Self-Perceptions of Aging Moderate Associations of Within- and Between-Persons Perceived Stress and Physical Health Symptoms

Abstract

Objectives

To examine how self-perceptions of aging (SPA) moderated within- and between-persons perceived stress associations with physical health symptoms.

Methods

A community-dwelling sample of 103 adults (Mean_age = 63, range = 52–88) participated in an online microlongitudinal study for 100 days (N_occasions = 7,064). Participants completed baseline surveys consisting of SPA, social connections, and demographics followed by 100 daily surveys including information about daily stress perceptions and physical health. Utilizing generalized multilevel models, we examined whether daily fluctuations and average levels of perceived stress over 100 days affected physical health symptoms and whether these associations varied by SPA.

Results

Adults who had higher perceived stress, on average across 100 days, reported significantly more physical health symptoms compared to individuals with lower perceived stress on average (p < .05). On days when individuals reported higher perceived stress than their own average, they had a higher likelihood of reporting more physical health symptoms compared to days when their perceived stress was lower than their own average (p < .05). Further, SPA significantly moderated associations between both within- and between-persons perceived stress and physical health symptoms (ps < .05). Individuals with more positive SPA were less affected by high levels of perceived stress—both on average and on days when perceived stress was higher than their own average.

Discussion

More positive SPA significantly dampened the impact of perceived stress, suggesting the importance of SPA as an individual characteristic within stress processes. Future work should examine how daily changes in SPA may exacerbate or mitigate the impacts of daily stress processes and health outcomes.

Microlongitudinal, Stress processes, Subjective age

Perception of stress—the subjective appraisal of an experience as stressful (Lazarus & Folkman, 1984)—is associated with numerous long-term health outcomes such as increased blood pressure (Elbert et al., 1994) and coronary heart disease (Richardson et al., 2012). Further, greater perceived stress is associated with poorer short-term outcomes such as poor eating and exercise behavior (Li et al., 2020), perceived health (Whitehead & Blaxton, 2020), physical health symptoms (DeLongis et al., 1988), and cognitive functioning (Boals & Banks, 2012). Salleh (2008) suggests characteristics of stressful experiences such as perceived stress exacerbates health symptoms and illness as it can build up, become chronic, suppress the immune system, and can activate and continue heightened physiological arousal; thus, understanding how perceived stress affects daily health outcomes is an important avenue of research.

Daily physical health symptoms such as headaches, fatigue, and soreness are common throughout the life span, and are thought to be increasingly common with age. Importantly, in older adulthood, daily physical health symptoms are often ignored or written off as a normal component of aging; indeed, physical health symptoms such as pain are often undiagnosed, untreated, or ignored by medical professionals (Noroozian et al., 2018). Not only are physical health symptoms an indicator of illness or diagnoses (e.g., Daltroy et al., 1999), but day-to-day physical health symptoms have been associated with reporting chronic conditions and functional disability nearly 10 years later (Leger et al., 2015). While knowledge of the short-term and long-term implications of higher perceived stress on health outcomes has been noted, less is known about what might moderate these associations on a daily basis and, further, how perceived stress might affect daily reporting of physical health symptoms. Elucidating what might buffer the impact of perceived stress and physical health symptoms may be useful in informing targets for interventions related to stress and health.

Awareness of aging (AoA) may be one such buffering agent. AoA has been noted through adolescence and into young adulthood as a particularly impactful indicator on health outcomes (Diehl et al., 2015). Diehl and colleagues (2015) suggested that AoA has strong daily and developmental implications as individuals grow older. Five constructs encompass AoA: subjective age, aging identity, self-perceptions of aging (SPA), aging stereotypes, and awareness of age-related change. Moreover, Stereotype Embodiment Theory (Levy, 2009) explains the processes by which aging stereotypes can be internalized with detrimental or beneficial impacts on outcomes for older adults. Specifically, Stereotype Embodiment Theory suggests that these internalized aging stereotypes, termed SPA, may affect health through pathways such as stress. As such, SPA may be useful in buffering or exacerbating the impact of daily experiences on health outcomes. SPA refers to the internalized appraisals of one’s own aging experiences or processes and is comprised of two factors (positive and negative SPA; Turner et al., 2021). Researchers have found that positive SPA is related to better health behaviors (e.g., physical activity, healthy eating; Hooker et al., 2019), reduced mortality (Kotter-Gruhn et al., 2009), and decreased health care costs (Levy et al., 2020). Importantly, because SPA is multidimensional, revolves around personal experiences, is processed implicitly but can be expressed explicitly (Diehl et al., 2015), and can be targeted for intervention through multiple levels, it is an ideal modifiable resilience factor (Busso et al., 2019). Thus, this study aims to examine how both daily perceptions of stress and SPA affect daily physical health (i.e., physical health symptoms).

Daily Perceived Stress and Health Outcomes

A plethora of research reveals connections between perceived stress and health (e.g., Hellhammer et al., 2004; Rueggeberg et al., 2012). As early as the 1960s scholars explored how these appraisals may influence health and well-being. Notably, Lazarus and Folkman’s (1984) stress and coping theory suggests that the perceived stress related to an experience is just as impactful to health and well-being as objective stress. Indeed, decades of research shows that perceived stress is related to health and well-being outcomes such as impaired cognitive function (Munoz et al., 2015), emotional reactivity to daily stress (Stawski et al., 2008), positive and negative affect (Watson et al., 1988), social interaction (daSilva et al., 2021), general mental and physical health (Shavitt et al., 2016), perceived health (Whitehead & Blaxton, 2020), physical health symptoms (DeLongis et al., 1988), and health-promoting behaviors (Homan & Sirois, 2017) in multiple samples, measurement points (e.g., longitudinal, daily), and ages.

Stress researchers have examined how stress may affect physical health symptoms (Costanzo et al., 2012; DeLongis et al., 1988; Hager, 2009; Hoffman & Stawski, 2009; Laganà & Reger, 2009). Studies specifically examining associations between stress and physical health symptoms on a daily level (e.g., Costanzo et al., 2012; DeLongis et al., 1988; Hoffman & Stawski, 2009) often focus on exposure to daily stressors and health associations rather than perceived stress. Both Costanzo and colleagues (2012) and DeLongis and colleagues (1988) report that exposure to daily stressors is related to more physical health symptoms in both cancer survivors and with marital couples, respectively. Further, Hoffman and Stawski (2009) suggest within- and between-persons daily stress exposure predicts daily physical health symptoms. In highly specialized populations, both Hager (2012) and Laganà and Reger (2009) reported that perceived stress did affect general and physical health symptomatology; however, these were cross-sectional data, so understanding whether and how daily fluctuations in perceived stress affects physical health was not possible. Importantly, in midlife and older adulthood, how individuals perceive, manage, and react to stressors has been related to constructs related to SPA (e.g., subjective age; Levy, 2003; Levy et al., 2000, 2016). Indeed, some research has suggested that both subjective age and aging perceptions were related to both perceived stress and physical health outcomes (Wettstein et al., 2021; Whitehead & Blaxton, 2020).

SPA Associations to Stress Processes

For decades, research has shown that chronological age moderates associations between stress processes and health (e.g., Charles, 2010; Lazarus & Folkman, 1984). Theoretical perspectives on aging, however, have suggested that chronological age is not the only aging construct that may affect associations on health (cf., Diehl et al., 2015). Perceptions of aging (i.e., SPA) have been noted to be a particularly important mechanism to affect health and well-being (Levy, 2009). Indeed, more positive SPA has been related to the reduction of cardiovascular stress (Levy et al., 2000) while more negative SPA is related to increased cortisol (Levy et al., 2016). Particularly, Levy and colleagues (2000) found that following an implicit aging stereotype paradigm meant to prime negative aging stereotypes older adults reported significant increases in systolic and diastolic blood pressure and skin conductance. Moreover, Levy and colleagues (2016) found that older people with more positive aging stereotypes experienced less cortisol increase over 30 years.

SPA has further previously been associated with subjective indicators of stress such as reactivity to stress related to aging (Neupert & Bellingtier, 2017) and reactivity to daily stressors (i.e., negative emotional reactivity; Bellingtier & Neupert, 2018). Both studies examined one common measure of SPA (attitudes toward own aging; ATOA) at baseline and measured AoA-related change or daily stressors, ATOA, and daily negative affect over 8 days. Findings from Neupert and Bellingtier (2017) suggested that ATOA were significantly associated with more reactivity to negative affect related to AoA changes. Specifically, for AoA-related losses (e.g., feeling slower cognitively), individuals with more positive ATOA reported more negative affect related to these losses compared to individuals with less positive ATOA. Further, findings from Bellingtier and Neupert (2018) suggest that more positive ATOA were related to lower negative affect regardless of stressor exposure, but those with more negative aging attitudes reported more negative affect regarding stressors. Importantly, however, while changes in negative affect (e.g., emotional reactivity) have been recognized as a crucial mechanism by which daily stress processes affect health outcomes (Almeida, 2005), neither of these studies tangibly link the complex associations between stress and ATOA to health outcomes. Moreover, in line with Almeida’s (2005) daily stress process model, emotional reactivity and stressor exposure are but a few of the components that make up daily stress processes. Daily perceived stress is an additional characteristic of the daily stress process that may moderate the association between daily stressors and health and may be further moderated by SPA.

Recently, two studies, from Wettstein and colleagues (2021) and Whitehead and Blaxton (2020) explored how SPA and subjective age, which has been theoretically and empirically connected to SPA, might moderate associations between perceived stress and health (i.e., functional, and perceived general health) over time. The first noted study (Wettstein et al., 2021) focused on subjective age over two time points, 3 years apart; it is important to note that these authors reported perceived stress was associated with larger declines in functional health (e.g., walking several blocks, bathing), and that subjective age buffered the associations. Particularly, individuals who felt younger were affected less by perceived stress than those who felt older. These authors, however, did not explore how SPA might additionally modulate these associations, nor did they explore how daily perceived stress may affect daily physical health symptoms. Whitehead and Blaxton (2020), however, specifically examined how between- and within-person measures of SPA moderated associations between perceived stress and perceived health. Over 14 days, these authors reported significant associations between perceived stress, SPA, and worse perceived health such that both between- and within-persons, better SPA was related to lower perceived stress and perceived health. While this study highlights the particularly important relationship between perceived stress, SPA, and health, it does not examine physical health symptoms; further, it only examines stress as an outcome of daily health and SPA and does not explore the potential impacts stress has on health. As physical health symptoms may additionally affect outcomes such as quality of life (Tveito et al., 2004), and predicts the occurrence of future conditions and impairment 10 years later (Leger et al., 2015), understanding how both perceived stress and SPA affect physical health symptoms is an important avenue of research. Therefore, we aim to extend the work of these authors to explore how daily perceived stress affects daily physical health symptoms and may be moderated by SPA.

Intraindividual Variability and the Current Study

Intraindividual variability highlights, “that at a given moment [the individual] is a complex configuration of characteristics and that some of these characteristics are changing from moment to moment, day to day” (Nesselroade & Ram, 2004, p. 10). This crucial perspective to understanding the continually changing daily context has provided the groundwork for studying other developmental phenomena (e.g., daily stress, indices of personality). Utilizing an intraindividual, person-centered perspective may be particularly fruitful for research on physical health symptoms because of their acute nature. Importantly, because physical health symptoms are acute, they come and go at shorter intervals such as daily (Larsen & Kasimatis, 1991), the current study examines the impact of both within- and between-person levels of perceived stress on physical health symptoms; to our knowledge this is the first study to explore these associations among older adults.

Our within-persons research questions are twofold. First, we ask, “Are perceived stress and physical health symptoms positively related within persons?” We hypothesize that on days when individuals report more perceived stress, they will report significantly more physical health symptoms. Additionally, our second research question is, “Are within-person perceived stress and physical health symptom associations moderated by SPA?” We predict that individuals with higher SPA (in contrast to those with lower SPA) will report fewer physical health symptoms when their perceived stress is higher than their own average—conferring a protective effect of SPA. Similarly, our between-persons research questions are twofold. First, we ask, “Are those who have more perceived stress, on average, also more likely to report more physical symptoms over 100 days?” We hypothesize that individuals who report more perceived stress on average over the 100-day study will report more physical health symptoms on average. Our second research question asks, “Does SPA moderate between-persons perceived stress associations with physical health symptoms?” We hypothesize that individuals with higher SPA and lower perceived stress on average will report fewer physical health symptoms on average compared to individuals with lower SPA and higher perceived stress on average.

Method

We utilize data from the Personal Understanding of Life and Social Experiences study (Hooker et al., 2013), an online microlongitudinal study of a sample of community-dwelling older adults aimed at exploring questions related to self-regulation, personality, and daily goal pursuit. This project consisted of an initial phase to measure a host of demographic and baseline variables, followed by a daily phase of 100 daily surveys. Participants received a daily email and were encouraged to respond to the survey at the end of the day in order to reflect on the full day. Each daily survey included a host of measures including perceived stress and physical health. The sampling frame consisted of 100 days to balance participant burden with research suggesting that: (a) new behaviors require 3–6 months to avoid relapse (Gruber, 2010) and (b) at least 3 months of data are needed to obtain a reliable measure of intraindividual variability (Estabrooke et al., 2012). For methodological reasons, a smaller portion of the sample (25%) was randomly selected to participate in a measurement burst design (Stawski et al., 2015) and completed four 7-day bursts of measurement across the 100 days. Preliminary t tests indicated the daily and burst groups did not differ on any demographic variables (i.e., age, general health, education, gender, race) in this sample (p values all greater than p = .05); thus, they were combined for analytic purposes.

Participants

One hundred and five participants between the ages of 52 and 88 (M_age = 63, SD = 7.80) were recruited to complete both the initial and daily components of the study. Between the daily and burst participants, there was a total of 8,744 days of data; following the exclusion of missing data on the dependent variable, 7,064 occasions were available for analysis. The majority of the sample was White (97%), women (88%), and highly educated (75% with a 4-year degree or higher). While this demographic makeup is similar to Census data for older adults in Oregon, it is important to note that we utilized a nonrepresentative sample of convenience.

Initial Survey Measures

Self-perceptions of aging

Perceptions of aging were measured with the five-item Attitudes Towards Own Aging subscale in the Philadelphia Geriatric Center Morale scale (Lawton, 1975). Four items were yes (1) or no (0) responses to statements such as, “As you get older, you are less useful.” The fifth item asked participants to state whether things were getting better (2), staying the same (1), or getting worse (0). Negatively worded items were reverse-coded. Then, items were summed, where a higher score indicated a higher self-perception of aging (α = 68).

Daily Survey Measures

Perceived stress

Perceived stress was measured with four items on a sliding analog scale from 0 to 100 based on Cohen and Williamson’s (1988) perceived stress measure. Participants were asked to respond about how they felt that day. An example of an item is, “Today, I felt difficulties were piling up so high I could not overcome them.” Responses were averaged, with higher scores indicating higher perceived stress that day (α _{within-persons} = 0.71, α _{between-persons} = 0.84; Scott et al., 2020).

Physical health symptoms

Physical health symptoms were reported daily through an adapted symptom checklist (Winter et al., 2007). Participants were asked to check yes (1) or no (0) if they had felt any of 13 symptoms pertaining to either physical health (e.g., headache, diarrhea) or cognitive (i.e., trouble focusing, forgetting something) health. Given the focus of this study on physical health symptoms, we summed the 11 physical health items each day and omitted the two items pertaining to cognitive health. These 11 items were: fatigue, shortness of breath, trouble with mobility, allergy symptoms, poor appetite, dizziness, heart pounding, nausea/upset stomach, tightness in chest, constipation/diarrhea, or stiffness/muscle soreness.

Covariates

Given previous known associations with either perceived stress, SPA, or physical health (e.g., Almeida, 2005; Bellingtier & Neupert, 2018; Turner et al., 2021; Wettstein et al., 2021), this study covaried for day in study, group (0 = daily, 1 = burst), race (0 = White, 1 = other), gender (0 = male, 1 = female), marital status (1 = married, 2 = divorced or separated, 3 = single, 4 = widowed), general health (1 = poor, 2 = fair, 3 = good, 4 = excellent), chronological age, and education (1 = eighth grade or less, 2 = high school graduate, 3 = partial college, 4 = college graduate, 5 = graduate or professional degree, 6 = other).

Analytic Strategy

We utilized generalized linear multilevel models with random intercepts using SAS PROC GLIMMIX (SAS Institute, 2013) with maximum likelihood estimation using Gaussian Quadrature for all research questions. Using multilevel modeling was necessary for the within-persons analyses of perceived stress as the data resulted in a nested structure (Hoffman & Stawski, 2009). Due to the dependent variable being count data, and the distribution being skewed, we utilized a Poisson distribution in all of our models. Thus, our estimates represent how an increase in the predictor variable(s) are related to a multiplicative increase in the dependent variable rather than a traditional odds ratio.

Our model for research questions 2 and 4 examined the 2 two-way interactions between within- and between-persons perceived stress and SPA on daily physical health symptoms, utilized the following equation with the additions of Level 2 covariates previously mentioned:

$${\displaystyle \begin{array}{l}{\displaystyle \begin{array}{l}\mathrm{P}\mathrm{h}\mathrm{y}\mathrm{s}\mathrm{i}\mathrm{c}\mathrm{a}\mathrm{l}\mathrm{H}\mathrm{e}\mathrm{a}\mathrm{l}\mathrm{t}\mathrm{h}\mathrm{S}\mathrm{y}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{o}\mathrm{m}{\mathrm{s}}_{di}=b_{00}+b_{01}(\mathrm{P}\mathrm{e}\mathrm{r}\mathrm{c}\mathrm{e}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{d}\mathrm{S}\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{s}{\mathrm{s}}_{di})\\ +b_{02}(\mathrm{P}\mathrm{e}\mathrm{r}\mathrm{s}\mathrm{o}\mathrm{n}\mathrm{M}\mathrm{e}\mathrm{a}\mathrm{n}\mathrm{\_}\mathrm{P}\mathrm{e}\mathrm{r}\mathrm{c}\mathrm{e}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{d}\mathrm{S}\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{s}{\mathrm{s}}_i)+b_{03}(\mathrm{B}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{S}\mathrm{P}{\mathrm{A}}_i)\\ +b_{04}(\mathrm{P}\mathrm{e}\mathrm{r}\mathrm{c}\mathrm{e}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{d}\mathrm{S}\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{s}{\mathrm{s}}_{di}*\mathrm{B}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{S}\mathrm{P}\mathrm{A})\\ +b_{05}(\mathrm{P}\mathrm{e}\mathrm{r}\mathrm{s}\mathrm{o}\mathrm{n}\mathrm{M}\mathrm{e}\mathrm{a}\mathrm{n}\mathrm{\_}\mathrm{P}\mathrm{e}\mathrm{r}\mathrm{c}\mathrm{e}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{d}\mathrm{S}\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{s}{\mathrm{s}}_i*\mathrm{B}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{S}\mathrm{P}\mathrm{A})\\ +U_{0i}+{\mathrm{e}}_{di}\end{array}}\end{array}}$$

(1)

The above equation represents the number of physical health symptoms on day d for individual i as a function of a conditional intercept on between- and within-persons perceived stress, and baseline SPA. b₀₀ is the intercept representing physical health symptoms on days when perceived stress was at the mean. b₀₁ represents the time-varying association of perceived stress where the slope represents the same-day change in physical health symptoms. b₀₂ represents the person mean frequencies of perceived stress and was included to model individual differences in perceived stress when estimating the time-varying associations (Hoffman & Stawski, 2009). b₀₃ represents the change in physical health symptoms by the individual difference average of SPA. b₀₄ and b₀₅ represent the interactions between the time-varying perceived stress and SPA association and the time-invariant (person mean) association between perceived stress and SPA, respectively.

Results

Averages and correlations for the main variables (i.e., perceived stress, SPA, physical health symptoms) are reported in Table 1. Correlations between average levels of physical health symptoms, perceived stress, and SPA show significant correlations of moderate strength in expected directions. On average, higher perceived stress was related to lower SPA and higher numbers of physical health symptoms (ps < .0001); more positive SPA was additionally related to fewer health symptoms (p < .0001).

Table 1.

Descriptive Statistics of Perceived Stress, SPA, and Physical Health Symptoms

	N	Mean (SD)	Range
Average over time
Perceived stress	103	11.43 (7.29)	0.44–30.67
Physical health symptoms	105	1.35 (1.28)	0.01–7.16
Baseline
SPA	103	4.28 (1.67)	0–6
Correlations	1.	2.	3.
1. Perceived stress	—	—	—
2. SPA	−0.48**	—	—
3. Physical health symptoms	0.52**	−0.44**	—

Notes: SD = standard deviation; SPA = self-perceptions of aging. Correlations represent the relationship between the means of each variable after creating z-scores.

**p < .0001.

Preliminary analyses additionally examined a fully unconditional multilevel model to examine the intraclass correlation of physical health symptoms (Raudenbush & Bryk, 2002). In this model, we ran an empty model where physical health symptoms were included as an outcome with no predictors allowing us to understand the within- and between-person variability of the measure. Results indicated a significant amount of variance at the within-person level; the Intraclass Correlation Coefficient (ICC) ICC was 0.68, suggesting that 68% of the variance in physical health symptoms was attributable to between-persons differences, while 32% of the variance was attributable to within-persons.

Within- and Between-Persons Perceived Stress and SPA Main Effects

Both between- and within-persons perceived stress was significantly associated with physical health symptoms (ps < .05). Unstandardized estimates for covariate unadjusted and adjusted models are available in Table 2. Model 1 represents the unique effect of within- and between-persons perceived stress on physical health symptoms. Model 2 includes the unique effects of SPA on physical health symptoms, and Model 3 is the combined effects of perceived stress (both within- and between-persons) and SPA. When exponentiated, these estimates represent the multiplicative increase associated with an increase in the variables. For between-persons perceived stress, for every one-unit increase in perceived stress above the sample average, there is a multiplied expected number of physical health symptoms by e^0.06 or 1.06. Further, on days when individuals report a one-unit increase in perceived stress above their personal average (e.g., 1 SD above their average), there is an e^0.02 or 1.02 multiplicative increase in expected physical health symptoms. The effect of SPA was not significant (estimate = −0.05, SE = 0.07, p = .44). The nonexponentiated estimates can be seen in Figure 1.

Table 2.

Main Effects of Within- and Between-Persons Perceived Stress and SPA on Physical Health Symptoms

	Unadjusted for covariates			Adjusted for covariates
	Model 1	Model 2	Model 3	Model 1	Model 2	Model 3
	Estimate (SE)	Estimate (SE)	Estimate (SE)	Estimate (SE)	Estimate (SE)	Estimate (SE)
Intercept	0.02 (0.001)**	—	0.07 (0.02)**	0.07 (0.01)**	—	0.06 (0.02)*
	—	−0.27 (0.06)**	−0.12 (0.07)	—	−0.19 (0.07)*	−0.05 (0.07)
Between-persons
Perceived stress	0.09 (0.01)**	—	0.02 (0.002)**	0.02 (0.002)**	—	0.002 (0.002)**
SPA	0.02 (0.001)**	—	0.07 (0.02)**	0.07 (0.01)**	—	0.06 (0.02)*
Within-persons
Perceived stress	—	−0.27 (0.06)**	−0.12 (0.07)	—	−0.19 (0.07)*	−0.05 (0.07)

Notes: SE = standard error; SPA = self-perceptions of aging. Model estimates represent unstandardized coefficients in the fixed-effects model. For interpretability, exponentiating these estimates will provide you with the multiplicative effect on physical health symptoms of a one-unit increase. Models adjusted for covariates (right-side models) are adjusted for day of study, burst or day group, race, gender, marital status, education, chronological age, and health. Full covariate models can be seen in Supplementary Table 1.

**p < .0001. *p ≤ .05.

Figure 1.

Main effects of self-perceptions of aging (SPA), and between- and within-persons perceived stress from Model 3. All estimates are unexponentiated and significant at p < .05. BP = between-persons; WP = within-persons.

Within- and Between-Persons Perceived Stress Moderated by SPA

Table 3, Model 1 represents the unadjusted and adjusted fixed effects for within- and between-persons perceived stress and SPA after including two interactions between within-persons perceived stress and SPA and between-persons perceived stress and SPA. The interaction between between-persons perceived stress and SPA was significant (estimate = 0.02, SE = 0.01, p = .01). Figure 2 shows the multiplicative increase of between-persons perceived stress at −1 SD, +1 SD, and at the mean of SPA. At −1 SD below the mean of SPA, the multiplicative increase in between-persons perceived stress is e^0.51 or 1.67 (95% CI: 1.18, 2.36). This suggests that when individuals have less positive SPA, if they are additionally one unit higher in perceived stress on average the expected number of physical health symptoms reported would be multiplied by 1.67. This increase substantially decreases when individuals are at the average level of SPA for the sample and for +1 SD above average SPA.

Table 3.

Interactions Between Within- and Between-Persons Perceived Stress and SPA on Physical Health Symptoms

	Unadjusted for covariates	Adjusted for covariates
	Estimate (SE)	Estimate (SE)
Intercept	0.20 (0.63)	1.68 (1.29)
Between-persons
Perceived stress	0.02 (0.04)	−0.02 (0.04)
SPA	−0.31 (0.13)*	−0.34 (0.13)*
Perceived stress × SPA	0.01 (0.01)	0.02 (0.01)*
Within-persons
Perceived stress	0.01 (0.003)*	0.01 (0.003)*
WP perceived stress × SPA	0.002 (0.001)*	0.002 (0.0001)*

Notes: SE = standard error; SPA = self-perceptions of aging; WP = within-persons. Model estimates represent unstandardized coefficients in the fixed-effects model. For interpretability, exponentiating these estimates will provide you with the multiplicative effect on physical health symptoms of a one-unit increase. Models adjusted for covariates (right-side models) are adjusted for day of study, burst or day group, gender, marital status, education, chronological age, and health. Full covariate models can be seen in Supplementary Table 2.

**p < .0001. *p ≤ .05.

Figure 2.

Between- and within-persons perceived stress at differing levels of self-perceptions of aging (SPA) from Model 4. All estimates are exponentiated and significant at p < .05, numbers in parenthesis represent confidence intervals at the 95%. BP = between-persons; WP = within-persons.

Moreover, the interaction between within-persons perceived stress and SPA was significant (estimate = 0.002, SE = 0.001, p = .03). Figure 2 additionally includes the multiplicative increase of within-persons perceived stress at −1 SD, +1 SD, and at the mean of SPA. At −1 SD below the mean of SPA, the multiplicative increase in within-persons perceived stress is e^0.57 or 1.78 (95% CI: 1.17, 2.70). Thus, for individuals with positive SPA, on days when perceived stress is one unit higher than their average, it is estimated that the number of physical health symptoms reported should be multiplied by 1.78. For individuals with average SPA, on days when perceived stress is one unit higher than their average, it is estimated that the number of physical health symptoms reported should be multiplied by 0.72 (estimate = −0.33, SE = 0.13, p = .01) and this multiplicative increase decreases to e^−0.55 or 0.56 for individuals who have more positive SPA (+1 SD above the mean; estimate = −0.55, SE = 0.21, p = .01).

Exploratory Analyses

Provided that some research suggests bidirectional effects of physical health symptoms and stress processes (e.g., Wettstein et al., 2021), we examined the potential associations between within- and between-persons physical health symptoms and SPA on perceived stress (see Supplementary Tables 3 and 4). We utilized two-level multilevel models in SAS PROC MIXED with maximum likelihood estimation; we included both a random intercept and a random slope for physical health symptoms. For main effects models (see Supplementary Table 3), SPA and both the between- and within-persons effects of physical health symptoms were significantly associated with perceived stress (ps < .05). After including interactions between SPA and both between- and within-persons physical health symptoms (see Supplementary Table 4), the only significant effects were that of SPA (p < .05) and within-persons physical health symptoms (p = .02). That is, the association between physical health symptoms and daily stress did not depend on SPA.

Discussion

Reporting physical health symptoms becomes increasingly common into old age and symptoms are related to reporting more chronic health conditions up to 10 years later (Leger et al., 2015). The aim of this study was to explore how perceived stress and SPA affect the daily reporting of physical health symptoms. More perceived stress on average, and days when individuals felt more perceived stress, were uniquely and significantly related to more physical health symptoms. Further, more positive SPA on average was significantly associated with reporting fewer physical health symptoms on average. Finally, SPA significantly moderated associations between both between- and within-persons perceived stress and reporting physical health symptoms. Specifically, on days when individuals reported higher perceived stress than usual, having more positive SPA significantly decreased the impact on reporting physical health symptoms. Similarly, when individuals reported more perceived stress on average there was a difference of almost three times the increase in physical health symptoms when SPA was less positive. These results provide important new information to current research pertaining to the impact of both SPA and perceived stress on physical health.

SPA as a Modifiable Resilience Factor

Our results are in line with previous literature reporting that subjective age and SPA moderate associations between daily perceived stress and functional and general perceived health (Wettstein et al., 2021; Whitehead & Blaxton, 2020). The aforementioned studies found that other indicators such as feeling younger (i.e., subjective age) buffered the impact on general perceived health. The current study further clarifies, however, that this buffering related to SPA extends to associations between daily perceived stress and daily physical health symptoms.

Research previously hypothesized that perceptions of aging are related to situation selection pertaining to stressful experiences (Neupert & Bellingtier, 2017) and could function as personal vulnerability factors to react negatively to stressful events (Bellingtier & Neupert, 2018). Theoretical perspectives developed to explain age differences in the impact of stress processes suggest that as individuals get older, they are better at avoiding stressful situations; however, when a stressful situation does occur, older adults are just as reactive to the stressors as a younger individual (Charles, 2010) and health consequences may be more likely in this age group. Interestingly, the current research might suggest that in addition to chronological age affecting the stress process as Charles (2010) suggests, AoA might also affect the stress process. While individuals with more positive SPA may not attribute their physical health symptoms or stressors to old age, individuals that feel as if their aging is not optimal may be more vulnerable to stressors and the stress process and the physiological reactions related to the stress process. Perhaps those with lower SPA have been more likely to experience age discrimination, which has been shown to be associated with SPA (Hooker et al., 2019).

Moreover, commonly acknowledged theoretical pathways suggest SPA affects health outcomes through health behaviors. For example, Hooker et al. (2019) found that worse SPA is prospectively related to a lower likelihood of engaging in regular physical activity and an increased likelihood of smoking. Our study, however, suggests that SPA may additionally buffer stress and health associations through psychological processes. Indeed, previous research reports that constructs related to SPA (e.g., subjective age) are related to resilience to stressors (Bellingtier & Neupert, 2018; Hoffman et al., 2015; O’Brien & Sharifian, 2019). Future research should explore this more fully as a means to understand how SPA may affect other aspects of the stress process, regardless of the focus of the stressor (e.g., pertaining to aging or not).

Between- and Within-Persons Perceived Stress

Our results are additionally in line with similar research on physical health suggesting that both the between- and within-persons perceived stress significantly affect physical health symptoms (Shavitt et al., 2016; Whitehead & Blaxton, 2020). The current study, however, extends research on perceived stress and physical health associations by focusing on the reporting of physical health symptoms. Daily physical health symptomatology is a critical means for physician diagnoses of chronic health conditions (Daltroy et al., 1999), and may be an indication of mental health illness (e.g., Tveito, 2004). Thus, understanding that perceived stress may increase the reporting of these symptoms may be important for physicians and other medical professionals. Moreover, as previously mentioned, physical health symptoms predict chronic health conditions not just within days, but over years (Leger et al., 2015). Understanding what affects the reporting of daily physical health symptoms, then, becomes an important avenue of research for potential intervention research on chronic health conditions.

Bidirectional Effects

Provided bidirectional associations between stress processes and physical health have been noted in previous literature (Wettstein et al., 2021), we explored the possibility that SPA additionally moderated physical health effects on perceived stress. Indeed, models suggested a bidirectional effect such that perceived stress was associated with physical health symptoms and vice versa. However, SPA did not equivalently moderate these associations. Specifically, SPA differentially affected the associations between perceived stress on physical health symptoms but did not significantly moderate the association between physical health symptoms on perceived stress. In the Stereotype Embodiment Theory, Levy (2009) suggests that SPA affects health through stress pathways. Though bidirectional associations may exist, our results in this study support the perspective that perceived stress and SPA are antecedent predictors of health outcomes.

These results may additionally highlight the impact of SPA on different types of predictors (e.g., psychological vs health). In particular, SPA may more effectively buffer psychological behaviors, such as perceptions of stress, compared to physical health symptoms. One reason for this may be the priming effect physical health symptoms have on acknowledging stereotypes about aging. For example, an individual may report a physical health symptom and be more likely to reflect on their aging process as a result (e.g., I’m not aging well). Conversely, stress may not be directly tied to aging processes or stereotypes of aging.

Implications

SPA are a modifiable resilience factor to target for interventions regarding daily health outcomes (Busso et al., 2019). Busso and colleagues suggest that even simple interventions such as reading reframed aging messages to focus on the contributions to society in older age, age as an accumulation of wisdom and energy, or the understanding of prejudice against older adults resulted in short-term changes to implicit bias surrounding aging. While interventions such as this have only begun to gain traction in the field (Busso et al., 2019), utilizing SPA as a mechanism by which to intervene may be fruitful for perceived stress and physical health associations. Promoting positive views of individual aging and decreasing daily levels of perceived stress may provide a means for promoting health from day to day.

Limitations and Future Directions

While this study has many strengths, it is important to acknowledge limitations. First, SPA was only measured at baseline; thus, examining the day-to-day fluctuations in SPA was not possible. Research has begun to explore how SPA and related constructs may vary across days (e.g., Bodner et al., 2021) and as such, understanding how daily fluctuations in SPA may affect other daily processes (e.g., stress, health) is an important avenue of research. Moreover, because this study only included SPA at baseline, we had a relatively small sample size for the between-persons analyses. Although we did find significant results in this study, larger and more representative samples may show even stronger relationships. Finally, symptom severity of the physical health symptoms was not measured. Future studies should include this additional information to better understand how these daily physical health symptoms might affect behaviors or motivation.

Additional future directions might explore the bidirectional effects of daily SPA and daily perceived stress on health outcomes. Importantly, perceived stress may affect individuals’ perceptions of aging negatively. For example, perceiving stress at work or home may result in an individual reflecting on how well (or how poorly) they are aging and that in turn may result in more perceived stress when experiences occur. Moreover, given age differences in SPA and the impact of perceived stress (Diehl et al., 2021 and Scott et al., 2013, respectively), it may be fruitful to extend this work to examine how chronological age might further moderate associations between SPA and perceived stress. Particularly, individuals who are chronologically older may see a stronger protective factor of SPA compared to older or younger adults who have lower SPA.

This study aimed to elucidate whether SPA moderated associations between within- and between-persons perceived stress and physical health symptoms. SPA clearly moderated associations and, as evidenced by these results, better SPA buffers the association between both within- and between-persons perceived stress on physical health symptoms. Our findings further suggest that daily and average levels of perceived stress do have detrimental impacts on physical health symptoms, which adds to the rich literature on perceived stress and functional and general health. To this end, identifying ways to promote better SPA (Diehl et al., 2020) has considerable promise for understanding daily stress and physical health over the life span.

Acknowledgments

We would like to thank and acknowledge the efforts of the participants of this study as well as those who contributed to data collection and data management for the project. Further, portions of this study were presented at the Gerontological Society of America’s (GSA) conference in Phoenix, AZ in 2021. This study was not preregistered. Researchers who would like access to the data can contact Dr. K. Hooker ([email protected]) to discuss the submission of an institutional review board proposal for a specific project. Additionally, analytic methods and study materials can be made available upon request of the first author.

Funding

This study was funded by the Center for Healthy Aging Research at Oregon State University and the Jo Anne Leonard Petersen Endowment in Gerontology and Family Studies.

Conflict of Interest

None declared.