Understanding variation in the adoption of a new technology in surgery

Despite data supporting the superiority of laparoscopy over conventional surgery for donor nephrectomy, prior work reveals that this technology's adoption has been slow. However, the underlying cause for its gradual adoption remains unclear. For this reason, a multilevel analysis was conducted using a population-based cohort to evaluate the amount of variation in laparoscopic donor nephrectomy use attributable to the individual, provider, and hospital levels.

Using the Nationwide Inpatient Sample (2000-2003), discharges for donor nephrectomies were identified, distinguishing between those surgeries performed by laparoscopic versus conventional approaches. To examine variation in laparoscopy use, we fitted generalized linear mixed models with random intercepts. The principal model had no explanatory variables included and allowed for an understanding of the basic partitioning of the data's variability. Patient-, provider-, and hospital-level factors were then added to the model to determine how much variability at each level might be explained by these covariates.

In total, 974 conventional and 516 laparoscopic donor nephrectomies were performed by 157 surgeons at 46 hospitals. Among the different levels of care, the provider and hospital, combined, accounted for the majority of the variation in laparoscopy use (27.6% and 35.2%, respectively). After adjustment, the attributable variation at the provider level was largely unchanged (27.3%), whereas the hospital-level estimate decreased to 26.8%.

These data reveal that the provider and hospital contribute substantially to the variability in laparoscopy use among kidney donors, suggesting that interventions focused at these levels would likely hasten laparoscopy's uptake in the community.