Perceived Stress and Elder Abuse: A Population-Based Study of Adult Protective Services Cases in Chicago

Abstract

OBJECTIVES:

To characterize the relationship between perceived stress and Adult Protective Services (APS) elder abuse cases in a population-based sample.

DESIGN:

Cross-sectional.

SETTING:

Chicago.

PARTICIPANTS:

community-dwelling, older adults (N = 8,558; mean age 74 ± 7, 62% female, 64% African American).

MEASUREMENTS:

Perceived Stress Scale (PSS), APS database linkage.

RESULTS:

Bivariate and adjusted analyses showed that perceived stress was significantly higher in APS clients than in participants without any APS interaction for various subtypes of abuse (e.g., abuse from a perpetrator and self-neglect). This relationship was strongest for those with a confirmed history of abuse from a perpetrator, with a medium-large effect size (t = −5.8, P < .001, Cohen D = −0.6). Those in the highest stress tertile had a likelihood of having confirmed history of abuse from a perpetrator that was nearly three times as great as that of those in lower stress tertiles (odds ratio = 2.7, 95% confidence interval = 1.2–6.2). Analyses of individual PSS items revealed a robust relationship between distress items and APS involvement. Items reflecting coping were inconsistently associated with elder abuse.

CONCLUSION:

Clients of APS have higher levels of perceived stress, and abuse from a perpetrator strengthens this relationship. Therefore, victims of abuse from a perpetrator may be at the highest risk of stress-related consequences and should be targeted for intervention efforts that enhance empowerment and effective coping strategies.

Elder abuse is a prevalent yet underappreciated civil rights violation. It has been estimated that one in 10 older adults in the United States experience some form of abuse, including physical, emotional, and sexual abuse, as well as financial exploitation and neglect.^1–3 In low- to middle-income neighborhoods, the prevalence of such abuse is expected to be even higher given greater socioeconomic pressures and poorer access to resources such as community support and health care. Elder abuse is a major public health concern and has been associated with physical and psychological morbidity and mortality.^4–7 Although the association between elder abuse and poorer health outcomes has been well established, the mechanisms underlying this relationship are unclear.

Despite its widespread effect, only a fraction of these cases reach the attention of Adult Protective Services (APS). APS agencies investigate cases of elder abuse and intervene when cases are substantiated to resolve abusive situations. The types of elder abuse cases that APS handles are diverse and include abuse from a perpetrator (physical, emotional, and sexual abuse; financial exploitation; neglect) and self-neglect. For conceptual clarification, the Administration on Aging⁸ and the Elder Justice Act⁹ include abuse from a perpetrator and self-neglect under the umbrella construct of elder abuse. The majority of elder abuse cases reported to APS are cases of self-neglect.¹⁰ Given the complex psychosocial and individual-level mechanisms underlying various types of abuse, it is important to examine elder abuse as a global construct and according to its specific subtypes.

It is likely that psychological stress is an important contributor to deleterious health outcomes related to elder abuse. Elder abuse can be conceptualized as an acutely and chronically stressful event for the victim. Perceived stress, a construct representing the subjective experience of distress, may be particularly salient. The transactional model of stress¹¹ posits that individuals make primary and secondary appraisals when presented with stressful environmental challenges. Primary appraisals relate to the significance of the stressor to one’s well-being. Secondary appraisals relate to the controllability of the stressor and one’s perceived ability to cope. Therefore, stressors that are appraised to be threatening and uncontrollable are likely to trigger the greatest psychological and physiological response. Therefore, one’s individual appraisal of how stressful the abuse is may be more important in the context of health than simply examining objective measures of such events (e.g., number of occurrences). Moreover, one can extrapolate from this theory that one’s level of perceived stress associated with elder abuse depends on the type of abuse. Specifically, a person victimized by a perpetrator may experience more perceived stress given that he or she has less perceived control over the situation than someone engaging in self-neglect.

The relationship between perceived stress and health has been well documented. High perceived stress levels have been associated with poor mental health outcomes such as depression^12,13 and anxiety.^14,15 They are also associated with poorer physical health, particularly coronary heart disease.¹⁶ Perceived stress has also been shown to be a predictor of mortality in older adults.^17–20 Pathways that have been proposed linking stress to deleterious downstream health outcomes include stress-related neuroendocrine dysregulation, poor attention to health, lack of engagement in healthy behaviors, and use of unhealthy coping strategies, including smoking, alcohol, and emotional eating.

The literature examining perceived stress and elder abuse is sparse. One study²¹ examining the association between psychological distress (an aggregate variable including perceived stress, depression, and anxiety) and elder abuse (self-reported in a study interview format) in more than 3,000 older adults found that those reporting elder abuse had higher levels of distress than those without such abuse. Other research has examined the association between elder abuse and psychological distress more generally or in the context of specific psychiatric diagnoses and symptoms, such as depression.^22,23

The present study aimed to characterize the relationship between perceived stress and elder abuse in a population-based sample of community-dwelling older adults in Chicago. It examined this relationship using the criterion standard method of linking study data to APS data sets to capture reported and confirmed cases of elder abuse. This is the first study to examine the relationship between perceived stress and elder abuse independently and controlling for theoretically and empirically salient covariates. It was hypothesized that perceived stress would be higher in APS clients with reported and confirmed abuse. It was also predicted that these relationships would be present in addition to the effects of demographic variables, medical comorbidities, cognitive function, and physical function and that perceived stress would be higher in APS clients with histories of abuse with a perpetrator than in those with histories of self-neglect, given that victims of abuse with a perpetrator may find the circumstance to be less controllable, thereby affecting their secondary appraisals of the event(s).

METHODS

Design and Participants

Participants were 8,558 community-dwelling older adults aged 65 and older enrolled in the Chicago Health and Aging Project (CHAP), a population-based study of individuals living in three adjacent Chicago neighborhoods. First, a complete census of the community was taken. Of all age-eligible residents, 79% participated in a baseline population interview, a higher participation rate than most recent epidemiological studies of this size and scope, which reduces bias. Details of the CHAP study design have been described previously.^24,25 In brief, participants completed in-home interviews including standardized questionnaires assessing demographic information, medical history, cognitive functioning, physical functioning, and psychosocial variables. The Rush University institutional review board approved this study protocol, and all participants provided written, informed consent for participation.

Elder Abuse

Elder abuse was examined in multiple ways. It was examined as a broad construct including all cases of abuse and neglect (from here on referred to as “elder abuse” as an umbrella term) reported to or confirmed by APS. It was also divided into groups based upon cases with the presence or absence of a perpetrator. It was decided to examine these groups separately given that elder self-neglect (neglectful behavior of an older adult that poses a risk to his or her own health and safety without the involvement of a caregiver or trusted other) is the most common subtype of elder abuse and has been conceptualized as a distinct construct with unique surrounding qualities.²⁶ Exploratory analyses also examined specific subtypes of abuse with a perpetrator.

Elder abuse with a perpetrator refers to cases of abuse imposed on an older adult by another individual (perpetrator), including physical abuse, emotional abuse, sexual abuse, financial exploitation, and caregiver neglect. Physical abuse was defined as incurring physical pain or injury from a perpetrator. Emotional abuse was defined as verbal assaults, threats, harassment, or intimidation in a coercive manner. Sexual abuse was defined as nonconsensual, forced, or threatened sexual activity. Financial exploitation was defined as the misuse or withholding of an older adult’s resources in a way that disadvantages the older adult and benefits the perpetrator. Caregiver neglect was defined as failure by a caregiver or trusted other to provide the older adult with necessities of life, including food, shelter, and medical care. Self-neglect was defined as the failure of an individual to provide him- or herself with necessities of life (in the absence of any perpetrator or trusted other).^27,28

Illinois APS data were linked to the CHAP data set. Details on APS data linkage to CHAP participants have been published previously.^4,29,30 Elder abuse cases were characterized as reported and confirmed. Reported abuse included all cases opened by or reported to APS. Confirmed cases were determined to have substantial evidence to support the claim. Trained APS workers confirmed abuse cases using a standardized process involving confirmation of the presence of multiple indicators for each type of abuse. APS data were linked to CHAP data using an algorithm that matched participants based on date of birth, sex, race, home address, and home telephone number.

Perceived Stress

Perceived stress was assessed using a shortened form of the Perceived Stress Scale (PSS).³¹ Six of the original 10 items (Box 1) were included in the CHAP clinical interview to reduce participant burden, given the large number of measures administered. This version of the PSS has been shown to correlate strongly with a validated four-item version and has demonstrated good reliability in the CHAP population.³² Items on the PSS were rated on a 4-point rating scale, with two reverse-scored items. This questionnaire asked participants to rate how often they had felt certain ways over the past month. Possible responses included never (0), rarely (1), sometimes (2), often (3). The total score ranged from 0 to 18, with higher scores indicating greater levels of stress. Perceived stress was examined as total PSS score, categorically based on sample tertiles (low, medium, high stress), and using the individual items.

In a recent study,¹⁴ the factor structure of various forms of the PSS was examined in a community sample of older adults without dementia. It concluded that a two-factor structure was optimal for the PSS. Items reflected distress and coping. Distress items used in this study included items 1, 2, and 6. Coping items included items 3 and 5. Item 4 loaded onto the distress and coping factors.

Covariates

Extent of disease burden, or medical comorbidity, was included as a covariate and was quantified as the sum of six self-reported chronic medical conditions (heart disease, stroke, diabetes mellitus, cancer, hypertension, hip fracture). A global cognition score was calculated by averaging the z-scores of four cognitive tests: the Mini-Mental State Examination (MMSE),³³ the Symbol Digit Modalities Test,³⁴ and the East Boston Memory test (immediate and delayed recall subtests).³⁵ Finally, physical function was assessed using direct performance testing, which consisted of tasks such as the tandem stand, timed walk, and repeated chair stands.³⁶ Performance on these tasks was rated on a scale of 0 to 5 (higher scores indicating better function), and the sum of these scores was used as a total physical function score (range 0–15).

Data Analysis

Bivariate analyses were first conducted to examine the associations between perceived stress and elder abuse variables. T-tests were used to determine group differences (abuse present or absent) in total PSS score. Chi-square tests were used to assess the bivariate associations between PSS tertiles and APS case groupings.

Logistic regression models were conducted with perceived stress variables entered as independent variables and APS case groupings as dependent variables. Two models were tested for each combination of perceived stress variable and elder abuse variable. Model A controlled for time of assessment, age, sex, race, education level, income, and medical comorbidities. Model B added global cognitive score and physical function to the model.

Exploratory bivariate analyses were conducted to examine perceived stress according to specific subtypes of abuse with a perpetrator (physical abuse, emotional abuse, sexual abuse, financial exploitation, caregiver neglect). Exploratory analyses were performed to examine associations between APS client involvement groups and self-reported medical conditions.

RESULTS

Participant Characteristics

Table 1 presents demographic characteristics of the study sample. The mean age was 74.0 ± 7.1. Sixty-two percent were women and 63% were African American.

Table 1.

Study Population Characteristics

Characteristic	Total Cohort, N = 8,558	Participants without APS Involvement, n = 6,977	Participants with Reported APS Elder Abuse, n = 1,581	Participants with Confirmed APS Elder Abuse, n = 1,204
Age, mean ± SD	74.0 ± 7.1	73.8 ± 7.1	75.1 ± 7.1	75.4 ± 7.1
Female, n (%)	5,303 (62.0)	4,224 (60.5)	1,079 (68.3)	839 (69.7)
African American, n (%)	5,433 (63.5)	4,054 (58.1)	1,379 (87.2)	1,051 (87.3)
≥12 years education, n (%)	6,092 (71.4)	5,195 (74.6)	897 (57.0)	681 (56.8)
Annual income ≥$15,000, n (%)	6,322 (73.9)	5,452 (78.1)	870 (55.0)	649 (53.9)
≥3 medical conditions, n (%)	725 (8.5)	513 (7.4)	212 (13.4)	162 (13.5)
Mini-Mental State Examination score, mean ± SD	26.3 ± 4.4	26.6 ± 4.2	24.9 ± 4.8	24.8 ± 4.8
Symbol Digit Modalities Test score, mean ± SD	30.2 ± 13.5	31.7 ± 13.2	23.6 ± 12.3	23.0 ± 12.1
East Boston Immediate Recall score, mean ± SD	8.5 ± 2.7	8.7 ± 2.6	7.8 ± 2.8	7.7 ± 2.8
East Boston Delayed Recall score, mean ± SD	8.0 ± 3.0	8.2 ± 3.0	7.1 ± 3.2	7.0 ± 3.2
Global Cognitive score, mean ± SD	0.2 ± 0.8	0.3 ± 0.7	−0.1 ± 0.8	−0.1 ± 0.8
Physical Function score, mean ± SD	10.0 ± 4.1	10.4 ± 4.0	8.2 ± 4.4	8.0 ± 4.4
Total Perceived Stress Scale Score, mean ± SD	5.4 ± 3.5	5.2 ± 3.4	6.3 ± 3.7	6.3 ± 3.7

APS = Adult Protective Services; SD = standard deviation.

The mean total PSS score for the total sample was 5.4 ± 3.5. Eighteen percent (n = 1,581) of participants were clients of APS, with at least one elder abuse case reported.

Bivariate Analyses

There were significant differences in total PSS score between reported and confirmed cases of elder abuse, abuse with a perpetrator, and self-neglect (Table 2). Specifically, PSS scores were significantly higher for clients of APS than for those without history of APS cases. T-scores ranged from −5.8 (confirmed abuse with a perpetrator) to −12.0 (any reported elder abuse). Cohen D effect sizes were also calculated. There was a moderate to large effect size for PSS group differences for confirmed abuse with a perpetrator. All other group differences had a small to moderate effect size.

Table 2.

Total Perceived Stress Scale (PSS) Score and PSS Tertile According to Type of Adult Protective Services (APS) Elder Abuse Cases

Type of Case	Total Cohort	Total PSS Score				PSS Tertile
		Mean ± SD	T	P-Value	Cohen D	Low Stress	Medium Stress	High Stress	Chi-Square	P-Value
	N						N (%)
Any APS reported elder abuse case
Absent	6,977	5.2 ± 3.4	−11.95	<.001	−0.33	2,385 (27.9)	2,285 (26.7)	2,307 (27.0)	118.23	<.001
Present	1,581	6.3 ± 3.7				387 (4.5)	447 (5.2)	747 (8.7)
Any APS confirmed elder abuse case
Absent	6,977	5.2 ± 3.4	−10.40	<.001	−0.33	2,385 (29.1)	2,285 (27.9)	2,307 (28.2)	80.41	<.001
Present	1,204	6.3 ± 3.7				300 (3.7)	350 (4.3)	554 (6.8)
Reported elder abuse with a perpetrator
Absent	6,977	5.2 ± 3.4	−6.25	<.001	−0.47	2,385 (33.4)	2,285 (32.0)	2,307 (32.3)	25.55	<.001
Present	171	6.8 ± 4.1				40 (0.6)	43 (0.6)	88 (1.2)
Confirmed elder abuse with a perpetrator
Absent	6,977	5.2 ± 3.4	−5.82	<.001	−0.62	2,385 (33.8)	2,285 (32.3)	2,307 (32.6)	22.44	<.001
Present	88	7.3 ± 4.0				17 (0.2)	21 (0.3)	50 (0.7)
Reported self-neglect
Absent	5,795	5.1 ± 3.4	−11.70	<.001	−0.34	2,036 (27.9)	1,872 (25.7)	1,887 (25.9)	113.89	<.001
Present	1,489	6.3 ± 3.7				368 (5.0)	422 (5.8)	699 (9.6)
Confirmed self-neglect
Absent	5,795	5.1 ± 5.0	−10.01	<.001	−0.21	2,036 (29.3)	1,872 (27.0)	1,887 (27.2)	76.31	<.001
Present	1,146	6.3 ± 6.0				290 (4.2)	335 (4.8)	521 (7.5)

Sample tertiles of PSS scores were used to create low-, medium-, and high-stress groups. The PSS tertile cutoffs for each stress group were 0 to 3 for low stress, 4 to 6 for medium stress, and 7 to 18 for high stress. Chi-square values ranged from 22.4 to 118.2. Approximately half of participants with any type of elder abuse history fell into the high-stress group. The most marked proportion of high-stress individuals had a history of APS-confirmed abuse with a perpetrator, with 56% falling into the high-stress group. Of those with other types of abuse, 45% to 51% fell into the high-stress group. Table 2 presents detailed results for PSS tertiles.

Adjusted Stress Models

Total PSS score was a significant predictor of reported and confirmed APS elder abuse cases in adjusted Models A (adjusting for demographic characteristics and medical comorbidities) and B (adjusting for demographic characteristics, medical comorbidities, cognitive function, and physical function). High-stress group membership was significantly associated with reported APS elder abuse cases in both models (Model A: odds ratio (OR) = 1.6, 95% confidence interval (CI) = 1.3–1.9, P < .001; Model B: OR = 1.3, 95% CI = 1.1–1.6, P = .006) but was associated with confirmed APS cases only in Model A (OR = 1.6, 95% CI = 1.3–1.9, P < .001). When controlling for cognitive and physical function, the association between high-stress group membership and confirmed APS elder abuse cases was no longer significant. Detailed results are presented in Table 3.

Table 3.

Regression Models of Total Perceived Stress Scale (PSS) Score, Highest PSS Tertile, and Individual PSS Items Predicting Types of Adult Protective Services (APS) Elder Abuse Cases

Perceived Stress Variable	Cases Reported to APS		Cases Confirmed by APS
	Model A	Model B	Model A	Model B
	Odds Ratio (95% Confidence Interval)
Total PSS score	1.07 (1.04–1.09)c	1.04 (1.01–1.06)b	1.06 (1.04–1.09)c	1.03 (1.00–1.06)a
Highest-stress tertile	1.59 (1.33–1.91)c	1.30 (1.08–1.57)c	1.55 (1.26–1.90)c	1.24 (1.00–1.54)
Item 1	1.17 (1.08–1.27)b	1.15 (1.06–1.25)a	1.14 (1.04–1.26)b	1.13 (1.03–1.24)a
Item 2	1.20 (1.12–1.29)c	1.10 (1.02–1.18)a	1.20 (1.11–1.30)c	1.09 (1.00–1.19)a
Item 3	1.18 (1.09–1.29)c	1.06 (0.96–1.16)	1.19 (1.08–1.31)c	1.05 (0.94–1.16)
Item 4	1.19 (1.10–1.29)c	1.12 (1.04–1.22)b	1.16 (1.06–1.27)b	1.10 (1.00–1.20)
Item 5	1.11 (1.02–1.22)a	0.98 (0.89–1.07)	1.11 (1.00–1.22)a	0.95 (0.86–1.06)
Item 6	1.17 (1.08–1.26)c	1.11 (1.02–1.20)a	1.14 (1.05–1.25)b	1.09 (1.00–1.20)a

Model A was adjusted for time, age, sex, race, education, income, and medical comorbidities.

Model B was adjusted for time, age, sex, race, education, income, medical comorbidities, global cognitive score, and physical function.

^aP < .05

^bP < .01

^cP < .001.

As for specific types of APS elder abuse cases (Table 4), total PSS score was consistently associated with reported and confirmed APS cases of abuse with a perpetrator and self-neglect in partially (Model A) and fully (Model B) adjusted models. ORs ranged from 1.0 to 1.2. The categorical perceived stress variable (highest-tertile stress group membership) yielded a more-detailed characterization of the association between stress and elder abuse cases. High-stress group membership was a significant predictor of all types of elder abuse variables in partially adjusted models, but the addition of cognitive and physical functioning to the model resulted in nonsignificant associations between high-stress group membership and reported abuse with a perpetrator and confirmed self-neglect. High-stress group membership was most strongly associated with abuse with a perpetrator (OR = 3.1 for Model A; OR = 2.7 for Model B). High-stress group membership was less strongly related to confirmed self-neglect (OR = 1.6 for Model A; OR = 1.2 for Model B).

Table 4.

Regression Models of Total Perceived Stress Scale (PSS) Score, Highest PSS Tertile, and Individual PSS Items Predicting Types of Adult Protective Services (APS) Elder Abuse Cases

Perceived Stress Variable	Cases Reported to APS				Cases Confirmed by APS
	Elder Abuse with a Perpetrator		Self-Neglect		Elder Abuse with a Perpetrator		Self-Neglect
	Model A	Model B	Model A	Model B	Model A	Model B	Model A	Model B
Total PSS score	1.11 (1.05–1.18)c	1.08 (1.01–1.14)a	1.07 (1.05–1.09)c	1.04 (1.01–1.06)b	1.16 (1.08–1.26)c	1.13 (1.04–1.23)b	1.06 (1.03–1.09)c	1.03 (1.00–1.05)a
Highest-stress tertile	1.89 (1.11–3.20)a	1.60 (0.92–2.76)	1.65 (1.36–1.99)c	1.33 (1.09–1.62)b	3.11 (1.38–7.01)b	2.69 (1.16–6.23)a	1.56 (1.26–1.94)c	1.24 (0.99–1.55)
Item 1	1.43 (1.13–1.79)b	1.39 (1.11–1.75)b	1.18 (1.09–1.29)b	1.16 (1.06–1.27)b	1.55 (1.13–2.13)b	1.54 (1.12–2.11)b	1.15 (1.05–1.27)b	1.13 (1.02–1.25)a
Item 2	1.20 (0.99–1.47)	1.10 (0.89–1.35)	1.22 (1.13–1.32)c	1.11 (1.03–1.21)b	1.26 (0.96–1.66)	1.16 (0.87–1.53)	1.21 (1.11–1.32)c	1.10 (1.00–1.20)a
Item 3	1.31 (1.05–1.64)a	1.10 (0.86–1.41)	1.16 (1.06–1.27)c	1.03 (0.94–1.14)	1.29 (0.95–1.76)	1.09 (0.77–1.54)	1.12 (1.06–1.29)b	1.03 (0.92–1.14)
Item 4	1.37 (1.09–1.73)b	1.28 (1.02–1.61)a	1.20 (1.10–1.30)c	1.13 (1.03–1.23)b	1.72 (1.24–2.04)b	1.61 (1.16–2.24)b	1.15 (1.05–1.26)b	1.08 (0.98–1.19)
Item 5	1.27 (1.01–1.59)a	1.06 (0.83–1.36)	1.10 (1.00–1.21)a	0.97 (0.88–1.07)	1.57 (1.17–2.10)b	1.37 (1.00–1.87)	1.08 (0.97–1.20)	0.93 (0.83–1.04)
Item 6	1.32 (1.08–1.63)b	1.26 (1.02–1.56)a	1.19 (1.09–1.29)c	1.13 (1.04–1.22)b	1.50 (1.14–1.98)b	1.47 (1.11–1.95)b	1.15 (1.05–1.26)b	1.10 (1.00–1.21)a

Model A was adjusted for time, age, sex, race, education, income, and medical comorbidities.

Model B was adjusted for time, age, sex, race, education, income, medical comorbidities, global cognitive score, and physical function.

^aP < .05

^bP < .01

^cP < .001.

Adjusted Models for Individual PSS Items

PSS items 1 and 6 (both reflecting distress) were consistently associated with every APS elder abuse subtype in partially and fully adjusted models. Items 1 (OR = 1.5, 95% CI = 1.1–2.1, P = .008) and 6 (OR = 1.5, 95% CI = 1.1–2.0, P = .008) were also strongly associated with confirmed abuse with a perpetrator in fully adjusted models. Items reflecting distress appeared to be consistently associated with reported and confirmed elder abuse cases, whereas items reflecting coping were insignificant when controlling for cognitive and physical functioning. All PSS items were associated with reported and confirmed elder abuse in partially adjusted models, but items 3 and 5 (reflecting coping) were insignificant when controlling for cognitive and physical function. Physical function primarily drove this association in these models (all ORs = 0.9, P < .001). With regard to specific abuse subtypes, item 2 was associated only with reported and confirmed self-neglect and not abuse with a perpetrator. Detailed results are presented in Tables 3 and 4.

Exploratory Analysis: Perceived Stress and Subtypes of Reported Cases of Abuse with a Perpetrator

Given that APS clients with a history of abuse with a perpetrator were highest in perceived stress, exploratory analyses were run to determine which subtypes of abuse with a perpetrator were specifically associated with higher stress. Results are presented in Table 5. APS clients with cases of emotional abuse (t = −3.36, P < .001), financial exploitation (t = −4.66, P < .001), and caregiver neglect (t = −5.58, P < .001) had higher total PSS scores than older adults who were never APS clients. Physical abuse and sexual abuse were unassociated with perceived stress, although the number of cases in these subgroups of abuse with a perpetrator was low (n = 30 for physical abuse; n = 4 for sexual abuse), so results should be interpreted with caution.

Table 5.

Total Perceived Stress Scale (PSS) Score by Subgroupings within Adult Protective Services (APS) Reported Abuse with a Perpetrator

Subtypes of Abuse with a Perpetrator	Total Cohort N	Total PSS Score
		Mean ± SD	T	P-Value
No APS involvement (nonabused or neglected: reference)	6,977	5.2 ± 3.4	n/a	n/a
Reported physical abuse	30	6.0 ± 3.5	−1.34	.18
Reported emotional abuse	79	6.5 ± 3.9	−3.36	<.001
Reported sexual abuse	4	5.5 ± 3.1	−0.17	.86
Reported financial exploitation	101	6.8 ± 4.4	−4.66	<.001
Reported caregiver neglect	80	7.3 ± 4.3	−5.58	<.001

Exploratory Analysis: Associations Between Elder Abuse Variables and Medical Conditions

Bivariate analyses were conducted to examine associations between APS client groupings and specific medical conditions included in the medical comorbidities variable. Specific APS groupings examined were non-APS clients (reference group); APS clients with reported history of abuse with a perpetrator, self-neglect, or both; APS clients with reported history of abuse with a perpetrator; and APS clients with reported history of self-neglect. Results suggested that all types of APS clients had a significantly higher prevalence of diabetes mellitus (chi-square 20.9–96.0, P < .001 for all analyses) and stroke (chi-square 17.1–59.8, P < .001 for all analyses) than non-APS clients. APS clients were more likely to report histories of hypertension (chi-square 34.4, P < .001) and heart disease (chi-square 26.6, P < .001) than non-APS clients. APS clients with histories of only abuse with a perpetrator did not have significantly higher rates of hypertension. APS clients with histories of only abuse with a perpetrator had marginally higher rates of heart disease (chi-square 3.6, P = .06). APS clients with histories of self-neglect had a significantly higher prevalence of hypertension (chi-square 49.2, P < .001) and heart disease (chi-square 20.9, P < .001) than non-APS clients. Cancer and hip fracture rates did not differ significantly between APS clients and non-APS clients.

DISCUSSION

The current study is the first to independently examine levels of perceived stress in a population-based study of older adults with and without histories of APS cases of elder abuse. Results suggested that APS clients with reported or confirmed elder abuse of any type were more likely to have higher levels of perceived stress than older adults without any such history. The strongest relationship was between perceived stress and history of APS confirmed abuse with a perpetrator, with a medium to large effect size. Moreover, when creating sample tertiles for total perceived stress scores, approximately half of older adults with any type of abuse history fell into the highest-stress tertile. Individuals reporting the highest levels of stress were three times as likely to have a confirmed history of abuse with a perpetrator as those without such abuse. Perceived stress continued to be a significant correlate of reported and confirmed elder abuse after adjusting for demographic characteristics, medical comorbidities, cognitive function, and physical function. These findings suggest a robust relationship between elder abuse and subjective experience of stress.

These finding are consistent with past work examining the associations between distress and elder abuse.^21,22 Although the literature on elder abuse and perceived stress is limited, there has been some relevant work in the area of child abuse. The current results are consistent with past work indicating that child abuse and neglect are associated with higher levels of perceived stress.^37,38 This relationship persists even into adulthood, as was demonstrated in a population-based sample of nearly 35,000 adults.³⁹ A study⁴⁰ found that perceived stress and emotion-focused coping mediated the relationship between history of child abuse and physical health in women. Given the strong association between elder abuse and perceived stress found, it is plausible that a similar mechanism might drive the association between elder abuse and adverse health outcomes.

Exploratory analyses examining specific types of abuse from a perpetrator revealed that APS clients with reports of emotional abuse, financial exploitation, and caregiver neglect had significantly higher perceived stress than non-APS clients. APS clients with reports of physical and sexual abuse had higher perceived stress scores, but these were not significantly higher than in non-APS clients. Although it would be expected that APS clients with reports of physical and sexual abuse would have significantly higher perceived stress, the sample size of APS reports of physical and sexual abuse are small. These analyses are intended to provide a preliminary characterization of how subtypes of abuse may be associated with stress, and the results should be interpreted with caution. It is also possible that physical and sexual abuse may go unreported more often than other types of abuse for several reasons (e.g., victim’s fear of possible repercussions associated with filing a report, feelings of shame associated with such abuse, fewer overt signs of abuse that might be recognized from an outside party). Future work can examine these types of abuse in more detail by supplementing APS data with self-reported data regarding experiences of abuse.

The associations between types of APS involvement and specific self-reported medical conditions were also examined. The results demonstrated that APS clients had higher rates of hypertension, heart disease, diabetes mellitus, and stroke. APS clients with self-neglect histories also had higher rates of these conditions, whereas APS clients with histories of abuse with a perpetrator had significantly higher rates only of diabetes mellitus and stroke. Cancer and hip fracture rates did not differ between APS clients and nonclients. These findings suggest that clients of APS are more likely to have cardiometabolic conditions, a finding that is consistent with a study finding that elder abuse is associated with metabolic syndrome.⁴¹ Although abuse from a perpetrator was not associated with higher rates of hypertension and heart disease, the sample size of this group was significantly smaller than that of self-neglect. It is possible that, with more data, the marginal association for heart disease might reach significance. Moreover, it makes sense that self-neglecters would have greater risk of these conditions, given the behavioral components involved. Self-neglectors, by definition, are more likely to pay less attention to or completely neglect care, including medication adherence, diet, physical activity, and medical appointments. Longitudinal analyses are warranted to examine whether perceived stress is a possible mediator of the relationship between elder abuse and cardiometabolic disease.

The current analyses also support the hypothesis that APS clients with abuse from a perpetrator reported higher stress than those with history of self-neglect. These findings initially seem to be consistent with the Transactional Model of stress, although individual examination of PSS item 2 (“In the last month, how often have you felt that you were unable to control the important things in your life?”) contradicts this conclusion. Specifically, the extent to which individuals felt that they had poor control over their lives was associated with self-neglect status but not abuse with a perpetrator, although this item of the PSS may not reflect a proxy of sense of control or secondary appraisals of stress, especially in the specific context of elder abuse. In isolation, the finding that self-neglecters endorse feelings of poor control over their lives is intuitive, given that they are in circumstances in which they are unable to care for themselves properly. To determine the specific reasons underlying the association between stress and types of elder abuse, future work might examine coping factors specific to abuse, such as level of empowerment and sense of victimization. Moreover, it is possible that APS intervention might have affected perceived stress ratings in some of these participants. That is, APS intervention may have increased an elder abuse victim’s sense of control just before completing the CHAP interview. Although specific information regarding time of APS intervention was not available, future longitudinal work might examine these associations longitudinally and examine psychological well-being as an outcome in elder abuse interventions.

Analyses examining individual items of the PSS suggested a robust relationship between distress items and elder abuse. Items reflecting coping were also associated with abuse, although this relationship was no longer significant when controlling for physical function. These findings provide support that the two PSS factors (distress and coping) probably represent two qualitatively distinct constructs and that they are differentially related to abuse. Distress items used negative language, and coping items used positive language. Therefore, an individual’s tendency to endorse negatively worded items may be related to abuse history, whereas the tendency to endorse positively worded items is less affected. This distinction may, in part, speak to the respondent’s level of negative and positive affect. Some work in the area of child adversity has found links between low positive affect and history of abuse,⁴² although such studies did not specifically compare positive and negative affect. Alternatively, it might be that coping and resilience reflect more-stable, trait-like constructs, and elder abuse is thus less likely to affect them. Future longitudinal work might focus on examining the specific relationships between positive and negative affect and coping styles in the context of elder abuse to determine what specific styles might be more affected.

The findings from this study have important clinical implications. These results suggest that older adults with history of abuse involving a perpetrator may be in the highest-risk group with regard to stress-related health consequences. Given the wealth of literature linking perceived stress to adverse mental and physical health outcomes, stress and coping represent important targets for clinical intervention. Specifically, increasing empowerment by teaching victims of elder abuse effective coping strategies might buffer the effect that such history has on disease outcomes. Focusing on improving coping strategies is particularly appealing because such interventions are nonpharmacological, have relatively long-lasting effects, and are efficacious. Cognitive behavioral interventions focusing on adaptive coping strategies have been shown specifically to decrease levels of perceived stress.^43–47 Recent investigations have also focused on brief cognitive behavioral interventions that are more accessible to a wide range of individuals in that they are deliverable in the primary care setting and on-line.^48,49 Future work might focus on developing such brief, accessible cognitive behavioral interventions targeting clients of APS to determine whether use of effective coping strategies might have a beneficial effect on disease outcomes.

Findings from the analyses regarding elder abuse and specific medical conditions suggest a robust association between self-neglect and cardiometabolic diseases. Although it is unclear whether stress plays a mediating role in this association, clinical intervention targeting physical and psychological well-being of self-neglectors is also warranted. Multicomponent interventions that affect self-care and coping outcomes, such as a recently published one,⁵⁰ may be useful for self-neglectors.

The current study has limitations. First, the findings were from cross-sectional analyses, and therefore inferences on causality cannot be made. It is plausible that there is a bidirectional association between perceived stress and elder abuse, such that perceived stress predisposes an individual to abuse and is also a consequence of abuse. Work in the area of depression, a construct related to perceived stress, has demonstrated such a bidirectional relationship.^23,51,52 Future work should focus on longitudinal studies examining perceived stress in APS clients to characterize the exact nature of this relationship.

Another limitation is the lack of validated measures assessing stress appraisals and coping, particularly specific to the context of elder abuse. Individual PSS item analyses were not meant to represent larger constructs of coping and appraisals, and these results should be interpreted with caution. Rather, the intention was to provide preliminary evidence of possible components of subjective stress that may be salient in the context of elder abuse. These results are meant to generate more-specific hypotheses regarding stress and abuse and to inform future studies that may examine this relationship in more detail.

In conclusion, the current study found a robust relationship between perceived stress and elder abuse in a population-based study of older adults in Chicago. This relationship was strongest for individuals with a confirmed history of abuse with a perpetrator, with a medium to large effect size. Given the well-established relationship between perceived stress and disease, it is plausible that psychological stress may constitute an important and modifiable pathway linking elder abuse to morbidity and mortality.

APPENDIX 1: SIX ITEMS FROM THE PERCEIVED STRESS SCALE USED IN THE CHICAGO HEALTH AND AGING PROJECT

In the last month, how often have you been upset because of something that happened unexpectedly?

In the last month, how often have you felt that you were unable to control the important things in your life?

In the last month, how often have you felt confident about your ability to handle your personal problems?

In the last month, how often have you felt that things weren’t going your way?

In the last month, how often have you felt that you were on top of things?

In the last month, how often have you felt difficulties were piling up so high that you could not overcome them?