STUDY OBJECTIVE: To develop the Pharmacy-Based Disease Indicator (PBDI), and to evaluate its performance versus the diagnosis-based Deyo version of the Charlson Index in predicting subsequent-year hospitalization in adults. DESIGN: Retrospective cohort analysis. DATA SOURCE: Longitudinal health insurance database derived from the national health insurance system in Taiwan. PATIENTS: Two adult populations were identified: 697,823 individuals who were at least 18 years of age on January 1, 2005 (dataset 2005), and 714,072 who were at least 18 years of age on January 1, 2006 (dataset 2006). MEASUREMENTS AND MAIN RESULTS: Based on the Chronic Disease Score framework and the Anatomical Therapeutic Chemical classification system, we developed the PBDI, a comorbidity measure that is a function of 37 drug categories that correspond to major diseases in Taiwan. The relationship between individuals' PBDI score and subsequent-year hospitalization was evaluated by use of logistic regression models. Covariates in the models included age group, sex, PBDI score, and Deyo score. Using the two overlapping adult populations, we calculated both the PBDI score and the Deyo score for each individual in each year. Using subsequent-year hospitalization as the outcome and each comorbidity measure as the predictor, we demonstrated that the c statistic of the PBDI versus the Deyo version of the Charlson Index was 0.72 versus 0.69 for both the 2005 and 2006 populations. The Akaike information criterion, Bayesian information criterion, model calibration, and reclassification measures also confirmed the utility of the PBDI. CONCLUSION: The PBDI demonstrated acceptable predictive performance for subsequent-year hospitalization. It can be used as a general comorbidity measure to describe the health status of populations based on data derived from population-based automated health care databases.
STUDY OBJECTIVE: To develop the Pharmacy-Based Disease Indicator (PBDI), and to evaluate its performance versus the diagnosis-based Deyo version of the Charlson Index in predicting subsequent-year hospitalization in adults. DESIGN: Retrospective cohort analysis. DATA SOURCE: Longitudinal health insurance database derived from the national health insurance system in Taiwan. PATIENTS: Two adult populations were identified: 697,823 individuals who were at least 18 years of age on January 1, 2005 (dataset 2005), and 714,072 who were at least 18 years of age on January 1, 2006 (dataset 2006). MEASUREMENTS AND MAIN RESULTS: Based on the Chronic Disease Score framework and the Anatomical Therapeutic Chemical classification system, we developed the PBDI, a comorbidity measure that is a function of 37 drug categories that correspond to major diseases in Taiwan. The relationship between individuals' PBDI score and subsequent-year hospitalization was evaluated by use of logistic regression models. Covariates in the models included age group, sex, PBDI score, and Deyo score. Using the two overlapping adult populations, we calculated both the PBDI score and the Deyo score for each individual in each year. Using subsequent-year hospitalization as the outcome and each comorbidity measure as the predictor, we demonstrated that the c statistic of the PBDI versus the Deyo version of the Charlson Index was 0.72 versus 0.69 for both the 2005 and 2006 populations. The Akaike information criterion, Bayesian information criterion, model calibration, and reclassification measures also confirmed the utility of the PBDI. CONCLUSION: The PBDI demonstrated acceptable predictive performance for subsequent-year hospitalization. It can be used as a general comorbidity measure to describe the health status of populations based on data derived from population-based automated health care databases.
Authors: Thomas R Radomski; Xinhua Zhao; Joseph T Hanlon; Joshua M Thorpe; Carolyn T Thorpe; Jennifer G Naples; Florentina E Sileanu; John P Cashy; Jennifer A Hale; Maria K Mor; Leslie R M Hausmann; Julie M Donohue; Katie J Suda; Kevin T Stroupe; Chester B Good; Michael J Fine; Walid F Gellad Journal: Healthc (Amst) Date: 2019-04-26