Risk Factors for Venous Thromboembolism in Nursing Home Residents

Abstract

OBJECTIVE To determine risk factors for venous thromboembolism (VTE) among nursing home (NH) residents. PARTICIPANTS AND METHODS This population-based historical cohort study used Minnesota Case Mix Review Program (MCMRP) Public Research Files to identify all unique Olmsted County, Minnesota, residents in any local NH from January 1, 1988, through December 31, 1994. Cases were defined as residents with a VTE diagnosis at any NH assessment. For each case, we identified 2 age- and sex-matched controls, defined as NH residents who did not have VTE but were assessed in the year of the case's earliest VTE diagnosis (ie, index assessment). Cases and controls were compared for preindex assessment characteristics. Accuracy of MCMRP VTE ascertainment was explored by comparing MCMRP cases with members of the Rochester Epidemiology Project incident cohort who were objectively diagnosed with VTE from January 1, 1988, through December 31, 1994, while residing in an Olmsted County NH. RESULTS At the preindex assessment, cases were more likely than controls to have returned from hospital; to require assistance with grooming, toileting, transferring, bed positioning, or wheelchair use; or to need physical therapy, rehabilitation, clinical monitoring, or wound care and were less likely to have neurologic disease. There were 53 MCMRP VTE cases vs 161 Rochester Epidemiology Project NH VTE cases; the proportion with pulmonary embolism was 21% vs 62%; 1-year mortality was 24% vs 55%. CONCLUSION Administrative data from NHs reveal important VTE risk factors not routinely documented in hospital or ambulatory records. However, ascertainment of VTE from NH administrative data appears biased toward surviving cases, highlighting concerns about using such data to assess provider quality and pointing to the need for studies that track individuals through multiple data sources across institutional settings. To determine risk factors for venous thromboembolism (VTE) among nursing home (NH) residents. This population-based historical cohort study used Minnesota Case Mix Review Program (MCMRP) Public Research Files to identify all unique Olmsted County, Minnesota, residents in any local NH from January 1, 1988, through December 31, 1994. Cases were defined as residents with a VTE diagnosis at any NH assessment. For each case, we identified 2 age- and sex-matched controls, defined as NH residents who did not have VTE but were assessed in the year of the case's earliest VTE diagnosis (ie, index assessment). Cases and controls were compared for preindex assessment characteristics. Accuracy of MCMRP VTE ascertainment was explored by comparing MCMRP cases with members of the Rochester Epidemiology Project incident cohort who were objectively diagnosed with VTE from January 1, 1988, through December 31, 1994, while residing in an Olmsted County NH. At the preindex assessment, cases were more likely than controls to have returned from hospital; to require assistance with grooming, toileting, transferring, bed positioning, or wheelchair use; or to need physical therapy, rehabilitation, clinical monitoring, or wound care and were less likely to have neurologic disease. There were 53 MCMRP VTE cases vs 161 Rochester Epidemiology Project NH VTE cases; the proportion with pulmonary embolism was 21% vs 62%; 1-year mortality was 24% vs 55%. Administrative data from NHs reveal important VTE risk factors not routinely documented in hospital or ambulatory records. However, ascertainment of VTE from NH administrative data appears biased toward surviving cases, highlighting concerns about using such data to assess provider quality and pointing to the need for studies that track individuals through multiple data sources across institutional settings.