Elder abuse in the oldest old: prevalence, risk factors and consequences

The group of the oldest old (aged over 80 years) is one of the fastest growing populations in Germany [3]. A widespread phenomenon among older people in general (60+ years) is called elder abuse (EA). The most common working definition of EA, which is used in most studies and adopted by the World Health Organization (WHO) describes EA as:

a single or repeated act or lack of appropriate action, occurring within any relationship where there is an expectation of trust which causes harm or distress to an older person. Elder abuse has serious consequences for the health and well-being of older people and can be of various forms: physical, verbal, psychological/emotional, sexual and financial. It can also simply reflect intentional or unintentional neglect [31, p. 1].

The occurrence of EA has been investigated in many national and international studies. Two different meta-analyses reported prevalence in community settings of 10% [13] and 15%, respectively. [33]. Within these two meta-analyses, the range of reported prevalence rates was between 0.8% [20] and 79.7% [26], which indicates heterogeneous definitions and operationalizations of EA that differ between analyzing solely illegal acts and addressing subjective experiences of abuse [8]. In a representative German study, a prevalence rate of 25.6% for psychological aggression and 1.5% for physical violence in community-dwelling people aged between 60 years and 85 years old was reported [8]. In contrast, EA in nursing homes is seldomly explored through the perspective of the victims, which might be due to the high sensitivity of the topic as well as difficult access to and limited communicative abilities of this group [8]. Moreover, a representative prevalence rate of EA in the oldest old (80+ years) is lacking on the international and national levels.

As prevalence rates are missing for the oldest old and consequences of EA are presumably more devastating for them [8], special attention for this group is needed. Firstly, higher need for care, reduction in the size of social networks, higher prevalence rates for dementia, and frailty are reported in this age group [23]. Therefore, psychological forms of violence, like intimidation, paternalism as well as shaming and blaming, should be investigated [21]. Other forms of abuse, like sexual abuse, are rarely reported in old age [33]. Secondly, the victim’s perspective is crucial when EA happens in social relationships with a high dependency between victim and perpetrator. Victim self-reports are of utmost importance when victims have less ability to defend themselves against intentional or unintentional violence, when they have fewer possibilities to reach for help and support, and when their access to social studies is limited [8], all of which is the case for the oldest old.

Risk factors for EA abuse have been explored in two recent systematic reviews [16, 28]. Risk factors for the victims can be categorized into sociodemographic, physical, and psychological aspects [16, 28]. Sociodemographic risk factors with consistent evidence are ethnicity and lower education [28]. Contradicting evidence exists for age > 75 years, gender (female), marital status, higher income, and living alone, which are protective factors in some studies and risk factors in others. Psychological constructs like cognitive impairment, aggressive behavior, loneliness, personality traits (e.g. antisocial personality), self-neglect, and stress-related coping processes are associated with EA [16, 28]. Health-related risk factors comprise various measures of functional impairments like ADL and IADL, frailty, multimorbidity, and incontinence as well as psychiatric illnesses (depression, alcohol abuse, past abuse) [16, 28].

EA affects central aspects of quality of life (QoL) in very old age. In a review of 25 cross-sectional studies, Dong et al. [4] pointed out that EA has a detrimental impact on the psychological well-being of the aged population. In addition, they reported that depression is one of the most prevalent psychological consequences of EA. Recently, Wang and Dong [30] showed that community-dwelling older people who report a greater degree of self-perceived loneliness have a greater probability of having experienced EA, especially psychological forms of abuse. A further aspect of quality of life that is influenced by EA is autonomy, often by use of custodial measures.

The first objective of this study is to estimate the prevalence of various facets of EA in very old age. In a second step, risk factors for experiencing EA are identified. Thirdly, we investigate the effects of EA on central outcomes of QoL in very old age.

The NRW80+ dataset is used to examine EA in the oldest old. The sample consists of n = 1863 persons over 80 years and includes n = 1687 self-report interviews and n = 176 proxy interviews. Details about the recruitment process, a description of the proxy interviews, and a general description of the participants of the NRW80+ study are presented elsewhere [10].

The challenges and potentials (CHAPO) model describes the theoretical framework of this study [29]. The definition of EA already implies a connection between life conditions and emotional results which are also contemplated in the broader CHAPO framework. With respect to this model, we have investigated the role of disposable resources, skills, and competencies relating to experiences of EA and linked these experiences to the model-inherent life results depression, loneliness, life satisfaction, and autonomy.

During the last 12 months, 54.1% of persons aged 80 years and older have experienced some kind of EA. This prevalence is higher compared to other studies for the elderly population in Germany [8]; however, the comparison of different prevalences of EA can be misleading when the difference in operationalizations is not accounted for. In this sample, the low-threshold approach in the wording of the questions and the emphasis of the EACS on emotional forms of abuse are expected to have contributed to higher prevalence rates. The higher age of the informants, which is associated with other risk factors like multimorbidity, also contributes to the higher prevalence. In old age, emotional and psychological abuse are the most common forms of abuse, which resembles the results of known the meta-analysis and reviews [13, 27, 33] in which emotional abuse was also identified as the most frequent form of EA in younger populations. Even though many very old people experience EA, it is nevertheless a rare event for most of them. In each dimension, less than 4% of the people are affected “often” or “very often” by EA.

In the regression model, multimorbidity, and the victim’s tendency for aggressive interactions are significant risk factors for EA, while a greater social network is a protective factor. An influence of cognition, nursing home status, IADL, education, gender, age group and income could not be shown in the multivariate logistic regression model. While evidence for the effect of gender, education, age group and income has already been inconsistent in existing literature [16, 28], reasons for the missing effect of nursing home status and cognition are not obvious. In a subsequent bivariate sensitivity analysis of our data, nursing home status has shown a significant association with EA, but this effect was moderated by cognition, functional health, and multimorbidity, which are also more common in nursing homes. Therefore, the effect was attenuated in the full model. The same applies to the constructs related to physical health, where IADL is not significant when multimorbidity is included in the model.

Risk factors for the abusers, like caregiver burden, especially for caregivers of people with dementia or substance abuse [16], cannot be investigated in this article. Likewise, risk factors lying within the victim-perpetrator relationship cannot be explored in this article. Therefore, the overall low amount of explained variance was expected.

On the basis of the CHAPO model, life events, such as experiences of EA, are associated with more generalized life results. In our sample and understanding, EA is associated with higher depression, increased loneliness, a reduction of autonomy, and life satisfaction. For depression, loneliness, and life satisfaction, the observed effect sizes are even higher than reported effect sizes in other populations [4, 24], supporting the proposition that the impact of EA is more severe for vulnerable groups [8]. Autonomy and loneliness are also significantly associated with EA. These two life results are more strongly affected by interactions with others, i.e. victim-perpetrator interaction in the case of EA. For a better understanding of underlying processes, the inclusion of the perspective of potential perpetrators in studies seems necessary. This could be done in future qualitative analyses but is not possible within the NRW80+ dataset. For loneliness, it might also be true that old people who have a smaller social network may feel more alone when they experience abuse in one or more of their few relationships. Causality might also work in the other direction, meaning that loneliness makes old people easier to be taken advantage of.

Based on associations found between aspects of QoL and EA, we conclude that our choice of a broad understanding and low-threshold operationalization of EA leads to results that are relevant for the oldest old and should be pursued in future research, especially in the context of quality of life.

Two limitations of the current study should be considered. Firstly, the exclusion of proxy interviews limits the representativity of the sample. The resulting bias has been addressed by a recalculation of sample weights that gives more importance to cases which are underrepresented; however, a complete elimination of this bias is not possible. Nevertheless, compared to surveys that do not include people with dementia and focus entirely on community-dwelling older adults, the use of adapted sample weights represents an improvement. Secondly, testing causality was not possible within this cross-sectional dataset and results should be replicated. As control group designs are no ethically possible alternative, longitudinal studies need to be performed.