(Unsuccessful) Binary Modeling of Successful Aging in the Oldest‐Old Adults: A Call for Continuum‐Based Measures

To the Editor: The underpinning premise upon which the successful aging (SA) paradigm is founded is the acknowledgment that health trajectories in later life are heterogeneous. Previous work1 differentiates usual aging from SA, highlighting alternative trajectories to linear decline. Going beyond the idea of merely “getting by,” SA focuses on achieving and maintaining a high level of functioning in a number of domains. Since the previous seminal work,1 these ideas have been further articulated, refined, and reframed into a myriad of SA conceptualizations. The current study aims to examine the prevalence of SA in a longitudinal population-based sample of older adults using a definition derived from recent systematic reviews of operational definitions2 and layperson perspectives of SA.3 The Cambridge City over 75 Cohort (CC75C)4 is a longitudinal population-based cohort study that began in 1985. Individuals in five selected primary care practices and one in three from a sixth practice in Cambridge aged 75 and older at baseline were eligible for inclusion. Follow-up interviews were conducted with surviving participants at 3, 7, 9, 10, 13, 17, 21, 23, and 25 years. Because of the absence of relevant variables, participants from the baseline survey were not included; all subsequent waves were included. A binary SA model was created using items identified by recent systematic reviews of operational definitions of SA2 and lay perspectives of SA.3 The most commonly identified components in both reviews were mapped onto the CC75C study using relevant variables included in the questionnaires and dichotomized based on reported level of functioning. Participants were identified as successfully aging if they felt positive or optimistic about the future, reported no loss of interest in usual activities, did not feel lonely, had good or very good self-rated health, demonstrated no disability in activities of daily living5 or instrumental activities of daily living,6 and had a Mini-Mental State Examination7 score of 27 or greater. Descriptive statistics were conducted in Stata 12 (Stata Corp., College Station, TX). The sample consisted primarily of women, with the percentage increasing with each wave of data collection. Fewer than 1% of participants met the criteria for SA in Waves 1, 2, and 3, and none met the criteria in Waves 4 through 8 (Table 1). Age, mean ± SD (range) 83.0 ± 4.1 (77.4–100.6) 85.9 ± 4.1 (81.1–102.7) 88.3 ± 3.3 (83.9–102.0) 91.3 ± 2.8 (87.6–105.3) 94.4 ± 2.4 (91.6–105.8) 97.4 ± 1.5 (95.4–101.4) 98.8 ± 1.1 (97.6–100.6) 101.1 ± 0.9 (100.2–102.4) Approximately 30% of participants fell short of the psychosocial SA criteria in each wave. The biomedical criteria increasingly excluded individuals, from approximately half of the participants in Wave 1 to all of the participants in Wave 8 (Table 1). The binary model used to examine the prevalence of SA in the CC75C identified fewer than 1% of study participants as successfully aging in each wave of data collection. These results highlight shortcomings in the dichotomization of SA and the inability of binary models of SA to articulate the heterogeneity of aging trajectories. The primary thrust of SA literature is the acknowledgment that certain individuals age particularly well (function at a particularly high level), but failure to meet the thresholds of these high levels of functioning led to the exclusion of the vast majority of study participants, suggesting that these expectations may be unrealistic for an older population. Although failure to meet the threshold for psychosocial items eliminated some individuals, failure to meet the biomedical components' thresholds eliminated the vast majority. Previous studies have suggested that the complete absence of functional disability as a criterion for SA in the oldest old adults is unrealistic;8 maintenance of “perfect” physical and cognitive functioning is rare. As a result of the inability of this binary model to articulate the nuances of SA in the oldest old adults, it is suggested that better methods of capturing SA be implemented: continuum-based SA models. These models allow individual changes to be identified more readily through the exploitation of granularity, which is not possible using binary models. Furthermore, these methods can be applied at the population level to more-accurately describe overarching trends in aging. The current study demonstrates that using a binary model of SA in the oldest old adults has several fundamental flaws that inhibit examination of the heterogeneity of SA trajectories. Individuals are eliminated from SA by being unable to achieve high levels of functioning, particularly in the biomedical domain. As a result of these methodological shortcomings, it is suggested that prospective models of SA be continuum based rather than disparate, dichotomous states. Conflict of Interest: The editor in chief has reviewed the conflict of interest checklist provided by the authors and has determined that the authors have no financial or any other kind of personal conflicts with this paper. Author Contributions: The study was designed and implemented by the CC75C. TDC conducted the statistical analysis and drafted the manuscript, which BS and CB edited and reviewed. Sponsor's Role: None.