A systematic review and meta-analysis of interventions designed to prevent or stop elder maltreatment

Abstract

Background: elder maltreatment is a major risk for older adults’ mental health, quality of life, health, institutionalisation and even mortality.

Objectives: to perform a systematic review and meta-analysis of interventions designed to prevent or stop elder abuse.

Methods: Studies that were posted between January 2000 and December 2014, written in English, specifically designed to prevent or stop elder maltreatment were included.

Results: overall, 24 studies (and four records reporting on the same participants) were kept for the systematic review and the meta-analysis. Studies were broadly grouped into three main categories: (i) interventions designed to improve the ability of professionals to detect or stop elder maltreatment (n = 2), (ii) interventions that target older adults who experience elder maltreatment (n = 3) and (iii) interventions that target caregivers who maltreat older adults (n = 19). Of the latter category, one study targeted family caregivers, five targeted psychological abuse among paid carers and the remaining studies targeted restraint use. The pooled effect of randomised controlled trials (RCTs)/cluster-RCTs that targeted restraint use was significant, supporting the effectiveness of these interventions in reducing restraint use: standardised mean difference: −0.24, 95% confidence interval = −0.38 to −0.09.

Interpretation: the most effective place to intervene at the present time is by directly targeting physical restraint by long-term care paid carers. Specific areas that are still lacking evidence at the present time are interventions that target (i) elder neglect, (ii) public awareness, (iii) older adults who experience maltreatment, (iv) professionals responsible for preventing maltreatment, (v) family caregivers who abuse and (vi) carers who abuse.

Introduction

Elder maltreatment or elder abuse is broadly defined 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’ [1]. Within this broad category, there are various types of elder abuse, such as physical (non-accidental acts that result in pain and injury), emotional (screaming, insulting or threatening), financial (illegal use of financial resources) or sexual abuse (unwanted sexual acts) and neglect, which is defined as ‘intentional or unintentional withholding of food, medication or other necessities that result in the older person's failure to thrive’ [2]. In addition, both physical and chemical restraints are considered abusive [3].

The prevalence of abuse and neglect varies based on the type of maltreatment assessed and the population. A recent systematic review reported a prevalence rate of significant elder abuse of 6% among the general population and 25% among vulnerable older adults [4]. Neglect and psychological and financial abuse are the most common types of elder maltreatment [5]. Elder maltreatment is particularly high among physically or cognitively vulnerable older adults [6].

Elder maltreatment is considered a violation of human rights and therefore should be prevented. The urgency to prevent elder maltreatment is further heightened given its well-known negative effects on the quality of life, well-being and physical health of older adults [7]. Elder maltreatment is a known risk for institutionalisation [8] and even for mortality [9].

Although several systematic reviews and/or meta-analyses on the topic exist, these are either narrowly focused on a particular area (e.g. restraint or intimate partner violence) [10, 11] or provide a broad qualitative overview of the field [12, 13]. In addition, some of the reviews [10] employed rigorous review criteria which might be appropriate for medical research, but are often deemed inadequate for public health interventions [14]. Whereas a randomised controlled trial (RCT) is the gold standard in medical research, this may not be the case for public health interventions, which aim to evaluate the effect of a new national policy for instance. Moreover, even when an RCT is conducted, some of the rigorous requirements proposed by medical science, such as treatment concealment, are impossible in social research [15].

The present study was designed to systematically review and appraise current knowledge based on interventions for the prevention of elder maltreatment. This review is designed to guide administrators, clinicians, family caregivers and older adults in making informed decisions concerning the prevention and elimination of elder maltreatment. To the best of our knowledge, this provides the first systematic review and meta-analysis of the entire field of elder maltreatment interventions.

Methods

The study was funded by ESHEL—The Association for the Planning and Development of Services for the Aged. Supplementary data, Appendix 1, available in Age and Ageing online provides a detailed description of the methods. Three independent researchers (S.L., O.G., U.N.) searched the selected databases, using predefined keywords. Subsequent snowballing searches were conducted. All titles and available abstracts were reviewed for relevance by at least two independent researchers (L.A., S.L., O.G.). Disagreements were resolved through a consensus, while relying on predefined selection criteria. Data extraction was conducted independently by at least two investigators (L.A., S.L., O.G.). Disagreements between reviewers were discussed and a consensus agreement was established. Predefined abstraction guidelines were used. We appraised the risk for bias, using pre-specified guidelines. We used Comprehensive Meta-analysis Version 2 [16] to calculate effect sizes and to assess for publication bias.

Results

Figure 1 demonstrates the study flow. A total of 24 studies and 4 additional records reporting on the same participants [17–20] were kept for the systematic review and the meta-analysis. These were broadly grouped into three main categories: (i) interventions designed to improve the ability of professionals to detect or stop elder maltreatment (n = 2), (ii) interventions that target older adults who experience elder maltreatment (n = 3) and (iii) interventions that target caregivers who maltreat older adults (n = 19) (plus four additional records on the same population). Given inherent differences between these three categories, all further analyses were conducted separately within each category. Because different research designs result in different effect sizes [21], pooled effects within each category were calculated based on study design: RCT/clustered RCT, non-RCT and pre–post.

Professionals responsible for preventing elder maltreatment

Two studies: one from Japan [22] and one from the United States [23] fell under this category. One study assessed the impact of a legal policy intervention [22], whereas the other assessed the impact of an educational intervention that advocated for the use of an integrated system of clinical assessment, service planning and outcome measurement in a community mental health centre [23]. Table 1 outlines their characteristics. One study employed a pre–post design [22] and one employed a non-RCT comparative group design, using clustering [23]. Both studies employed a fidelity check (i.e. implementation assessment). Blind assessment was not reported. Effects were non-significant.

Interventions that target older adults who experienced maltreatment

Three studies: from Canada [24], the United States [25] and Iran [26] fell under this category (See Table 1). These studies offered psychoeducational/case management interventions to older adults at risk. One study tested the presence or absence of abuse as an outcome [24], one study assessed readiness to make a change among older adults who had experienced maltreatment [25], whereas the third study focused on perceived threat [26]. Two studies used an RCT design [24, 26] and one relied on a pre–post design [25]. Studies were ranked as high risk or unclear risk with regard to blind assessment and fidelity check. One study [24] had a non-significant effect and another study [26] had a significant effect, disfavouring the intervention. The third study had a significant effect favouring the intervention [25].

Interventions that target carers who maltreat

This was the largest category with 19 studies and 4 additional records based on the same populations [17–20]. Five studies examined interventions that targeted psychological abuse by paid carers [19, 20, 27–31] and one examined physical abuse by family members [32]. The remaining 13 studies examined educational/consultation/structural interventions to reduce physical restraint in institutions [17–20, 31, 33–45]. Because of the inherent differences between studies that target family members as carers versus studies that target paid carers, these two groups were examined separately. In addition, we examined separately studies that targeted psychological abuse versus studies that targeted restraint use as outcomes.

The study that targeted family caregivers was conducted in the United Kingdom and employed an RCT design. This study conducted a fidelity check and blind assessment and accounted for the therapist's effect. Its effect was non-significant [32].

As for the five interventions that targeted paid carers to reduce psychological abuse, three were conducted in the United States [28–30], one in Taiwan [27] and one in the Netherlands [19, 20, 31]. Two employed a pre–post design [28, 30], two employed a non-RCT comparative group design [27, 29] and one relied on a cluster-RCT design that was potentially contaminated [19, 20, 31]. Blind assessment was conducted by two studies [19, 20, 30, 31]. The studies did not conduct a fidelity check. The effects of four interventions were statistically significant [19, 20, 28–31, 46].

Of the studies that assessed restraint use, one study employed a stepped-wedge approach [45], one employed a pre–post design [41], two employed a quasi-experimental design [18, 38, 39] and the remaining studies employed cluster-RCT design. Of the cluster-RCTs, three adjusted for the clustering design [33, 39, 42]. One study conducted a fidelity check/implementation assessment [18]. Blind assessment was reported by most studies [17–20, 31, 33–38, 43, 45]. It is important to note that even though the interventions targeted carers, the studies mainly focused on the details of the care recipients and with very few exceptions [17, 34, 36, 37, 42, 44] have failed to provide information about carers’ characteristics. See Table 1 for study characteristics.

The stepped-wedge design resulted in a non-significant effect [45], whereas the pre–post design [41] resulted in a significant reduction of restraints. The effects of the quasi-experimental studies were significant [18, 38, 39]. As for the cluster-RCTs, three had a significant effect [17, 33, 42, 43], whereas the remaining studies had a non-significant effect, after adjustment for clustering. The pooled effect of these nine studies was significant: standardised mean difference: −0.24, 95% confidence interval = −0.38 to −0.09, indicating that the intervention results in a reduction in restraint use. Homogeneity was low: (Q[df] = 10.76 [8], P = 0.22, I² = 25.7, T² = 0.01). See Figure 2 for details.

There was no evidence for publication bias. The Begg and Mazumdar’s correlation was not significant (Kendall’s τ = −0.25, P = 0.34) and so was Egger’s intercept (0.32, 95% CI [−2.08 to 2.72], P = 0.76). The trim-and-fill method suggested that one study should be imputed, but the overall effect following imputation (−0.25, 95%CI [−0.41 to −0.11]) remained consistent with the overall effect without imputation, providing no support to publication bias.

Discussion

There is a growing body of literature attesting to the importance of preventing elder maltreatment, given its negative effects on the quality of life, well-being, health and mortality of older adults [9]. The present study was designed to systematically review and appraise current knowledge based on interventions for the prevention or reduction of elder maltreatment. The most notable finding of the present review is that although a few thousand studies that address elder maltreatment exist, only 24 studies met our inclusion criteria and contributed to current evidence on the topic. Of these studies, two targeted professionals responsible for preventing maltreatment, three targeted older adults who experienced maltreatment and the remaining 19 studies targeted carers who maltreat. These 19 studies were also quite diverse, with one study targeting family carers, five targeting psychological abuse among paid carers and the remaining 13 studies targeting the use of physical restraint among paid carers. Current findings raise important issues with regard to the need for more funding in this area of research, as well as for better quality monitoring of the research conducted.

A website search of the most prominent elder abuse prevention organisations (e.g. National Center on Elder Abuse, Action on Elder Abuse) clearly demonstrates that one of the most central areas for intervention is the education and training of professionals responsible for the prevention of elder maltreatment. Nonetheless, these strategies clearly lack evidence at the present time as the two studies that targeted professionals responsible for preventing elder maltreatment were both non-significant.

Three of the reviewed studies targeted older adults who experienced maltreatment. These studies were quite diverse in terms of their outcomes. Hence, our current review provides no evidence for such interventions. Limited evidence also exists with regard to interventions that target unpaid caregivers who maltreat, with only one study meeting our inclusion criteria.

Evidence is also limited with regard to interventions that target psychological abuse among long-term care (LTC) paid carers. These interventions offered educational components to increase awareness to communication patterns and some also had specific contents about dementia care, with one study providing training in the use of snoezelen as a means to improve elder care. However, four of the studies employed a very weak design (e.g. pre-post, non-RCT comparison groups) and none of the studies conducted an implementation assessment.

It is also important to note that the outcomes assessed by these studies were not necessarily clear indicators of psychological abuse. The use of ‘elder-speak’ or diminutive language could be hurtful and disrespectful, but its interpretation as abusive is quite subjective. This is contrasted with clearer signs of elder abuse, such as acts of violence, which were not assessed in these studies.

The strongest evidence exists for interventions that target physical restraint among LTC paid carers. Most interventions addressed older adults with dementia. There was no concern for publication bias with regard to these interventions. These studies have largely relied on educational means to improve the care provided to older residents by helping carers to identify alternatives to restraint use and by providing information about the care of older adults with dementia. Some of these interventions also offered a change-agent or an expert who was available for ongoing consultations.

Although these studies suggest a promising direction for intervention, we concur with a recent review on the topic that has questioned the quality of these intervention studies [10]. Most of the cluster-RCT did not account for a clustering effect in their analysis. Moreover, the cluster-randomisation employed by some of these studies is questionable given the low number of clusters that participated in the randomisation process. Most studies did not assess the quality of implementation of the intervention employed. In addition, even though interventions specifically targeted carers, carers’ characteristics were not provided by many of these studies making future replications challenging.

As with any meta-analysis, it is possible that we had missed studies that could have otherwise met our inclusion criteria. Our reliance on studies available in English, given our limited resources, is another limitation. Nevertheless, our study provides a first attempt to systematically review, appraise and analyse the emerging field of elder maltreatment. Our findings call for more rigorous, better quality research in the field. Specific areas that are still lacking evidence at the present time are interventions that target (i) elder neglect, (ii) public awareness, (iii) older adults who experience maltreatment, (iv) professionals responsible for preventing maltreatment, (v) family caregivers who abuse and (vi) carers who abuse. The most effective place to intervene at the present time is by targeting restraint use in institutions.

Future research directions

There are significant national and international efforts to increase awareness to the topic of elder maltreatment (e.g. national elder abuse day) and to make legislative changes to reduce or prevent elder maltreatment, but the effectiveness of these efforts has not been examined scientifically. Given current investment in training programmes for professionals responsible for reducing elder maltreatment and in awareness campaigns, there is an urgent need to systematically assess the effectiveness of these interventions.

In addition, future research should specifically address elder neglect. Although this is the most prevalent form of elder maltreatment according to some statistics [5], none of the interventions reviewed specifically addressed this form of maltreatment. Another area that requires further attention is the care of older adults at home. The majority of older adults live at home [47] and as a result, abuse, most often occurs behind closed doors [48]. Nonetheless, most interventions that targeted paid caregivers did so in institutions and only one study addressed abuse by family caregivers. Hence, the most vulnerable population that receives home care services or unpaid care by family members has received almost no research attention. Further attention should also be paid to the cognitive status of the care recipients. Past research has shown that cognitive status is a major risk for elder maltreatment [49]. However, because the cognitive status of care recipients was not clearly detailed in all studies reviewed, we did not evaluate whether intervention effects vary by cognitive diagnosis.

Conflicts of interest

None declared.

Funding

This research was funded by ESHEL—The Association for the Planning and Development of Services for the Aged.

References