International Collaboration in Pediatric Critical Care RCTs
Duffett M, Choong K, Hartling L, Thabane L, Cook D.
Aims & objectives
Our objective was to describe the degree and patterns of international collaboration among pediatric critical care researchers who co-authored at least one randomized controlled trial (RCT).
We considered coauthors or study sites from 2 or more countries to indicate international collaboration. We used the Evidence in Pediatric Intensive Care database (epicc.mcmaster.ca) to identify RCTs, the authors’ affiliations from trial publications to identify country of origin, and the World Bank’s classification of country income levels.
We included 276 RCTs and 1507 researchers from 35 countries. 54 (19.6%) trials reported international collaboration. Of the 51 (18.5%) trials with coauthors from 2 or more countries, 31 (60.8%) were from only high-income countries (HIC), 19 (37.3%) were from at least one HIC and a middle-income country (MIC) or low-income country (LIC) and 1 (2.0%) was from only a MIC or LIC. 19 (6.9%) trials included study sites in more than 1 country. Although 77 (27.9%) trials included study sites in at least one MIC or LIC, only 2 (2.6%) of these trials included also included at least one study site in a HIC. Trials with international collaboration were more often multicentered (50.0% vs. 11.3%, p<0.001), but not larger (median [IQR] children randomized: 49 [26, 119] vs. 50 [30, 84], p=0.56) than other trials. In a network analysis of coauthorship, we identified a large cluster of 512 (33.9%) researchers (85.2% from HIC) publishing 81 (29.3%) of the trials. There were 22 smaller unconnected clusters of researchers publishing a median of 3 (minimum 2, maximum 10) trials each. There were also 106 (38.4%) trials published by researchers who were not connected by co-authorship to any other researchers.
A minority of trials in pediatric critical care are the product of international collaboration. Co-authorship networks are fragmented. Although there is a large cluster of researchers who are linked by coauthorship, most trials are conducted by researchers who have limited or no coauthorship links to researchers who have published other RCTs.