Simulation Study to Investigate the Impact of Pre-Surgical Cost Analysis on Estimating Bariatric Surgery Cost-Effectiveness in the VA System
*Sandra Woolson, Center for Health Services Research in Primary Care Durham VA Medical Center
Keywords: Bariatric surgery, causal inference, collider variables, health care expenditures, cost effectiveness, simulation
Bariatric surgery induces significant weight loss among the severely obese, but the long-term cost-effectiveness of bariatric surgery remains uncertain. Obtaining unbiased effects on expenditures is challenging because bariatric surgery is often preceded by increased costs due to pre-surgical workup, which creates a collider variable in the causal path. Controlling for the inflated costs could produce results biased in favor of surgery. To examine this, we simulated cost data at one year prior and five years after surgery under various scenarios of confounding and pre-surgical workup costs. Log-normal regression models were fit with and without adjusting for pre-surgical costs. With no confounding, adjusting for pre-surgical costs created substantial bias and lower coverage probability for the effect of surgery. With both confounding and pre-surgical workup, adjusting for pre-surgical costs had varying results on bias and coverage based on the magnitude of workup and confounder effects. Therefore, it is crucial for the analyst to carefully consider the relationship among confounders and pre-surgical workup costs to accurately estimate long-term cost-effectiveness.