The research team assembled a population-weighted sample of all residents of New York State aged 60 and older. Age 60 was selected as the cutoff for the study because many official service systems that serve elder abuse victims (e.g., Adult Protective Services) use the same age criterion, and in subsequent analyses the aim of the study was to compare self-reported and officially reported cases meaningfully.
The sample was created using a random digit dialing strategy derived from census tracts targeting the older than 60 demographic. To assure adequate representation of minority populations, Hispanics and African-American subjects were intentionally oversampled; similarly, the sample was augmented with older subjects (age greater than 80) to assure adequate representation of the oldest elders. Subsequent weighted analyses adjusted for this sampling strategy.
Eligible subjects were those that were (1) at least 60 years of age, (2) living in a community residence (not a long term care facility) in New York State, (3) English or Spanish-speaking (all instruments were translated into and conducted in Spanish when this was the primary language of the respondent) and (4) who had enough cognitive ability to participate in the full interview. With regard to the latter, the investigators considered several protocols to assess cognition ranging from brief screening to formal mental status testing. Formal mental status testing was deemed as either too onerous given the overall length of the interview or potentially off-putting to subjects given the already sensitive nature of the study topic. Ultimately, a simple cognitive screening procedure was employed consisting of three questions: marital status, date or year of birth and age.
If the subject was unable to provide answers to any of these questions, or if their reported age differed by more than one year from their age as computed from the birth date, they were deemed cognitively impaired and therefore not eligible to participate in the interview.
The actual surveys were conducted by telephone by the Cornell Survey Research Institute (CSRI). Located at the Cornell University campus in Ithaca, NY, CSRI has been providing survey research, data collection and analysis services since 1996 to a wide range of academic, non-profit, governmental and corporate clientele.
CSRI interviewers were highly diverse in age, gender and ethnic background; two of the investigators (Mark Lachs, Art Mason) conducted on-site training and education on the general topic of elder abuse with the survey staff prior to the initiation of the study. The training specifically included techniques and referral information on how to address potential situations in which a subject was deemed to be in immediate danger;
Dr. Lachs made himself available by phone and pager through the self-reported study to assist interviewers and or subjects with any urgent clinical issue that might arise. Fortunately, no such situation occurred during the telephone interviews.
Instruments
The survey instrument used for this component of the study was designed to capture elder mistreatment in four general domains: (1) Neglect by a responsible caretaker in meeting ADL (Activities of Daily Living) and/or IADL (Instrumental Activities of Daily Living) needs, (2) Financial Exploitation, (3) Emotional (Psychological or Verbal) Abuse and (4) Physical Elder Abuse (including Sexual Abuse).
Wherever possible, existing instruments were employed to measure mistreatment so as to permit comparisons to previous studies; this was a formal recommendation of the National Academy of Sciences Panel on Elder Abuse, which suggested use of instruments like the Conflict Tactics Scale to measure physical abuse (Bonnie, R.J., Wallace, R.B., 2002). However, that panel also suggested modification and adaptation of instruments for elder abuse, and the study investigative group (who collectively has many decades of experience in direct elder abuse service, administration, research and policy) found serious shortcomings in some aspects of virtually all instruments. For example, many older financial exploitation instruments fail to take into account many newer forms of “technologically-mediated” financial abuse, such as theft or misuse of an ATM card.
The Conflict Tactics Scale (and its actual preamble) assert that family violence occurs when families resolve stressful situations dysfunctionally but many cases of serious elder abuse have no such predicate (for example, a demented man who strikes a spouse with no provocation whatsoever). Additionally, although the intent of the study was to exclude stranger-mediated violence and crime, the study partners wished to capture abuse, neglect or exploitation conducted by non-family actors (such as a paid home health attendant) as situations such as this constitute a violation of an important trusting relationship and are highly relevant to victims, responders and policy makers. Accordingly, modest modifications were made to existing instruments as described below for each domain.
Additionally, following the precedent of Pillemer and Finkelhor in their adaptation of the Conflict Tactics Scale (Pillemer and Finkelhor, 1988), both a frequency and severity rating were assigned to all items. For frequency, subjects were asked if since turning age 60 they had ever experienced mistreatment; if they endorsed the mistreatment item, they were then asked how many times they had experienced the event in the past year.
The possible response categories were “never, once, two to ten times, or greater than ten times.” This strategy permitted a calculation of both incident events (those occurring in the past year) versus prevalence (those occurring at any time since age 60). When a subject endorsed a form of mistreatment as having occurred at least
once in the past year, he or she was also asked how serious a problem the event(s) was for them; the ordinal choices for this were: “not serious, somewhat serious, very serious.”
Frequency and severity ratings of an item were then combined to determine whether abuse was deemed to be “present” or “absent.” However, the criteria for individual items necessarily varied by item. For example, neglect of an ADL by a responsible caregiver on a single occasion in the course of a year (e.g., failure to provide assistance with dressing) would not be considered abuse by many, whereas kicking or hitting would be, irrespective of how many times the event occurred or how severe the victims perceived the assault.
To resolve this and establish frequency and severity criteria for each variable, the research partners conducted a consensus meeting and discussed each variable individually; there was high agreement on all items after robust discussion. The final frequency and severity criteria were then vetted with several external experts with similar experience in elder abuse. The following is a summary of the domains, variable items and criteria for positivity.
A detailed version of the final questionnaire can be found in Appendix A.
Neglect of IADLs and ADLs Domain. Subjects were asked if they needed assistance with any of four higher functional IADL-type activities: shopping, meal preparation, basic housework, or taking medicine, and six more basic ADL-type activities: eating, bathing, dressing, toileting, transferring or walking. When individuals said assistance was required, they were asked (1) who was responsible for providing assistance with the activity and what their relationship was with the respondent, (2) whether they had failed to provide assistance with the activity in the past year and how many times and (3) how serious a problem it was when and if the care failure occurred. An ADL or IADL was deemed to be neglected when the neglect occurred more than ten times in the previous year, and/or the subject described the neglected care need as being “somewhat” or “very serious”
for them.
Financial Exploitation Domain. Several financial exploitation instruments were reviewed; the research team found instruments used for a Canadian study and a study in the UK to be the most salient for the purposes of our study (Podnieks, 1992; Manthorpe et al., 2007). Some items from the existing instruments were adopted;
the team also drew on the collective practice experience and expertise of the research study partners in financial exploitation to update the instrument and to add items. The scale was reviewed by member agencies of the New York City Elder Abuse Network for review and to ensure that all domains were being captured.
Ultimately, the research partners arrived at five financial exploitation items, three of which were designated
“high severity items” and two of which were designated “modest severity.” The high severity items were:
(1) whether or not someone stole or misappropriated money or property without permission, (2) whether someone forced you or misled the subject into surrendering rights or property or coerced them into signing or changing a legal document (such as a will, deed or power of attorney) and (3) whether someone impersonated the subject to obtain property or services. The modest severity items were (1) whether someone who had agreed to contribute to household finances had stopped doing so, and (2) whether someone who had been contributing to household finances had stopped doing so, even though they had the ability to continue, such that there was no money left at the end of the month for necessities such as food or rent.
The high severity items were deemed as “positive” when endorsed with any frequency, irrespective of the severity ranking. The modest severity items were deemed positive when endorsed as occurring greater than ten times annually and rated as very serious by the subject.
Elder Mistreatment Domain (Emotional, Physical and Sexual Abuse). The version of the Conflict Tactics Scale (CTS) as modified by Pillemer (Pillemer and Finkelhor, 1988) was adopted for use in the study, using a frequency and importance ranking for each item. Two of the CTS items were assigned to what is termed in the instrument as a “verbal” (emotional/psycholological) abuse domain (when another party either (a) insulted or swore at the respondent, or (b) threatened to hit or throw something at the respondent) while the rest were attributed to a “physical abuse domain;” the prevalence of mistreatment within those two domains is reported separately. The physical abuse items were deemed as “positive” when endorsed with any frequency, irrespective of the severity ranking. The first emotional abuse item (insulted or sworn at the respondent) was deemed
“positive” when endorsed as occurring greater than ten times annually and rated as very serious by the subject, and the second emotional abuse item (threatening to throw something at the respondent) was deemed
“positive” when occurring with any frequency or seriousness.
The CTS and its preamble assert that conflict occurs in all families, and that it is the way families choose to resolve those conflicts that potentially results in family violence. Because the objective of the study was to capture elder mistreatment perpetrated by any individual in a trusting relationship with the subject (e.g., paid home attendants, neighbors) in addition to family members, the preamble of the instrument was modified to include such individuals. In addition to specific categories of mistreatment, information was collected on the individuals alleged to have been the perpetrators in elder mistreatment episodes as described by the subjects.
Our coding strategy permitted the classification of up to two perpetrators in mistreatment episodes.
Finally, an open-ended query was included at the completion of the CTS in which subjects were asked if they had been victims of elder mistreatment in any manner that was not ascertained in the existing instrument, so as not to miss cases of elder abuse not included in the predetermined categories chosen for the study.
Subjects were also encouraged to provide qualitative detail about any and all episodes of mistreatment; these were recorded in open text fields.
The survey instrument was pilot tested with ten elder abuse victims served by the New York City Department for the Aging. The ten known victims were administered the survey in a telephone interview by the Cornell Survey Research Institute interview staff employed for the subsequent study. All but one individual endorsed elder mistreatment by at least one item in the instrument. That individual has modestly advanced cognitive impairment.
The study also captured data elements regarding individual subjects that were potential covariates of elder mistreatment. (Variables that may be predictive of risk for elder abuse.) The survey protocol collected basic demographic information about age, gender, marital status, household composition (number of individuals in household and relationship to subject) and income. Clinical variables included ADL and IADL status as well as self-reported health status. Cognition was assessed through the brief screening procedure described above.
The research instrument and the protocol for the Self-Reported Prevalence component of the study were approved by Cornell University’s Institutional Review Board.