| Description |
Hospital-associated venous thromboembolism (HA-VTE) is a serious complication of total hip (THA) and knee arthroplasty (TKA). HA-VTE can cause significant economic burdens. The economic burden of HA-VTE in THA/TKA patients might be inaccurately estimated in prior studies due to time-dependent bias. To investigate the impact of time-dependent bias, we estimated the impact of HAVTE on inpatient cost and length of stay (LOS) in THA/TKA patients using three different approaches including the "conventional" analysis, the "post-VTE" analysis, and the exposure incidence density sampling analysis. Furthermore, to investigate the impact of time-dependent bias on the costeffectiveness of different pharmacologic prophylaxis, we developed a decision analytic model to compare the cost-effectiveness of enoxaparin and rivaroxaban among THA/TKA patients. Without correcting time-dependent bias, the inpatient HA-VTE attributable costs were $21,304 (95% CI $16,692 to $25,916, P < 0.0001) and LOS was 6.39 days (95% CI 4.72 to 8.06, P <0.0001). After correcting for time-dependent bias, these estimates were $16,694 (95% CI $3,117 to $30,271, P = 0.0160) and 5.07 days (95% CI -3.51 to 13.66, P = 0.2465) in the "post-VTE" analysis, respectively. In the exposure incidence density sampling analysis, these estimates were $20,289 (95% CI $15,607 to $24,970, P < 0.0001) and 6.10 days (95% CI 4.39 to 7.82, P < 0.0001), respectively. iv The cost-effectiveness model showed that compared with enoxaparin, rivaroxaban is associated with cost savings and less VTE events regardless of the method used to estimate the cost of HA-VTE. We found that ignoring the impact of time-dependent bias increased the cost attributable to HA-VTE by 4.8% to 8.9%, and the LOS attributable to HA-VTE by more than 4.5%. The results of the cost-effectiveness analysis indicated that the impact of timedependent bias on a cost-effectiveness model might be limited. This study reveals that time-dependent bias could artificially inflate the results of cost of illness studies. Researchers should take the time-dependent nature of an exposure into consideration when estimating the impact of these types of exposures on health utilization outcomes. More research is needed to define the impact of time-dependent bias on the results of cost-effectiveness models. |
