Elder abuse (EA), a single or repeated act including lack of appropriate action occurring within any relationship where there is an expectation of trust and causing harm or distress to an older person,1 is often classified as neglect, psychological, financial, physical, or sexual abuse among other forms.2,3 The elderly are often more susceptible to harm due to their inherent human vulnerability, which is more likely to manifest with advancing age.4,5 Frailty is an age-related clinical state that results from the higher likelihood of developing adverse health-related outcomes under exposure to a stressor that in younger counterparts would be easily tolerated.5,6 Estimates suggest that at a global level, one in six older people is affected by abuse (pooled prevalence of 15.7% according to a large-scale meta-analysis),3 and it is related to higher levels of depression,7–9 suicidal ideation,9 anxiety disorders,8 post-traumatic stress disorder,8 sleep disturbances,9 chronic pain,9 metabolic syndrome9 and poor general health.8,9 There is more frequent hospitalization,10–12 nursing home placement,12 and increased disability and mortality9,13 among older adults who suffer abuse. Furthermore, according to the available data, in the institutional setting, the estimate of the overall abuse rate self-reported by staff was 64.2%.14 Despite the associated health impact, the problem often goes unrecognized with little awareness,15,16 and health care professionals report a lack of training and knowledge on the matter.16 A qualitative study showed that a large percentage of health care professional students did not perceive all subtypes as a case of abuse.15

Several systematic reviews have attempted to estimate the worldwide prevalence of EA, but a PubMed search for reviews of the topic in January 2023 did not reveal any overview of reviews published to synthesize all relevant studies. Overviews of reviews, also known as umbrella reviews, are essential tools in evidence synthesis, making it possible to determine if the findings of systematic reviews on a certain topic are consistent or contradictory, by providing a broader picture of the topic.17 Such an umbrella review on the prevalence of EA could give us a better understanding of the magnitude of and the variation in the global health burden of the problem to help policymakers.

This umbrella review aimed to estimate the worldwide overall prevalence rate of EA in community and care settings. As an additional objective, we aimed to stratify the estimation according to the type of abuse.

The umbrella review was prospectively registered in PROSPERO (registration number: CRD42021281866), and written to meet the guidance covered in the Preferred Reporting Items for Overviews of Reviews (PRIOR) statement18 (Appendix A).

Our umbrella review covered 16 systematic reviews and drew data from 136 primary studies to synthesize scientific evidence. We chose this method, as various systematic reviews had already been published, but no overview consolidated the findings of these. This umbrella review on the prevalence of EA is the first of its kind, prospectively registered and reported using a checklist.28 Overviews of reviews are often considered as the highest level of scientific evidence,46 thus filling in an important research gap. The reviews were largely of low quality. EA was a prevalent problem (1.1–78%), with a wide dispersion on a global level, both in the community and care setting. A large difference was observed between the prevalence of the different subtypes.

As this is the first umbrella review of the prevalence of EA, it is not possible to compare our results with measures obtained by the same research design. When comparing with previous studies, it is natural that our reported prevalence range is wider, as our estimate includes all the extreme values reported by any previous systematic reviews (and primary studies indirectly). Prevalence ranges reported by previous studies, such as the commonly cited values of Yon et al. (15.7%, 95% CI: 12.8–19.3)3 are covered in our overall range. Greater precision in findings can be achieved with meta-analysis, but this technique is not so far developed for umbrella reviews. The added value of our work is that it provides a broader picture of the phenomenon. An important aim of overviews is to determine if systematic reviews come to the same or contradictory conclusions. Thus, to interpret our results, it is worth not only to examine the overall range of EA, but the included reviews separately as well. The differences between the design of the included systematic reviews could be a potential source of heterogeneity, as they were aiming to capture different phenomena, such as elder abuse in community setting, in care setting, in rural areas, perpetrated by intimate partners, measured in specific geographical locations, etc. Some did perform meta-analysis, while others were reporting data ranges. They also used different techniques to assess the risk of bias, half of which (8/16) did not meet the criteria set by the modified AMSTAR-2 checklist, while the other half scored partial yes. Apart from the differences between reviews, other potential sources of heterogeneity could be the differences between primary studies (e.g. sample size, sampling method, data collection, EA definition, prevalence measure period). Reviews reported an association between abuse and geographical area (e.g. WHO regions, income levels) and cultural context.

The exhaustive search strategy we applied can be considered as one of the main strengths of this study. We searched eight electronic databases, covering a wide range of relevant disciplines, including biomedical-, health-, social- and behavioural sciences. Some of the databases apart from peer-reviewed journals also included citations for dissertations and book chapters, thus covering grey literature as well. To avoid language bias, neither our search strategy, nor the inclusion criteria included language restrictions. By overviewing texts in Chinese and Portuguese, we were able to increment the number of included systematic reviews by three. In addition, by not excluding systematic reviews of special populations of elderly (disabled, dementia), we were able to broaden the spectrum of the included evidence, while by reporting those separately from the other reviews we aimed to capture a prevalence estimate generalizable to the whole population.

One of the limitations of our review is that the quality appraisal instrument we used (AMSTAR-2) was primarily designed to evaluate systematic reviews of clinical trials.28 A modification of the original checklist was necessary for our research. AMSTAR-2 for example does not have any item to measure if systematic reviews reported the sampling methods used by the primary studies, which in prevalence studies is equally important to research design. In our modified checklist, we included this aspect in the assessment of risk of bias as a subitem. We must also note that while AMSTAR-2 evaluates the quality of systematic reviews, it does not evaluate the quality of the included primary studies. The overall evaluation of the included reviews depended on the number of critical weaknesses, not on the global score. As seen above, none of the review authors provided a list of excluded studies after full-text reading, and neither reported the funding of the primary studies. It is important to interpret the acquired estimates of prevalence data with caution, because of the low confidence in the results of the included reviews (in most cases assessed as “critically low”, and in two cases as “low”). We also acknowledge that subjectivity is always involved in critical appraisal by its nature, and other authors might have gotten different results using the same checklist. The low quality of the included reviews is one of the main limitations of our overview. Data extraction revealed that some authors even reported their results with contradictory data, a mistake that should be avoided in scientific writing (see footnotes of Table 1). AMSTAR-2 – which is a checklist to evaluate methodology – did not assess these trivial mistakes. Another difficulty we encountered during our research was defining systematic reviews. During our screening for relevant studies we excluded some reviews that were denominated as systematic reviews by the authors,47,48 however, they did not meet our operational definition detailed in the “Methods” section. We also identified some studies that were originally not published as systematic reviews but met our inclusion criteria. In the latter case, terms like “review”42 or “analytic review”41 were used. We recommend authors and journals to use an operational definition based on an expert consensus when we talk about systematic reviews.49

Despite the high variation in the results, we can conclude that EA is an alarmingly frequent phenomenon, and according to Yon et al.3 it affects 141 million people on a global level. With the ageing of the global population, authors predict a further rise in the magnitude of the problem. The fact that the transcendence and public health impact of the problem keeps being high despite the recent attention of the scientific community, should be a call for action for policymakers. As we saw, EA affects elderly people both in the community and the healthcare setting, and as the complex problem it is, it requires a multidisciplinary solution from experts in various areas.50 We would like to emphasize the importance of primary care providers and family physicians in this process, as they are in frequent contact with seniors, and they often build a deep relationship of trust with them.51 The education of health care professionals on the matter must be bettered both in the curricular and institutional levels, and population wide prevention strategies and screening programmes based on scientific evidence should be developed and implemented. We think primary care physicians and nurses will have a key role in such strategies.

Although we can consider that the phenomenon is widely studied (between the years 2007 and 2022 we identified 16 systematic reviews and 136 primary studies), there are still research gaps. Primary studies from low- and middle-income countries were reported to be scarce. In regions where the phenomenon is under-researched, there is a need for reliable primary studies using validated instruments. In the future, already existing systematic reviews should be updated with recent findings. We only identified one review focusing specifically on the general population in the institutional setting,14 according to which institutional abuse is also widespread, but under-researched. Regarding the methodological aspects of evidence synthesis methods, our recommendation for future research is to make a new instrument or a validated adaptation of AMSTAR-2 for prevalence studies. As we observed, poor reporting of findings was also a common problem, including data discrepancies. The solution to better the quality of reporting would be to strengthen the integrity of biomedical research. We further endorse the rising attention of the scientific community in recent years dedicated to the development of integrity guidelines, which could help authors to reduce erroneous writing.52–54

This umbrella review is the first overview of systematic reviews measuring the prevalence of EA, and identifies important research gaps and implications for decision-makers. It demonstrates clearly that EA is an alarmingly frequent problem, although the exact prevalence data is hard to estimate, due to the overall low quality of the identified systematic reviews. Prevalence was different throughout the subtypes of EA, neglect being the most common form, followed by psychological-, financial-, physical- and sexual abuse. Evidence showed that the problem of elder abuse in care setting is just as relevant as in the community.

Our results imply that the education of health care professionals on the matter, and the prevention and early detection of EA should be bettered, ideally starting at the level of primary health care.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

The authors declare that they have no conflict of interest.