Idealization of Youthfulness Predicts Worse Recovery Among Older Individuals

Abstract

This study examined whether stereotypes about an out-group could influence physical health. It had been previously shown that positive stereotypes held by older individuals about their in-group benefited physical health. However, the potential impact on physical health from idealizing their out-group, the young, through positive stereotypes had not been studied. The cohort consisted of 189 participants, aged 60 and older, who experienced a cardiovascular event: a myocardial infarction (MI). Participants reported their stereotypes about the young and the old at baseline. Their MI recovery was assessed with a physical-performance battery that was administered at 4 time points across 1 year following the event. As hypothesized, positive stereotypes about the young predicted significantly worse recovery and positive stereotypes about the old predicted significantly better recovery, after adjusting for relevant covariates. Considering out-group idealization as a risk factor could provide an innovative research and clinical tool.

Previous research has shown that positive stereotypes held by older individuals about their in-group can benefit physical health; for example, they have been found to prevent cardiovascular response to stress and benefit recovery from disability (Levy, Hausdorff, Hencke, & Wei, 2000; Levy, Moffat, Resnick, Slade, & Ferrucci, 2016; Levy, Slade, Murphy, & Gill, 2012; Westerhof et al., 2014). However, there had been no known investigation of whether stereotypes about an out-group could influence physical health. Because we considered it likely that positive stereotypes about the out-group, the young, would be stressful for older individuals, we examined whether these stereotypes would adversely affect their myocardial-infarction (MI) recovery. This cardiovascular outcome was selected because it tends to: reflect psychosocial factors, is susceptible to stress, shows considerable variability, and occurs within the 1-year time frame of the current study (Cohen, Edmondson, & Kronish, 2015; Levy, Slade, May, & Caracciolo, 2006).

The phenomenon of out-group preference by marginalized persons has a long research tradition. In one of their pioneering studies, Clark and Clark (1950) found a tendency of Black children to express a preference for White children. The authors explained that this was stressful because it generated “a fundamental conflict at the very foundation of the ego structure” (Clark & Clark, 1950, p. 350).

Similarly, older individuals tend to express an implicit preference for their out-group over their own age group (e.g., Chopik & Giasson, 2017; Hummert, Garstka, O’Brien, Greenwald, & Mellott, 2002; Jost, Banaji, & Nosek, 2004) that is potentially stressful. Concomitantly, older individuals are unique in holding implicit attitudes about their in-group that are as negative as the attitudes about it that are expressed by the out-group (Levy & Banaji, 2004; Nosek, Banaji, & Greenwald, 2002). Furthermore, older individuals tend to identify with their out-group as strongly as the young tend to identify with their own age group; this out-group identification is not found to the same extent in other marginalized group members, such as women and Blacks (Nosek et al., 2002). The lack of in-group identification was illustrated by a study that found older individuals tended to oppose federal programs that were designed to benefit them; the opposition was predicted by youth preference (Levy & Schlesinger, 2005). A misalignment of membership and reference groups can result in “personal conflict, uncertainty, or insecurity” (Sherif, 1953, p. 222).

These various anomalies may be partially attributed to a culture that facilitates out-group preference by stigmatizing the aging process and glorifying youthfulness. An example is provided by the widespread marketing of the anti-aging industry that uses positive stereotypes about the young and negative stereotypes about the old to promote the sale of biological and cosmetic products that are alleged to resist or reverse the aging process (Clarke, 2011; Levy, 2017). This type of advertising reflects and reinforces prevailing societal motifs of age stereotypes: debilitation of the old and invigoration of the young (Koppel & Berntsen, 2016; Levy, Slade, Kunkel, & Kasl, 2002).

The out-group preference of older individuals may also result in part from their being members of one of the rare marginalized groups with an out-group that is comprised of its preceding in-group. For the positive stereotypes about younger individuals that were internalized when they were the in-group and the negative stereotypes about older individuals that were internalized when they were the out-group are likely to be transported to old age (Levy, 2009). This process is facilitated when older individuals are unable to develop timely and adequate defense mechanisms against the carried-over stereotypes.

Taking into account these various factors that could contribute to the stressfulness of positive stereotypes about the young, together with the known health benefits of positive stereotypes about the old (e.g., Levy et al., 2012), we hypothesized that among older individuals: (a) more positive stereotypes about the young will lead to worse recovery from an MI; and (b) more-positive stereotypes about the old will lead to better recovery from an MI.

Method

Participants

The cohort consisted of 189 individuals who were recruited from, and initially interviewed at, four hospitals in southern Connecticut. The study was approved by the Yale Human Investigation Committee. Participants met the following criteria: age 60 years or older; experienced an MI, as defined by the Enhancing Recovery in Coronary Heart Disease study (ENRICHD) (Berkman et al., 2003; ENRICHD Investigators, 2000); doctors consented to their participation; responded to the stereotype measures within 2 weeks of the MI; agreed to provide physical-function measures at months 1, 4, 8, and 12 after the MI; and cognitively intact, as defined by receiving a score of 6 or better on the 10-item Short Portable Mental Status Questionnaire (Pfeiffer, 1975).

The baseline demographics of the participants were as follows: mean age of 70 years (range from 60 to 95 years, SD = 7.5); 41% female; 10% African American; and, on average, greater-than-high-school education (range from 6 to 20 years, M = 13, SD = 3.63). The baseline health of the participants consisted of: an average self-rated health of 3.04 (SD = .96) on a scale ranging from 1 to 5, with a higher score indicating better health; an average depressive-symptom score of 19.44 (SD = 6.23) on the 11-item Center for Epidemiologic Studies Depression (CES-D) scale (Radloff, 1977) ranging from 11 to 44, with a higher score indicating more symptoms; an average Charlson Comorbidity Index score of 5.11 (SD = 1.98) on a scale ranging from 2 to 12, with a higher score indicating more comorbidities (Charlson, Pompei, Ales, & MacKenzie, 1987); and an average premorbid-physical-function score of 4.22 (SD = 1.73), which indicates an ability to perform at least four of the six items comprising the Health Scale for the Aged (Rosow & Breslau, 1966). Approximately half (45%) of the participants experienced a more severe MI (ST-elevation MI, as recorded by an electrocardiogram; Kumar & Cannon, 2009; O’Gara et al., 2013). (See Table 1 for more details about baseline variables.)

Table 1.

Correlation Matrix of Baseline Variables

Variables	Mean (SD) or %	1	2	3	4	5	6	7	8	9	10	11
1. YS	3.02(1.11)	1.00
2. OS	2.65 (1.24)	−.08	1.00
3. Age	70.24 (7.50)	−.01	.01	1.00
4. Education	13.43 (3.63)	.09	.08	.07	1.00
5. Female	40.57%	.03	.08	.07	−.12	1.00
6. African American	10.05%	−.10	.07	.01	.05	.04	1.00
7. Dep	19.44 (6.23)	−.09	.06	−.12	.05	.12	−.05	1.00
8. MIS	45.00%	.03	.03	.03	.13	.11	.02	.04	1.00
9. SPPB	8.57 (2.68)	−.10	.10	−.37	.27*	−.28*	−.10	−.04	.03	1.00
10. CCI	5.11 (1.98)	.003	−.01	.40*	−.08	.08	.04	.05	.23*	−.14	1.00
11. SRH	3.04 (.96)	.02	.02	.08	.22*	.08	.10	.27	.08	.21*	.06.	1.00

Note. YS = young stereotype; OS = old stereotype; Education = years of education; Female = scored as 1 = male and 2 = female; African American = scored as 1 = not African American and 2 = African American; Dep = depressive symptoms; MIS = myocardial infarction severity, measured according to whether participants experienced an ST-elevation MI; SPPB = Short Physical Performance Battery; CCI = Charlson Comorbidity Index; SRH = self-rated health.

^*p < .05.

Participants with either positive or negative stereotypes about the young did not significantly differ on those demographic and health factors at baseline. Additionally, participants with either positive or negative stereotypes about the old did not significantly differ on those demographic and health factors at baseline.

Measures

Stereotype predictors.

At baseline, stereotypes about the old and the young were assessed with open-ended questions. To assess stereotypes about the old, participants were asked, “When you think of old people in general, what are the first five words or phrases that come to mind?” This assessment has been validated in a number of previous studies in terms of predicting an outcome among older persons in the expected direction (e.g., Levy & Langer, 1994; Levy et al., 2006, 2012). Stereotypes about the young were assessed with a new measure: “When you think of young people in general, what are the first five words or phrases that come to mind?”

Stereotypes about the young and the old were assessed for all participants. Three raters (aged 20, 51, and 82 years), who were not aware of the recovery of participants, rated responses on a scale from 1 (very negative) to 5 (very positive). The scores were averaged to create scores for both the young and the old stereotypes. The raters achieved an interrater reliability of .94. The young stereotypes had a mean of 3.01, range of 1–5, and SD of 1.11. The old stereotypes had a mean of 2.59, range of 1–5, and SD of 1.24. The young stereotypes were more positive than the old stereotypes, t = −3.47, p < .001.

The calculation of both the young- and old-stereotype measures as an average of five responses was found to be an effective approach to assess the age stereotypes. Compared with the final model that included average scores for both types of stereotypes, there was no significant improvement, in terms of explaining recovery variance, with the addition of: (a) standard deviations of the two age-stereotype measures; (b) interaction of the standard deviations of the two age-stereotype measures with their respective age-stereotype measure; (c) valence homogeneity (whether or not stereotype-item valences were all positive or all negative) of the two age-stereotype measures; and (d) interaction of the valence homogeneity with the respective age-stereotype measure.

The most prevalent themes of the various groups, with examples of extreme ratings, were: successful physical functioning for the positive-young-stereotype group (e.g., vitality scored 5); unsuccessful cognitive functioning for the negative-young-stereotype group (e.g., clueless scored 1); successful cognitive functioning for the positive-old-stereotype group (e.g., wisdom scored 5); unsuccessful physical functioning for the negative-old-stereotype group (e.g., frail scored 1).

Outcome:

Ml recovery.

The outcome of recovery from an MI was assessed by the Short Physical Performance Battery (SPPB; Guralnik et al., 1994) at months 1, 4, 8, and 12. The SPPB, which was created to assess physical functioning in older individuals, consists of timed balance, chair stands, and walking components, with performance scores ranging from 0 to 12 (Guralnik et al., 1994). This measure has been shown to have good reliability and validity in diverse samples of older individuals (Freire, Guerra, Alvarado, Guralnik, & Zunzunegui, 2012; Guralnik et al., 1994; Levy et al., 2006).

Covariates.

The covariates included in the models have been found to predict MI recovery (Khot et al., 2003; O’Gara et al., 2013; Rosengren et al., 2004). They consisted of the demographic variables of age, sex, race, and years of education as well as the health variables of MI severity, self-rated health, overall physical health derived from the Charlson Comorbidity Index (Charlson et al., 1987), and depressive symptoms that were assessed with the 11-item CES-D scale (Radloff, 1977). The CES-D, the Health Scale for the Aged, and the Charlson Comorbidity Index have been found to be reliable and valid health assessment tools (Karim, Weisz, Bibi, & ur Rehman, 2015; Quan et al., 2011; Reuben, Siu, & Kimpau, 1992). In addition, all models were adjusted for baseline physical function, assessed with the SPPB at 1 month, and premorbid physical function, defined as the response to how many of the six items comprising the Health Scale for the Aged (Rosow & Breslau, 1966) the participants were able to do “two weeks before coming to the hospital.” MIs were categorized as severe if they met the definition of an ST-elevation MI (Kumar & Cannon, 2009; O’Gara et al., 2013).

Procedures

Nurses affiliated with the Yale Program on Aging collected health data by medical-chart abstraction. They collected stereotype and demographic information by in-person interviews at baseline, within 2 weeks of the participants experiencing an MI. Without awareness of stereotype responses, the nurses assessed physical function in the homes of participants at months 1, 4, 8, and 12 after the baseline interviews.

Statistical Analysis

The young- and old-stereotype groups were compared according to background variables, using either Pearson χ² or Fisher’s exact tests for categorical variables and t tests for continuous variables. A longitudinal analysis that took advantage of all time points was performed with a generalized linear model to examine the relationship between the stereotype predictors at baseline and the outcome of physical-functioning slope during 1 year, based on assessments of physical function at months 1, 4, 8, and 12. That is, for each participant a physical-function slope was generated and this served as the dependent variable. In the model, the young and old stereotypes were included as continuous variables to achieve higher sensitivity. In addition, the model included both predictors and controlled for all the covariates. Because we had predictions about the direction of effects based on previous studies (e.g., Levy, 2009; Marques, Lima, Abrams, & Swift, 2014; Meisner, 2012; Westerhof et al., 2014), the p values were presented as one-tailed tests (Cumming, 2012). Also, to compare the relative contribution of variables to MI recovery, we examined the model with all variables standardized. All analyses were performed with SAS, version 9.3 (SAS Institute Inc., Cary, NC).

To illustrate the pattern of physical function over time (Rucker, McShane, & Preacher, 2015) as predicted by the young and old stereotypes, we plotted the linear functions of the SPPB scores of participants 1 SD above the mean (positive stereotype) and 1 SD below the mean (negative stereotype) of both the young- and old-stereotype predictors, adjusting for all covariates. In the old-stereotype model, we adjusted for young stereotypes, and in the young-stereotype model, we adjusted for old stereotypes.

Results

As hypothesized, stereotypes about the young and the old predicted MI recovery. When both types of stereotypes were included as continuous predictors in the same model, more-positive young stereotypes led to significantly worse MI recovery, [β = −.03, p = .01, ${\eta }_{\text{p}}^2=.03$ and more-positive old stereotypes led to significantly better recovery, β = .02, p = .03, ${\eta }_{\text{p}}^2=.02$, adjusting for relevant covariates (see Figure 1).

Figure 1.

Contrasting associations of positive age stereotypes with myocardial-infarction (MI) recovery. The figure is based on models with all covariates included. The lines in both figure panels represent the participants 1 SD above the mean (positive stereotypes) and 1 SD below the mean (negative stereotypes) of either the young-stereotype or old-stereotype predictor on the outcome of the Short Physical Performance Battery, assessed at months 1, 4, 8, and 12 after baseline.

The young and old stereotypes operated independently of each other. These variables were not significantly correlated, r = −.08, p = .19. Also, when the interaction term of the two types of age stereotypes was added to the overall model, it did not reach significance, β = .02, p = 08.

An examination of the relative impact of all the variables in the model that predicted recovery showed the positive-young-stereotype variable was the third strongest predictor, preceded by age and the 1-month physical-function assessment, and immediately followed by the positive-old-stereotype variable (see Table 2). Thus, the impact of the young and old stereotypes on MI recovery was stronger than MI severity, self-rated health, comorbidities, depression, education, premorbid physical function, race, and sex. The inverse impacts of the positive young and old stereotypes on recovery did not significantly differ from each other in absolute value of effect size when included in the same model.

Table 2.

Impact of Positive Young and Old Stereotypes on 1-Year Recovery Following MI

Variables	Beta (SE)	t	p value
Positive young stereotypes	−.03 (.01)	−2.27	.01
Positive old stereotypes	.02 (.01)	1.85	.03
Age	−.05 (.03)	−2.91	<.01
Education	.02 (.04)	1.00	.32
Female	.01 (.03)	.71	.48
African American	−.004 (.05)	−.30	.76
Depressive symptoms	−.01 (.02)	−.81	.42
MI severity	.02 (.03)	.76	.45
SPPB at 1 month	−.06 (.01)	−3.47	<.01
Premorbid function	−.01 (.01)	−.34	.73
Comorbidity index	−.002 (.01)	−.14	.89
Self-rated health	.003 (.02)	.18	.86

Note. MI = myocardial infarction; SPPB = Short Physical Performance Battery. One-year recovery was calculated as slope of the Short Physical Performance Battery, based on scores at months 1, 4, 8, and 12 for each individual. Female scored as 1 = male and 2 = female. African American scored as 1 = not African American and 2 = African American. Betas are standardized estimates.

Two composite variables that combined stereotypes about the young and old predicted the outcome in the expected way. We found that the first composite measure (based on dichotomizing the age stereotypes into those below and those at or above the mean) that ordered the age-stereotype combinations from the most likely to lead to recovery (negative young, positive old) to the least likely to lead to recovery (positive young, negative old) significantly predicted better physical recovery, β = .05, p = .007, ${\eta }_{\text{p}}^2=.03$. The second composite measure, composed of adding the values of the reverse-scored positive young stereotypes and the positive old stereotypes as continuous variables, also showed the expected pattern. A higher score on the composite measure predicted better physical recovery, β = .03, p = .002, ${\eta }_{\text{p}}^2=.05$.

Discussion

The current study demonstrated for the first time that stereotypes about an out-group can influence physical health. Specifically, we found, as hypothesized, that positive stereotypes about the young held by older individuals predicted worse MI recovery, while their positive stereotypes about the old predicted better MI recovery (see Figure 1). Operating in opposite directions, the positive stereotypes about the young and the old had equivalent and robust magnitudes of effect on worse MI recovery. Both types of stereotypes were stronger predictors of MI recovery than demographic variables, such as sex and race, and health variables, such as MI severity and comorbidities.

It is likely that the mechanism by which positive young stereotypes exerted their influence on MI recovery involved a level of stressfulness comparable with what previous research has shown can be generated by negative old stereotypes (Levy et al., 2000, 2016). The phenomenon of nostalgia, which entails romanticizing the past by means of distorted perceptions (Leboe & Ansons, 2006), may help to explain the stressfulness associated with young stereotypes. For participants holding positive young stereotypes may have experienced a stressful conflict between romanticization of youthfulness and reality of the present, which often includes exposure to incidents of ageism that proliferate in society. The impact of nostalgia could have been compounded by a sense of loss, as shown when participants followed their reporting positive young stereotypes with comments such as, “I wish I was young again.” Further, the prevailing themes of the positive young stereotypes suggest nostalgia was often directed at consequential matters: physical and cognitive health.

The results indicate that it is adaptive to simultaneously hold images of old age that are affirmative and to acknowledge that youthfulness can have challenges. The participants who showed the strongest recovery across the 1 year of follow-up were the participants that reported holding positive old stereotypes and negative young stereotypes. These two types of stereotypes contributed to recovery in an additive way.

The lack of a significant difference between the positive- and negative-age-stereotype groups at baseline is consistent with the participants responding to health measures right after they experienced an MI. It appears the life-threatening event was transcendent at that point. Yet, this was transitory: 1 month later the age-stereotype effects on recovery were found.

In conclusion, the same stereotype valence was shown to have an equally detrimental or beneficial impact on physical health, depending on whether the stereotype referred to an out-group or an in-group. The finding of an adverse consequence for recovery because of positive stereotypes about the young (e.g., invincible) identifies a previously unrecognized risk factor for poor health outcomes among older individuals. It has the potential to expand research and clinical options.