Clinical and laboratory predictors of all-cause mortality in older population

Although some clinical and laboratory tests have been studied on their individual relationship with total mortality or cause-specific mortality such as cardiovascular mortality, the overall effect of these indicators on mortality has rarely been evaluated. The purposes of this study were to assess the relationship of clinical and laboratory measures and all-cause mortality and to evaluate their potential clinical importance in mortality prediction in older adults. A sample of 2086 persons aged 65 and older participating the population-based health examination in 1995 and 1996 in Kaohsiung City, Taiwan was followed until the end of 2003. All participants completed medical history and underwent clinical assessment and laboratory tests. Measures selected for analysis were pulse rate, blood pressure, height, weight, serum level of cholesterol, triglyceride, creatinine, and uric acid, fasting blood glucose (FBG), hemoglobin (HG) and red (RBC) and white blood cell (WBC) counts. Cox regression was used to select measures significant to total mortality. All participants were further classified into risk groups, based on disease history and values of measures identified from analyses, to evaluate mortality risk. A total of 409 deaths occurred during an average of 8.2 years of follow-up time. Among all 14 measures assessed individually, five (systolic blood pressure = SBP, creatinine, uric acid, FBG, and HG) were statistically related to total mortality. SBP (hazard ratio (HR) = 1.22; 95% confidence interval (CI) = 1.09-1.36), FBG (HR = 1.18; CI = 1.08-1.29), and HG (HR = 0.81; CI = 0.73-0.91) were further identified to have independent effect on total mortality in the multivariate analysis. Age- and sex-adjusted total mortality HRs for disease risk (with disease history but with normal biomedical values), biomedical risk (without disease history but with abnormal biomedical values), and combined risk groups (with disease history and with abnormal biomedical values) were 1.94 (CI = 1.22-3.10), 2.08 (CI = 1.57-2.76), and 2.45 (CI = 1.83-3.27) compared with low risk group (without diseases history and with normal biomedical values). Results from this study reveal the importance of incorporating clinical and laboratory measures on the assessment of mortality in older adults. Establishing mortality risk profile based on both diseases conditions and inexpensive biomedical measures (for example, SBP, FBG and HG identified in the study) may help physicians in evaluating older persons' prognosis. (C) 2007 Elsevier Ireland Ltd. All rights reserved.