Validation of REAGERA-S: a new self-administered instrument to identify elder abuse and lifetime experiences of abuse in hospitalized older adults

ABSTRACT

This study aimed to develop and validate REAGERA-S, a self-administered instrument to identify elder abuse as well as lifetime experiences of abuse in older adults. REAGERA-S consists of nine questions concerning physical, emotional, sexual, financial abuse and neglect. Participants were recruited among patients (≥ 65 years) admitted to acute in-hospital care (n = 179). Exclusion criteria were insufficient physical, cognitive, or language capacity to complete the instrument. A semi-structured interview conducted by a physician was used as a gold standard against which to assess the REAGERA-S. The final version was answered by 95 older adults, of whom 71 were interviewed. Sensitivity for lifetime experiences of abuse was 71.9% and specificity 92.3%. For elder abuse, sensitivity was 87.5% and specificity was 92.3%. REAGERA-S performed well in validation and can be recommended for use in hospitals to identify elder abuse as well as life-time experience of abuse among older adults.

KEYWORDS:

• Screening and assessment instruments
• violence
• neglect
• polyvictimization
• elderly

Introduction

Geriatric care is based on a person-centered, multi-dimensional approach to promote better health for older adults (Besdine et al., 2005). Health can be negatively affected by experiences of abuse, both earlier and later in life (Baker et al., 2009; Campbell, 2002; Dong et al., 2009; Dong & Simon, 2013; Ellsberg, Jansen, Heise, Watts, & Garcia-Moreno, 2008; Felitti et al., 1998; Schofield, Powers, & Loxton, 2013; Simmons, Wijma, & Swahnberg, 2015). Abuse in later life is often referred to as elder abuse and can be defined as “a single or repeated act, or lack of appropriate action occurring within any relationship where there is an expectation of trust, which causes harm or distress to an older person” (WHO, 2002). Elder abuse can be categorized according to five different areas: physical, emotional, sexual, financial abuse, and neglect (Lachs & Pillemer, 2015; WHO, 2002).

Concerning children and youth, research on poly-victimization has underlined the importance of multiple victimization. Victims reporting several forms of violence are more likely to report renewed victimization as well as worse health outcome than victims of single forms of violence (Finkelhor, Ormrod, & Turner, 2007a, 2007b; Turner, Finkelhor, & Ormrod, 2010). Studies suggests similar findings also among older adults, albeit rarely investigated (Hamby, Smith, Mitchell, & Turner, 2016; Ramsey-Klawsnik, 2017; Teaster, 2017; Williams, Racette, Hernandez-Tejada, & Acierno, 2017). Recently, experiences of childhood abuse have also been found to be associated with reporting abuse as well as ill-health in older age (Kong & Easton, 2018; Maschi, Baer, Morrissey, & Moreno, 2013; Wang & Dong, 2019). Hence, from a health care standpoint, a life-course perspective is well motivated.

The health care system is considered an important institution for identifying and managing elder abuse (Cohen, Levin, Gagin, & Friedman, 2007; X. Dong, 2015, X. Q. Dong, 2015; Ministry of Health and Social Affairs, 2014). Despite elder abuse being a known and serious health problem, many health care professionals hesitate to ask about experiences of abuse (Mohd Mydin & Othman, 2017; Schmeidel, Daly, Rosenbaum, Schmuch, & Jogerst, 2012). Furthermore, older adults seldom self-report abuse and are often unaware of where to seek help (Eriksson, 2001; Erlingsson, Saveman, & Berg, 2005; Fulmer, Guadagno, Bitondo Dyer, & Connolly, 2004). To be able to support the older adult to better health, the health care system requires better methods to identify abuse (Brijnath, Gahan, Gaffy, & Dow, 2018).

Screening for abuse in hospitalized patients may have advantages compared with screening in an outpatient environment or in the general population. The in-hospital setting may allow opportunities for enquiry in private, not always easily available in some other settings. The patient is also at a safe distance from the potential current abusive environment; for example, if living with an abusive spouse, other family member or in a long-term care facility. To improve identification of abuse, the use of brief screening instruments has been suggested (Cohen, 2011; X. Dong, 2015; X. Q. Dong, 2015; Fulmer et al., 2004). These instruments are often constructed as checklists for health care professionals, including risk factors and observations, or a set of questions for the professionals to ask the patient (Cohen, 2011; Dong, 2017; Dong & Simon, 2014; Gallione et al., 2017).

Another way of screening for elder abuse is the use of a self-administered instrument, followed by a more thorough assessment if abuse is indicated (Murphy, 2011). In intimate partner violence research, self-administration has been shown to be preferred compared to a face-to face approach by women in acute care and in an outpatient gynecology setting (MacMillan et al., 2006). A self-administered instrument might be difficult in some cases of limited cognitive, physical, or language capacity, and in such instances another mode of assessment should be considered. Self-administered instruments for identifying elder abuse have been shown to be feasible in an outpatient context (Yaffe, Weiss, & Lithwick, 2012). However, screening with a self-administered instrument has not been tested in acute inpatient care for older adults and therefore needs to be further investigated.

To identify experiences of abuse, a suitable self-administered instrument should have certain qualities. It should cover all five areas of abuse and cause minimal side effects and negative experiences. It should be feasible in the sense of being short and easily understood, and as many hospitalized older adults as possible should be able to complete it. Previously developed instruments, such as HS-EAST (Neale, Hwalek, Scott, & Sengstock, 1991), VASS (Schofield & Mishra, 2003), and FREDA-kortfrågor (National Board of Health and Welfare, 2014) which is recommended in Sweden, have different strengths and shortcomings. However, none has been shown to meet the all the qualities mentioned previously (Brijnath et al., 2018; Cohen, 2011; Dong, 2017; Feltner et al., 2018; Gallione et al., 2017; National Board of Health and Welfare, 2014). Most commonly, one or more areas of abuse are not covered, or the instrument has not been validated against a gold standard. EASI (Yaffe, Wolfson, Lithwick, & Weiss, 2008), a commonly used instrument, is an exception in that it covers all five areas of abuse; however, the validation, conducted on the non-self-administered version, revealed a low sensitivity of 47% for detection of abuse. Thus, there is a need for a self-administered instrument that covers different areas of abuse sufficiently, can be used in the hospital setting, and satisfies validity measures. The main aim of this study was to develop and validate REAGERA-S (Responding to Elder Abuse in GERiAtric care-Self-administered), an instrument for identifying elder abuse and life-time experiences of abuse among older adults. Further, we aimed to 1) evaluate the feasibility of using the instrument for older adults admitted to acute care hospital 2) evaluate self-administration of a likert-type scale regarding suffering from abuse, and 3) evaluate experiences of answering questions about abuse.

This study is presented as recommended by STARD, Standards for Reporting of Diagnostic Accuracy Studies (Cohen et al., 2016). It was conducted as part of the larger REAGERA project in Linköping, Sweden. “Reagera” in Swedish means “reacting” and the main aim of the project is to improve the health care response to older victims of abuse.

Materials and methods

Initial development of the REAGERA-S instrument

The REAGERA-S instrument (Table 1) was constructed after reviewing previous screening instruments in both English and Swedish concerning elder abuse as well as intimate partner violence (Erlingsson, Carlson, & Saveman, 2003; Kristensen & Lindell, 2013; National Board of Health and Welfare, 2014; Neale et al., 1991; Schofield & Mishra, 2003; Swahnberg, 2011; Swahnberg & Wijma, 2003; Yaffe et al., 2008). Based on the literature an initial set of questions was constructed by the first and last authors. The process was guided by the aim to cover all five areas of abuse with as few questions as possible to increase readability. Also, when available we wanted to use wordings similar to previously used Swedish instruments, and questions were especially inspired by the NorVold Abuse Questionnaire (Swahnberg & Wijma, 2003) and FREDA kortfrågor (National Board of Health and Welfare, 2014). Because the most common form of elder abuse in Sweden is emotional abuse (Eriksson, 2001; Kristensen & Lindell, 2013) we decided to use several questions to try to capture this complex phenomenon. Based on the poly-victimization framework and well-known long-standing consequences of abuse on health (Campbell, 2002; Ellsberg et al., 2008; Felitti et al., 1998; Maschi et al., 2013; Simmons et al., 2015; Teaster, 2017), we decided to include response categories for experiencing abuse both before and after 65 years, even though ongoing elder abuse was our primary focus.

Table 1. The REAGERA-S instrument.

Area of abuse	Item	Version 2 (Final)	Modifications from version 1
Headline		Vulnerability of older people throughout life	“Throughout life” was included in the headline
Introductory text		It is quite common during one’s life to be subjected to negative actions committed by people close to you, care workers or others. Such experiences can have a negative impact on health, even if they happened a long time ago. We would therefore like you to please answer the following questions:	New introductory text to clarify that experiences earlier in life also matter
Emotional	Q1	Have you felt uncomfortable with or afraid of someone close to you or a care worker?	Changed the order of the examples to increase sensitivity
Emotional	Q2	Has anyone repeatedly spoken to you in a disrespectful manner or attempted to put you down, degrade you or humiliate you?	No modifications
Emotional	Q3	Has anyone attempted to control you, limit your contact with others, or decide what you may or may not do?	Changed “make your decision for you” to “control you”
Emotional	Q4	Has anyone threatened to harm you or someone or something that you care about, for example other people, pets or possessions?	No modifications
Economic	Q5	Has someone close to you or a care worker exploited you financially, for example by stealing from you, taking control of your finances, cheating you out of money or objects, or demanding these from you?	Clarified that we meant by next of kin or care workers
Physical	Q6	Have you been subjected to any form of physical violence, for example being shoved, pinched, held down, hit or kicked?	No modifications
Sexual	Q7	Have you been subjected to any form of sexual abuse, for example someone touching your body against your will or forcing you to perform sexual acts?	Shortened, removed “behaved violently toward you in sexual situations”
Neglect	Q8	Have you ever not been given the help that you need, for example with taking medicine, going to the toilet, getting out of bed, getting dressed or getting food or drink?	Changed from 2 questions: “Has anyone prevented you from using an accessibility aid you needed” and “has anyone refused to help you with … (same examples as in version 2)”
Neglect	Q9	Has someone close to you or a care worker who should have been helping you treated you badly, for example by speaking to you in a disrespectful manner or handling you roughly?	New question
	Q10	How much do you suffer today as a result of what you have been subjected to? (Place a cross on the scale below) Visual analog scale ranging from “not suffering at all” (= 0) to “suffering very severely” (= 10)

To ensure content validity, a group of researchers and clinicians in geriatric care met to discuss and revise the initial questions. When agreement was reached, the content and form were further elaborated with the guidance of a research consultant specializing in survey design and respondents’ perspectives in survey research. Adjustments were made to the response categories, layout, and wording of the questions. The first reference group was then asked again for their final input.

To ensure face validity and to identify problems with understanding and answering the questions, six cognitive interviews were conducted (Willis, 1994). The six respondents were selected among patients admitted to the same wards as the study population and were recruited a few weeks before the study commenced. Their responses in the interviews were not included in the validation study, and they were only used to guide construction of the REAGERA-S. Cognitive interviews were performed using the “think aloud technique”, meaning that the respondents were asked to comment on the questions while filling out the instrument. The respondents were also observed for their behavior while filling out the instrument. Overall, the questions were well understood and no revision of the instrument itself was needed. The layout with Arial font size 13 and lines between all questions were well tolerated. However, background questions such as educational level and living situation were often misinterpreted and therefore moved from being an appendix to the instrument to be a part of the oral interview.

The instrument was validated in Swedish and translated to English by a professional translator. The translation was then translated back to Swedish by another professional translator and compared with the original Swedish version. The English version was then adjusted in agreement with the translator and authors.

Setting

The study was conducted in one acute medical ward and one acute geriatric ward belonging to the same university hospital clinic. Patients are most often admitted directly from the emergency department and the mean duration of hospital stay for patients over 65 years was 9.9 days at the geriatric ward and 4.1 days at the medical ward during the study period. Common reasons for admission to the wards are pneumonia as well as other infectious diseases, congestive heart failure and chronic obstructive pulmonary disease. Both wards have a holistic approach and conduct a multi-professional assessment of the patient.

Recruitment of participants

Participants were recruited consecutively and there were no recruitments during the weekends. Adults older than 65 years who were admitted between January and June 2018 were eligible for inclusion. Exclusion criteria were insufficient physical, cognitive, or language capacity to answer the questions presented in the instrument. Patients excluded for physical reasons were most often very ill, sometimes in palliative care.

To minimize the risk of recruitment bias, a member of the research team (one of the first three authors) visited each ward every weekday to make sure all patients were considered for inclusion. Participants were included after a subjective assessment by the nurse responsible for the patient on the day of recruitment. Hence, all nurses working at the wards helped in the recruitment of participants. Before the study started all personnel attended a brief (45 minute) introduction to explain what elder abuse is and to inform them of the study. Nurses were instructed to invite patients 65 years and older to participate if they believed the patient was able to fill out REAGERA-S. The reason for using a subjective assessment instead of a standardized instrument for assessing, for example, cognitive capacity, was the variable status of the possible participants during their hospital stay. An inpatient was often too ill to participate during the first days of their hospital stay. When he or she had recovered physically, they could sometimes experience lingering confusion. Many participants were included as late as the day before discharge, and during this brief period it was not considered possible to do any comprehensive cognitive assessment. Also, REAGERA-S is intended for use in the same busy acute care environment as it was tested, where the cognitive capacity of the respondent is often not known. Instead, daily subjective assessments by the nurses were made. Also, participants were interviewed (process described later) by one of the three first authors, all licensed physicians trained to detect confusion or other cognitive impairments. If in doubt about the cognitive status of the respondent the third author was consulted, who is both a specialist in geriatrics and psychiatry. When confusion or confabulation was detected in the interview the participant was excluded. A flowchart of participant selection is shown in Figure 1.

Figure 1. Flowchart for participants.

Written instructions on how to ask the patient for participation were as follows: “At our clinic, there is a study currently being conducted about older adults’ experiences of violence and abuse. We would like to include both patients subjected to abuse and patients who have never been abused in the study. To participate entails answering a brief questionnaire and thereafter participating in an interview. Here is some more information and the questionnaire”. The instrument was administrated together with three pages of follow-up questions and an informed consent form. The informed consent clarified that the topic being examined was abusive experiences, underlined that participation was voluntarily, and it assured participants of confidentiality. If the participant was able to fill out the form independently, the instrument was left in the room and picked up a few hours later; therefore, the participant could choose not to participate after reviewing the questions.

Some participants required assistance due to, for example, severe visual impairment or arthritis in the hands. In such cases staff were instructed to give the required assistance, for example reading the questions and the informed consent information aloud. It was important not to exclude these patients considering that physical impairment is a risk factor for elder abuse (Johannesen & LoGiudice, 2013). There was also an ethical reason not to exclude these participants who were not able to perform full self-administration, as the participants in these cases often requested the assistance themselves.

The same day or the day after answering REAGERA-S, the participants were interviewed by a member of the research team. The time span between answering REAGERA-S and participating in the interview was short because many participants were not included until the last days of their hospital stay. We considered interviewing the participants a few weeks after discharge, but we decided against it for two reasons. First, and most importantly, it was deemed too risky as confidentiality could not be assured in the homes of participants who might be living in an abusive environment. The alternative to invite participants back to the hospital for the interview a few weeks later was deemed unethical considering that for a large proportion of the study population (who were recently hospitalized older adults) returning to the hospital would constitute a substantial effort. Second, we wanted to minimize the number of participants lost to follow up.

To calculate the required sample size, we estimated a 10% prevalence of elder abuse (Eriksson, 2001; Ho, Wong, Chiu, & Ho, 2017; Kristensen & Lindell, 2013; Yon, Mikton, Gassoumis, & Wilber, 2017). We aimed for 90% sensitivity. With a null hypothesis of 50% sensitivity, a sample size of 120 participants is required to gain sufficient power for the prevalence rate 10% (Bujang & Adnan, 2016). To stratify the results by sex we needed a sample size of 300 patients. These numbers are consistent with experiences of similar previous validation procedures (Swahnberg, 2011; Swahnberg & Wijma, 2003).

The interview

The gold standard against which to assess REAGERA-S was an interview with each participant by one of the first three authors. All interviewers are licensed physicians, well-trained in clinical interviewing and with qualitative research experience and education. The interviewers were blinded to the result of the instrument, which meant that participants with negative screening results were also interviewed. The purpose of the interview was to assess if the participant had experienced abuse or not, and if so to determine the category of abuse (i.e., physical, emotional, sexual, financial abuse, or neglect). Confidentiality was assured by interviewing all participants in private. We started the interview by asking for background information, for example about their family, functioning in activities of daily living (ADL) and their living situation. Thereafter we explained that abuse is common and includes a broad variety of experiences, and that we were interested in recent experiences as well as what had happened a long time ago. We asked if the participants had been exposed to any such events. Depending on each participant’s response, this was followed by a series of open-ended questions. If abusive experiences were not revealed with these questions, we proceeded to more direct questions. The interview guide used is presented as an online supplement.

When there was any doubt in the assessment of whether the participant’s experience should be classified as abuse or not, there was a discussion between the first two or three authors to form a consensus. These researchers were still blinded to the result of the screening instrument and discussions were often regarding experiences that happened a long time ago. For example, some participants shared stories from their childhood about mild physical punishment by a teacher or a parent. If the participant regarded the experience as normative of that time and social context rather than as abusive, it was classified as negative. However, if there were any indications that the participant had experienced the event as abuse or expressed they had suffered the aftermath of it, we classified that participant as positive for abuse. Difficulties concerning classifications of experiences in later life most often occurred concerning mild forms of neglect or emotional abuse that the participant tended to downplay. In these discussions we were more prone to code participants as positive for abuse, often supported by indications that the experience still affected the participant. Also, we wanted to be sure not to overestimate the sensitivity of REAGERA-S for elder abuse.

All participants rated positive for abuse were assessed regarding whether they still suffered from their experience. The assessment was in part based on patients’ own accounts of suffering, but we also considered current psychiatric illness in the aftermath of abuse as well as if the patient had changed their behavior due to abusive experiences. The interview was completed by asking the participant how they experienced participating in the study.

Revision of the instrument

Sensitivity was monitored continually during data collection and initially found to be low. After 84 participants and 63 interviews, the instrument was therefore revised. We realized during the interviews that the participants often trivialized their experiences. The questions were therefore generally broadened and mitigated as can be seen in Table 1. As the intended clinical use of the instrument is to do a follow-up interview in case of indications of abuse, the risk of false positives was not as relevant as the risk of false negatives.

The most extensive revisions were done on two questions to enquire about neglect and to highlight our interest also about abusive experiences occurring earlier in respondents’ lives. Regarding neglect, the original questions, “Has anyone prevented you from using an accessibility aid you needed” and “Has anyone refused to help you with taking medicine, going to the toilet, getting out of bed, getting dressed or getting food or drink?” were taken from the Swedish instrument “Freda-kortfrågor” (National Board of Health and Welfare, 2014). No true positives for neglect were found in the first 63 interviews, and considering the content of the interviews it was evident that the wording of the questions was too harsh. Also, this wording does not correlate with the WHO definition of neglect (WHO, 2002). The WHO definition does not require the caregiver’s intention to abuse which is suggested by the words “prevent” and “refuse”. This is consistent with the development of the long-term care version of EASI, where a similar adjustment was made after discussion in focus groups (Ballard, Yaffe, August, Cetin-Sahin, & Wilchesky, 2019). This mitigation is further supported by the low sensitivity of 4% on the very similar question from EASI: “Has anyone prevented you from getting food, clothes, medication, glasses, hearing aids or medical care, or from being with people you wanted to be with?” (Yaffe et al., 2008). Therefore, the wording was adjusted as can be seen in Table 1.

Abuse that occurred earlier in life was initially a secondary focus of this study. It became recurrently obvious during the interviews that abuse earlier in life could still affect the participants to the extent that it was clinically relevant. The final decision to do the revision was made after a false negative interview were the participant had suffered extensive trauma during childhood and had recently been diagnosed with post-traumatic stress disorder. But due to the focus on elder abuse in the instrument this participant had not screened positive. Thus, the instrument was revised to better capture life-time experiences of abuse.

In version 1, there were three alternatives to each question: “no”, “yes, before 65,” and “yes, after 65.” However, the three alternatives were often misinterpreted by the respondent which is why the response categories in the final version were changed to “yes” and “no.” Since the intended clinical use is to do a follow-up interview in case of a “yes”, we concluded that it was not necessary to get more specific information with the instrument about when in life the abuse had occurred.

Statistical analysis

The calculations were conducted in SPSS Statistics 24. Validity measures such as sensitivity and specificity were the main outcome measures. The outcome of the index test, the REAGERA-S, was dichotomous (positive/negative). One “yes” on Q1–Q9 was considered a positive result. Separate validity analyses were made for lifetime abuse and elder abuse. In the analysis concerning lifetime abuse, all respondents were included and categorized as never abused or abused at some time during their lifetime. Likewise, in the analysis concerning elder abuse, respondents were categorized as never abused or subjected to elder abuse (≥65 years). Respondents reporting abuse only before the age of 65 years were excluded from the analyses on elder abuse. This is because they would otherwise be positive on the instrument and negative in the interview (false positive), whereas in reality they would be rated as true positives of abuse occurring before the age of 65 years. Confidence intervals (CI) for sensitivity and specificity were calculated with Clopper-Pearson CI. For all analyses, 95% CIs were used.

Background information, such as age, living situation, education and ADL was used to explore differences between the interviewed group and the non-interviewed group as part of a dropout analysis. Comparisons of mean age differences were calculated with the t test, and differences in categorical data were calculated with Fisher’s exact test or Pearson’s chi-squared test.

Ethical aspects

The regional ethical review board in Linköping, Sweden, approved the study (Registration no. 2017/181-31 and 2017/564-32). The participants gave written informed consent to participate. Confidentiality was ensured by conducting all the interviews in private. The risk of causing the participant harm by asking about abuse was handled with a series of follow-up strategies. First, all participants were given a card with contact information to resources that assist older adults subjected to abuse. Second, all participants were asked how they experienced participating in the study, both in terms of answering the instrument and regarding the interview. Finally, participants reporting elder abuse or effects from abuse earlier in life were followed up on the phone by the interviewer a few weeks after the interview.

Results

Validity measures of the instrument

Cross tabulations and validity measures for lifetime abuse and elder abuse are presented in Tables 2 and 3. For lifetime abuse, the sensitivity of the instrument was found to be 71.9% (95% CI 53.3–86.3) and specificity 92.3% (95% CI 79.1–98.4). For elder abuse, the sensitivity of the instrument was found to be 87.5% (95% CI 61.7–98.5) and specificity 92.3% (95% CI 79.1–98.4).

Table 2. Cross tabulation of the outcome of screening with REAGERA-S compared with the gold standard (the outcome of the interview) (N = 71).

Lifetime abuse	Positive interview	Negative interview	Elder abuse	Positive interview	Negative interview
Positive REAGERA-S	23	3	Positive REAGERA-S	14	3
Negative REAGERA-S	9	36	Negative REAGERA-S	2	36

Table 3. Validity measures calculated from Table 2.

	Lifetime abuse		Elder abuse
		95% CI		95% CI
Sensitivity (%)	71.9	53.3–86.3	87.5	61.7–98.5
Specificity (%)	92.3	79.1–98.4	92.3	79.1–98.4
Positive likelihood ratio	9.3	3.1–28.3	11.4	3.8–34.3
Negative likelihood ratio	0.3	0.2–0.5	0.1	0.04–0.5
Positive predictive value (%)	88.5	71.7–95.9	82.4	60.8–93.4
Negative predictive value (%)	80.0	69.5–87.5	94.7	83.1–98.5

In calculations on elder abuse, cases positive for abuse before 65 years but not after 65 years (n = 16) were excluded. This is because they would be positive on the instrument and negative in the interview (false positive), whereas in reality they are true positives for abuse occurring before the age of 65 years.

False-negative cases

For lifetime abuse, there were nine false negatives, i.e., these patient’s answers on the REAGERA-S classified them as negative, but the interview classified them as positive for lifetime abuse. One of them was found to still suffer from these experiences and was positive for both abuse before 65 years (emotional abuse) and after 65 years (physical abuse and neglect). This participant was not perceived to have a need for interventions, because the abuse was not ongoing, and because she had already received therapy and had strategies for dealing with the aftermath of her experiences. Also, the participant did not request any further help or follow-up.

Background characteristics

Background characteristics of the respondents are presented in Table 4. The only significant differences between the interviewed group and the non-interviewed group were a higher percentage of married participants in the non-interviewed group (p =.01) and a higher percentage of widowed participants in the interviewed group (p = .002).

Table 4. Background characteristics of interviewed participants (n = 71) Compared with non-interviewed participants (n = 24).

	Interviewed n (%)	Non-interviewed n (%)
Age (years)
Mean (standard deviation)	80.2 (8.4)	78.8 (8.0)
Range	66–98	66–98
Sex
Female	39 (54.9)	17 (70.8)
Male	32 (45.1)	7 (29.2)
Children
Yes	64 (90.1)	9 (90.0)
Living situation
Married*	35 (50)	13 (86.7)
Divorced	8 (11.4)	1 (6.7)
Widow/widower**	28 (40.0)	0 (0)
Assisted living	2 (2.9)	1 (5.9)
Living alone	34 (48.6)	6 (33.3)
Educational level
6–9 years	41 (58.6)	Missing data
10–12 years	11 (15.7)
>13 years	18 (25.7)
Help with instrumental ADL	28 (41.2)	5 (33.3)
Help with basic ADL	12 (18.8)	2 (13.3)
Help with medications	12 (17.6)	2 (15.4)
Need a wheelchair or stroller	29 (44.6)	8 (44.4)
Item non-response	0–7	0–14

Background information on the participants interviewed was collected during the interview and information on the participants not interviewed was collected from medical records. ADL, activities of daily living. Significant differences between interviewed and not interviewed, calculated with Fishers’ exact test: *p =.01 **p =.002.

Feasibility and participant completion

For recruitment of participants see Figure 1. When including participants, 66.5% (n = 444) of the admitted patients were considered eligible for inclusion and among those 68.9% (n = 306) were approached. Of the 306 approached, 191 patients (62%) accepted participation. Among the 138 patients who were eligible but not approached, 27% (n = 37) were moved to another department at the end of their hospital stay and 31% (n = 43) were discharged during weekends. Because we mainly included patients during the last days of their hospital stay these patients were missed. There is no evident reason for why the remaining 42% (n = 58) were not approached, but a likely cause is time restraint among staff.

The instrument was completed independently by 88.4% (n = 61) of the participants. Eight participants (11.4%) needed assistance, meaning a staff member at the ward read the questions and the informed consent information aloud without explaining further. Data on the need of assistance were missing for 26 participants, 24 of whom were not interviewed. Validity measures with these participants excluded can be found in a supplementary file online.

Out of the 95 participants included in the study, 24 (25.3%) were never interviewed. There were two reasons for this: declining the interview (n = 14) and being discharged before the interview could take place (n = 10). Of the 14 participants who declined to be interviewed, three were positive for abuse according to the instrument. Those not interviewed were less likely to report abuse in the REAGERA-S compared with those who were interviewed, but the difference was not statistically significant (non-interviewed 16.7%, n = 4; interviewed 36.6%, n = 26; p = .07).

Suffering

For elder abuse, nine of 16 participants subjected to abuse (56.3%) were assessed in the interview to suffer currently from abuse. For lifetime abuse, the proportion of suffering was 12 of 32 participants (37.5%). In REAGERA-S, participants were asked to assess their own current suffering of abuse on a scale from 1 to 10 (Q10, Table 1). Of the participants assessed to suffer according to the interview, all but one had also reported suffering on Q10.

Experiences of participating in the study

The experience of answering REAGERA-S was reported as “mostly positive” by 20% (n = 12) and “neither positive nor negative” by 78% (n = 46). Participating in the interview was reported as “mostly positive” by 39% (n = 23) and “neither positive nor negative” by 58% (n = 34). Older adults reporting abuse were more likely to experience the interview as “mostly positive” than patients classified as not experiencing abuse (p = .006) (Table 5).

Table 5. Experiences of participating in the study: responses to questions asked at the end of the interview (N = 59).

	”How did you experience filling out the instrument?”					”How did you experience participating in the interview?”
	Lifetime abuse		Never abused			Lifetime abuse		Never abused
	n	%	n	%	p	n	%	n	%	p
Mostly positive	8	27.6	4	13.3	.15	16	55.3	7	23.3	.006
Neither positive nor negative	20	69.0	26	86.7		11	38.0	23	76.7
Both positive and negative	1	3.5	0			2	6.9	0
Mostly negative	0		0			0		0
Item non-response	3		8			3		8

p values calculated with Pearson’s χ² test for differences between never abused and lifetime abuse, and differences between participants reporting a “mostly positive” or “neither positive nor negative” experience.

All participants subjected to elder abuse and those with lifetime abuse with current effects were offered a phone follow-up. Several declined follow-up and thought it was unnecessary, and we failed to get in touch with a few participants despite multiple attempts. Only one participant reported a negative experience of study participation in the follow-up phone call. This was due to feeling bad about sharing such a personal story in the interview.

Discussion

In this study, we have described the development and validation of a new self-administered instrument for older adults’ experiences of abuse that covers all five areas of abuse, is feasible for use in acute care hospitals, and has satisfactory validity measures. REAGERA-S may be a step toward better identification of abuse and consequently better opportunities to support the health and security of older adults.

REAGERA-S is intended to be used for in-hospital screening to identify experiences of abuse. Positive cases should be followed up with an interview to assess further the older adults’ experiences of abuse and what kind of follow-up and/or referral is needed. The life-time perspective of REAGERA-S is in line with research on poly-victimization and is a strength of the instrument, as it may help with: 1) Identification of older adults who are suffering in the aftermath of abuse and that can be helped by for example be referred to psychiatric care, regardless of when the abuse occurred. 2) Identification of ongoing abuse so that actions can be taken to help victims. Screening for elder abuse could potentially be included as a routine part of a comprehensive geriatric assessment of hospital in-patients (Parker et al., 2018). Health professionals could choose to screen with REAGERA-S, with another instrument better suited to older adults with substantial cognitive impairment, or if preferred, ask the patient directly about abuse. It is important to have several options, including a self-administered option, to give older adults the chance to speak for themselves regarding these matters whenever possible.

There is no established gold standard for identifying elder abuse. We chose an interview as a gold standard to give the older adults their own voice and the chance to define their experiences as abusive or not. An interview could have the disadvantage that it is dependent on the interviewer’s ability to gain participants’ trust. There is also a risk of researcher allegiance bias (Manea, Boehnke, Gilbody, Moriarty, & McMillan, 2017; Singh, Grann, & Fazel, 2013), especially when assessing exposure to abuse in the interview situation. This was compensated for by blinding the interviewers to the result of the instrument and by using an interview guide to guarantee that all questions were asked. The interview which followed closely after instrument completion also came with the risk of short-term recall bias, in the sense that the participants may not have wanted to disclose abuse in the interview if they did not do so in the instrument. To minimize this risk, the interview guide started with open-ended questions before proceeding to closed questions. The opposite effect of this form of recall bias, that answering REAGERA-S would prime participant to report abuse also in the interview, is more acceptable as it can be considered to strengthen the interview as a gold standard. Also, the validation process conforms with the intended clinical use of the instrument which is in conjunction with a follow-up interview for all patients who are found to be positive for abuse according to the REAGERA-S. However, the short time span between answering the instrument and participating in the interview could partly explain the quite high specificity of 92.3%.

The sensitivity of 87.5% for elder abuse was close to the target detection rate of 90%. A high sensitivity of a screening instrument is fundamental to its use in clinical practice. If the sensitivity is not adequate, the instrument cannot be trusted as a way of identifying older adults in need of help and support.

Regarding feasibility, as many as two-thirds of hospitalized older adults were assessed as being able to complete a self-administered instrument, three out of five of these accepted participation, and nine out of ten participants could fill out the instrument independently. Only four patients were excluded due to confusion at the time of the interview (Figure 1). This indicates that using a subjective assessment by nurses concerning whom to include is an acceptable strategy. In conclusion, we believe that hospital is a suitable environment to screen for abusive experiences among older adults and that REAGERA-S is a feasible instrument to be used for patients with the cognitive capacity to answer the questions, either using self-administration or administration with assistance from staff.

The question on suffering seems to perform well in validation with only one false negative. It is hard to draw any conclusions from this finding due to the low number of participants. However, it can be seen as an observation of value to be considered in future research.

The possible side effects and negative experiences were assessed with follow-up questions. Participants subjected to abuse experienced the interview as “mostly positive” to a greater extent than the participants that had not been subjected to abuse. In the follow-up phone call, only one participant revealed negative feelings about disclosing abusive experiences in the study. Thus, screening with REAGERA-S and a follow-up-interview, which corresponds to the intended clinical use, was perceived as predominantly positive.

Limitations

A few limitations must be mentioned. First, the intended sample size of 300 participants was not achieved. The main reasons for this were the revision of the instrument, which was not envisaged from the beginning, and an unexpectedly high number of patients were not asked to participate. Reasons for the latter were mainly time constraints among staff, patients moving to another ward before being asked and patients being discharged during weekends (Figure 1). None of these reasons indicates a systematic bias. To mitigate the lower sample size, calculations were not stratified by sex as originally planned, and as such an acceptable number of participants was obtained. The sample size could still be considered small (n = 71). This was compensated to some extent by the chosen methodology. Because most participants who screened negative were interviewed, we know that only very few of them were false negatives and that only one false negative case was considered to be still suffering from the abuse. Also, although not intended at the beginning of the study, the revision of the instrument can also be considered a strength in the instrument’s development. Based on the initial interviews we learned how to broaden the questions in relevant ways and thereby to increase sensitivity. Also, despite the rather small sample size, the sensitivity of REAGERA-S has 95% confidence intervals that even at the lower range (61.7 for elder abuse and 53.3 for lifetime abuse) exceed the shown sensitivity of EASI at 47% (Yaffe et al., 2008). Comparisons with EASI should however be done with caution since the version of EASI, for which sensitivity data is available, is not a self-administered instrument and furthermore was tested in a very different sample and setting using a different gold standard.

A second limitation was that one in four participants filled out the REAGERA-S but did not participate in the interview. Though not statistically significant, the non-interviewed group had a tendency to reported less abuse in the instrument than the interviewed group, which could have affected the results of the validation. The lower prevalence in this group is however in accordance with research concerning non-response bias where non-exposed individuals are reported to be less inclined to participate (Galea & Tracy, 2007; Halbesleben & Whitman, 2013). Several of the participants explained that they declined the interview with the comment that the study was not applicable to them, as they had not been exposed to any abuse. Similarly, in a study concerning non-response bias in interpersonal violence research, one of the most common reasons for declining participation, “does not concern me”, indicated a lack of interest (Simmons & Swahnberg, 2019). Also, known risk factors for elder abuse, such as poor functioning in ADL and female sex, were evenly distributed between the groups. Significant differences between the groups were found regarding marital status where married people were less likely and widows/widowers were more likely to be interviewed. Marital status is a factor where a difference in the risk for elder abuse has not been fully investigated, but the few studies that exist point toward a higher prevalence of abuse in divorcees than in married older adults. (Dong, 2017; Johannesen & LoGiudice, 2013) Altogether this indicates that the risk of overestimating sensitivity due to limited participation in the interview is low. Also, we believe that the risk of under-sampling of participants subjected to abuse was decreased by conducting our study in hospital, at a distance from the abusive environment. This underlines the importance of the context when screening for abuse.

The final version of REAGERA-S is not able to distinguish between elder abuse and abuse that occurred earlier in life, as the only response categories are “yes” and “no”. In validity calculations (Tables 2 and 3) data concerning age at abuse is derived only from the interviews. However, abuse is well known to have long-standing consequences for victims. As previously mentioned, it was obvious in the interviews that also abuse earlier in life caused suffering for participants. As REAGERA-S is intended for clinical use with a follow up interview, details of the abuse including timing of the event can be elaborated upon in the interview. However, our data concerning the different time spans, show that REAGERA-S adequately captured both lifetime experiences of abuse and elder abuse.

Further research is intended to determine how to identify abuse in older adults where REAGERA-S is not feasible; for example, in older adults with severe cognitive impairment. It should, for example, be evaluated whether REAGERA-S could be used by proxy, where health or social care professionals or a next of kin to an older adult could answer the questions. REAGERA-S also needs to be validated in different settings, such as long-term care facilities and outpatient care. Also, it would be valuable to replicate this study with larger sample sizes to narrow the confidence intervals and to test the validity and feasibility of the REAGERA-S in different cultural settings. For future research, we can also learn from this study that older adults do not perceive it as negative to be asked about abuse. The concept of suffering, and that self-reported suffering from abuse seems to correspond well to what can be evaluated in an interview, is interesting and should also be further studied.

Conclusion

REAGERA-S is a nine-item self-administered instrument showing good properties for screening for elder abuse. It is feasible for use in acute inpatient care for patients able to complete self-administration or complete the instrument with assistance from staff. Negative experiences of answering questions about abuse, both in writing and in the interview, were rare. The self-administered scale question about suffering should be included in further similar studies as it showed promising qualities. The instrument is novel in the sense that it also captures clinically relevant lifetime abuse in older adults. This study indicates that REAGERA-S is more sensitive than previous instruments, which makes it an attractive and promising alternative for in-hospital screening of abuse among older adults. When abusive experiences are disclosed in health care, an opportunity to give help and support is revealed, and the quality of life of older adults may be improved.

Acknowledgments

Special thanks to: Researchers participating in reference group for in the development of the instrument: Elisabet Classon, Linköping University; Maria Johansson, Linköping University; Britt-Inger Saveman, Umeå University; Marika Wenemark, Linköping University; Ewa Wressle, Linköping University. All the nurses and nursing assistants who conducted data collection in addition to their daily work on the ward. “Tantjouren”, part of the women’s shelter in Norrköping supporting older women, for all your valuable input.

Disclosure statement

No potential conflict of interest was reported by the authors

Supplementary material

Supplemental data for this article can be accessed here.

Additional information

Funding

The project was supported by the Swedish Crime Victim Fund under grant no. 3322/2017 and 2944/2018 and the County Council of Östergötland under grant no. LIO-626861.