BACKGROUND: Many states and municipalities are trying to estimate how many more services they will need in order to accommodate the so-called “silver tsunami.” These calculations are complicated by not only the changing numbers and demographics of older adults, but whether the health burden of near-future seniors will differ from those who are seniors now. A few cohort studies have used longitudinal data to identify changes over time in seniors’ health profiles but these have provided only national estimates, of limited value to state Departments of Elderly Affairs, community-based organizations, and other stakeholders in home- and community-based services (HCBS) trying to assess their planning needs. We show how states can use readily-available state-level data to conduct more specific assessments for planning purposes, rather than relying on national estimates.
METHODS: We created two datasets of the state’s Behavioral Risk Factor Surveillance System 15 years apart by pooling 1998-2000 and 2013-15 data. We identified key social determinants, HRQOL, and health outcome measures common to both datasets and used multivariate logistic regressions for two sets of analyses: first to compare the middle-aged (those 45-64) to adults over 65 within each cohort, and then to evaluate change between cohorts by calculating the adjusted odds ratios (AORs) for outcomes in the relevant age groups in 2011-13 compared to their peers 15 years earlier. Finally, we assessed whether racial/ethnic disparities had decreased at all between cohorts.
RESULTS: Among people 65 and older, the cohort effect was mixed, with higher AORs for some but not all outcomes relative to 15 years earlier. By contrast, “upcoming” seniors (those 45-65) in 2013-15 had higher AORs compared to their peers 15 years ago for all poor HRQOL measures and multiple other measures (e.g. AORs of 1.72 for diabetes [95% CI 1.36-2.19] and 2.95 for class II obesity [95% CI 2.20-3.96]).
CONCLUSIONS: As with previous cohort studies, ours yielded some contradictory conclusions regarding whether “incoming” seniors are likely to be healthier or worse off than current older adults. Overall, however, both cohort effects for the middle aged and intracohort relationships between middle-aged and older adult health measures provided guidance on how much more assistance the state’s next wave of seniors will need in order to remain in their homes, compared to seniors now. While imperfect, this model is a useful middle ground between using demographic data alone and relying on scarce and costly longitudinal health data.