediatrics journal. We included all RCTs published between 1990 and 2009 that had randomized 100 or more patients and reported mortality and/or length of stay as an outcome. We excluded cluster RCTs and secondary publications from a parent study. We determined the funding model (industry vs peer review vs mixed) and the administrative model (single center, 2-5 multicenter, multicenter run by ad hoc group, multicenter led by a formally-constituted trials group, or industry run). Citation rates were obtained from ISI Web of Science in July 2013, and expressed as citations/year. Data are expressed as means ± SD or medians + IQR; statistical analysis used GraphPad Prism. results: We included 327 trials that recruited 159,869 patients (median 280; IQR 146-532). Both the number of trials (31; 74; 100; 122) and the median citation rates (3.9; 6.5; 10.6; 11.9) increased significantly over time, expressed in 5 year increments. Comparative effectiveness trials were more common than trials evaluating novel technologies (57.4 vs 42.6%), and cited more frequently (27.6+/-51.7 vs 15.4+/-27.0 citations, p=0.01). Industry funding was the primary support for 87 (26.8%) studies, while 159 (48.9%) received peer review funding or were unfunded; the remaining 79 (24.3%) were funded through blended models; data were not obtainable for 2 trials. The funding source did not influence citation rates. Citation rates were, however, strongly influenced by the organizational model. Reports from formally constituted trials groups were cited a median of 44.9 (IQR 12-77) times a year, significantly more frequently (p<0.0001) than single center studies (5.4; IQR 3-10), small multicenter studies (9.6; IQR 4-20), studies conducted by ad hoc groups (16.1; IQR 6-25), or studies conducted by industry (10.6; IQR 5-23). Of 11 trials cited more than 100 times per year, 5 were conducted by trials groups, whereas only 1 was led by industry. Conclusions: Investigator-initiated research is the most common model for critical care RCTs, with a primary focus on comparative effectiveness research. Studies conducted by dedicated trials groups have the highest impact reflected in annual citation rates. Expansion of this model should be actively promoted, and new funding models developed to reflect the impact that investigator-led research has in defining best practice in the care of critically ill patients.