Concepts, Definitions, and Guidelines for Measurement
As noted in Chapter 1, one of the complexities of research on elder mistreatment is that researchers have used varying definitions of mistreatment. To some extent, this problem has been traceable to statutory definitions that are highly variable and ambiguous. Because legal definitions of abuse and neglect vary widely from state to state, efforts to match research definitions in any given state with the statutory definitions tend to undermine efforts to achieve comparability in research designs. In addition, legal definitions ultimately depend on value judgments (initially by clinicians and then by judges and juries) about the seriousness of the perpetrator’s conduct; these value judgments are contingent social facts that are themselves subject to empirical investigation.
One of the most urgent challenges confronting the field is the need to develop objective, uniform research definitions that are disentangled, to the greatest feasible extent, from state statutory variations, as well as from the contingent and subjective value judgments that inevitably characterize the application of vague statutory language. At the same time, however, the research definitions should also be logically connected to common statutory concepts so that they can inform policy and practice. Accordingly, the panel has reviewed state laws on elder abuse and neglect for the purpose of identifying common patterns and providing a concrete context for thinking about the core concepts and boundaries of the field of elder mistreatment. (Appendix B summarizes state statutes as of December 2001.)
Most state statutes include in some form (either under an umbrella definition of “abuse” or as separately defined elements) at least the follow-
ing types of mistreatment: (1) physical acts causing pain or injury; (2) conduct inflicting emotional distress or psychological harm; (3) sexual assault; (4) financial exploitation; and (5) neglect. Some states also include other conduct and conditions, such as “isolation” or “unreasonable confinement.” In most, but not all, states, abuse and neglect of elders fall under the general adult protection statute, whereas some states have enacted specific provisions for elder protection. In almost all states, protective interventions are authorized or required only if the adults (or elders) are mentally or physically impaired. In some contexts, the nature of the relationship between the elder person and the alleged perpetrator also matters. On one hand, “neglect” (by definition) usually is associated exclusively with persons who have a legal duty to provide care, but some states direct or authorize intervention in cases involving adults found to be neglecting their own basic needs (“self-neglect”). On the other hand, in most states, anyone, even a stranger, can be found to have committed “abuse.” The common statutory patterns in definitions of abuse, neglect, and financial exploitation are depicted in Figures 2-1, 2-2, and 2-3.
TOWARD A SCIENTIFIC VOCABULARY
The scientific vocabulary and measures that are used to study elder abuse and neglect must diverge from the legal definitions in three important respects: First, the conduct (by a perpetrator) and harms (to the elder) being studied must be objectively ascertainable based on observation, record review, or direct questioning of relevant parties. Second, although abuse and neglect represent dichotomous (yes/no) judgments from a legal standpoint, most of the underlying behaviors fall along a continuum and must be analyzed empirically as dimensional variables in terms of frequency, intensity, and severity (or riskiness)—even though the data may often be subjected to dichotomous judgments. Third, the range of conduct being measured should be more inclusive than the behaviors or harmful consequences that would indisputably amount to abuse or neglect under the applicable law.
In other words, researchers should investigate all the conduct and harms that could amount to abuse or neglect if the perpetrator had the necessary intention or culpability and if other statutory conditions are met. Some subset of this all-encompassing category could be disaggregated in data analysis to represent “core” cases of abuse or neglect, based on suppositions about the presence of the necessary intention and other conditions. The main point, however, is that the ideal empirical strategy would define the category of interest broadly in terms of conduct and harmful consequences, leaving further narrowing to the analytical and interpretive stages.
Elder Abuse and Neglect
In order to avoid unnecessary confusion, the panel has developed a research terminology to refer, descriptively, to the behaviors, relationships, interactions, and conditions of scientific interest, reserving the terms “abuse” and “neglect” to refer primarily to the legal category (recognizing that the statutory definitions vary). Box 2-1 presents the panel’s glossary of terms. The term being used in this report to encompass the conduct and harmful consequences of scientific interest is “elder mistreatment.” Although “mistreatment” is itself a value-laden term (and is used in some state statutes), the panel has selected it because it appears to have been least used as a statutory category.
The panel’s definitions of elder mistreatment (and its constituent elements) have been guided entirely by scientific considerations. We have asked: What definitions will be most useful for facilitating advances in knowledge? It bears repeating that whether “mistreatment,” as the panel is defining it, amounts to “abuse” or “neglect” in a legal sense depends on the statutory definitions in a particular jurisdiction, the actor’s state of mind, and other factors.
Abuse. Conduct by responsible caregivers or other individuals that constitutes “abuse” under applicable state or federal law.
Caregiver. A person who bears or has assumed responsibility for providing care or living assistance to an adult in need of such care or assistance.
Harm. Injuries or unmet basic needs attributable to acts or omissions by others.
Mistreatment. (a) Intentional actions that cause harm or create a serious risk of harm, whether or not intended, to a vulnerable elder by a caregiver or other person who stands in a trust relationship to the elder, or (b) failure by a caregiver to satisfy the elder’s basic needs or to protect the elder from harm.
Neglect. An omission by responsible caregivers that constitutes “neglect” under applicable federal or state law.
Trust Relationship. A caregiving relationship or other familial, social or professional relationship where a person bears or has assumed responsibility for protecting the interests of the older person or where expectations of care or protection arise by law or social convention.
Vulnerability. Financial, physical or emotional dependence on others or impaired capacity for self-care or self-protection.
“Elder mistreatment” is defined in this report to refer to (a) intentional actions that cause harm or create a serious risk of harm (whether or not harm is intended) to a vulnerable elder by a caregiver or other person who stands in a trust relationship to the elder or (b) failure by a caregiver to satisfy the elder’s basic needs or to protect the elder from harm. “Mistreatment” conveys two ideas: that some injury, deprivation, or dangerous condition has occurred to the elder person and that someone else bears responsibility for causing the condition or failing to prevent it.
Two features of this definition merit emphasis. First, the term “mistreatment” is meant to exclude cases of so-called self-neglect—failure of an older person to satisfy his or her own basic needs and to protect himself or herself from harm. Self-neglect may often be a proper occasion for intervention, at least of a temporary nature—for the purpose of determining whether the elder has the capacities for self-care and, if appropriate, of designating a caregiver, but the panel regards self-neglect as a separate domain of elder protection, not as a component of mistreatment.
Second, elder mistreatment, as defined by the panel, excludes victimization of elders by strangers. In the panel’s view, ordinary predatory victimization of elders merits empirical attention as a species of criminal behavior, but it should not be regarded as a component of the distinct domain of elder mistreatment. We say this because the nature of the relationship between the elder and the perpetrator lies at the heart of common understanding of the concept of mistreatment (and in most statutory definitions of abuse and neglect) and therefore should guide the definitions used in empirical research.
Caregiving and Other Trust Relationships
Although we have excluded ordinary victimization by a stranger, thereby narrowing the boundaries of the field, what types of relationships are relevant? In the panel’s view, the range of relevant relationships depends on whether the victim’s condition was caused by an intentional act (typically causing an injury) or by a failure to satisfy a legal duty of care (leading to unmet needs). If the elder has been injured—we refer here to financial injury as well as physical and emotional injury—or otherwise put at risk by the actor’s intentional conduct, the category of relevant relationships includes not only caregivers, but also other family members or even unrelated people (e.g., lawyers) who are aware of the elder’s vulnerability and exploit it. The panel uses the phrase “trust relationships” to denote the relevant relationships. Financial exploitation is illustrative: the conduct of interest is exploitation by family members and others who may have as-
sumed fiduciary obligations for elders with diminished capacity for financial decisions. As noted above, however, it does not include exploitation by other predatory parties; these victimizations would amount to legal harms (financial injuries) but not to mistreatment.
By contrast, if the presenting condition relates to the elder’s unmet needs, a de facto caregiving relationship (or expectation of care) is required in order to preserve the boundary between neglect by responsible others (mistreatment) and self-neglect. Professionals who are clinicians, such as physicians, nurses, psychologists, or social workers, are de facto in trust relationships with elders for whom they care. In this context, the relevant relationships include only those people who have assumed the responsibility for caregiving or are expected to do so. Obviously this characterization ultimately depends on highly contextual social facts that are not easily ascertained in surveys or observations.
Vulnerability Associated with Aging
Vulnerability is another core concept in elder mistreatment. Its importance can be seen by asking whether intimate partner violence constitutes elder mistreatment simply because the victim is older than a designated age (e.g., 65). In the panel’s view, the answer is “no” (although the issues may overlap when the victim is older and vulnerable). A predicate feature of elder mistreatment is that the victim has a diminished capacity for self-care or self-protection. Thus, a chronic pattern of intimate partner violence that has persisted into older age is not, by itself, “elder mistreatment.” Conversely, if violence against an intimate partner is initiated or becomes more frequent or severe due to the older partner’s age-associated vulnerability, then it is properly characterized as “elder mistreatment.”
Although vulnerability is a core concept in the definition of elder mistreatment, the panel concluded that further specification would be premature at this time. Some aspects of vulnerability are indisputable, including financial dependence and impairments of mobility (being wheelchair-bound) or cognition (dementia). However, other factors that diminish capacity for self-care or self-protection have not been well characterized. For this reason, the panel regards the meaning of vulnerability as an empirical question—as a referent for the cluster of clinical or psychosocial risk factors associated with increased likelihood of mistreatment. For most research purposes, vulnerability should not be used as a selection criterion; instead, data bearing on vulnerability should be routinely collected and analyzed in most studies of elder mistreatment.
Finally, another boundary issue relates to the age cut off for being an elder. This is a complicated issue. Conceptually speaking, vulnerability, not age, is the determinative concept. There seems to be no important
difference (conceptually or morally) between caretaker neglect of a 35-year-old with mental retardation and of a 65-year-old with dementia. So too with financial exploitation and other forms of abuse. Setting children to one side, the relevant population of vulnerable adults includes all persons with impairments or disabilities, such as mental retardation or impairments of capacity for mobility, associated with diminished capacity for self-protection. Not surprisingly, most adult protective services statutes include elder abuse and neglect within the broader category of vulnerable adults. Also, legislation designed to protect institutionalized persons (which includes psychiatric hospitals and mental retardation facilities as well as nursing homes) typically codify the right to be free of “abuse and neglect.”
Having said this, however, the panel recognizes that the categorical channeling of research funding (as well as protective legislation) along the path of aging gives particular salience to vulnerability associated with aging (as opposed to other conditions). The National Institute on Aging has commissioned the panel’s study, and a special focus on elders establishes the policy framework within which we are working. Accordingly, within the larger domain of adult protection, this report gives special attention to the aspects of research that focus on people who are vulnerable to mistreatment due to aging. However, this does not mean that “vulnerability associated with aging” should be defined categorically in terms of some particular age cutoff, such as 65. Even for legal purposes, the age of eligibility for benefits tied to older age varies—e.g., 65 for social security, 60 for programs funded under the Older Americans Act, 60 or 65 under adult protective services statutes. (Interestingly, the threshold age of protection under the Age Discrimination in Employment Act is 40.) For research purposes, a category defined as persons 65 or over would be both overinclusive and underinclusive—since many people over 65 are not vulnerable and some younger than 65 are vulnerable due to aging (e.g., dementias). In sum, the panel regards “older age” as one of the risk factors that should be explored empirically under the rubric of “vulnerability associated with aging.” As the field of elder mistreatment develops, surveillance and research must attend specifically to age as well as other indicators of vulnerability. (The panel’s conceptual vocabulary is depicted in Figures 2-4 and 2-5.)
GUIDELINES FOR MEASURMENT
In science, good measurement has several prerequisites. The first is a concept of what is being measured. In this case the object of measurement is the occurrence of elder mistreatment using the vocabulary and definitions presented above. The second prerequisite is an operational definition of the concept being explored so that it is objectively ascertainable in the field. Operational definitions in the domain of elder mistreatment are compli-
cated, in part because the relevant concepts are poorly developed, and in part because researchers’ aims vary widely across studies. Operationalization answers questions such as: “How do we measure this or that aspect of mistreatment?” or “How will we know whether we should count this as a case of mistreatment?” Ideally, operationalization leads to the development of a set of criteria for answering this question and a process by which these criteria can be applied in the field—the measurement method.
Failure to provide care
Unmet [basic] needs
Vulnerability Associated with Aging
Impairment of capacity for self-protection or self-care
Person in trust relationship
(depends on definitions under applicable law)
(depends on definition under applicable law)
Given the complexity of the definition of mistreatment, its operationalization is best approached in a stepwise fashion, with each step addressing a different aspect of the concept.
The next prerequisite involves the standardization of measures prior to their use in research. In many research settings, such as population surveys, it will be sufficient for the researchers to describe the conduct and other variables that have been measured in a way that reflects the relevant dimensions of mistreatment. As researchers continue the iterative process of conceptualizing and operationalizing the measures, a consensus will gradually emerge regarding the relevance and significance of the measures being used in the field for different aspects of mistreatment and its correlates and outcomes. In some clinical research contexts, however, it will be necessary for researchers to classify whether or not the data represent a “case” of mistreatment. Under optimal circumstances, there would be a method of measurement to definitively assess the presence or absence of elder mistreatment in such cases. Such a “gold standard” could be used to judge the value of other measures as well as to definitively determine the presence or absence of mistreatment for research purposes. Given the nature of the object of measurement, such a standard is not possible in the elder mistreatment field. There are two basic problems—contested facts and contested values. Irrespective of research methodology, uncertainties will arise regarding the conduct of the alleged perpetrator and the effects on the elder. Also, as already indicated, characterization of particular conduct as mistreatment requires value judgments, contestable at the margin if not at the core.
Using the example of other fields confronted with a similar problem, a “LEAD standard” (longitudinal, experts, all data) could be developed to serve in the place of a gold standard (Spitzer, 1983). A LEAD standard would use longitudinal observation, all relevant data, and the review by experts in the field to determine the presence or absence of mistreatment. A LEAD methodology typically involves two components. The first component is collection of data on the case that is to be classified (as mistreated or not mistreated). An expert in the area who investigates the case thoroughly collects the data. The investigation might include taking history from several sources, interviewing the person who may have been mistreated, interviewing the possible perpetrator(s), as well as reviewing medical and other pertinent records. The data collection focuses, as much as possible, on whatever longitudinal information is available on the case.
The second component of the methodology involves evaluation of the data by a panel of experts, who are asked to make a collective judgment about whether or not the case meets an a priori definition of mistreatment or a specific type of mistreatment. The definition is made available to the panel, often in the form of operational criteria. The panel, typically small—
5–7 members—for the sake of efficiency, includes interdisciplinary representation from professional backgrounds with expertise in the area of mistreatment. The expert who collected the data presents each case verbally and in the form of a structured case summary.
The panel then deliberates case-by-case to decide if a specific case meets the a priori definition. The decisions the panel makes in each case, such as judgments about which specific examples meet the definition and which do not, are recorded and eventually summarized in a workbook/minutes book reflecting the consensus process as applied to real cases. The latter can then be used to improve upon the definition, as precedent for future panels, or to train professionals in the recognition of elder mistreatment. This should be a process that (almost) everyone would agree is able to classify correctly individuals as mistreated or not mistreated without worrying about the resources or cost needed to make the determination. Put another way, if resources were not an issue, what would be done to decide if someone has been mistreated or not? Once a LEAD standard is in place, then several potential methods of determining mistreatment can be tested against this standard. And the LEAD method itself can be used in the context of research to definitively assess the presence or absence of elder mistreatment. Such an approach is already under way in the work of Fulmer and Wetle (1986).
The next step is the development of the measure. This involves deciding on the specific purpose of the measure and the measurement method, followed by an assessment of its reliability and validity. Measures may have different purposes, such as to ascertain occurrence of mistreatment in the population for research purposes or for surveillance, to assess the risk of mistreatment for early intervention, to screen for mistreatment in different settings (e.g., emergency department, long-term care), to determine whether mistreatment occurred in a given circumstance, to differentiate different types of mistreatment (e.g., physical or financial), to quantify the severity of mistreatment, and so forth. Clearly, different measures will need to be developed for use in these different research contexts.
Similarly, measures are likely to vary according to the method used to elicit the data. These include self-report, proxy or informant report, direct examination of the elder’s physical and/or mental state, clinical observation, or a composite of these. The choice of assessment method will depend on the purpose of the measure, the risks of error associated with each method, the tolerance for error in measurement, and the research resources available.
The measure should be both reliable and valid. Reliability assesses how much agreement there is if different people are conducting the measurement (interobserver) or if a measure is applied at different points in time (test-retest). High interobserver reliability should be pursued. Test-retest reli-
ability should also be pursued for all measures; short time intervals of test-retest are optimal, since the occurrence of mistreatment is transient in some cases.
Validity assesses how accurate a measure is of what we want it to measure, in this case some aspect of mistreatment. There are several types of validity: content, criterion, predictive, and construct validity. Content validity is an assessment of the measurement method’s ability to measure mistreatment using logic and special expertise, typically by expert opinion. Criterion validity assesses the measure against a widely agreed-upon standard, in this case a LEAD standard. Predictive validity assesses the ability of the measure to make predictions about the future, such as predicting response to interventions or the course of mistreatment. Finally, any research evidence that tends to illuminate exactly what the instrument measures adds to its construct validity.
In sum, research measurement in the field of elder mistreatment is complicated for several reasons. First, several elements require measurement. Second, observations necessary to make a determination of mistreatment are usually not directly available to the researcher and must be inferred indirectly. Third, even with the necessary observations available, a determination of mistreatment is not immediately apparent but rather requires human judgment to assess whether these observations meet (a priori) definitions (operational criteria) of mistreatment derived from common sense, consensus, or law. Fourth, the definitions against which the observations are assessed appear to be variable in research conducted thus far. These issues greatly limit the ability of researchers to develop measurement tools that meet high standards of reliability and validity.
OPERATIONALIZING THE ESSENTIAL ELEMENTS
In the context of elder mistreatment, several variables are a target of measurement. These are listed in Box 2-2.
The items in this box merit initial comment prior to later detailed discussion. With regard to the first item, while the issue of who is an older person has no definitive answer for all purposes, the demographic category of interest needs to be defined explicitly for the purposes of research. The second item is the existence of a trust relationship between an elder and another person. As indicated above, the concept of elder mistreatment is predicated on the existence of such a relationship. Thus, a definition for what constitutes a trust relationship is needed. For both these items, operationalization is straightforward in the sense that an a priori definition, whatever its strengths and weaknesses, can be applied in the process of research so that a particular situation can be assessed against that definition (e.g., “Does this particular person meet the definition of being an older
person?” and “Does the relationship meet the definition of being a trust relationship?”)
It is necessary to measure the relevant conduct (what was done or not done) of the other person, to assess whether or not the elder has been harmed, and, if so, to determine whether what the other person has done or not done has caused the harm. The definition of the relevant conduct and harm is a complex undertaking in that it is not always possible to anticipate in advance all conduct and consequences that might be of interest. Furthermore, in the process of assessment while in the field, it is highly unlikely that the conduct in question will be directly observable to those conducting the research and is therefore likely to be evaluated indirectly. This is often true for harms as well. Thus an operational definition of conduct and harm should contain a general description of the kinds of conduct and consequences that may be of interest as well as a description of the process used to determine the occurrence of relevant conduct and harms, constructed so that they can be assessed both directly and indirectly.
Determining whether a particular conduct caused a particular harm will not be necessary in most studies, especially those using survey methods. However, this task may be necessary in some studies, especially those relating to the factors that differentiate, clinically, between inadvertent injuries and intentional ones (see Chapter 6.). In some cases, when direct observation is available, it is possible to state unequivocally that a specific conduct caused the injury. For example, if a caregiver hits an elder on her upper right arm and there is a bruise where there the elder was struck, causality for the bruise is clear. However, if the elder also is anxious and scared, under what circumstances can one conclude that the striking of the arm caused these psychological consequences? Furthermore, if the elder has
bruises elsewhere on her body, when might it be concluded that the same caregiver striking her at other times caused these other bruises? As should be apparent, the causal inference connecting conduct to consequences in many instances necessitates a judgment on the part of the researcher. Thus, for some studies, the researcher may have to specify an operational definition of causality and a process by which the determination of causality is made.
The next item in the box deals with the evaluation of information from the previous items in judging whether or not the combination of conduct and harm constitutes mistreatment. This requires operational definitions of the types of mistreatment against which the specific circumstances can be assessed. The panel encourages researchers to be as specific as possible in identifying which combinations of conduct and harm are being defined as mistreatment for purposes of the study. It may be helpful to define the category of mistreatment separately for (1) physical mistreatment, (2) sexual mistreatment, (3) emotional mistreatment (including isolation), (4) financial exploitation, and (5) failure to provide needed care, including abandonment.
As explained earlier, the panel recommends that researchers try to avoid stand-alone, unmodified use of the terms “abuse” and “neglect” because these terms require legal interpretations and community value judgments that inevitably vary across states and localities. In addition, it should be recognized that whether a case is found by the applicable authorities to constitute abuse or neglect also depends on the purpose of the intervention in a particular case and other aspects of the social context. In some situations, for example, the question being asked is whether an intervention or treatment might be implemented to help the elder. In other situations, however, a determination of abuse or neglect might lead to criminal prosecution of the perpetrator. These determinations are all rooted in value judgments made initially by the examining clinicians and subsequently by public officials and courts. These judgments are themselves subject to empirical study. In such investigations, what must be defined and measured are the variety of possible clinical, social, and legal responses that might be made to particular cases. If this type of research were added to the box, the question of interest would be “Was this combination of conduct and harm characterized as ‘abuse’ or ‘neglect’ by the relevant decision-makers?”
The sixth element in the box involves the operationalization and measurement of factors that increase or decrease the likelihood of elder mistreatment. Measurement of risk and protective factors is as important as measurement for mistreatment itself. Risk factors are factors that increase the probability of mistreatment, while protective factors are ones that decrease its probability. Their measurement is critical from the public health
point of view for several reasons. Their identification is crucial to the detection of who is at risk for mistreatment so that preventive interventions can be applied. As well, their identification promotes understanding of the mechanisms leading to mistreatment. Because vulnerabilities associated with aging (and with disabilities) are of special concern in the fields of elder mistreatment and adult protective services, these risk factors require careful attention. (Research on risk and protective factors is discussed in Chapter 5.)
The final element in the box relates to the outcomes and consequences of mistreatment. Mistreatment has been associated with a series of consequences and adverse outcomes. For example, mistreatment can cause physical and mental morbidity that is at times sustained. It can lead to serious financial strain. As well, social isolation, loss of dignity, impaired quality of life can result. Research on the consequences of mistreatment is critical to understanding its individual and societal impact and to targeting and assessing the benefit of interventions. (A theoretical model linking mistreatment to its outcomes is discussed in Chapter 3.)
With this overview in mind, the discussion now turns specifically to the measurement of several of the elements involved in research on mistreatment as identified in Box 2-2.
If an age cutoff is to be used, then the operationalization and measurement of who is an older person are straightforward and merit no further discussion. If, however, the definition is broadened to include other groups of vulnerable adults, operational definitions and specific measurement methods may be needed. For example, if the definition is broadened to include “adults with developmental disabilities,” or “adults with mental illness,” or “adults with physical disabilities,” then a definition of each of these terms is necessary for research to go forward, as is a method of determining whether a specific individual meets the definition. It seems fair to assume that definitions and measurements for various types of disabilities exist in the relevant fields and can be imported with appropriate modification to research on elder mistreatment.
The panel favors specific definitions of disability if the population being studied is chosen on this basis, rather than use of a generic and vague category of all “vulnerable adults.” Objective criteria of inclusion, such as cognitive impairment or frailty or disability impairing locomotion, should be used. However, if the study population is defined by age (e.g., everyone over 18 or 40 or 55, etc)., then the elements of vulnerability can be defined empirically according to the personal characteristics that emerge as risk factors for mistreatment.
A trust relationship is at the center of research in elder mistreatment. In the simplest terms, such a relationship exists when one party is charged with, or has assumed, the responsibility for caring for or protecting the interests of the older person, or when the relationship (in its social context) creates the expectation of care or protection. There are therefore at least two participants in such a relationship, the elder and the person—or persons—responsible for care or protection or expected to provide care or protection for the elder. Such a relationship may arise formally or informally and may be voluntarily undertaken or imposed by operation of law or social custom.
For example, the relationship may arise from a formal guardianship or a durable power of attorney in which the trusted person agreed to serve in that role in the event of the elder’s incapacitation. Even in the absence of any formal designations, the relationship may arise out of kinship or friendship or professional roles. For example, a relative, caregiver, or other person may find himself or herself in the position of making health care or living decisions for the elder without any formal agreement or designation. Furthermore, someone may take on the responsibility to assist the older person in financial matters, such as an accountant, financial adviser, or friend with special knowledge in this area. Under certain circumstances, the existence of a trust relationship may be predicated on the fact that the other person is a health care professional who has taken on the care of the elder, as would happen in a nursing home, assisted living, or hospital with a nursing aide or licensed professional.
Different types of relationships may have different bearings on elder mistreatment, depending on the type of mistreatment. The threshold of involvement that constitutes a trust relationship may vary for physical mistreatment, emotional mistreatment, financial exploitation, or failure to provide needed care. For example, failure to provide needed care (neglect) depends on the existence of a de facto caregiving relationship and therefore would require a narrower range of relationships than the other categories of mistreatment. For example, any family member would be in a trust relationship for purposes of the basic expectation that they will not exploit or harm the vulnerable elder person. However, a family relationship clearly will not always amount to a relationship sufficient to give rise to a caregiver obligation; this is why the panel has distinguished between these two concepts and has defined trust relationships as a broad category that includes, but is not limited to, caregivers.
In some circumstances, of course, existence of a trust relationship is unambiguous and harm caused by the other person would always constitute mistreatment. These include legal guardians and professionals who
enter into formal professional relationships with an older person. The case for a legal guardian should be obvious. Similarly, paid professionals, whether they be clinicians, attorneys, financial advisers, or accountants, enter into trust relationships by virtue of their professional activities. It should be apparent that relationships with paid professionals in all health care settings, such as hospitals, nursing homes, assisted living homes, adult day care programs, and the like, enter into trust relationships when they come into contact with older people. This includes not only licensed professionals such as doctors or nurses but also personal care workers (nursing aides), janitorial staff, escort staff, etc.
In other circumstances, whether a particular relationship amounts to a trust relationship for purposes of elder mistreatment research may be unclear. In the first instance, researchers should make every effort to determine the point of view of the older person regarding their trust relationships. In some situations, this may be determinative. However, in many situations, the elder person’s point of view will not be ascertainable or will be superseded by social conventions or legal duties. For example, the older person may be suspicious of, and have no expectation of protection from, a home health care aide who has assumed a caregiving obligation. Conversely, the older person may develop a trusting relationship with a door-to-door vacuum cleaner salesman who, by law and social convention, bears no obligation to protect the older person’s interests. Accordingly, applying the concept ultimately requires objective assessment and judgment.
In sum, empirical knowledge is lacking about the kinds of trust relationships that older persons enter into, the other parties involved in these relationships, the foundations of these relationships, and their association with different types of mistreatment. Therefore, an early priority of research in the field ought to be the conceptual and empirical development of different operational definitions of trust relationships.
Conduct of the Other Person in the Trust Relationship
The relevant conduct of the other party that may be of interest includes direct physical contact (hitting, pushing, shoving, etc.), verbal mistreatment (yelling, threatening verbally, criticism, etc.), placing restrictions on the older person (isolating to a room, unnecessary use of physical or chemical restraints), social embarrassment (berating the elder in public), depriving the older person of material possessions (restricting access to money, stealing from the elder, etc.), not providing necessary care (e.g., not providing medications, bathing infrequently, feeding a limited diet), and many more. The challenge for researchers is to define the conduct with maximum possible specificity to facilitate analysis and interpretation. Whether any such conduct amounts to mistreatment requires a value judgment based on con-
text. For example, restricting access to money may be entirely appropriate conduct in caring for a person with dementia. (See further discussion of this point in the section on “mistreatment” below.) As noted earlier in this chapter, however, the most sensible strategy for research is to define the category overinclusively (with reference to the expected definition of mistreatment) for purposes of data collection and measurement and to refine it thereafter in analysis and interpretation.
Measurement of conduct is subject to a number of significant methodological limitations. Briefly, much of this conduct is not observed directly and relies for its detection on report by the elder, by the other party in the trust relationship—who may be the perpetrator of mistreatment— or by a third party, such as a colleague or supervisor of the other party in an institutional setting. Indeed, in the absence of direct observation, conduct is harder to assess than harm, since it may not leave evidence in the form of readily observable physical or emotional consequences, since it may be forgotten by the elder if she is cognitively impaired, since the older person may be reluctant to report the occurrence of such conduct, or since the other person may not report it out of conflict of interest.
The investigator is faced with the difficult task of detecting a “latent variable” requiring a research methodology that optimally employs several modalities of assessment and takes repeated observation. As with assessment of harm, there is a dearth of basic descriptive studies of conduct involved and of measurement methods.
As already noted, mistreatment (under any consensus definition) will include some types of conduct that have not actually caused harm—perhaps because harm was not intended or because the conduct creates an unacceptable risk of harm. However, many types of mistreatment do involve actual injury or harm, most notably physical assault and financial mistreatment (loss of property). To the extent that the definition includes harm, the measure of harm must be operationalized and measured.
The importance of measuring harm varies according to the type of research being conducted. For example, survey research and other studies in nonclinical settings (or not using clinical or legal records) are likely to focus mainly on the possible perpetrator’s conduct; the presence of harm is likely to be ascertained on the basis of a few specific indicators (e.g., “Were you hurt?” “Did you have to go to the hospital?” “Did you lose any money?”) However, in the context of research in clinical settings, such as identification of forensic markers for mistreatment, or development of improved screening tools, the assessment of harm may be a particularly important element of the study.
Pertinent types of harm to older persons include physical injury, emotional injury, and financial harm. Physical harm is the most straightforward, since its presence typically can be assessed reliably through examination, through laboratory tests (e.g., X-rays), or upon forensic assessment. Examples of physical harm include lacerations, burns, fractures, bruises, malnutrition, and others (Dyer et al., in this volume). Similarly, financial harm can typically be assessed reliably if access to the older person’s financial records is available. Emotional harm is more difficult to assess. This may take on the form of mental distress and other psychological responses, post-traumatic symptomatology (social withdrawal, reexperiencing traumatic events, trouble sleeping or eating, etc.), or the onset (at times recurrence) of a psychiatric disorder, such as major depression, post-traumatic stress disorder, panic disorder, agoraphobia, among others. Measurement of harm must be able to determine the presence or absence of different consequences in the various domains above. Since many harms may not be anticipated prior to the initiation of the research, the measurement method must be general and flexible so as to detect a wide range of consequences that may be specific to the specific elder-trust relationship, and to the setting involved (home, hospital, long-term care etc.). The measurement method must also evaluate temporal aspects of the consequences (onset, frequency, duration) and quantify the severity of these consequences.
A key methodological issue in the measurement of consequences is that some of the harms involved are not always accessible to direct measurement. This is true for several reasons. First, in many cases, harms are transient and remit by the time an assessment occurs, as in the case of a bruise or a laceration. Second, many older persons who are mistreated are cognitively impaired and cannot recall past harms. Third, many older people are reluctant to report conduct of others who may have harmed them or that may constitute mistreatment out of embarrassment or for other reasons. Fourth, often the only other source of information about past harms may be the other person in the trust relationship, who has a conflict of interest regarding disclosure of the harm. Therefore, the best a researcher can do, as is customary when latent variables are being investigated, is to employ methods of assessment that are multimodal (e.g., self-report, observer report, data review, direct examination, laboratory studies, forensic assessment) and that are repeated with sufficient frequency to minimize the likelihood that relevant consequences are missed. The corollary to this is that measurement methods should include checks and balances so that false positives are minimized as well.
In general, methods to assess the presence or absence of physical injury, emotional disturbance, or financial injury are available and have been adapted to the elder mistreatment context. These methods generally are able to evaluate the presence or absence of injury (harm) and are also able
to approach its temporal aspects and severity. The vast majority of measures have focused on the determination of whether elder mistreatment has occurred using a wide range of methods and definitions. Many studies have assessed the occurrence of mistreatment by review of protective agency records, study of sentinel reports (reports of professionals serving older adults), or criminal justice system statistics. These have used typically unstandardized or vague definitions of abuse and neglect, many based on the wording of state statutes, and have significant methodological weakness.
A few researchers have tackled the problem directly, but they have used definitions or measures that have varied from study to study. In some cases, methods have been developed to assess the occurrence of elder abuse using telephone or in-person interviews of family members or proxies or direct assessments of samples of older adults. However, there is a dearth of such measures, and most existing measures have had limited assessment of their measurement characteristics, reliability, or validity. Furthermore, the measures used have almost always been direct adaptations of measures intended for other purposes or for other settings (e.g., the Conflict Tactics Scale was intended to measure interpersonal violence for married couples and was modified by Pillemer and Finkelhor to assess abuse of older people by their caregivers). As well, existing measures are inadequately differentiated or specialized. They do not, for example, distinguish clearly the types of harm they are measuring (physical, emotional, etc.) or differentiate the measures according to whether they are intended to screen for harm, define its occurrence, or measure its severity.
The elder mistreatment field is lacking in descriptive methodological research on how to measure consequences that are related to mistreatment. As Dyer et al. (this volume) note, no studies have carefully described the different types of harms that mistreated older persons may suffer, the interrelationship of the different harms (e.g., relationship of physical to emotional to financial), the severity of harms, their characteristics, and their clinical course. In addition, few studies have compared different approaches to the measurement of harms. The greatest gap relates to psychological consequences of elder mistreatment. This sort of information is key to the ability to develop measures that are methodologically sound. Basic methodological research should also be an early priority in the field.
Whether certain facts, collected using the methods discussed so far in this chapter, constitute mistreatment is a matter of definition and judgment. The researcher’s main goal should be to make the process as transparent as possible. The facts, collected as above, must be assessed against an a priori
TABLE 2-1 Cross-Tabulation of Conducts and Harms
definition. (The panel’s preferred definition is set forth above.) Much of the time, the facts fit the definition with little doubt. In other cases, whether the facts should be characterized as mistreatment requires judgment. Thus the researcher must specify a process for making this determination.
This process works best if it has at least two characteristics. The first is a set of rules or guidelines specifying whether anticipated combinations of circumstances and conduct will constitute one form or another of mistreatment if detected in the field. A different cross-tabulation of conducts and harms is necessary for each type of mistreatment. This is illustrated in Table 2-1.
This approach illustrates several issues. While it is not possible to anticipate all combinations of conduct and harms that could be encountered, prior consideration of the types of issues that will come up in the course of the research will help standardize decisions in the conduct of the study. Also, the frequency of harms and conducts must be taken into account in the table. For example, a single instance of pushing an older person down the stairs may constitute mistreatment regardless of whether an injury occurs, whereas forgetting to feed her a meal now and then may not. Furthermore, the absence of both specific conduct and specific harms may, under certain circumstances, constitute mistreatment.
For example, as already mentioned, specific harms that could be due only to mistreatment (such as certain types of fractures) might be classified
as mistreatment even if the relevant conduct were not detected. But conduct that has not resulted in any apparent harm may also at times be properly characterized as mistreatment. If a trusted person attempts to push an older person down the stairs but fails to succeed, thus causing no obvious harm, most would agree that this is physical mistreatment. What distinguishes this situation is the intent of the trusted person, so that if there is clear intent to harm, mistreatment may be present even if the harm did not occur.
While intent is important in limited circumstances, it should be kept in mind that intent to harm, which is very hard to determine (even in a courtroom, much less for research purposes), is not a necessary element of the definition of mistreatment, as defined in this report. Of course, when there is intent to harm that can be determined unequivocally, then in all likelihood mistreatment has occurred. While it would be interesting to study the relationship between intent of the trusted person and conduct or harms that constitute mistreatment (and that are characterized as neglect and abuse by social authorities) this is not a core aspect of the measurement of mistreatment.
An important caveat should be added at this point. One of the difficulties in elder mistreatment research so far has been the use of overly inclusive definitions of what constitutes mistreatment. Some writers have included the entire range of harms and conducts in Table 2-1 as mistreatment, including, in some cases, “social embarrassment” that may have been transient, or minor physical injuries that were clearly an accident. It is important for researchers to focus their attention on serious forms of mistreatment and not to define “normal” negative human interactions as pathological. The purpose of the cross tabulation is not to lay out all harms and consequences involved in interactions between an older person and another in a trust relationship; rather it is to limit the definition of elder mistreatment to a specific number of intersecting sets that have been defined by the researcher as plausible forms of mistreatment.
The second part of the assessment of mistreatment is the specification of a process used to apply the definition and guidelines to all cases under study. This process has as its starting point the facts in each case, the definitions of mistreatment being employed in the study, and the rules/ guidelines derived from cross-tabulation of anticipated examples. This information is reviewed by designated individuals using a specified process. Typically, a trained expert looks at each case and decides the simple ones. A consensus panel then reviews the more complex cases, plus a sampling of the simpler ones classified by the individual reviewers. It bears emphasis that this classification is needed only if a dichotomous classification is required by the goals of a particular study. As already noted, the panel encourages study of a broad range of conducts and harms beyond those
that could be classified as cases of mistreatment. The presence (or absence) of some of these might be predictive of the onset of mistreatment in the future. Also, even though they may not be mistreatment, some of these could affect the quality of life of the older person. Finally, these may be of interest to social scientists studying the relationship between conduct and consequence in older persons and their caregivers.
Risk and Protective Factors
A major purpose of research in elder mistreatment is to understand its causes so that it can be prevented, treated, or managed effectively. This line of inquiry typically begins with the study of factors that increase (risk) or decrease (protective) the probability that mistreatment will occur. As discussed earlier, it is the panel’s view that one of the crucial risk factors involved is the vulnerability of the elder person. More research is needed on risk factors, including vulnerability, and protective factors for mistreatment. This issue is of such significance to the field that Chapter 5 has been devoted to it.
Further basic research on the phenomenology of elder mistreatment is a critical early step in the further development of the field. Such research will lead to a much better understanding of the key elements of elder mistreatment (see Box 2-1), which in turn will facilitate the development of broadly accepted operational definitions and the development of research and clinical measures for these phenomena. Examples of such research include studies of: (1) the kinds of trust relationships that older persons enter into, the other parties involved in these relationships, the foundations of these relationships, and their association with different types of mistreatment; (2) the different types of harms that mistreated older persons may suffer, the interrelationship of the different harms (e.g., relationship of physical to emotional to financial), the severity of harms, their temporal characteristics, and their clinical course; (3) the injurious conduct or omissions of other parties in trust relationships, how they manifest themselves, and their clinical course; (4) the psychological effects of mistreatment, including types of psychological harm, their presentation, and their clinical course; and (5) the circumstances under which harm is most likely to have been caused by the acts or omissions of another person.
The development of widely accepted operational definitions and validated and standardized measurement methods for the elements of elder mistreatment is urgently needed to move the field forward. The field must develop widely accepted operational definitions of the elements of elder
mistreatment, its different forms, and associated risk factors and outcomes. The field must also develop a series of measures for these elements, with good (and known) reliability and validity. A menu of measures is necessary for each of the multiple contexts of research, including screening and case identification in clinical settings as well as studies of elder mistreatment in populations. To the extent that dichotomous classifications of mistreatment are needed, agreement must be reached on what LEAD-type methodology will be used in place of a gold standard for such studies.
Agreement on operational definitions for research may provide a useful foundation for developing standard definitions and classification criteria for surveillance and reporting. It is conceivable that a consensus conference could be convened to propel this process forward, although such an effort may be premature in the absence of greater experience in the field developing the approach outlined in this chapter. Another possibility would be to initiate a consensus process in a more limited domain—such as defining mistreatment in the context of developing definitions and measures of quality in long-term care, as recommended in a recent report on this subject by the Institute of Medicine (2001). (see Chapter 7 for further discussion).