Elder Abuse Workshop: Measurement and Data: Measuring Prevalence and Incidence of Elder Abuse in the Community - Reports from Current or Recently Completed Research Projects

Introduction — Sid Stahl, NIA

In the first cohort of grantees, a grant by Dr. Mark Lachs, a geriatrician, for a very little bit of money looked at the New Haven Study and compared APS (adult protective services) reports — he found that death rates among those with confirmed cases of mistreatment were three times higher than those with no APS report. This was very dramatic. In 1999, wondering why so little on elder mistreatment was in the literature, I went to look at the science. It took very little time because there was so little research in elder mistreatment. We began talking with leaders in the field and learned that research on elder mistreatment was exquisitely difficult because so few people were working in this area. That is still true today — only 4/10 of 1 percent of NIA funding goes to elder mistreatment. We convinced the director of NIA to give $350,000 to the National Academy of Sciences to do the 2003 study — the result was “the Bible” that included measurement issues as well as prevalence and incidence issues.

Incidence and prevalence studies of elder abuse are still all over the map, and estimates range from ½ of 1 percent all the way to 10 percent. It's tough to create social policy if you are not sure whether 200,000 or 4 million people need help. One reason that estimates vary so widely is measures. It is dangerous to have a single measure, but we are not sure what we are measuring.

We then got the director to award nine elder mistreatment grants on measurement at NIA, and another three measurement grants funded by NIJ are the basis of what will be covered this morning. The goals of the grants were to set the groundwork for national incidence and prevalence, but a barrier was the difficulty in defining elder mistreatment — financial, institutional, sexual, community-based — all have different risk factors. In addition, cultural diversity in the U.S. may defy our ability to come up with a single measure for elder abuse. So, let's not put all our eggs in one basket. There are multiple levels of data available to see if we can make reasonable estimates of elder mistreatment, and we may need ways other than surveys to obtain measures of incidence and prevalence of elder mistreatment in the U.S.

One objective is to seed the field with dynamic investigators. We need to get good people in other areas of research to join in the enterprise.

Panel Presentations:

Ron Acierno — Population-Based Assessment of Elder Mistreatment in Non-Institutionalized Adults


  • NIA Study: To demonstrate the feasibility of a method for epidemiological assessment of physical, sexual and emotional mistreatment of an elderly person who has a diminished capacity for self-care or self-protection perpetrated by someone in a position of trust.
  • NIJ Study: To conduct a national epidemiological study of elder mistreatment, defined generally as physical, sexual, emotional, neglectful or financial mistreatment, using a multi-method assessment strategy with 4,500 randomly selected households.

NIJ Study Design:

  • To implement a nationwide population-based study methodology validated by our prior research with younger populations and with a pilot sample of older adults.

  • To augment this methodology with an alternative method of assessment of caretakers for those noninstitutionally based elders with significant cognitive impairment.

  • To document prevalence and characteristics of other traumatic events that might interact with risk factors to increase or decrease the likelihood of experiencing mistreatment, including violent assault by strangers.

  • To include measures of financial mistreatment, as well as measures of relevant risk factors for financial mistreatment.

  • To use the knowledge gained about the scope, intensity and character of elder abuse, including risk and protective factors, to inform public policy and community-based interventions to reduce elder mistreatment and treat its negative effects.

Observations: An artificial distinction has been made between abuse and assault, which might be measured effectively using the same methodology. Cognitive status is conceptually important to the selection of mistreatment assessment methods. In considering models from other areas, the domestic violence model is more helpful and on point than the child abuse model. In order to avoid case identification failure, it is necessary to ask a certain type of question in a certain way: first a contextually orienting preface statement, then a behaviorally specific question. Feasibility data from a pilot study have shown: 1) explicit questions didn't really bother older participants, according to their own report; 2) physical health barriers (e.g., hearing, fatigue) were not insurmountable; 3) participants felt that they were helping others by completing the survey; and 4) the victimization rates did not differ according to interview context. These findings inform the current NIA and NIJ studies. As a final, ethical note, screening for dementia in the two studies is critical; if the interviewer has the slightest concern, the interview is terminated.

Conclusions: After field testing, the first 200 interviews indicate:

  • Only 14 calls lasted more than 25 minutes. Most of these were associated with a greater incidence of abuse.

  • The call duration did not appear to vary greatly between adult and proxy interviews.

  • Offering a $10 incentive did not appear to have any impact on response rates or call duration.

  • The data appear to be very complete for those who participated. There was not a great degree of item nonresponse. The majority of those contacted who did not participate were screened out due to the lack of an elder adult in the home. Of the 1,610 records contacted, 1,136 were screened out or ineligible. Most of the call break-offs occurred during the introduction. There were very few break-offs once the interview began.

Scott Beach — Testing Survey Methods for Collecting Data on the Prevalence of Elder Mistreatment


  • Test the feasibility, acceptability and cost implications of survey methodologies for collecting self-report data on elder mistreatment.

  • Test the reliability and validity of prevalence estimates derived from survey methodologies for collecting self-report data on elder mistreatment.

  • Collect data regarding elder abuse from in-home elder service providers in order to: a) test feasibility, b) provide prevalence estimates to compare with the elder self-report data, c) incorporate data from severely cognitively impaired elders, and d) capture more severe forms of elder mistreatment.

Study design: Randomized experiment varying the mode of data collection and presence and absence of privacy-enhancing technology. Total n = 903 (called 35,000 numbers to get 900 participants; sample included moderately cognitively impaired participants). Four methods tested:

  • Computer-assisted telephone interview (CATI), n = 228
  • CATI/interactive voice response (IVR) hybrid telephone, n = 227
  • Computer-assisted personal interviewing (CAPI) in person, n = 224
  • CAPI/audio computer-assisted self-interviewing (A-CASI), n = 224

Observations: Elderly people are not as apt to hang up on interviewers, but there was some reluctance to use headphones. The definitions of abuse (financial, neglect, sexual) were very important and will continue to be refined. Very low prevalence was found for neglect and sexual abuse (less than 1 percent since turning age 60). The in-home provider survey was less successful because of resistance/reluctance to collaboration by directors at the Area Agency on Aging (AAA) level.

Tentative Conclusions:

  • Direct victim surveys of elder abuse are feasible, although neglect and sexual abuse may require other approaches.

  • Older adults are generally able and willing to use new privacy-enhancing survey technologies (A-CASI and IVR).

  • A-CASI and IVR appear to increase reporting of psychological abuse.

  • A-CASI also seems to increase reporting of financial abuse. Results are less clear for physical abuse.

  • The feasibility and usefulness of in-home provider surveys is questionable.

Laura Mosqueda — Developmental Research on Elder Mistreatment Measurement

Goal: To develop a standardized measurement tool for evaluating elder abuse.

Study Design:

  • Phase I (complete) Define and operationalize five types of elder abuse (physical, psychological, sexual, financial and neglect) using a multidisciplinary advisory board.

  • Phase II (complete) Develop and field test the survey (Older Adults and Conflict Behaviors [OACB]) using focus groups (English and Spanish).

  • Phase III (ongoing) Recruit and enroll 250 participants ages 65+, either English- or Spanish-speaking, with cognitive impairment allowed when a caregiver is available. One-time home visit for interview, questionnaire and survey that is evaluated via LEAD Panel assessment (longitudinal, experts, all data).

Observations: Successful collaboration with APS with approximately 50 percent Mexican enrollment. Challenges include: 1) difficulty in recruiting participants with cognitive impairment and confirmed abuse, particularly in the Mexican-American population, and 2) getting LEAD Panel assessments that are timely and consistent.

Conclusions: None yet. Next steps include validating the survey/statistical analysis and disseminating results.

Kate Wilbur — Toward a Better Understanding of Elder Mistreatment in Community Settings


  • Enhance conceptual clarity through literature review and focus groups.
  • Develop and test an instrument to measure abuse in diverse communities.
  • Pilot the instrument using promotores as interviewers.

Study Design: Focus on elder abuse in diverse communities through door-to-door surveys using promotores (people who are knowledgeable about the culture, values and language of the population and can provide a culturally sensitive link and trusting relationship for those traditionally overlooked in research [e.g., immigrants, low-income and isolated individuals]).

Observations: The promotores model has been used in public health as a way to reach underserved populations. Highly diverse communities will be identified through the LA County Service Planning Areas (SPAs), and we will build on community structure by utilizing neighborhood councils to help recruit promotores. The challenges include inspiring trust so that participants are willing to answer the door, and getting the right definitions so that our participants will know what we are talking about.

Conclusions: None yet. The study began in September 2007. The key question is whether this approach will work as a means to identify elder abuse in diverse communities and include people who might usually be left out of traditional surveys.

Larry Branch — Florida Elder Abuse Survey

Goal: To construct a self-report questionnaire that can correctly identify cases of elder abuse or neglect by a caregiver or exploitation as confirmed by APS field workers (i.e., our criterion) compared to other elders receiving home care services, but not identified as victims of elder mistreatment by case managers.

Study Design: Focus was on community residents who can self-report.

  • Phase I – Shadowed Florida Elder Abuse Hotline to understand what information is used to designate a case as elder mistreatment.

  • Phase II – Shadowed APS workers as they made their decision whether elder mistreatment was present or not.

  • Phase III – Developed a questionnaire and conducted a survey with 26 confirmed cases of second-party elder abuse or neglect and 69 elder services clients who were not abused. Ten AAAs sent names of 168 clients to be interviewed, of those, 117 were contacted.

Observations: Age should not be used as a surrogate for vulnerability, and in considering the issue of “person of trust” we need to consider how long there has been a trust relationship — is the con man who promised house repairs a person of trust? Elder mistreatment is not equal to mistreatment of an elder (i.e., an elder who is a victim of crime). Elder mistreatment requires intentional actions that cause harm or a serious risk of harm. According to the National Research Council definition, an intentional act by a trusted advisor that caused unintended financial harm to a vulnerable elder (e.g., a financial advisor who intentionally sells a stock, resulting in a loss) meets the definition of abuse. We continue to argue that it does not.

Conclusions: The instrument was statistically significant with a sensitivity of 77 percent (identified 20 out of 26 known cases) and specificity of 42 percent (identified 44, 20 were confirmed, 24 were not). The instrument needs to be tightened to distinguish elder abuse from abuse of an elder. We still must define a set of risk factors so that we can get to a national prevalence rate.

Lori Jervis — Exploring the Mistreatment of Native Elders

Goal: To lay the groundwork for a pilot study of Native elder mistreatment to inform a set of recommendations for further examining the phenomenology and prevalence of Native elder mistreatment.

Study Design: Assemble a network of tribally diverse community experts with backgrounds in Native elder advocacy and abuse prevention/intervention and an interdisciplinary team of researchers to conduct developmental work on the phenomenology and prevalence of mistreatment among older American Indians/Alaska Natives.

Observations: The project has been under way for five months and the following themes have emerged: 1) elder abuse receives much less attention than child abuse or domestic violence in Indian country; 2) there is no centralized mechanism for collecting data on elder abuse in Indian country; 3) criminal justice in this area is inadequate; 4) there is a need to know about more than prevalence – starting with prevalence is premature; 5) we need better definitions of elder abuse, including considering the issue of coerced caring for grandchildren; 6) existing measures need revision; and 7) there is a need for open-ended questions to provide qualitative perspective from elders, and participants must be treated in a culturally relevant manner.

Conclusions: The collaborative approach will ensure that scientific approaches are grounded in local realities. Next steps include measure modification and development, selection of data sites, a pilot study and development of recommendations.

Date Created: August 11, 2008