Time-varying nature of risk factors for the longitudinal development of disability in older adults with arthritis

Abstract

Background: To investigate changes over time in risk factors for the development of Activities of Daily Living (ADL) disabilities in older adults with arthritis.

Methods: The data were obtained from the Longitudinal Survey of Health and Living Status of the Elderly in Taiwan (1989-1999). The major analytic cohort comprised 977 older adults (458 men and 519 women) with arthritis and without ADL limitation at study baseline. A generalized estimating equations (GEE) model was used to analyze all temporally correlated errors, population-averaged estimates, and longitudinal relationships.

Results: Overall, the cumulative incidence of ADL disability in the analytic cohort was 17.4% during an observation period of 11 years. With respect to baseline risk, ADL disability was associated with older age, presence of comorbid chronic conditions, and poor self-rated health. However, the findings changed after accounting for the time-varying nature of risk factors and the temporal sequence of possible cause-and-effect relationships. In addition to the baseline predictors, a high score on the Center for Epidemiologic Studies Depression Scale, lack of regular exercise, and becoming widowed were associated with an increased risk of ADL disability and a decreased chance of recovery.

Conclusions: An understanding of the time-varying nature of risk factors for the disabling process is essential for the development of effective interventions that aim to maintain functional ability and prevent limitations among older adults with arthritis.

Figure 1.. Illustration of 3 longitudinal models used for analysis of risk factors for functional decline in older adults with arthritis. Model 1: Baseline measurements of risk factors were assumed to be related to functional decline reported at 3 follow-up points. Model 2: All risk factors were modeled using data collected at the examination preceding the assessment of the outcome variable (ADL disability). In this model, longitudinal causal relationships between ADL disability and risk factors were analyzed by using all available data, not only with respect to correlation in time sequence, but also with regard to correlation between time-dependent and time-independent variables. Model 3: Changes between 2 consecutive measurements of outcome and predictor variables were studied. We investigated relations between changes in values between different time points, rather than the actual values.

Figure 2.. Changes in mean activities of daily living (ADL) score during the 11-year follow-up period, by baseline arthritis status (higher scores indicate a greater number of ADL limitations; a score of 0 indicates no difficulty on any item).