Risk Factors for Elder Mistreatment
Why do family members (or others in a trust relationship) mistreat elderly persons? What factors place older persons at risk? These are critically important questions, but finding answers poses a number of difficult challenges for researchers. Some of the difficulties are methodological: obtaining information on a hidden—and for most people, shameful—phenomenon is a daunting task at best. The search for risk factors has also been clouded by a 20-year history of elder mistreatment as a social problem: early assertions, founded on faulty data (or no data at all), have been frequently repeated and widely believed, despite the lack of evidence.
Thus, although fairly extensive research on risk factors for child abuse and intimate partner abuse has been conducted, the risk factor literature on elder mistreatment is both limited and inconsistent. It is important to remedy this situation, for several reasons. First, an understanding of associated factors and antecedents of elder mistreatment is necessary for the development of screening methods. Victims of elder mistreatment infrequently seek help for the problem on their own; therefore, by the time the case has progressed to the point at which it is detected by a service agency, it is often very complex and difficult to treat. Effective screening could result in a reduction of the negative effects of elder mistreatment and reduce the need for extensive treatment.
Second, specification of risk factors for elder mistreatment is needed to provide a rational basis for prevention programs. If the risk factors for elder mistreatment can be uncovered, we may be able to reduce or eliminate
those factors and thus prevent the development of new cases of elder mistreatment or deter the progression of existing cases. Third, understanding risk factors is critical to the development of public policy initiatives. It is necessary to identify populations at higher risk, and the causes of that heightened risk, before the costs and benefits of reducing exposure can be determined (Gordis, 1996).
A note on terminology will be useful at the outset. For the purposes of this chapter, following Timmreck (1998), risk factors are defined as experiences, behaviors, aspects of lifestyle or environment, or personal characteristics that increase the chances that elder mistreatment will occur. Increased risk factor exposure increases the probability of the occurrence of elder mistreatment. As noted in Chapter 2, a distinction can be made between risk factors (factors that increase the probability that a problem will occur) and protective factors (factors that decrease the probability of occurrence). To simplify the discussion, in the rest of the chapter we refer to risk factors only, in part because most published work involves variables associated with an increased probability of mistreatment. The discussion of risk factors, however, may also hold for protective factors. In fact, research on protective factors may be as important as study of factors that increase risk, since it may suggest factors that can be put in place as a means of preventing elder mistreatment.
PROBLEMS IN THE RESEARCH BASE
Prior to summarizing the available findings, it is important to review briefly the problems in using existing research to establish risk factors for elder mistreatment. Problems exist in two areas: (1) the nature of the phenomenon of elder mistreatment itself creates challenges for risk factor research and (2) specific methodological limitations of existing studies limit the ability to integrate findings.
Nature of Elder Mistreatment
With some diseases or conditions, attribution of cause can be fairly simple and straightforward; a salmonella outbreak serves as an example. Other conditions have very complex causation, and indeed the condition itself may be difficult to define and identify. Elder mistreatment clearly fits the latter pattern.
The complexity of elder mistreatment can be highlighted by reference to the concept of a “geriatric syndrome”—that is, common clinical problems that typically do not have a single underlying pathophysiological process, but instead have several contributing factors that shape presentation
(Lachs and Pillemer, 1995). Examples of geriatric syndromes include falls, urinary incontinence, and functional decline.
Geriatric syndromes share several characteristics: environmental factors play an important role; interventions must be multifaceted and directed at both specific pathophysiological problems as well as at contributing factors in the environment; and such syndromes are often underdiagnosed and undermanaged by health and social service providers. Elder mistreatment shares these characteristics of a geriatric syndrome. Most important for the purposes of this chapter, contributing etiologies can be related to the relative (or person in a trust relationship), to the elder, or to the environment. Thus, the search for risk factors is both complex and challenging and necessarily must look for sources of risk in the host (the elderly person), the agent (the perpetrator), and the environment. As well, it must study the interplay of factors in these three domains in affecting the risk of elder mistreatment.
Weaknesses of Existing Studies
The first major limitation of previous risk factor research results from unclear definition of the object of study. Findings from most studies are confused in that they do not differentiate the various types of abuse and neglect articulated earlier in this report. It is likely that the etiology of these elder mistreatment types differs. Second, different criteria have been used to determine the population at risk of elder mistreatment. Some researchers have included persons under age 60 in their studies, while most others have chosen 60 or 65 as the entry point. Some researchers have restricted their studies to caregivers of elderly persons or to persons sharing a residence, while others have included all categories of elderly people.
Third, studies of risk factors have employed widely differing sampling methods, including random sample surveys, interviews with patients in medical practices or caregivers in support programs, and reviews of agency records. Fourth, few studies that have purported to address risk factors have in fact included control groups in their designs. In the absence of controls, the validity of associations between elder mistreatment and putative risk factors cannot be assessed. Furthermore, even those studies that have included control groups have often failed to ascertain that the controls were actually free of elder mistreatment. Fifth, a number of studies have not employed reliable and valid measurement of the indicators of risk.
Sixth, with one exception (Lachs et al., 1994, 1997a), prospective studies of elder mistreatment do not exist. As Lachs and colleagues (1994) point out, retrospective research designs contain several potential biases: recall bias—the respondent reinterpreting key facts or feelings from a later vantage point; information bias—the respondent (especially if cognitively
impaired) may not be able to recall or provide valid information about exposure to maltreatment; and the failure of studies to take into account the timing and duration of events and their progression over time.
For these reasons, a clear framework of known risk factors for elder mistreatment cannot be derived from previous research on elder abuse. Despite a large number of review articles over the past two decades, it must be acknowledged that any statements about relative risk among the elderly should be viewed with caution.
However, the small number of studies using acceptable research designs do reveal some associations that are of interest. In this chapter, findings from these studies are summarized. For the purposes of the discussion, an attempt has been made to focus on studies that meet two criteria. First, priority was given to studies that involve a comparison group of some kind. In such studies, elderly victims (or perpetrators) have been compared to nonabuse cases uncovered in a survey or to a comparison group of some kind. Because the literature is so sparse, however, in a few cases, studies are referred to that have an “implied” comparison group. Second, the study must have collected information directly from victims and perpetrators and not from agency records (for problems with using agency records in elder mistreatment research, see Chapter 2).
FRAMEWORK FOR ELDER MISTREATMENT RISK FACTORS
The theoretical model for understanding the risk factors for elder mistreatment presented in Chapter 3 includes both the microprocess of generation of elder mistreatment risk, involving the individual and the trusted other(s), as well as the environing sociocultural context in which victims and perpetrators are embedded (such as living environment and social and economic characteristics). This model indicates the wide array of variables that could be included in risk factor research.
To date, a small number of this wide range of potential risk factors has been addressed in research. These factors, for each of which there is at least one study, fall into the following categories of the framework depicted in Figure 3-2:
Social Embeddedness/Context (subject): social isolation.
Social Embeddedness/Context (trusted other): social isolation.
Individual Level Factors (subject): gender, race, dementia, physical health status, personality characteristics.
Individual Level Factors (trusted other): mental illness, hostility, alcohol abuse, experience of violence or aggression in childhood.
Relationship Type: Shared living arrangement, relationship to victim (spouse or child).
Power and Exchange Dynamics. Abuser dependency, victim dependency/caregiver stress.
For the purposes of this chapter, we review these risk factors based on the supporting evidence. It is possible to categorize risk factors for elder abuse into three general groups:
Risk factors validated by substantial evidence, for which there is unanimous or near-unanimous support from a number of studies.
Possible risk factors, for which the evidence is mixed or limited.
Contested risk factors, for which potential for increased risk has been hypothesized, but for which there is a lack of evidence.
As the earlier discussion makes clear, however, these categories are only loosely constructed. All findings should be taken with caution, due to methodological shortcomings in the studies. Furthermore, the evidence is generally too limited to make clear distinctions among abuse types. When such information is available and relevant, it is mentioned below.
RISK FACTORS VALIDATED BY SUBSTANTIAL EVIDENCE
Both clinical accounts and limited empirical research suggest that a shared living situation is a major risk factor for elder mistreatment, with older persons living alone at lowest risk (Pillemer and Finkelhor, 1988). Paveza et al. (1992) found that risk of mistreatment of Alzheimer’s disease patients by caregivers was greatest when the patient resided with immediate family members (other than the spouse). Lachs et al. (1997a) found living alone to be an important protective factor against mistreatment. Pillemer and Suitor (1992) also found a shared living arrangement to be a risk factor for violence by Alzheimer’s disease caregivers.
The mechanisms for the effect of living arrangement are straightforward. A shared residence increases the opportunities for contact, and thus conflict and mistreatment. Furthermore, tensions that may be relieved by simply leaving the immediate situation can escalate into mistreatment (see Wolf and Pillemer, 1989). Exploration of the differential role of living arrangement according to type of elder mistreatment needs to be conducted. For example, neglect (as the panel has defined it) by its very nature suggests a shared living situation, but financial exploitation may occur even when abuser and victim live apart.
Social isolation has been found to be characteristic of families in which other forms of domestic violence occur. This is in part because behaviors that are considered to be illegitimate tend to be hidden. Detection of abusive actions can result in informal sanctions from friends, kin, and neighbors and formal sanctions from police and the courts. Thus, elder mistreatment is hypothesized to be less likely in families embedded in strong social networks.
Research provides support for this view. In the Lachs et al. (1994) prospective, community-based study of risk factors for elder abuse, having a “poor social network” significantly increased risk of mistreatment. Compton et al. (1997) found low levels of social support to be associated with verbal and physical abuse by caregivers, as did Wolf and Pillemer (1989). Grafstrom et al. (1993) found both caregivers and care recipients to be more socially isolated in families in which abuse occurred. The case comparison by Phillips (1983) also found abused elder persons to be more socially isolated.
There are two types of evidence that implicate Alzheimer’s disease or related dementia as a risk factor for the mistreatment of elderly persons. First, several studies have estimated prevalence rates of elder mistreatment in samples of dementia caregivers; these rates can then be compared with rates in general population surveys. Coyne et al. (1993) found that 11.9 percent of the dementia caregivers in their sample reported having committed physical abuse. Paveza et al. (1992) found a rate of severe physical violence toward care recipients of 5.4 percent, which is close to Pillemer and Suitor’s (1992) finding of 5 percent in a similar sample. Homer and Gilleard (1990) found physical abuse occurring in 14 percent of caregivers to Alzheimer’s disease patients in a respite care program. Given the prevalence findings of rates of physical abuse in the 1-3 percent range in the general population, dementia patients would appear to be at greater risk of such mistreatment.
Second, a few studies have contrasted abusive and nonabusive caregivers, examining dementia in the victim as one among a number of risk factors. The results are contradictory. Lachs et al. (1994) did not find cognitive impairment to be a risk factor, and Reis and Nahmiash (1998) did not find dementia to discriminate between probable abuse and nonabuse cases. However, Lachs et al. (1997a) found that dementia predicted identification as an abuse victim.
One explanation for this contradictory set of findings comes from
Pillemer and Suitor’s (1992) finding that Alzheimer’s caregiver violence is strongly related to experience of violence from the care recipient, and Compton et al.’s (1997) finding that behavior problems are related to both verbal and physical abuse. It may be that dementia itself is not the risk factor, but rather disruptive behaviors that result from dementia. Such an explanation would be consistent with research that has shown disruptive behaviors by Alzheimer’s disease patients to be an especially strong cause of caregiver stress. Future research of the relationship between dementia and elder mistreatment should differentiate the cognitive, functional, and behavioral effects of dementia and examine the independent association between each and the risk for elder mistreatment.
Intraindividual Characteristics of Abusers
Intraindividual theories of mistreatment locate the causes of abuse in some pathological characteristic of the abuser, usually mental illness, personality characteristics, or alcohol or drug abuse. This approach has a lengthy history in the study of child and intimate partner abuse, including a long-standing debate over the role of intraindividual factors as risk factors for the forms of mistreatment. In the field of elder mistreatment, there is compelling evidence that certain characteristics of perpetrators constitute major risk factors for elder mistreatment, with surprising unanimity on this issue among studies using different methods.
Wolf and Pillemer (1989) found that 38 percent of abusers in three related samples had a history of mental illness and 39 percent had alcohol problems. Reis and Nahmiash (1998) attempted to validate a screening tool using a sample of 341 agency cases in which caregivers could be interviewed. The cases were classified as “likely” or “not likely” to involve abuse of the care recipient. They found that the caregivers’ mental health and behavior problems were strong predictors of likely abuse. Pillemer and Finkelhor (1989) found in a case-comparison study that abusers were substantially more likely to have experienced psychiatric hospitalization than nonabusers.
These studies did not differentiate particular forms of mental illness. Several studies have specifically pointed to depression as characteristic of perpetrators of elder mistreatment. Paveza et al. (1992), in their study of Alzheimer’s caregivers, found that caregiver depression predicted physical abuse. Coyne et al. (1993) compared physically abusive and nonabusive caregivers who called into a telephone helpline for family members of Alzheimer’s disease patients; abusive caregivers were more depressed.
Homer and Gilleard (1990) found that among caregivers referred to a respite service, abusive ones scored higher on a depression scale. In a recent study, Williamson and Shaffer (2001) conducted structured interviews with 142 spousal caregivers regarding “potentially harmful behaviors”; these 10 items included verbal aggression, threats, and physical violence. Multivariate analyses found that more depressed caregivers were also more likely to treat their dependent spouses in potentially abusive ways. This finding is also supported by Fulmer and Gurland (1996). Of course, it is possible that depressed individuals may be more likely to report their own behavior as abusive. All of these studies were retrospective; prospective research will be needed to establish the causal direction.
In the only study to distinguish between abuse types, Reay and Browne (2001) found that physical abusers scored significantly higher on a depression scale than did perpetrators of neglect; thus, there may be a difference by type of mistreatment, which needs to be assessed in future research.
A study of a sample of Alzheimer’s disease caregivers found that abusive caregivers (a category that combined “emotional and/or physical abuse”) scored higher on a hostility scale (Quayhagen et al., 1997).
Several studies of elder mistreatment suggest that alcohol abuse on the part of perpetrators was relatively common. For example, Greenberg and colleagues (1990) reviewed 204 substantiated cases of elder abuse; 44 percent were identified as having alcohol or drug abuse problems. Case-control studies have supported this assertion, finding that elder mistreaters were disproportionately more likely to be identified as having an alcohol use problem (Bristowe and Collins; 1989; Homer and Gilleard, 1990; Wolf and Pillemer, 1989).
In a study funded by the National Institute on Aging that directly addressed this issue, Anetzberger et al. (1994) compared a group of 23 adult children identified by agencies as perpetrators of domestic violence against an elderly parent with a group of 39 nonviolent caregiving children. Both alcohol use and abuse were more common among the perpetrators; for example, daily alcohol consumption was more than twice as likely among perpetrators.
It is possible that the role of alcohol abuse may differ by abuse type. Reay and Browne (2001) found that alcohol abuse by the caregiver (consumption of over 21 units of alcohol per week) occurred in seven out of nine of the physical abuse cases, but only one of the neglect cases.
Related to the previous risk factor, findings from early research on elder mistreatment suggests that perpetrators tended to be dependent on the individual they were mistreating. In 1982, Wolf and colleagues surveyed community agencies in Massachusetts regarding elder abuse cases they had encountered (Wolf et al., 1982). The authors identified a “web of mutual dependency” between abuser and abused. In two-thirds of the cases in that study, the perpetrator was reported to be financially dependent on the victim. Another early study by Hwalek et al. (1984) also reported that financial dependence on a relative was a risk factor in abuse. Other studies, without control groups, have found substantial percentages of financially dependent abusers (Anetzberger, 1987; Greenberg et al., 1990).
A number of studies have confirmed this finding. Pillemer (1986; Wolf and Pillemer, 1989) found that abusers were substantially more dependent on the victim for housing and financial assistance than were members of a comparison group. In Pillemer and Finkelhor’s (1989) analysis of cases from a random-sample survey, nearly identical results emerged.
POSSIBLE RISK FACTORS
Adult protective services reports and other studies of agency samples universally find that the majority of victims are female (Wolf, 1997b). However, it is not clear whether this is due to higher risk for victimization or to women’s greater numbers in the population of seniors. Pillemer and Finkelhor’s (1988) survey suggested that the latter may be the case; in their study, they found that the victimization rate for men was higher at 5.1 percent, compared with 2.5 percent for women. They attributed this in part to the fact that elderly women are much more likely to live alone, which reduces their risk. Furthermore, their sample included intimate partner abuse among the well elderly, in which the victim would not necessarily have been classified as vulnerable according to our definition.
Pillemer and Finkelhor (1988) also noted the important caveat that women tended to sustain more serious abuse and to suffer greater physical and emotional harm from mistreatment. This may in turn explain their greater representation as victims in adult protective services caseloads.
Relationship of Victim to Perpetrator
Despite suggestions that adult children are the most likely perpetrators of elder abuse, the only survey-based study of this topic found that spouses
were more likely to be abusers (Pillemer and Finkelhor, 1988). However, there are insufficient data on this risk factor to make a determination.
Personality Characteristics of Victims
Comijs et al. (1998) found that certain personality traits of elderly persons increased their risk of being an abuse victim. In a community survey conducted in the Netherlands, they examined whether hostility and coping style were related to being a victim of chronic verbal aggression, physical aggression, and financial mistreatment. Victims of chronic verbal aggression scored lower on a locus of control scale than did the nonabused members of the sample and higher on one indicator of hostility. Victims of all three abuse types showed higher levels of aggression as measured by the hostility scales and were generally more likely to use passive and avoidant ways of coping, rather than active problem-solving strategies. Because of the cross-sectional nature of this study, it is impossible to determine whether these characteristics are indeed risk factors, or whether they are consequences of the abuse. However, the findings are sufficiently suggestive to merit further exploration of personality factors in longitudinal studies.
Lachs and colleagues (1994, 1997a) found that being black was a risk factor for reported elder mistreatment. However, they noted that this may be an artifact of the definition of elder mistreatment, which was “being reported to an [adult protective services] agency.” No other study has found significant differences in elder abuse risk based on race.
CONTESTED RISK FACTORS
Physical Impairment of the Older Person
The role of victim health and functional status as a risk factor for elder abuse is a complex one. For the purposes of this report, some degree of physical vulnerability is considered to be a necessary component of the definition of elder mistreatment. That is, mistreatment necessarily implies a weaker individual who is mistreated by a stronger one. Greater impairment diminishes the individual’s ability to defend himself or herself or to escape the situation. It therefore is reasonable to consider physical health problems as a predisposing factor for elder mistreatment, which increases the likelihood of abuse in the presence of other risk factors.
However, research has generally failed to find support for the view that frailty of elderly persons is in itself a risk factor for elder mistreatment.
That is, case-control studies have not found a direct relationship between elder mistreatment and functional impairment or poor health. Reis and Nahmiash (1998) did not find impairment in activities of daily living to be associated with elder abuse. Neither Cooney and Mortimer (1995), Paveza et al. (1992), Bristowe and Collins (1989), nor Phillips (1983) in case-comparison studies found functional impairment to be a risk factor for abuse by caregivers. Lachs et al. (1997a) found that impairment in activities of daily living was associated with being an abuse victim, but the researchers acknowledged that the dependent variable—protective services intervention for elder abuse—may have led to these results, and that findings may differ for elder mistreatment that is not detected by an agency.
There is as yet no evidence as to whether this pattern of nonfindings holds for all types of mistreatment. In the only study to address this issue Wolf and Pillemer (1989) found that victims of elder neglect were more likely to be impaired than victims of either physical or psychological abuse.
Victim Dependence and Caregiver Stress
If there can be said to be a “traditional” view in the field of elder mistreatment, then it can be summed up in the following way. Elderly people become frail, difficult to care for, and sometimes demanding. These characteristics cause stress for their caregivers; as a result of this stress, the caregivers become abusive or neglectful toward the elder. In this view, elder mistreatment is seen as an outgrowth of the aging process, which leads to the need for care by others. Much early writing emphasized the dependence of the elderly person and resulting caregiver stress as the pre-dominant (and sometimes sole) cause of elder abuse (Davidson, 1979; Hickey and Douglass, 1981; Steinmetz, 1988).
However, there is a lack of evidence that an older person’s need for assistance or that caregiver stress in fact lead to greater risk for elder mistreatment. First, it is clear from the gerontological and geriatric literature that a substantial number of elderly persons are dependent on relatives for some degree of care. However, findings about the prevalence of elder mistreatment indicate that only a small minority of the elderly is mistreated. Since abuse occurs in only a small proportion of families, no direct correlation can be assumed between the dependence of an elderly person and abuse, as sometimes has been done.
Second, case-comparison studies have generally failed to find either higher rates of elder dependence or greater caregiver stress in elder abuse situations. Bristowe and Collins (1989), Homer and Gilleard (1990), Phillips (1983), Pillemer (1985), Wolf and Pillemer (1989), Pillemer and Finkelhor (1989), Pillemer and Suitor (1992), and Reis and Nahmiash (1997) did not find greater dependence or caregiver stress among victims
and their family members, when compared with nonvictims. One exception, a study by Coyne et al. (1993), found that callers to a help line who had committed abuse had been providing care for a longer time and for more hours a day than nonabusers and had higher burden scores.
Social learning theory gives rise to the hypothesis that when individuals experience violent behavior from parents or other role models in childhood, they tend to revert to these learned behaviors when provoked as adults. Indeed, it has by now become a commonplace that victims of child abuse may grow up to themselves become child abusers, a pattern often described as the “cycle of violence.” The cumulative research evidence supports this hypothesis, with experiencing violence from parents or witnessing violence between parents in childhood strongly related to perpetrating child or intimate partner abuse (See Newberger, 1998; Stark and Flitcraft, 1998).
Despite this evidence from other fields, the only two studies that have addressed this issue (Anetzberger et al., 1994; Wolf and Pillemer, 1989) found no evidence of intergenerational transmission of physical violence against elderly relatives. This issue, however, is worthy of further study, given the importance of childhood experience of aggression as a risk factor for other forms of interpersonal violence. The importance of early childhood experiences of perpetrators as risk factors for types of elder mistreatment other than physical violence should be explored. In addition, given that in the elder mistreatment field the victim and perpetrator have been in a long-standing personal relationship, as spouses or as parent and child, it may be more important to assess the type of relationship between the abuser and the victim as a risk factor for elder mistreatment.
ELDER MISTREATMENT IN INSTITUTIONAL SETTINGS
Despite the likelihood that elder mistreatment in nursing homes is equally or more prevalent than abuse in domestic settings, only one study has been conducted that specifically addressed risk factors. Pillemer and Bachman-Prehn (1991) analyzed data from a survey of staff regarding self-reported psychological and physical abuse. Predictors of psychological abuse were staff burnout, experiencing physical aggression from residents, negative attitudes toward residents, and age of the staff member, with younger staff more likely to engage in psychological abuse. Risk factors for physical abuse were again staff burnout and resident aggression, as well as the reported amount of conflict with residents. This study is limited by the self-report method used to assess the occurrence of elder mistreatment. Self-report may be subject to bias, especially since the staff would often
have to report themselves or their colleagues as abusers, which may well have affected ascertainment of occurrence of elder mistreatment.
A number of other potential risk factors can be derived from the more general literature on quality of care in nursing homes. Pillemer (1988) proposed four sets of variables that may be related to maltreatment: exogenous factors (including the availability of nursing home beds and the unemployment rate in an area); characteristics of the nursing home environment (such as size, reimbursement rates, ownership status, staff-resident ratio, and turnover rate); staff characteristics (including age, gender, education level, and burnout), and resident characteristics (health and functional status, social isolation, and gender). A full-scale test of this model remains to be conducted.
In her background paper for this panel, Hawes suggests risk factors derived from surveys of stakeholders in long-term care. She proposes three risk factors from these studies: stressful working conditions, particularly resulting from staff shortages; staff burnout; and the joint effects of resident aggression and poor training of staff in management of challenging behaviors. Pillemer (2001) combined insights derived from long-term care practice with the limited data on nursing home mistreatment to suggest four key factors: poor hiring and staff screening practices; chronic staffing problems; lack of administrative and supervisory oversight; and inadequate training. Taken together, these approaches suggest a number of avenues for studies of risk factors, at both the structural and the individual levels.
CONCLUSIONS AND RECOMMENDATIONS
Although risk factors at times are causes of mistreatment, this is not always the case. Some risk factors (preferably called “risk indicators”) may be “markers” for unmeasured/unobserved causes (confounders); or risk factors may modify the relationship between causal factors and elder mistreatment (effect modifiers). For example, depression in a caregiver may be a causal risk factor in that a depressed caregiver may be more likely to neglect the care of an elder by virtue of the fatigue, social withdrawal, and uninterest associated with depression. Living with others has been associated with an increased probability of mistreatment. However, this may not be a direct causal relationship, because living with others is a contextual factor in which mistreatment is more likely to occur; it would be possible to reduce the risk of mistreatment by modifying other factors associated with living with others and not changing the living circumstances of the older person (which is often difficult and disruptive). To provide another example, frailty—a form of vulnerability—may be an effect modifier, such as at very high levels of frailty the probability of mistreatment may be much
higher than at lower levels of frailty. Further study of issues such as these is absolutely critical to a research agenda on elder mistreatment.
A research agenda for risk factor research on elder mistreatment is to some degree straightforward, because it parallels our general recommendations for research on this topic. Studies using larger and more representative samples, as well as scientifically accepted epidemiological techniques, must be conducted before risk factors can be more accurately specified. The importance of case-control designs and cohort studies cannot be over-emphasized. We do not reiterate all of these recommendations here.
The following are specific additional recommendations to advance knowledge of risk factors for elder mistreatment.
Studies are urgently needed that examine risk and protective factors for different types of elder mistreatment. Studies are needed to advance understanding of what places older and vulnerable adults at risk for mistreatment and what places persons at risk for becoming abusive. These studies can be carried out using well-established methods for determining risk factors, including epidemiological case-control studies. These studies must, however, use common measures that allow for comparison across studies. Moreover, they must focus not only on the composite concept of elder mistreatment, but also on its various forms.
Intensification of epidemiological research to establish risk factors will be facilitated by greater collaboration between researchers and the adult protective services and elder services systems. Researchers in the field have generally been hampered in their efforts to establish risk factors because they have often needed to find both mistreatment cases and “controls” in general population samples. Most retrospective epidemiological research has used readily available case populations for studies, while attempting to sample well-defined control groups. Many of the advances in understanding risk factors associated with child abuse and intimate partner violence occurred as the result of participation and engagement between the research communities and the service provider agencies. In the panel’s view, advances in risk-factor research in elder mistreatment will require cooperation between adult protective services agencies and the research community, such as exhibited in the work of Dyer and her colleagues in Texas. By using persons clearly identified by some external source as victims of mistreatment, the focus can shift from concern about sample size to the identification of an appropriate group of controls.
Such studies have the potential for providing vital information for both policy makers and program developers to help define target behaviors for intervention. In addition, this information can be used to develop profiles of persons at risk for being mistreated, as well as to develop forensic mark-
ers. However, studies to identify protective factors from elder mistreatment should not be neglected.
A particularly critical need exists for studies of risk indicators and risk and protective factors for elder mistreatment in institutional settings. The available evidence reviewed by Hawes (this volume), combined with extensive professional and public concern about serious quality problems in long-term care (Institute of Medicine, 1996), suggests that a vast reservoir of undetected and unreported elder mistreatment in nursing homes may exist. Because nursing home residents as a class are both extremely physically vulnerable and generally unable either to protect themselves or report elder mistreatment they experience, the physical and emotional costs of elder mistreatment in such environments are likely to be very high. Prevention programs exist (see Pillemer and Hudson, 1993), but they have not been informed by rigorous risk factor research. Understanding the causes of mistreatment of this extremely fragile population is of the highest priority.
Research on risk and protective factors should be expanded to take into consideration the clinical course of elder mistreatment. Although longitudinal data are absent, it seems probable that elder abuse situations may follow a pattern similar to disease progression, which would include lead time prior to the manifestation of active signs and symptoms of elder mistreatment; periods of “remission” from elder mistreatment; and critical points in which elder mistreatment becomes more intensive or acute. Some have speculated that elder mistreatment typically increases in severity and intensity over time (Breckman and Adelman, 1988), but no empirical data exist that demonstrate this pattern or individual differences in progression. Clinical accounts suggest that elder mistreatment situations include cases that resolve on their own; cases in which mistreatment intensifies; and cases in which the situation remains abusive but stable. It is therefore both possible and important to identify risk factors for an increase or intensification in elder mistreatment.
For these reasons, cohort studies are of great importance in determining risk factors for elder mistreatment. Although prospective cohort studies would be ideal, the lengthy period needed for cases of elder mistreatment to develop is a deterrent. In the near term, retrospective cohort, or nested case-control studies using established study populations may be preferable, in which a preexisting data set is used and elder mistreatment measured at a later point (the technique used by Lachs et al., 1994, 1997a). There are a number of existing datasets involving elderly persons that could be used for such a purpose (for example, existing panel studies of caregivers could be assessed for incidence of elder mistreatment in a follow-up study).
Advances in measurement in risk and protective factor research are needed. The measurement of risk factors many times can be accomplished
adequately by importing measures for suspected risk factors from other settings. For example, good measures exist in the literature for cognitive impairment, dementia, handicap, and frailty. As well, some characteristics of individuals that may place them at increased risk of mistreatment, such as personality, stress, and the burden of caregiving, have been developed in the child mistreatment field and could be adapted to this research setting. For the most part, observational and hypothesis-driven research in the elder mistreatment field will have to develop measures that are specific to the field, such as measures of risk characteristics in trust relationships and aspects of settings that may be of interest. This is in addition to adapting measures of risk factors that have been developed in the child mistreatment field and in other research.