Help-Seeking Among Victims of Elder Abuse: Findings From the National Elder Mistreatment Study

Abstract

Objectives

The vast majority of elder abuse (EA) victims remain hidden from formal institutional response systems. Guided by the Behavioral Model of Health Services Use, this study examined factors that facilitate or impede formal help-seeking among victims of elder emotional abuse, physical abuse, and sexual abuse.

Methods

Data came from a national, population-based EA study in the United States with a representative sample (n = 304) of past-year victims. Gold-standard strategies were used to assess EA subtypes. Multivariate logistic regression was conducted to identify help-seeking facilitators/barriers.

Results

Help-seeking through reporting to police or other authorities occurred among only 15.4% of EA victims. Help-seeking was higher among victims of physical abuse, poly-victimization, or those with a perpetrator having prior police trouble. Help-seeking was lower among victims who were dependent upon their perpetrator and in cases where the perpetrator had a large friendship network.

Discussion

This study highlights the hidden nature of EA as a problem in our society and the need to develop strategies that incorporate victim, perpetrator, and victim–perpetrator relationship factors to promote greater help-seeking among victims.

Elder abuse (EA) refers to an intentional act or lack of action by a person in a relationship involving an expectation of trust that causes harm or risk of harm to an older adult (Centers for Disease Control and Prevention, 2016). It is comprised of several subtypes, including financial emotional/psychological, physical, and sexual abuse, or neglect by others (National Research Council [NRC], 2003). Approximately 9.5% of community-dwelling older adults in the United States experience EA each year (Acierno et al., 2010; Pillemer, Burnes, Riffin, & Lachs, 2016), and victimization is associated with serious consequences such as premature mortality, hospitalization, and poor physical/mental health (Acierno, Hernandez-Tejada, Anetzberger, Loew, & Muzzy, 2017; Dong, 2015).

Understanding EA victim help-seeking is a critical challenge in the field. Population-based EA studies find that only 4%–14% of cases come to the attention of formal response systems, such as law enforcement, legal/justice, or adult protective services (APS) (Acierno et al., 2018; Amstadter et al., 2011; Lachs & Berman, 2011), which is problematic insofar as these agencies often represent primary access points to helping-related resources. Thus, the vast majority of victims living in the community remain hidden from authorities and endure EA without receiving assistance to reduce the risk and magnitude of revictimization.

Theoretical Framework and Literature Review

The Behavioral Model of Health Services Use (BMHSU) was designed to predict and explain the use of formal health services including help/care-seeking (Andersen, 1995) and is widely used in social gerontology scholarship (Alley, Putney, Rice, & Bengtson, 2010). The BMHSU has been found as useful for conceptualizing help-seeking among victims of intimate partner violence (Fleming & Resick, 2017) and has been used to understand service usage among EA victims (Barker & Himchak, 2006; Burnes, Breckman, Henderson, Lachs, & Pillemer, 2018). The model proposes that service utilization, here represented by a call for help in the form of a report to police or other authorities, is predicted by a person’s predisposing characteristics, enabling resources, and level of need. Predisposing characteristics reflect demographic factors and sociostructural/status inequalities that may limit/enhance access to services. The enabling dimension represents personal, interpersonal, social and community resources that facilitate/impede service utilization. The need dimension reflects the nature and magnitude of the presenting problem, as well as secondary morbidities that might exacerbate need for services (e.g., health vulnerabilities). The BMHSU proposes that greater predisposing sociostructural/status advantage, enabling resources, and need predict higher levels of service utilization. Since EA occurs in the context of a relationship, application of BMHSU dimensions must incorporate characteristics attached to the victim, perpetrator, and victim–perpetrator relationship to fully understand help-seeking facilitators/barriers (Barker & Himchak, 2006; Burnes et al., 2018; Burnes, Rizzo, Gorroochurn, Pollack, & Lachs, 2016).

The literature includes only a few studies on the topic of EA victim help-seeking determinants. Aligning with BMHSU expectations related to predisposing sociostructural/status inequalities, prior research has found that EA victims of African American or Hispanic/Latino origin are less likely to accept or pursue formal services (Burnes et al., 2016; Roberto, Teaster, & Duke, 2004). In regard to enabling characteristics, victims with greater informal support resources demonstrate heightened formal help-seeking behavior (Burnes et al., 2018; Yan, 2015), while victim–perpetrator relationships characterized by a shared living arrangement, abuser control tactics, or close familial ties/loyalty impede seeking assistance from authorities (Burnes et al., 2018; Newman, Seff, Beaulaurier, & Palmer, 2013). Finally, in regard to the BMHSU need dimension, help-seeking is more likely when victims can no longer tolerate the mistreatment (e.g., going on too long, too severe) and they exhibit secondary vulnerabilities such as poor health and functional impairment (Barker & Himchak, 2006; Burnes et al., 2016; Yan, 2015). To date, research on EA victim service utilization and help-seeking is largely limited by qualitative, retrospective agency case record, convenience sampling, and/or nonvictim proxy respondent study designs. Building on prior research, the current study used a nationally representative sample of EA victims to explore factors that facilitate or impede help-seeking with formal authorities.

Methods

Data

The National Elder Mistreatment Study used a stratified, random-digit-dialing sampling strategy derived from census-defined size-of-place parameters to conduct computer-assisted telephone interviews with a representative (age, gender) sample (n = 5,777) of cognitively intact, English- or Spanish-speaking older adults (age ≥ 60) from February 6 to September 9, 2008. The sampling strategy yielded a cooperation rate of 69%. Further details on study methodology are well-documented (Acierno, Hernandez-Tejada, Muzzy, & Steve, 2009). The current paper analyzed the subsample (n = 304) of older adults assessed as victims of elder emotional, physical, or sexual abuse. To minimize the effect of recall bias, this study was limited to older adults reporting EA within the previous year.

Independent Variables

Mistreatment

In accordance with recognized procedures to maximize sensitivity in epidemiological interpersonal violence research (NRC, 2003), this study assessed EA subtypes with contextually oriented, multiple, closed-ended (no/yes) behaviorally defined items describing specific mistreatment events. Emotional abuse was assessed using four items related to verbal attacking/scolding/yelling, humiliation, harassment/coercion, and ignoring. Physical abuse was assessed with three items reflecting hitting/slapping/threatening with weapon, restraining, and injury. Sexual abuse was measured using four items related to forced intercourse, molestation, undressing, and photography violation. EA caseness was defined by an affirmative response to one or more items across subtype assessments. Specific EA assessment questions are provided in Supplementary Appendix A. Consistent with accepted EA definitions (Pillemer et al., 2016), this study considered EA as events perpetrated by a person in a conventional relationship of expected trust (family, friend, neighbor, professional, coworker, caregiver) and excluded incidents perpetrated by a stranger.

Predisposing

Predisposing variables included victim gender, race/ethnicity (White-non-Hispanic/African American/White-Hispanic-Latino/Other), age (continuous), marital status (married-partnered/widowed/divorced-separated-single), and education (continuous; some high-school [1], high-school [2], some college [3], associate degree [4], bachelor’s degree [5], some graduate/professional degree [6], graduate/professional degree [7]), and perpetrator employment status (unemployed/employed).

Enabling

Enabling factors included victim household income (continuous; ≤$10,000 [1], $10,001–$20,000 [2], $20,001–$35,000 [3], $35,001–$50,000 [4], $50,001–$75,000 [5], $75,001–$100,000 [6], >$100,000 [7]), social support (continuous), and social service engagement (no/yes; one or more of senior centers or senior day programs; physical rehabilitation; meals on wheels or other meal service; social services or health service visits; home health nurse visits; hospice visits; senior friends or other home visits; church group home visits; or other programs/services), and perpetrator friend network size (none/very few [1–3]/some [4–6]/a lot [7+]), as well as victim–perpetrator relationship type (spouse-partner/adult child/other family/nonfamily), victim–perpetrator living arrangement (victim lives without/with perpetrator), and victim dependence on perpetrator for daily activities (no/yes). Victim social support was measured using a modified five-item version of the Medical Outcomes Study module for social support, with higher scores (range: 5–20) reflecting greater social support (Sherbourne & Stewart, 1991). Perpetrator friend network size was included in this study because a perpetrator’s social embeddedness is conceptualized as a key analytic dimension in examining EA scenarios (NRC, 2003), and the current paper sought to explore whether perpetrator social resources served as a facilitator to victim help-seeking.

Need

Need variables included the nature of mistreatment (emotional/physical/sexual abuse), mistreatment severity indicators (poly-victimization [no/yes], lifetime frequency [continuous]), victim health status, and level of victim functional impairment, as well as perpetrator characteristics that could exacerbate the need for formal services (substance use problem [no/yes], police history [no/yes], mental health problems [no/yes]) (DeLiema, Yonashiro-Cho, Gassoumis, Yon, & Conrad, 2018). Victim health status was assessed using the general health question from the World Health Organization Short-Form 36 Health Questionnaire (continuous [1–6]; very poor–excellent) (Ware & Gandek, 1998). Functional impairment was measured as the number of daily activities that required assistance. Poly-victimization was operationalized as experiencing more than one of the emotional/physical/sexual abuse subtypes. Lifetime EA frequency was assessed by summing the number of times each reported mistreatment event had occurred in the respondent’s lifetime. The full survey instrument is available online (Acierno et al., 2009).

Dependent Variable

The dependent variable was help-seeking in the form of reporting EA. After endorsing past-year emotional abuse, physical abuse, or sexual abuse, participants were asked whether they reported the most recent incident to police or other authorities (no/yes): Thinking about the most recent incident where someone [emotionally/physically/sexually] mistreated you, was this incident reported to the police or other authorities?

Analytic Plan

Bivariate/unadjusted logistic regression was conducted individually with each independent variable to explore relationships with EA victim help-seeking. Multivariate logistic regression was subsequently carried out simultaneously on independent variables reaching borderline significance (p < .10) in bivariate/unadjusted analysis and satisfying tolerance/variance inflation factor diagnostics. Missing data were managed with a fully conditional specification multiple imputation method using 10 pooled data sets. Data were weighted to correct for sampling and nonresponse biases.

Results

Supplementary Table 1 presents sample (n = 304) descriptive characteristics. The sample was mostly White/non-Hispanic (82.9%) and female (64.8%) with mean age 68.0 (range: 60–94). Emotional abuse was the most common mistreatment subtype (76.7%) followed by physical (25.7%) and sexual (10.5%) abuses, while 11.2% of victims experienced poly-victimization.

Help-seeking from police or other authorities occurred among 15.4% of all EA victims. Help-seeking rates varied according to subtype: emotional abuse (10.7%), physical abuse (24.9%), and sexual abuse (22.6%). Among EA victims, help-seeking was significantly higher among those experiencing physical abuse (odds ratio [OR] = 2.92, CI: 1.26–6.80), poly-victimization (OR = 4.67, CI: 1.72–12.72), and when the perpetrator had a history of police involvement (OR = 3.01, CI: 1.08–8.44). Help-seeking was significantly lower when the victim was dependent upon the perpetrator for daily activities (OR = 0.32, CI: 0.12–0.88) or when the perpetrator had a large friend network (OR = 0.08, CI: 0.01–0.45) (Table 1).

Table 1.

Bivariate and Multivariate Logistic Regression Models Predicting Elder Abuse (EA) Victim Formal Help-Seeking

	EA victim help-seeking (n = 304)
	Bivariate	Multivariate
BMHSU characteristic	OR (95% CI)	OR (95% CI)
Predisposing
Gender female (ref. male)	1.55 (0.78–3.09)
Race/ethnicity (ref. White/non-Hispanic)
African American	0.09 (0.24–3.40)
White/Hispanic/Latino	0.79 (0.15–4.21)
Other	2.30 (0.81–6.53)
Age (cont.)	1.02 (0.98–1.06)
Marital status (ref. married/partnered)
Widowed	1.36 (0.56–3.27)	1.19 (0.40–3.55)
Divorced/separated/single	1.91 (0.96–3.82)†	1.66 (0.68–4.09)
Education (cont.)	0.91 (0.77–1.09)
Perpetrator employed (ref. unemployed)	0.82 (0.38–1.74)
Enabling
Household income (cont.)	0.77 (0.62–0.96)*	0.90 (0.70–1.17)
Social support (cont.)	0.94 (0.87–1.01)
Social service use (ref. no)	1.13 (0.60–2.14)
Friends network size (ref. none)
Very few	0.58 (0.15–2.27)	0.48 (0.09–2.49)
Some	0.36 (0.10–1.33)	0.29 (0.06–1.52)
A lot	0.12 (0.0300.46)**	0.08 (0.01–0.45)**
Relation type (ref. spouse)a
Adult child	0.49 (0.16–1.47)	0.46 (0.11–1.90)
Other family	0.45 (0.19–1.12)†	0.34 (0.10–1.09)†
Nonfamily	0.74 (0.35–1.59)	0.88 (0.26–2.96)
Victim lives with perpetrator (ref. lives without)	1.89 (0.97–3.69)†	1.88 (0.60–5.90)
Victim depends on perpetrator (ref. no)	0.46 (0.21–1.01)*	0.32 (0.12–0.88)*
Need
Health status (cont.)	1.35 (1.06–1.73)*	1.20 (0.83–1.72)
Functional impairment (cont.)	1.22 (0.99–1.49)†	1.17 (0.87–1.59)
Substance use problem (ref. no)	2.26 (1.18–4.31)*	1.49 (0.60–3.75)
Police history (ref. no)	3.52 (1.72–7.21)**	3.01 (1.08–8.44)*
Mental health problems (ref. no)	1.48 (0.69–3.19)
Emotional abuse present (ref. not present)	0.63 (0.32–1.25)
Physical abuse present (ref. not present)	3.06 (1.61–5.83)**	2.92 (1.26–6.80)*
Sexual abuse present (ref. not present)	2.12 (0.90–4.99)†	2.41 (0.78–7.45)
Poly-victimization present (ref. not present)	4.22 (1.94–9.19)***	b4.67 (1.72–12.72)**
Lifetime mistreatment frequency (cont.)	1.00 (1.00–1.01)

Note: OR = odds ratio; CI = confidence interval; BMHSU = Behavioral Model of Health Services Use. Final multivariate models satisfied the Omnibus Test of Model Coefficients (p < .01) and Hosmer–Lemeshow test (p > .05). Independent variables had tolerance of 0.59 or above and variance inflation factor of 1.69 or below, indicating little concern of multicollinearity.

^aIn poly-victimization cases with different perpetrators across EA subtypes, proximal relationship spouse or child perpetrators were selected.

^bPoly-vicitimization was entered into a separate multivariate model in place of physical and sexual abuse subtype variables, since these variables were correlated and generated multicollinearity.

*p < .05. **p < .01. ***p < .001. ^†p < .10 (borderline).

Discussion

This study identified factors that facilitate or impede EA victim help-seeking in the formal system (e.g., law enforcement, legal/justice, APS). Consistent with prior research, only a small minority of victims (approximately 1 out of 7) in this study sought help.

Using the BMHSU framework, need characteristics were the strongest facilitators of EA victim help-seeking. In particular, victims experiencing more than one type of EA (poly-victimization) were nearly four times as likely to seek help from police or other authorities compared to victims experiencing only one EA subtype. Poly-victimization can be viewed as an indicator of EA severity (Hamby, Smith, Mitchell, & Turner, 2016) and, in turn, reflects elevated need in relation to the problem. When considering separate EA subtypes, victims of physical abuse were most likely to seek formal EA assistance, which may reflect the egregious and dangerous nature of this subtype. Cases of sexual abuse also demonstrated elevated levels of help-seeking, although the borderline significance was lost after controlling for other factors (note also low power associated with sexual abuse analysis). Emotional abuse is a more common form of EA compared to physical and sexual abuse (Pillemer et al., 2016), and the relatively low level of help-seeking among emotional abuse victims may reflect a higher level of tolerance/acceptance of this mistreatment type among older adults and in society in general. Finally, within the BMHSU need dimension, victim help-seeking was higher when the perpetrator had previous trouble with the police. Familiarity with police interactions may facilitate EA reporting, and may suggest the benefit of community police programs wherein contact between older adults and police is more frequent.

Barriers to victim help-seeking were observed in the BMHSU enabling dimension. Victims who depended upon their perpetrator for assistance with daily activities were less likely to seek help. Prior research shows that EA victims who are dependent upon their perpetrator appraise their mistreatment situation with lower levels of seriousness (Burnes, Lachs, Burnette, & Pillemer, 2017) and can engage in a process of “tacit exchange” whereby they accept the abuse in exchange for a perceived benefit such as care, companionship, and/or the ability to remain living in the community (Enguidanos, DeLiema, Aguilar, Lambrinos, & Wilber, 2014). We also found that larger perpetrator friendship networks deterred victim help-seeking. Larger perpetrator social networks may elevate the level of perceived power that victims attach to their perpetrator and, in turn, disempower help-seeking.

Findings carry implications for efforts designed to promote help-seeking among EA victims. Education/awareness programs should help older adults understand the seriousness of emotional abuse as a form of mistreatment and the importance of seeking help before mistreatment situations reach heightened levels of severity. Accessible homecare services are required to provide care support for EA victims who are dependent upon their perpetrator for daily needs. To resonate with EA victims who are both dependent upon their perpetrator and wish to maintain a relationship with them, formal support programs need to offer a client-centered, harm-reduction model that allays victims’ fear that seeking formal help will result in a severance of the victim–perpetrator relationship or implicating the perpetrator in the legal/justice system (Burnes, 2017).

As limitations, the current study excluded older adults with cognitive impairment, which likely represents an impediment to help-seeking. Our outcome alluded to help-seeking with authorities and, thus, may have excluded reports of help-seeking in less formal community programs or victim services. Multivariate analyses were not conducted among separate EA subtype samples due to low power/sample size restrictions, and help-seeking data were unavailable for financial abuse and neglect EA subtypes.

Despite limitations, the current study represents the largest examination of EA victim help-seeking to date. Findings highlight an urgent need to elevate help-seeking among EA victims. Further research is required to understand why the majority of EA victims do not seek help and how policy and awareness efforts might address this issue.

Funding

This work was supported by the Social Sciences and Humanities Research Council (SSHRC) (435-2017-0578 to D. Burnes); National Institute of Justice (2007-WG-BX-0009 to R. Acierno); and National Institute on Aging (R21AG030667 to R. Acierno).

Author Contributions

D. Burnes led paper conceptualization, analysis, and writing. R. Acierno and M. Hernandez-Tejada designed the original survey, oversaw data collection, assisted in editing the current document, and provided citations to support edits.

Conflict of Interest

None reported.