Predictive validity of measures of comorbidity in older community-dwellers. The ICARe Dicomano Study

OBJECTIVES: To compare the ability of five measures of comorbidity to predict mortality and incident disability in basic activities of daily living (BADLs) in unselected older persons. DESIGN: An assessment of the data obtained from the Insufficienza Cardiaca negli Anziani Residenti a Dicomano (ICARe Dicomano) Study, a longitudinal epidemiological survey on heart failure in older people. SETTING: Dicomano, a small, rural town near Florence, Italy. PARTICIPANTS: The entire population aged 65 and older living in Dicomano, Italy, was enrolled in the ICARe Dicomano Study. MEASUREMENTS: At baseline (1995), comorbidity was quantified in 688 participants, based on clinical diagnoses, using disease count (DC), Charlson Comorbidity Index (CCI), Index of Co-Existent Diseases (ICED), and Geriatric Index of Comorbidity (GIC), or on drug use, using Chronic Disease Score (CDS). Incident ADL disability was assessed in 1999 and vital status in 2004. RESULTS: Mortality increased with the severity of comorbidity, with hazard ratios around 2 when comparing the highest and the lowest quartiles of DC, CCI, and ICED in Cox regressions adjusted for age, sex, and physical and cognitive performance. Prediction of mortality with GIC and CDS was only borderline significant. All measures predicted incident ADL disability; the strongest risk gradient (hazard ratio = 8.2 between the highest and lowest quartiles) was observed with ICED. Physical and, to a minor extent, cognitive performance added significantly to predicting mortality and incident BADL disability. CONCLUSION: All the measures of comorbidity predicted death and BADL disability in older community dwellers. DC, CCI, and ICED performed better than GIC and CDS. Physical performance measures are strong, independent contributors to the prediction of these outcomes