Association Between Depressive Symptoms, Multiple Dimensions of Depression, and Elder Abuse: A Cross-Sectional, Population-Based Analysis of Older Adults in Urban Chicago

Abstract

Objective:

Depression is conceptualized as both a risk factor for and a consequence of elder abuse; however, current research is equivocal. This study examined associations between elder abuse and dimensions of depressive symptoms in older adults.

Method:

Participants were 10,419 older adults enrolled in theChicago Health and Aging Project (CHAP), a population-based study of older adults. Regression was used to determine the relationships between depressive symptoms, depression dimensions, and abuse variables.

Results:

Depressive symptoms were consistently associated with elder abuse. Participants in the highest tertile of depressive symptoms were twice as likely to have confirmed abuse with a perpetrator (odds ratio = 2.07, 95% confidence interval = [1.21, 3.52], p = .008). Elder abuse subtypes and depression dimensions were differentially associated.

Discussion:

These findings highlight the importance of routine depression screening in older adults as a component of abuse prevention and intervention. They also provide profiles of depressive symptoms that may more accurately characterize risk for specific types of abuse.

Introduction

Elder abuse is a pervasive and often underestimated public health and civil rights concern. Population studies estimate the prevalence of elder abuse in the United States to be between 7.6% and 10% (Acierno et al., 2010; Lifespan of Greater Rochester, Inc., Weill Cornell Medical Center of Cornell University, & New York City Department for the Aging, 2011). An overwhelming number of these cases go unreported, despite Adult Protective Service (APS) agencies in every state and mandated elder abuse reporting legislation (Lifespan of Greater Rochester et al., 2011). Elder abuse is a multifaceted construct, encompassing physical, emotional, and sexual abuse, financial exploitation, as well as neglect (both caregiver neglect and self-neglect). Above and beyond the acute impact of such abuse, elder abuse has been associated with increased morbidity and mortality (Dong et al., 2009; Friedman, Avila, Shah, Tanouye, & Joseph, 2014; Lachs, Williams, O’Brien, Pillemer, & Charlson, 1998; Schofield, Powers, & Loxton, 2013). Given the psychological and physical toll that elder abuse can have on an individual, research is needed to characterize possible factors that can identify those at increased risk for abuse.

Psychological well-being has been the focus of emerging research in the area of elder abuse. One important indicator of psychological well-being is depression. Clinical depression and depressive symptom severity have been identified as both risk factors for and consequences of elder abuse (Dong, Chen, Chang, & Simon, 2013). Prior research demonstrates that caregiver depression is associated with increased harmful behavior (Beach et al., 2005; Williamson & Shaffer, 2001). Moreover, depression has been shown to be associated with elder abuse in the victims, themselves. For example, a cross-sectional study by Pillemer and Moore (1989) found that elder abuse was a strong predictor of depression in a study of more than 2,000 older adults. Higher levels of depressive symptoms have also been reported in self-neglecting older adults (Dong et al., 2009). However, a recent systematic review of psychological well-being and elder abuse suggests that research examining the link between depression and elder abuse is rather equivocal (Dong et al., 2013). Some investigations have found that this relationship was attenuated when controlling for other psychosocial covariates, such as social support (Dong, Beck, & Simon, 2010).

One factor contributing to the variability in this literature is the scarcity of large sample, community-based studies that systematically focus on elder abuse and associated risk factors. Moreover, there is a great deal of variability in how the construct of elder abuse is measured. More research is needed using the gold standard method of assessing elder abuse, which involves the linkage of APS datasets to study data to capture both reported and confirmed cases of abuse. Finally, the vast majority of the studies to date have examined depression as a unidimensional construct. This approach fails to account for the marked variability in what factors trigger and exacerbate depression and how depressive symptoms manifest. That is, it is possible that specific dimensions of depression (i.e., negative affect vs. somatic complaints) might be more strongly associated with elder abuse than others. Addressing these pitfalls would greatly improve our understanding of the association between depression and elder abuse.

Indeed, the paucity of research examining multiple dimensions of depression has been cited as a primary contributor to variability in depression research in the context of various social, clinical, and demographic correlates (Hayes et al., 1998). The research that has been done reveals a wealth of information regarding processes that differentially affect specific dimensions of depression. For example, negative affect has been shown to be more reactive to health, internal attributes, and the presence of negative life events (Clark & Watson, 1991; Zautra & Reich, 1983), compared with positive affect, which is primarily associated with the presence of positive life events (Zautra & Reich, 1983). Interpersonal problems are more heavily associated with variables related to social support network (Hayes et al., 1998).

The current study aimed to elucidate the relationship between depression and elder abuse by addressing the aforementioned issues in the existing literature. Specifically, the purpose of this study was to examine the link between depression and elder abuse in a population-based sample of community-dwelling, older adults living in urban Chicago. Moreover, we also aimed to examine these relationships among specific dimensions of depressive symptoms.

Importantly, we aimed to draw from a solid conceptual framework in the design of our analyses. Although no one comprehensive conceptual framework has been developed to explain elder abuse, several existing models have been applied. One such model that has been used to guide hypotheses regarding elder abuse is Bronfenbrenner’s (1979) Ecological Model of human development. This multilevel model identifies nested environmental systems with which the individual interacts. The Ecological Model as applied to elder abuse provides an organizational schema that highlights multiple levels of variables that are important to consider when characterizing elder abuse, but does not make predictions regarding the causality or directionality of relationships of these variables. Although we were unable to assess each of these characteristics directly, we aimed to strategically choose variables that capture components of multiple system levels. As applied to the context of elder abuse (Burnight & Mosqueda, 2011), variables in the ontogenetic system (the innermost level) include those related to physiology, affect, and behavior, such as age, sex, and mental and physical health (including medical comorbidities, cognitive status, and physical function). Microsystem characteristics refer to groups and institutions that directly affect the individual, such as family and church. Exosystem variables include those related to the socioeconomic environment, access to social services, and so on. Finally, macrosystem variables refer to ethnic and cultural attitudes and contexts. In the context of elder abuse, this might involve ethnic and cultural beliefs about age and gender inequality, cultural beliefs about family responsibility and aging, and society aggression norms.

We hypothesized that depressive symptom severity and prevalence would be higher among clients of APS services, both those with reported and confirmed abuse. We predicted that these relationships would be present above and beyond the effects of covariates representing multiple levels of Ecological Model. Finally, we hypothesized that there would be differential associations between depression dimensions and various subtypes of abuse. Specifically, we predicted that reports of interpersonal problems might be more salient in the context of abuse with a perpetrator (caregiver or trusted other neglect) compared with self-neglect.

Method

Design and Participants

Participants were 10,419 older adults who were enrolled in the Chicago Health and Aging Project (CHAP), a population-based study of individuals aged 65 years and older living in three adjacent neighborhoods in Chicago. Study design elements have been published in detail in previous publications (Bienias, Beckett, Bennett, Wilson, & Evans, 2003; Evans et al., 2003). Briefly, all participants completed in-home interviews, including standardized questionnaires assessing demographic information, medical history, cognitive functioning, and psychosocial variables. This study protocol was approved by the Rush University Institutional Review Board, and all participants provided written, informed consent for participation.

Study Variables

Elder abuse definitions and adult protective services record linkage.

Elder abuse encompasses physical abuse, sexual abuse, emotional abuse, neglect (caregiver neglect and self-neglect), and financial exploitation. Physical abuse is defined as infliction of physical pain or injury on an older adult. Sexual abuse is defined as touching, fondling, or any other sexual activity with an older adult when the older adult is unable to understand, unwilling to consent, threatened, or physically forced into the act. Emotional abuse is defined as verbal assaults, threats of abuse, harassment, or intimidation that compels the older adult to engage in conduct from which he or she has the right to abstain, or to refrain from conduct in which the older person has the right to engage. Neglect is the failure by a caregiver, trusted other, or by one’s self, to provide the older adult with necessities of life, including, but not limited to food, clothing, shelter, medical care, and so on. Financial exploitation is the misuse or withholding of an older adult’s resources to the disadvantage of the older adult and/or the profit or advantage of another person (“Illinois Adult Protective Services Act”; Illinois Department on Aging, 2006).

We chose to examine elder abuse in multiple ways: first, as a broad construct including each of the aforementioned forms of abuse and second, divided into groups based upon the presence or absence of a perpetrator. We chose to examine these subgroups separately, given that elder self-neglect (neglect without the involvement of a caregiver or trusted other) is the most common form of elder abuse and has been conceptualized as a distinct construct with unique surrounding characteristics.

From this point forward, the following terminology will be used to describe specific types of abuse. “Elder abuse” is an all-encompassing term referring to instances of either abuse perpetrated by another individual and/or self-neglect. “Elder abuse with a perpetrator” refers to instances of abuse that were imposed on an older adult by another individual (perpetrator). “Self-neglect” refers to the failure of an individual to provide himself or herself with necessities of life (in the absence of any perpetrator or trusted other).

Elder abuse data were obtained from 1993 by linking CHAP participants to the social services database of elder abuse cases. Abuse cases are confirmed by trained APS workers using a standardized process involving confirmation of the presence of multiple indicators for each type of abuse.

Depressive symptoms.

Depressive symptoms were assessed using a modified version of the Centers for Epidemiologic Studies Depression Scale (CES-D; Kohout, Berkman, Evans, & Cornoni-Huntley, 1993; Radloff, 1976). This version of the CES-D consisted of 10 “yes/no” items with two reverse-scored items. The total score ranged from 0 to 10 based on the number of “yes” responses endorsed indicating agreement with statements, such as “I felt sad” and “My sleep was restless.”

Depressive symptoms were examined in multiple ways: (a) continuously, as a total sum score (“total CES-D”); (b) as a binary, “categorical depression,” variable (total CES-D score ≥ 4 indicated “depressed,” CES-D < 4 indicated “non-depressed”), which was based upon established cutoffs to screen for the presence of depression in older adults (Irwin, Artin, & Oxman, 1999) and is also optimally associated with elder abuse (Dong & Simon, 2014); and (c) as a categorical variable based on tertiles calculated from this study population indicating low, medium, and high depression severity (“depression tertiles”).

Depressive symptoms were also examined in terms of four specific depression dimensions or factors: “depressed affect,” “positive affect,” “somatic complaints,” and “interpersonal problems.” Depressed affect (Factor 1) consisted of three items: “I felt depressed,” “I felt lonely,” and “I felt sad.” Positive affect (Factor 2) consisted of two items: “I was happy” and “I enjoyed life.” Somatic complaints (Factor 3) consisted of three items: “I felt that everything I did was an effort,” “My sleep was restless,” and “I could not get going.” Interpersonal problems (Factor 4) consisted of two items: “People were unfriendly” and “I felt that people disliked me.” This four-factor structure of depression has been validated in older adults in prior studies (Kohout et al., 1993). Individual depression factor scores reflected the sum of items endorsed in each factor category. The positive affect factor was reverse scored, such that higher scores indicated poorer positive affect.

Cognitive function.

A global cognition score was calculated by averaging the z scores of four different cognitive tests, including the Mini-Mental State Examination (MMSE; Folstein, Folstein, & McHugh, 1975), the Symbol Digit Modalities Test (Smith, 1984), and the East Boston Memory Test (immediate and delayed recall subtests; Albert et al., 1991).

Physical function.

Physical function was assessed using direct performance testing, which consisted of multiple tests, including tandem stand, timed walk, and repeated chair stands (Guralnik et al., 1994). Performance on each of these three tasks was rated on a scale of 0 to 5 (higher scores indicating better function), and the sum of these scores across the three tasks was used as a total physical function score (range = 0–15).

Data Analysis

First, unadjusted analyses were conducted to examine the bivariate associations between depression variables and elder abuse variables. Wilcoxen rank sum tests were performed to determine group differences in the total CES-D, given that depression was not normally distributed. Chi-square tests were used to assess the bivariate associations between categorical depression, depression tertiles, and APS elder abuse groupings.

Depression variables were then examined as independent variables in logistic regression models with respect to the dependent variable of APS elder abuse groupings. Four models that were progressively more adjusted were tested for each depression variable and elder abuse variable combination. All models controlled for duration of time between the depression assessment and the date of APS elder abuse report. Importantly, data from the CHAP study interview (i.e., CES-D scores, cognitive function, physical function, etc.) were selected based upon the measurements taken closest to the elder abuse report date. Analyses for Model A controlled for age and sex. Model B added race, education level, and income. Model C added medical comorbidities to the model. Finally, Model D added global cognitive score and physical function to the model. The choices of covariates entered in these models were guided by multiple levels of the Ecological Model as applied to elder abuse, with age, sex, medical comorbidities, cognitive function, and physical function reflecting the ontogenetic level, income as a proxy for socioeconomic status (SES) at the exosystem level, and race as a proxy at the macrosystem level.

Results

Study Population Characteristics

Table 1 presents demographic characteristics of the CHAP study population. The mean age of the total cohort was 78.59 ± 7.52 years. Sixty-one percent of the total population was female and 63% was African American. Table 1 also presents the mean total CES-D scores as well as mean scores for each of the four depression factors for the total population and elder abuse groupings. The prevalence of any reported elder abuse in this cohort was 1,892 (18%). Of these cases, 1,467 were confirmed. The vast majority of these cases were self-neglect cases (93% of reported cases, 95% of confirmed cases). Of the total cohort, 2,077 (20%) were “depressed” (i.e., screened positive for depression based on a CES-D cutoff of ≥4. The prevalence of depression in the CHAP study is consistent with other population-based estimates of depression in older adults, although slightly on the high end, likely due to the urban, ethnically diverse population (Blazer, 2003; Fiske, Wetherell, & Gatz, 2009; National Alliance on Mental Illness, 2009).

Table 1.

Characteristics of the Study Population.

	Total cohort (N = 10,419)	No elder abuse (N = 8,527)	Reported elder abuse (N = 1,892)	Confirmed abuse (N = 1,467)
Age in years, M (SD)	78.59 (7.52)	78.76 (7.49)	77.80 (7.62)	78.06 (7.66)
Female gender, n (%)	6,390 (61.33)	3,421 (40.12)	1,284 (67.86)	1,014 (69.12)
African American, n (%)	6,604 (63.38)	4,956 (58.12)	1,648 (87.10)	1,278 (87.12)
Graduated high school (≥12 years education), n (%)	7,109 (68.51)	6,068 (71.41)	1,041 (55.40)	806 (55.24)
Annual income ≥ US$ 15,000, n (%)	7,410 (71.12)	6,383 (74.86)	1,027 (54.28)	776 (52.90)
≥ three medical conditions, n (%)	1,405 (13.49)	1,074 (12.60)	331 (17.49)	266 (18.13)
MMSE total score, M (SD)	24.88 (5.82)	25.13 (5.74)	23.73 (6.04)	23.59 (6.07)
Symbol Digit Modalities Test, M (SD)	25.36 (14.28)	26.33 (14.33)	20.96 (13.23)	20.33 (13.18)
East Boston Immediate Recall, M (SD)	7.92 (3.13)	8.05 (3.1 1)	7.33 (3.16)	7.27 (3.17)
East Boston Delayed Recall, M (SD)	7.25 (3.56)	7.41 (3.53)	6.56 (3.63)	6.48 (3.64)
Global cognitive score, M (SD)	−0.06 (0.93)	−0.01 (0.93)	−0.29 (0.92)	−0.32 (0.92)
Physical function score, M (SD)	8.1 1 (4.47)	8.42 (4.41)	6.73 (4.48)	6.44 (4.47)
Total CES-D score, M (SD)	1.86 (2.22)	1.73 (2.15)	2.47 (2.43)	2.51 (2.44)
Depressed (CES-D ≥ 4), n (%)	2,077 (19.93)	1,533 (17.98)	544 (28.75)	437 (29.79)
Depressed affect (Factor 1), M (SD)	0.60 (0.98)	0.55 (0.94)	0.81 (1.08)	0.83 (1.09)
Positive affect (Factor 2), M (SD)	0.24 (0.53)	0.23 (0.52)	0.32 (0.59)	0.32 (0.59)
Somatic complaints (Factor 3), M (SD)	0.88 (0.99)	0.81 (0.96)	1.17(1.06)	1.18 (1.06)
Interpersonal problems (Factor 4), M (SD)	0.12 (0.40)	0.12 (0.39)	0.15 (0.44)	0.16 (0.45)

Note. MMSE = Mini-mental State Examination; CES-D = Centers for Epidemiologic Studies Depression Scale.

Total CES-D, Categorical Depression, and Elder Abuse Groupings

Bivariate associations between the total CES-D and categorical depression variables by all six elder abuse groupings were all significant (Table 2). With regard to the categorical depression variable, the proportion of older adults with a record of APS-documented abuse and/or neglect who were “depressed” ranged from 28% to 37%, depending on the type of reported and/or confirmed abuse/neglect documented. The proportion of older adults without any documented abuse or neglect who were “depressed” ranged from 18% to 20%. Thirty-seven percent of older adults with a history of APS-confirmed elder abuse were “depressed,” making them the group with the highest proportion of depression per the vulnerability risk index CES-D cutoff.

Table 2.

Total CES-D Score and Categorical Depression by APS Elder Abuse Groupings.

	Total cohort	Total CES-D score (continuous)			Depressed (categorical depression)
	n (%)	M (SD)	Wilcoxen rank sum test	p	n (%)	χ 2	p
Reported elder abuse			1 1,385,271.5	<.0001		112.62	<.0001
Absent	8,527 (81.84)	1.73 (2.15)			1,533 (17.98)
Present	1,892 (18.16)	2.47 (2.43)			544 (28.75)
Confirmed elder abuse			8,580,762.0	<.0001		110.32	<.0001
Absent	8,527 (85.32)	1.73 (2.15)			1,533 (17.98)
Present	1,467 (14.68)	2.51 (2.44)			437 (29.79)
Reported elder abuse with a perpetrator			1,234,269.0	<.0001		35.42	<.0001
Absent	8,527 (97.41)	1.73 (2.15)			1,533 (17.98)
Present	227 (2.59)	2.83 (2.68)			76 (33.48)
Confirmed elder abuse with a perpetrator			661,797.5	<.0001		28.40	<.0001
Absent	8,527 (98.62)	1.73 (2.15)			1,533 (19.98)
Present	1 19 (1.38)	3.05 (2.79)			44 (36.97)
Reported self-neglect			9,210,945.0	<.0001		95.03	<.0001
Absent	7,344 (80.64)	1.74 (2.16)			1,330 (18.11)
Present	1,763 (19.36)	2.44 (2.40)			502 (28.47)
Confirmed self-neglect			7,042,388.0	<.0001		92.96	<.0001
Absent	7,344 (84.09)	1.74 (2.16)			1,330 (18.11)
Present	1,389 (15.91)	2.47 (2.41)			408 (29.37)

Note. CES-D = Centers for Epidemiologic Studies Depression Scale; APS = Adult Protective Service.

Depression Tertiles and Elder Abuse Groupings

Low, medium, and high depression groups were based off of sample tertiles of CES-D scores. The approximate CES-D tertile cutoffs for each depression severity group were as follows: “low depression” = 0, “medium depression” = 1–2, and “high depression” = 3–10. Bivariate associations between depression severity groupings and all elder abuse groupings were significant (Table 3). Chi-square values ranged from 30.61 to 161.10. Older adults with a history of elder abuse with a perpetrator had the highest proportion of high depression reporters, with 45% of those having reported elder abuse with a perpetrator endorsing high levels of depressive symptoms and 48% of those having confirmed abuse with a perpetrator reporting high levels of depressive symptoms. In fully adjusted models (Model D), older adults with high depression levels were significantly more likely to have APS-documented history of abuse in all abuse categories (odds ratios [ORs] ranged from 1.41 to 2.07). The strongest association was for confirmed abuse with a perpetrator, with those reported having high depression being twice as likely to have a history of such abuse (OR = 2.07, 95% confidence interval [CI] = [1.21, 3.52], p = .008).

Table 3.

Low, Medium, and High Depression Groups (Tertiles of CES-D Scores) by APS Elder Abuse Groupings.

	Total cohort	Depression tertiles
	n (%)	Low depression n (%)	Medium depression n (%)	High depression n (%)	χ 2	p
Reported elder abuse					161.10	<.0001
Absent, n (%)	8,527 (81.84)	3,338 (39.15)	2,957 (34.68)	2,232 (26.18)
Present, n (%)	1,892 (18.16)	504 (26.64)	642 (33.93)	746 (39.43)
Confirmed elder abuse					143.51	<.0001
Absent, n (%)	8,527 (85.32)	3,338 (39.15)	2,957 (34.68)	2,232 (26.18)
Present, n (%)	1,467 (14.68)	381 (25.97)	499 (34.01)	587 (40.01)
Reported elder abuse with a perpetrator					42.31	<.0001
Absent, n (%)	8,527 (97.41)	3,338 (39.15)	2,957 (34.68)	2,232 (26.18)
Present, n (%)	227 (2.59)	56 (24.67)	69 (30.40)	102 (44.93)
Confirmed elder abuse with a perpetrator					30.61	<.0001
Absent, n (%)	8,527 (98.62)	3,338 (39.15)	2,957 (34.68)	2,232 (26.18)
Present, n (%)	1 19 (1.38)	26 (21.85)	36 (30.25)	57 (47.90)
Reported self-neglect					136.44	<.0001
Absent, n (%)	7,344 (80.64)	2,869 (39.07)	2,536 (34.53)	1,939 (26.40)
Present, n (%)	1,763 (19.36)	427 (26.77)	604 (34.26)	687 (38.97)
Confirmed self-neglect					122.05	<.0001
Absent, n (%)	7,344 (80.09)	2,869 (39.07)	2,536 (34.53)	1,939 (26.40)
Present, n (%)	1,389 (15.91)	364 (26.21)	477 (34.34)	548 (39.45)

Note. CES-D = Centers for Epidemiologic Studies Depression Scale; APS = Adult Protective Service.

Depression, Depression Dimensions, and Elder Abuse Groupings in Adjusted Models

Adjusted logistic regression models assessing the relationships between depression variables and elder abuse groupings are presented in Tables 4 and 5. With regard to reported elder abuse, total CES-D and categorical depression were significant predictors in all adjusted models. Depressed affect (OR = 1.10, 95% CI = [1.04, 1.16], p < .001 in fully adjusted models) and somatic complaints (OR = 1.18, 95% CI = [1.12, 1.25], p < .0001 in fully adjusted models) were significant predictors in all adjusted models. Positive affect was associated with reported elder abuse after controlling for time, age, sex, race, education, income, and medical comorbidities (OR = 1.19, 95% CI = [1.08, 1.31], p < .001). This relationship was not significant after controlling for global cognitive score and physical function. The interpersonal problems factor was only a significant predictor of reported elder abuse when controlling for time, age, and sex (OR = 1.19, 95% CI = [1.05, 1.35], p = .007). There was a consistent pattern of significant relationships found for confirmed elder abuse and depression factors.

Table 4.

Regression Models of Total CES-D, Categorical Depression, and Depression Factors predicting Elder Abuse (Reported and Confirmed).

	Reported e1der abuse				Confirmed e1der abuse
	A	B	C	D	A	B	C	D
Tota1 CES-D, OR (95% CI)	1.15 [1.12, 1.17]***	1.08 [1.05, 1.10]****	1.07 [1.05, 1.10]****	1.05 [1.03, 1.08]****	1.15 [1.12, 1.18]****	1.08 [1.05, 1.11]****	1.07 [1.05, 1.10]****	1.05 [1.02, 1.08]***
Categorica1 depression, OR (95% CI)	1.85 [1.64, 2.08]****	1.40 [1.23, 1.59]***	1.36 [1.20, 1.55]****	1.27 [1.11, 1.45]***	1.90 [1.66, 2.17]****	1.44 [1.25, 1.65]****	1.39 [1.21, 1.60]****	1.29 [1.11, 1.50]***
Factor 1: Depressed affect, OR (95% CI)	1.28 [1.21, 1.34]***	1.14 [1.09, 1.21]***	1.13 [1.08, 1.I9]****	1.10 [1.04, 1.16]***	1.29 [1.22, 1.36]****	1.15 [1.09, 1.22]****	1.14 [1.08, 1.21]***	1.10 [1.04, 1.17]**
Factor 2: Positive affect, OR (95% CI)	1.34 [1.22, 1.46]***	1.22 [1.10, 1.34]***	1.19 [1.08, 1.31]***	1.11 [1.00, 1.23]	1.32 [1.19, 1.46]****	1.20 [1.08, 1.34]***	1.17 [1.05, 1.30]**	1.08 [0.96, 1.21]
Factor 3: Somatic comp1aints, OR (95% CI)	1.41 [1.34, 1.49]***	1.24 [1.17, 1.30]****	1.22 [1.16, 1.29]****	1.18 [1.11, 1.25]****	1.42 [1.35, 1.51]****	1.25 [1.18, 1.32]****	1.23 [1.15, 1.30]****	1.17 [1.10, 1.25]****
Factor 4: Interpersona1 prob1ems, OR (95% CI)	1.19 [1.05, 1.35]**	0.91 [0.80, 1.04]	0.91 [0.80, 1.04]	0.90 [0.78, 1.03]	1.20 [1.05, 1.38]**	0.93 [0.81, 1.08]	0.93 [0.80, 1.07]	0.91 [0.78, 1.05]

Note. CES-D = Centers for Epidemiologic Studies Depression Scale; OR = odds ratio; CI = confidence interval.

A. Time, age, sex

B. Time, age, sex, race, education, income

C. Time, age, sex, race, education, income, medical comorbidities

D. Time, age, sex, race, education, income, medical comorbidities, global cognitive score, physical function.

^*p < .05.

^**p < .01

^***p < .001.

^****p < .0001.

Table 5.

Regression Models of Total CES-D, Categorical Depression, and Depression Factors Predicting Elder Abuse With a Perpetrator and Self-Neglect (Reported and Confirmed).

	Reported elder abuse with a perpetrator				Reported self-neglect				Confirmed elder abuse with a perpetrator				Confirmed self-neglect
	A	B	C	D	A	B	C	D	A	B	C	D	A	B	C	D
Total CES-D, OR (95% CI)	1.18 [1.12, 1.24]****	1.12 [1.06, 1.18]***	1.11 [1.05, 1.17]***	1.09 [1.03, 1.16]**	1.14 [1.12, 1.17]****	1.08 [1.05, 1.11]****	1.07 [1.05, 1.10]****	1.06 [1.03, 1.08]***	1.22 [1.14, 1.31]****	1.16 [1.08, 1.24]****	1.15 [1.07, 1.23]***	1.13 [1.05, 1.22]**	1.14 [1.12, 1.17]****	1.08 [1.05, 1.11]****	1.10 [1.05, 1.10]****	1.05 [1.02, 1.09]***
Categorical depression, OR (95% CI)	2.07 [1.54, 2.80]****	1.61 [1.19, 2.19]**	1.56 [1.15, 2.13]**	1.45 [1.04, 2.01]*	1.82 [1.61, 2.07]****	1.42 [1.24, 1.62]****	1.38 [1.20, 1.57]****	1.29 [1.12, 1.49]***	2.53 [1.72, 3.74]****	1.95 [1.31, 2.91]**	1.87 [1.25, 2.79]**	1.75 [1.15, 2.68]**	1.88 [1.63, 2.16]****	1.46 [1.26, 1.69]****	1.41 [1.22, 1.64]****	1.32 [1.13, 1.54]***
Factor 1: Depressed affect, OR (95% CI)	1.34 [1.18, 1.51]****	1.21 [1.06, 1.37]**	1.19 [1.05, 1.35]**	1.17 [1.02, 1.34]*	1.27 [1.20, 1.34]****	1.15 [1.09, 1.21]****	1.13 [1.07, 1.20]****	1.10 [1.04, 1.17]***	1.44 [1.22, 1.69]****	1.30 [1.10, 1.53]**	1.27 [1.08, 1.50]**	1.23 [1.03, 1.46]*	1.28 [1.21, 1.36]****	1.16 [1.09, 1.23]****	1.14 [1.08, 1.22]****	1.11 [1.04, 1.18]**
Factor 2: Positive affect, OR (95% CI)	1.44 [1.16, 1.79]**	1.35 [1.08, 1.69]**	1.33 [1.06, 1.66]*	1.14 [0.90, 1.46]	1.33 [1.21, 1.46]****	1.23 [1.11, 1.36]***	1.20 [1.08, 1.33]***	1.12 [1.01, 1.25]*	1.30 [0.96, 1.77]	1.22 [0.89, 1.67]	1.18 [0.86, 1.62]	1.06 [0.76, 1.49]	1.32 [1.19, 1.47]****	1.23 [1.10, 1.37]***	1.20 [1.07, 1.34]**	1.11 [0.99, 1.25]
Factor 3: Somatic complaints, OR (95% CI)	1.41 [1.23, 1.60]****	1.23 [1.07, 1.41]**	1.21 [1.05, 1.38]**	1.19 [1.03, 1.38]*	1.41 [1.34, 1.49]****	1.25 [1.18, 1.32]****	1.23 [1.17, 1.31]****	1.19 [1.12, 1.26]****	1.54 [1.30, 1.84]****	1.35 [1.13, 1.62]**	1.32 [1.10, 1.58]**	1.33 [1.10, 1.61]**	1.42 [1.33, 1.50]****	1.26 [1.18, 1.34]****	1.23 [1.16, 1.31]****	1.18 [1.10, 1.26]****
Factor 4: Interpersonal problems, OR (95% CI)	1.82 [1.42, 2.33]****	1.49 [1.15, 1.93]**	1.47 [1.14, 1.91]**	1.46 [1.11, 1.91]**	1.11 [0.97, 1.27]	0.87 [0.76, 1.00]*	0.86 [0.75, 0.99]*	0.86 [0.74, 0.99]*	2.23 [1.66, 3.00]****	1.82 [1.33, 2.49]***	1.79 [1.30, 2.45]***	1.78 [1.28, 2.47]***	1.13 [0.98, 1.31]	0.90 [0.77, 1.05]	0.89 [0.76, 1.04]	0.87 [0.74, 1.03]

Note. CES-D = Centers for Epidemiologic Studies Depression Scale; OR = odds ratio; CI = confidence interval.

A. Time, age, sex

B. Time, age, sex, race, education, income

C. Time, age, sex, race, education, income, medical comorbidities

D. Time, age, sex, race, education, income, medical comorbidities, global cognitive score, physical function.

^*p < .05.

^**p < .01.

^***p < .001.

^****p < .0001.

The subgroupings of elder abuse from a perpetrator and self-neglect were also examined as separate outcomes in adjusted models (Table 5). Total CES-D score (OR = 1.09, 95% CI = [1.03, 1.16], p = .004) and categorical depression (OR = 1.45, 95% CI = [1.04, 2.01], p = .030) were associated with reported abuse from a perpetrator in fully adjusted models. All depression factors were significantly associated with reported elder abuse from a perpetrator in fully adjusted models with ORs ranging from 1.17 to 1.46, with the exception of positive affect. Positive affect was a significant predictor of reported elder abuse from a perpetrator until global cognitive score and physical function were added to the model.

In models predicting confirmed elder abuse from a perpetrator, total CES-D score and categorical depression were associated with confirmed abuse with a perpetrator in all models. Older adults who were identified as “depressed” based on the categorical depression variable were nearly twice as likely to have confirmed abuse from a perpetrator as those scoring in the non-depressed range (OR = 1.75, 95% CI = [1.15, 2.68], p = .010). Depressed affect, somatic complaints, and interpersonal problems were associated with confirmed abuse from a perpetrator in fully adjusted models with ORs ranging from 1.23 to 1.78. Positive affect was unassociated with confirmed elder abuse from a perpetrator.

Total CES-D score, categorical depression, and all depression factors were associated with reported elder self-neglect in fully adjusted models. Total CES-D score and categorical depression were significant predictors in all confirmed self-neglect models. Depressed affect, positive affect (except when controlling for global cognitive score and physical function), and somatic complaints were associated with confirmed self-neglect. The interpersonal problems factor was not associated with confirmed self-neglect in any model.

Discussion

In sum, the current study found that depressive symptoms are consistently associated with elder abuse in a population-based sample of older adults in urban Chicago. This association was present for multiple subtypes of abuse, including both APS reported and confirmed instances of elder abuse. Moreover, depression was also associated with elder abuse subgroups (i.e., abuse from a perpetrator as well as self-neglect). These findings provide further confirmation that depression is a robust correlate of elder abuse in the context of the rather equivocal evidence that has been published to date. Importantly, this study improves upon the current research examining the relationships between depression and elder abuse, in that it utilizes a large, population-based sample with APS-linked elder abuse data, the gold standard in assessing elder abuse.

The current findings highlight the need for prospective, population-based studies with elder abuse assessment built into the design. To the authors’ knowledge, there have only been three prospective studies to date that have examined depression as a risk factor for elder abuse. These studies had conflicting results. One study by Lachs, Williams, O’Brien, Hurst, and Horwitz (1997) found that depression was not significantly associated with increased risk for reported elder abuse and neglect. Conversely, another study found that depression was a significant risk factor for incident self-neglect (rate ratio = 2.38, 95% CI = [1.26, 4.48], p = .007; Abrams, Lachs, McAvay, Keohane, & Bruce, 2002). Moreover, a 2014 study by the Chicago Health and Aging Project (CHAP) group included depression as an indicator in a vulnerability risk index profile for elder abuse, which was significantly associated with increased risk for abuse (OR = 1.82, 95% CI = [1.19, 2.79], p < .050; Dong & Simon, 2014). Even fewer prospective studies have been published examining depression as a direct consequence of elder abuse. Our findings offer strong support that depression is a consistent predictor of elder abuse in our cross-sectional analyses, even after controlling for theoretically salient covariates reflecting multiple levels in the Ecological Model.

What is particularly novel about this study is the examination of multiple dimensions of depressive symptoms. As predicted, the results suggest that there are differential associations between depression dimensions and elder abuse subtypes. Older adults reporting higher levels of depressed affect, somatic complaints, and interpersonal problems were significantly more likely to have confirmed abuse from a perpetrator. Positive affect was unassociated with confirmed abuse from a perpetrator. In contrast, the depressed affect and somatic complaints factors were significant predictors of confirmed self-neglect, while the interpersonal problems factor was not. Positive affect significantly predicted self-neglect until adjusting for cognitive and physical function.

The finding that the level of interpersonal problems is unassociated with self-neglect is intuitive, given that self-neglecters typically have lower social support (Alexandra Hernandez-Tejada, Amstadter, Muzzy, & Acierno, 2013; Dong & Simon, 2008; Fulmer et al., 2005) and social engagement/interaction (Dong, Simon, & Evans, 2010). As a result, self-neglecters may have less opportunity for both positive and negative interactions with others. Moreover, individuals who self-neglect may also be less interested in social connectedness, leaving them more susceptible to self-neglecting behaviors that go undetected by others who might otherwise intervene.

The finding that positive affect was unassociated with confirmed abuse from a perpetrator and self-neglect (after adjusting for cognitive and physical function) is somewhat counterintuitive on the surface. One possible explanation for these findings is that positive affect may represent a more stable, “trait”-like construct compared with the other depression dimensions examined. Indeed, past research has examined “trait” variance of positive affect, with results suggesting that positive affect is a relatively stable construct in a large sample of mood and anxiety disordered patients (Naragon-Gainey, Gallagher, & Brown, 2013). Therefore, positive affect may not be as sensitive to the occurrence of elder abuse and neglect as the other depression dimensions. This pattern of findings is also consistent with Zautra and Reich’s (1983) model that proposes that while negative events primarily influence negative affect, positive affect is primarily affected by positive events (or lack thereof), rather than negative events. Conversely, another study by Etter, Gauthier, McDade-Montez, Cloitre, and Carlson (2013) demonstrated links between positive affect, physical abuse, sexual abuse, and low social support in the context of child abuse and neglect, complicating this interpretation. Another possible explanation may be that use of spiritual coping in our primarily African American population may serve as a protective factor regarding positive affect. Indeed, past research has shown an association between spiritual coping and positive affect in African American parishioners (Holt, Lewellyn, & Rathweg, 2005). It is important to note that analyses of depression dimensions are meant to be preliminary in nature and are limited by the use of the CES-D as the only measurement instrument available to examine these constructs. Future studies may clarify these findings by using measures developed to specifically assess such dimensions, such as the Positive and Negative Affect Schedule (PANAS; Watson, Clark, & Tellegen, 1988).

A study by Hayes and colleagues (1998) examined differential associations between depression factors and various demographic, health, and psychosocial variables in an elderly population. Their results suggested differential associations between depression factors and several established elder abuse risk factors. For example, the elder abuse risk factor of functional disability was associated with all dimensions of depression (negative affect, positive affect, somatic complaints, and interpersonal problems). Cognitive impairment, however, was only associated with somatic complaints. Moreover, social network size was negatively associated with interpersonal problems, whereas satisfaction with the amount of social interaction was not. These findings in combination with our results highlight possible mechanisms in an otherwise complex system of relationships between elder abuse and psychosocial well-being.

The current study has valuable clinical implications. First, this study provides further support for the importance of routine assessment of depression in older adults in medical settings. Doing so may be a key component in both elder abuse prevention and intervention. Our results with regard to specific depression dimensions also suggest specific targets for mental health interventions that may optimize treatment of depression for victims of abuse as well as for those vulnerable to abuse. Prior research has examined the impact of mental health interventions for caregivers with the goal of reducing harmful behaviors (Livingston et al., 2013); however, less work has been done to assess the impact of such interventions for older adults at risk for abuse. Future work to tailor mental health interventions for elder abuse victims is also needed, given the emotional and physical toll that such abuse can have on the victim.

Future work should also focus on specific mechanisms linking depression to elder abuse. Although multiple studies have examined depression as a possible risk factor for elder abuse (Choi & Mayer, 2000; Dong, Simon, Odwazny, & Gorbien, 2008; Fulmer et al., 2005; Lachs et al., 1997), the specific mechanisms underlying this relationship remain unclear. Depression and more general psychosocial consequences of abuse may also represent important pathways linking elder abuse to downstream morbidity and mortality. Elder abuse can be characterized as a stressful life event, or even as a traumatic stressor in extreme cases. Such abuse can also be acute and chronic in nature, despite the number of abusive events one experiences, particularly if the victim manifests posttraumatic stress symptoms from the abuse. A great deal of literature has examined the impact of psychological well-being on morbidity and mortality (Edmondson, Kronish, Shaffer, Falzon, & Burg, 2013; Fan et al., 2014; Pan, Sun, Okereke, Rexrode, & Hu, 2011), and therefore, treatment of depression and possible trauma symptoms may reduce one’s risk for such adverse outcomes.

There are limitations of the current study that are worthy of mention. First, this study was cross-sectional in nature, thus precluding our ability to determine causality. Based upon the current literature and available conceptual frameworks, it is plausible that depression can be characterized as both a risk factor for elder abuse and a consequence for victims of elder abuse. Cross-sectional analyses only capture a snapshot of one’s mood at one point in time, and do not provide details with regard to trajectory or variability of mood symptoms over time and how that might be related to elder abuse outcomes. Longitudinal analyses assessing depression and elder abuse are currently underway and can help clarify these questions.

Second, it is well established that elder abuse is vastly underreported (Lifespan of Greater Rochester et al., 2011). Therefore, there are likely several older adults identified in our “non-abused” reference group who may indeed have a history of abuse. Self-report assessment of abuse may identify more cases of abused older adults, possibly those who are most vulnerable to abuse consequences, given their lack of support and services resultant from being unidentified by APS. This approach may also allow for more details regarding nuances of abuse that might be differentially associated with depression and depression dimensions. That said, APS data linkage is the gold standard in the assessment of elder abuse, and our results likely reflect the most conservative estimates of the associations between abuse and depression, which remain robust.

Third, there are limitations regarding the availability of data to capture each level of the Ecological Model in our analyses. Although we have several variables reflecting ontogenetic level factors, we have relatively less data to control for microsystem, exosystem, and macrosystem level factors. One particular microsystem variable that would be important for future studies to examine in the context of victim depression is perpetrator characteristics. Past research suggests that elder physical abuse is perpetrated by a partner or a spouse the majority of the time (National Elder Mistreatment Study, 2009). It is possible that depression among elder abuse victims may vary based upon who is perpetrating the abuse. Future work examining how characteristics of the perpetrator may differentially affect psychological well-being can have implications for tailoring interventions for the victim.

Moreover, race (largely African American and Caucasian in this population) was used as a proxy to reflect macrosystem influence. Future studies can improve on this by extending such analyses to other cultures and by including measures reflecting cultural values, acculturation, and so on. Indeed, accumulating research has examined elder abuse, depression, psychological well-being, and cultural beliefs in Chinese (Chang, Beck, Simon, & Dong, 2014; Dong, 2014; Dong, Beck, Simon, 2010; Dong, Chen, Chengyue, & Simon, 2014; Dong, Chen, Fulmer, & Simon, 2014; Simon, Chang, Zhang, Ruan, & Dong, 2014) and Latino populations (DeLiema, Gassoumis, Homeier, & Wilber, 2012; Otiniano & Herrera, 1999). Continued work in this area can help clarify how culture and ethnicity can affect psychological well-being and elder abuse.

It is important to note that there are several theoretical frameworks that may also be considered when conceptualizing elder abuse, for example, the Sociocultural Model (National Research Council, 2003), Life-course Perspective Model (Kuh & Ben-Shlomo, 1997), and Risk and Protective Factors Model (Coie et al., 1993; Hawkins, Catalano, & Miller, 1992). We chose this particular model for the current analyses because it does not assume causality/directionality of any of the covariates included and it recognizes the impact of multiple levels of contextual influence. Systematic research of elder abuse that is carefully guided by a solid conceptual framework is needed at the study design stage to model important covariates as well as risk and protective factors.

In sum, the current study provides support that depression is linked with reported and confirmed elder abuse, including abuse from a perpetrator and self-neglect. These associations are robust and present above and beyond the effects of covariates reflecting multiple system levels characterized by the Ecological Model. This study also suggests that there are differential associations between specific depression dimensions and type of abuse. Although more work is needed to understand the specific mechanisms operating in the association between depression and elder abuse, these findings offer preliminary targets of mental health intervention in older adults who are either vulnerable to elder abuse or who have already been victimized. They also provide profiles of depressive symptoms that may more accurately characterize the risk for specific types of abuse.