Elder Abuse Screening Instruments in Primary Care: An Integrative Review, 2004 to 2011

Elder abuse screening instruments are currently held to the same standards as disease screening tests and are determined to be valid if they meet the following criteria: (1) are sensitive (effectively identify individuals with the disease); (2) are specific (effectively identify individuals who do not have the disease as not having the disease); (3) demonstrate a positive predictive value (PPV); and (4) demonstrate a negative predictive value (NPV).^6,7 It should also be understood that, as with all disease screening tests, the screening process results in the label of “positive” or “negative” but is not diagnostic and warrants additional testing and assessment before conclusions are drawn.⁷ 

In 2004, Nelson and colleagues⁵ conducted a review of abuse screening instruments for the USPSTF. In this review, abuse screening is defined as an “assessment of current harm or risk for harm from family and intimate partner violence in asymptomatic persons in a healthcare setting.” The authors found 1045 articles on elder abuse, of which three met the inclusion criteria of elder abuse screening instruments. Of these three studies, one sought to validate the Caregiver Abuse Screen (CASE),⁸ one sought to validate the Hwalek-Sengstock Elder Abuse Screening Test (HSEAST),⁹ and one sought to examine the performance of the HSEAST among elderly adults living in public housing.¹⁰ Based on their findings, Nelson and colleagues⁵ concluded that very few EA instruments had been developed to identify EA. They also noted that while the three instruments they reviewed performed fairly well, none of them were developed or tested within traditional healthcare settings. In this article, we examine the development of EA screening instruments between 2004 and 2011 for use in primary care. 

Background

A universal definition of EA does not currently exist, and a variety of definitions have been developed for use in research or to meet legal needs.¹¹ In this article, we use the World Health Organization’s definition: “A single, or repeated act, or lack of appropriate action, occurring within any relationship where there is an expectation of trust that causes harm or distress to an older person.”⁶ It consists of five forms: physical, psychological/emotional, financial/material, sexual, and neglect.⁶ It does not include random acts of criminal violence or actions of self-neglect, in which an elder refuses to address his or her own basic needs.^6,11

It is important to note that the age at which a person becomes an “elder” also varies widely in the literature, with the defined age range of 50 years to 65 years or older, depending on the researcher or jurisdiction.^6,12 In the United States, the Older Americans Act set age 60 as the point at which persons found to be abused become eligible to receive aging services, making this age the most frequently cited cut point.¹² The Centers for Disease Control and Prevention is currently working to establish a standardized definition for EA to aid in prevention, monitoring, and intervention efforts.¹¹ 

The lack of a standardized definition and differences in research methodologies and theoretical frameworks has made it difficult to establish the prevalence of EA. In the most recent nationally representative sample, Acierno and colleagues¹³ determined an 11.4% prevalence rate (n=5777) of abuse among community-dwelling, cognitively intact elders. Due to the increased vulnerability of elders, it is possible for a single incident of abuse to initiate a downward trend, culminating in loss of autonomy, serious health issues, and even death.¹⁴ In 2005, the most extreme form of EA, eldercide, claimed the lives of 386 men and 331 women aged 65 years and older.¹⁵ 

In 2008, more than 96% of elders reported they had a regular place to go for medical care, and only 2.4% said that they failed to obtain needed medical care during the previous 12 months due to financial barriers.¹⁶ Elders are seen in primary care for common chronic health conditions associated with aging, providing a valuable opportunity for EA screening. While a direct correlation between self-reported poor health and EA in elders has yet to be substantiated by research, residual mental health issues associated with EA (anxiety and depression) may represent an important focus for improving the long-term health and quality of life of elders who screen positive for EA during a primary care visit.¹⁷ It has been established that individuals who experience abuse make more frequent visits to obtain healthcare services, have more health problems, and experience decreased quality of life and lost productivity.¹⁸

Although the effects of EA can be devastating, there are  barriers to screening for it in the primary care setting, including lack of a simple, short, universal screening and assessment methodology; lack of time for screening and intervention; lack of knowledge regarding intervention and resources; fear of retaliation or escalating violence; lack of knowledge regarding mandatory reporting requirements; a desire to honor the wishes of the elder or family not to report EA; lack of knowledge regarding EA, what comprises abuse, and how to recognize it; difficulty in determining elder capacity; and fear of the consequences of involving government agencies in the elder’s life.^19,20 Another ongoing issue in EA screening instrument development and research is the lack of a reference standard or “gold standard” comparison by which to establish the validity of EA screening instruments.⁵ 

Methods

To conduct our review, we selected the integrative review method described by Whittemore and Knafl²¹ for its flexibility and ease of handling a variety of evidence types. We also used the evidence synthesis methodology used by Nelson and colleagues⁵ as a way to provide consistency in addressing EA.However, unlike these authors, who examined screening and interventions for all forms of family and intimate partner violence, the focus of our article is limited to EA screening; we do not review EA interventions. The following six questions guided our literature review: 

1. With regard to EA screening tools developed since 2004, who is the EA screening tool designed to screen: the elder (care recipient), the caregiver, or both? 

2. What types of EA do the screening tools developed since 2004 identify? 

3. How well do the EA screening tools developed since 2004 identify current harm or risk of harm from EA (sensitivity and specificity)?⁵ 

4. Were the potential adverse effects and ethical issues of EA screening addressed? 

5. Does screening for EA reduce harm and risk of premature death and disability?⁴

6. What is the quality of the studies conducted from 2004 to 2011 to validate the EA screening tool when compared with USPSTF criteria?⁵

To answer these questions, we performed a literature review between May 2011 and August 2011 using PubMed; Ovid/MEDLINE; PsycINFO; CINAHL; Academic Search Premier, Health Source: Nursing/Academic Edition, and Psychology and Behavioral Sciences Collection; and ProQuest. This search yielded a total of 211 articles. After removing duplicates and applying our inclusion and exclusion criteria, we were left with 10 articles. All of these articles were written in English; reviewed abuse or violence towards older adults; were conducted in the primary care setting or contained a statement recommending an EA instrument for use in this setting; were performed by a healthcare provider or by researchers who made a recommendation regarding healthcare provider use; and were focused on evaluating the implementation of the EA instrument. 

Determination of Quality and Data Synthesis

The USPSTF criteria, available at www.uspreventiveservicestaskforce.org, were used to determine the quality of the individual studies.²² The international scope of EA as a worldwide public health issue became evident while conducting our review and performing the analysis. Of the 10 studies included in our review, four were conducted by two primary investigators: Cooper and colleagues (n=2) and Conrad and associates (n=2).^23-26 These investigators were from the United Kingdom and the United States, respectively, but the studies in our assessment also included research from Canada (n=1), Australia (n=1), Taiwan (n=1), Israel (n=1), and the Federation of Bosnia and Herzegovina (n=1). There was also an additional study from the United States, which was conducted by Wiglesworth and colleagues.²⁷ Cumulatively, the EA assessment instruments used included the Expanded Indicators of Abuse (E-IOA)²⁸; Older Adult Financial Exploitation Measure (OAFEM)²⁵; Older Adult Psychological Abuse Measure (OAPAM)²⁶; Modified Conflict Tactic Scale (MCTS)^23,24; The Hwalek-Sengstock Elder Abuse Screening Test (H-S/EAST)²⁹; Vulnerability to Abuse Screening Scale (VASS)³⁰; Elders’ Psychological Abuse Scale (EPAS)³¹; modified versions of the Conflict Tactics Scales (CTS2) and Elder Abuse Instrument (EAI)²⁷; and the Elder Abuse Screening Index (EASI).³² The OAFEM, OAPAM, VASS, EPAS, and EASI were new instruments that had been developed after 2004. What follows are our findings with regard to our research questions.

With regard to EA screening tools developed since 2004, who is the EA screening tool designed to screen: the elder (care recipient), the caregiver, or both? 
When reviewing the 10 studies as a whole, six screened only the elder,^25,26,29-32 two screened only the caregiver,^23,24 and two screened both the elder and the caregiver.^27,28 The sample sizes of the studies ranged from 8623 to 10,421.³⁰ Persons screened were aged 50 years and older,²⁷ 65 years and older,^28,29,32 had mean ages older than 60 years,^23,24,31 or had narrow age ranges older than 70 years.^25,26,30 Nine of the studies^23-30,32 undertaken since 2004 were used to screen elders and/or caregivers who were community-dwelling (noninstitutionalized) with only the study by Wang and associates³¹ including both community-dwelling and institutionalized elders.  

What type of EA is identified by the screening tools developed since 2004?
The studies captured in this review used a variety of instruments to identify single and multiple types of EA. Two studies used instruments to screen for both psychological and physical abuse.^23,24 Three studies used an instrument to screen for only one specific type of abuse: financial abuse²⁵ or psychological abuse.^26,31 Five studies used instruments to screen for more than two forms of abuse or risk for abuse.^27-30,32 

How well do the EA screening tools developed since 2004 identify current harm or risk of harm from EA (sensitivity and specificity)? ⁵
Currently, the validity of EA screening instruments is assessed like all disease screening tests through the determination of sensitivity and specificity. PPV and NPV were also calculated in some of the studies reviewed. With the United States having an estimated EA prevalence rate of 11.4%¹⁴ (pre-test probability), relatively low PPV and high NPV values should be expected. The MCTS²⁴  had a sensitivity of 87%, specificity of 70%, PPV of 18%, and NPV of 99%. The E-IOA²⁸ had a sensitivity of 92% and a specificity of 98%. The combination of the CTS2 and EAI²⁷ had a sensitivity of 75% and a specificity of 70.6%, whereas the EASI³² had a sensitivity of 47% and a specificity of 75%. The H-S/EAST²⁹ had a 2% false positive and 3% false negative rate. 

An ongoing issue in EA screening instrument development and research is the lack of a reference standard by which to establish the validity of EA screening instruments. To establish internal validity of diagnostic accuracy studies, the USPSTF requires the reference standard to be interpreted apart from the interpretation of the screening test. In three studies, the observer or expert documentation reviewer was provided with the screening test result,^23,27,29 resulting in compromised USPSTF internal validity ratings of poor and fair, respectively. Other studies reported estimates of reliability for the EA screening instruments by examining internal consistency among items on the instrument scales. The EPAS,³¹ MCTS,^23,24 and OAPAM²⁶ all demonstrated good internal consistency, whereas the OAFEM²⁵ demonstrated excellent internal consistency. 

Were the potential adverse effects and ethical issues of EA screening addressed?
None of the 10 studies addressed the possible adverse effects of EA screening. Cooper and colleagues²³ did, however, examine the acceptability of EA screening among caregivers of elders with dementia, and they reported that 83.7% found the interview using the MCTS acceptable. Other researchers reported a lack of willingness to admit to perpetrating or experiencing EA after being informed about mandatory reporting requirements during the consent process as a potential limitation.^24,27 In another study, the participants were blinded to the fact that the purpose of the study was to investigate the development and usefulness of a screening instrument to identify psychological abuse.³¹ 

Does screening for EA reduce harm and risk of premature death and disability?  ⁴
Similar to the findings of Nelson and colleagues,⁵ none of the studies we reviewed provided any evidence to answer this question. 

What is the quality of the studies conducted from 2004 to 2011 to validate the EA screening tool when compared with USPSTF criteria?
All of the studies reviewed were descriptive,^24,29,31,32 descriptive cross-sectional,²⁷ descriptive correlational,^23,25,26,28 or predictive³⁰ per Burns and Groves’ descriptions of these study types.³³ Two were conducted as part of other longitudinal studies and met the USPSTF criteria for level II-3 studies; such studies take multiple time-series measurements.^23,30 Schofield and Mishra’s study was the only one that used a randomized sampling method.³⁰ No qualitative studies were identified for inclusion. Only four studies demonstrated good internal validity using the USPSTF criteria (ie, evaluated a relevant available screening test; used a credible reference standard; interpreted the reference standard independently of the screening test; assessed reliability of the test; had few indeterminate results or handled them in a reasonable manner; and included a larger number of broad-spectrum patients [≥100] with and without disease)^25,26,28,32 and only one study demonstrated good external validity using USPSTF criteria (ie, the study differs minimally from the US primary care population/situation/providers and only in ways that are unlikely to affect the outcome; it is highly probable [>90%] that the clinical experience with the intervention observed in the study will be attained in the US primary care setting).³² 

Discussion

While 10 studies met the inclusion criteria for this review, compared with only three when Nelson and colleagues⁵ conducted their review in 2004, the level of evidence has not improved. What is heartening, however, is that four of the 10 studies in our review were conducted in places other than the United States, United Kingdom, and Canada. This demonstrates a growing global awareness of EA as a public health concern that is worthy of ongoing investigation. 

The other important finding from our review is the identification of the development of five novel EA instruments: EASI,³² EPAS,³¹ OAPAM,²⁵ OAFEM,²⁶ and VASS.³⁰ When Nelson and colleagues⁵ had conducted their review, only CASE, HS-EAST, and Indicators of Abuse were available. In addition, the OAFEM focuses on identifying financial exploitation, demonstrating a continuing evolution in EA knowledge by acknowledging that financial exploitation is also a form of abuse.²⁶ With researcher commitment to the development and psychometric testing of new EA screening instruments, there are more EA screening instruments to consider. 

Unlike in Nelson and colleagues’ study,⁵ our review included two studies conducted in primary care settings^29,32 and one study conducted in hospital inpatient units.²⁸ Since Nelson and colleagues’ review,⁵ the number of caregiver screens eligible for review doubled from one to two, and the number of elder-only screens increased from two to six.⁵ From 2004 to 2011, two studies examined EA instruments designed to screen both caregivers and elders (care recipients).^27,28 

While the majority of the 10 studies included only English-speaking participants, some progress has been made to expand the language options of EA screening instruments to include French (EASI) and Hebrew (E-IOA).^28,32 Unfortunately, no studies were found in this review evaluating the use of EA screening instruments in Spanish, but this may be expected since the review was limited to only those studies written in English. 

Only one study included fewer than 100 participants.²³ Nine studies had more than 100 broad-spectrum participants, meeting the USPSTF internal validity criteria for “good” sample size. In the quality ratings performed by Nelson and colleagues,⁵ all three studies reviewed had sample sizes of 100 or more but received only “fair” ratings for sample size. This demonstrates an inconsistency between the quality ratings of Nelson and colleagues and that of this review. Another noteworthy difference is that the studies we assessed included participants with dementia,^23,24,27,29 demonstrating the burgeoning need to identify abuse in this population.

Our review examined several studies that evaluated the sensitivity and specificity of EA screening instruments to identify harm or risk of harm from EA. Sensitivity and specificity data gathered on the weighted scoring of the MCTS²⁴ (100%; 98%), E-IOA²⁸ (92%; 98%), CTS2/EAI²⁷ (75.4%; 70.6%), HS-EAST²⁹ (98%; 97%), and EASI³² (47%; 75%), respectively, could support adoption of these EA screening instruments. PPVs and NPVs for several of these instruments were also provided and found to be strong.^24,27,28 These data, too, could be considered robust evidence for adopting such instruments. No studies used likelihood ratios, which have been espoused by some clinicians to be a more practical method to evaluate diagnostic testing.³⁴ Although no EA instrument should be adopted for use based entirely on high sensitivity, specificity, or PPV or NPV, the determination of these values represents a significant step in EA screening instrument development.   

The USPSTF criterion for the evaluation of diagnostic studies requires a reference or “gold” standard by which to compare the accuracy of new screening tests.²² In several of the studies, varying types of reference standards were used because no gold standard for EA screening instruments currently exists. The fundamental errors that impacted the quality ratings of the studies reviewed centered on a poorly performing reference standard (MDSA used by Cooper and colleagues)²³; the interpretation of the reference standard using the EA screening instrument result^24,27,29; and failure to establish interrater reliability among the individuals using the reference standard.^28,29 A significant methodological flaw in which the caregiver was given the opportunity to respond to EA screening items for the care recipient resulted in a poor rating for the study investigating the use of the EPAS.³¹ Until an EA screening instrument is derived that provides evidence of truth of abuse (gold standard), EA researchers can improve the quality of the internal validity of their studies by careful selection and application of a reference standard. The studies conducted by Conrad and colleagues^25,26 and Yaffe and associates,³² which evaluated the performance of the OAPAM/
OAFEM and the EASI, respectively, provide excellent examples of good internal validity according to the USPSTF criterion, as there was conscientious selection and rigorous application of reference standards. 

The evaluation of external validity involves a comparison of the study population, situation, or healthcare provider to that found in the US primary care setting, and a determination of the likelihood that similar results could be obtained with the EA screening instrument in US primary care.²² In some studies, the EA instruments received fair to poor quality ratings because they lacked information regarding administration time.^24-27,29 Others received such ratings because of significant differences among their study populations or situations due to participation in a longitudinal study,^24,30 whereas other samples were noted to have significantly higher²⁷ or lower education and/or income³¹ than generally encountered among US primary care patients. The EPAS, which was developed and tested by Wang and associates in Taiwan,³¹ may have limited application in US primary care due to cultural differences, whereas the lengthy administration times (>2 hours) due to semi-structured formats limits the applicability of the E-IOA investigated by Cohen and colleagues²⁸ for the US primary care environment.The feasibility of using the OAPAM and OAFEM^25,26 requires further testing for sensitivity and specificity in a sample more representative of US primary care. The abuse-specific nature of these instruments may also limit their applicability within US primary care. These instruments may be more useful within a comprehensive geriatric assessment performed by social services. The EASI was the only instrument administered by clinicians in a primary care environment, and it received good external validity quality ratings because of its rapid administration time and acceptability among clinicians.³²

Study Limitations

While every attempt was made to capture and review studies meeting the inclusion criteria, it is possible studies were missed. In addition, the review and quality ratings were conducted by one investigator and were not confirmed by another reviewer. While this review attempted to expand on Nelson and colleagues’ systematic review of EA screening instruments,⁵ slightly more broad inclusion criteria and the addition of descriptive questions undoubtedly resulted in important differences in both the number of studies included and their subsequent analysis. Every effort was made to prevent hasty analytic conclusions regarding the nature and quality of the studies reviewed, as recommended by Whittemore and Knafl.²¹ 

Areas for Future Research

This review highlights the need for additional research to address the gaps in EA screening. Similar to Nelson and colleagues’ assessment,⁵ no studies investigated the possible adverse effects of EA screening. Future qualitative research could examine the lived experience of elders who screen positive for abuse using an EA screening instrument within primary care and are referred to Adult Protective Services. The longitudinal nature of this type of study presents many challenges, but the exploration of both the diagnostic and therapeutic impact of EA screening on elders deserves consideration. Feasibility studies are needed to examine the time requirement, impact on clinic processes, implementation fidelity, and costs of EA screening. 

The majority of the instruments were administered by researchers or experts. Additional research is needed to examine the impact of the instrument delivery method on the self-report of abuse by elders. Delivery methods to explore include administration by healthcare providers other than physicians, such as nurses or medical assistants, and the self-administration by elders. The use of technology to deliver EA screening instruments within the primary care environment also remains unexplored in the literature. While the inclusion of EA screening instrument studies on elders with dementia is significant, more investigation of the use of EA screening instruments needs to be conducted with various ethnic and cultural groups. More effort also needs to be made to validate instruments in other languages, especially Spanish. 

Conclusion

The purpose of this integrative review was to explore the state of EA screening instruments developed between 2004 and 2011. Ten studies were included and six questions were examined. We found substantial growth in the nature and number of studies evaluating EA screening instruments for caregivers, caregivers and elders (care recipients), and elders. We also found that noteworthy knowledge has been gained in the development of new EA instruments and in the ability of EA screening instruments to identify current harm or risk of harm using sensitivity, specificity, and PPVs and NPVs.The studies continue to provide lower-level evidence that is descriptive and cross-sectional in nature. Unfortunately, due to ethical concerns, higher level evidence involving randomized controlled trials may not be realistic. Three of the 10 studies were administered in healthcare environments and many investigated the use of short and less complicated EA screening instruments, as recommended by Nelson and colleagues.⁵ Four studies included participants with dementia, which may signal a reduction of barriers to conducting EA screening instrument research in this vulnerable patient population. Although there are many gaps left to explore with regard to EA screening, this review provides evidence that the challenge made by Nelson and colleagues in 2004 to “establish a research agenda to develop, test, and implement” effective EA screening instruments was heard and is on its way to being answered.