Roman Hayda, MD, COL (ret)

We read with interest the article by Maxwell and Mirza.¹ We appreciate using the large National Surgical Quality Improvement Project (NSQIP) database to assess comanagement outcomes, although we have concerns about the study design. Propensity score–matching (PSM) studies are limited; PSMs generate an average effect that neither establishes whether a treatment is optimal for a given patient nor control for unmeasured confounders.² Some baseline characteristics suggest that the comanaged and noncomanaged populations are quite different and, therefore, likely had unmeasured confounders that contributed to not detecting true effects. Also, as suggested by the authors, the NSQIP definitions of comanagement and standardized hip fracture program are broad. Recent studies in hip fracture comanagement attribute best outcomes to an organized program, shared decision making, expert comanagers, and each service having full responsibility including writing their own orders.^3-5 As no large database captures this distinction, it is not yet possible to perform a large, multicenter analysis. This type of comanagement cannot be studied in a randomized controlled trial. We recommend caution in overinterpreting the conclusions because there is substantial evidence in favor of optimized comanagement.

We read with interest the article by Maxwell and Mirza.¹ We appreciate using the large National Surgical Quality Improvement Project (NSQIP) database to assess comanagement outcomes, although we have concerns about the study design. Propensity score–matching (PSM) studies are limited; PSMs generate an average effect that neither establishes whether a treatment is optimal for a given patient nor control for unmeasured confounders.² Some baseline characteristics suggest that the comanaged and noncomanaged populations are quite different and, therefore, likely had unmeasured confounders that contributed to not detecting true effects. Also, as suggested by the authors, the NSQIP definitions of comanagement and standardized hip fracture program are broad. Recent studies in hip fracture comanagement attribute best outcomes to an organized program, shared decision making, expert comanagers, and each service having full responsibility including writing their own orders.^3-5 As no large database captures this distinction, it is not yet possible to perform a large, multicenter analysis. This type of comanagement cannot be studied in a randomized controlled trial. We recommend caution in overinterpreting the conclusions because there is substantial evidence in favor of optimized comanagement.

We read with interest the article by Maxwell and Mirza.¹ We appreciate using the large National Surgical Quality Improvement Project (NSQIP) database to assess comanagement outcomes, although we have concerns about the study design. Propensity score–matching (PSM) studies are limited; PSMs generate an average effect that neither establishes whether a treatment is optimal for a given patient nor control for unmeasured confounders.² Some baseline characteristics suggest that the comanaged and noncomanaged populations are quite different and, therefore, likely had unmeasured confounders that contributed to not detecting true effects. Also, as suggested by the authors, the NSQIP definitions of comanagement and standardized hip fracture program are broad. Recent studies in hip fracture comanagement attribute best outcomes to an organized program, shared decision making, expert comanagers, and each service having full responsibility including writing their own orders.^3-5 As no large database captures this distinction, it is not yet possible to perform a large, multicenter analysis. This type of comanagement cannot be studied in a randomized controlled trial. We recommend caution in overinterpreting the conclusions because there is substantial evidence in favor of optimized comanagement.