Collaboration in Hand Surgery: Experiences From Silicone Arthroplasty in Rheumatoid Arthritis, Finger Replantation and Amputation Challenges in Assessing Impairment, Satisfaction, and Effectiveness, Wrist and Radius Injury Surgical Trial, and Surgery of the Ulnar Nerve

Collaborative hand surgery clinical research without sharing individual patient data; proof of principle study

High-quality research in hand surgery is increasingly important. A vital component is national and international multicenter collaborative research because of better generalizability and larger sample sizes. However, sharing patient data between centers can be hampered by regulations and privacy issues or reluctance to share patient data. Therefore, in this paper, we illustrate an approach for collaborative clinical research without sharing patient data while obtaining similar outcomes. To illustrate that this collaborative clinical research approach without sharing patient data leads to similar outcomes compared to aggregating all individual patient data in one database, we simulate an approach of performing meta-analyses on summary statistics of individual-center data. In the simulation, we compare the results to conventional analyses in an existing multicenter database of patients treated for Dupuytren's disease at three different centers with either limited fasciectomy (LF) or needle aponeurotomy (PNF). We share example data and all analysis code in a public GitHub Library. We found similar results for the meta-analysis approach without sharing individual patient data as in the conventional approach for 1) the proportion of patients treated for recurrences, 2) the Total MHQ score after both treatments, 3) the comparison of Total MHQ score after both treatments, and 4) the comparison of both treatments when correcting for confounders with regression analysis. CLINICAL SIGNIFICANCE: We illustrate how collaborative studies can be performed without sharing individual patient data while obtaining similar results as with conventional analyses. This approach can help speed up collaborative research without losing precision in outcome analysis.