| The 1984-1986 Longitudinal Study of Aging was used to evaluate respondent assessed health as a predictor of mortality and formal health service use in a sample of 3,697 participants aged seventy years and older. Based on previous studies, the perceived need for care was hypothesized to make the greatest contribution to explain variations in mortality and health care use. Logistic regression showed independent and predictive effects with early (1984-1986) and late (1986-1988) mortality for age, low health assessments, baseline functional status, sex, health behavior, and satisfaction with social life. Only age was a stronger predictor of mortality than low health assessments. The increased relative risks due to low health assessments for early mortality was 1.79 (1.43-2.16, 95% Confidence Interval) and late mortality was 1.58 (1.22-1.95, 95% Confidence Interval). These results are consistent with previous studies showing that low health assessments were equally predictive of early and late mortality.;Low health assessments were observed to have stronger longitudinal associations with mortality than with health service use. Low health assessments were equally predictive of mortality in both follow-up periods (1984-1986 and 1986-1988), yet the predictive effects with health care use were stable only in the first (1984-1986). Finally, cultural and individual differences exist among health perceptions that may differentially affect the perceived need for health care.;Furthermore, after controlling for many factors that affect the use of geriatric health services, low health assessments made independent contributions to explain the variations in the use of: 1985 Home Health Care, 1986 Hospitalizations, Physician Visits, and Nursing Home Admissions. Again, only age rather than low health assessments, was a better predictor of service use. The increased adjusted odds of service use due to these assessments were: 2.17 (1.54-2.81, 95% Confidence Interval) for 1985 Home Health Care; 2.01 (1.75-2.26, 95% Confidence Interval) for 1986 Hospitalizations; 1.67 (1.29-2.05, 95% Confidence Interval) for 1986 Physician Visits; and 2.47 (1.88-3.06, 95% Confidence Interval) for 1986 Nursing Home Admissions. The associations found between the predictor variable and 1986 Hospitalizations and Physician Visits were stronger than those determined in a previous study. |
