Health needs of older populations affected by humanitarian crises in low- and middle-income countries: a systematic review

Background The combination of global demographic changes and a growing number of humanitarian crises in middle-income countries that have a higher life expectancy has led to an increase in the number of older populations affected by humanitarian crises. The aim of this review was to systematically examine evidence on the health needs of older populations in humanitarian crises, including both armed conflicts and natural disasters, in low- and middle-income countries (LMICs). Methods A systematic review methodology was used. The search strategy used terms related to older populations and humanitarian crises in LMICs. Five bibliographic databases were used, along with relevant grey literature sources. Descriptive analysis was used, and a quality assessment conducted using the Newcastle-Ottawa Scale and CASP instruments. Results A total of 36 studies were eligible for review. The majority of the studies were cross-sectional, three were cohort studies, and four used qualitative methodologies. The main health outcomes were mental health, physical health, functioning, and nutrition. Vulnerability factors included older age, female gender, being widowed, increased exposure to traumatic events, prior mental health problems, low income and education, and rural residency. Ten studies addressed the responsiveness of health systems and access to such services. The quality of the included studies was generally low. Conclusions There is an urgent need to strengthen the evidence base on the health needs of older populations in humanitarian crises. Electronic supplementary material The online version of this article (10.1186/s13031-017-0133-x) contains supplementary material, which is available to authorized users.


Background
The proportion of older people is growing faster than any other age group globally. Approximately 12% of the world's population is aged 60 and over and the number of older people is estimated to surpass 1 billion by 2020. By 2050, there will be nearly as many people aged 60 and over as children aged under 15 [1]. Currently, twothirds of the world's older people live in low-and middle-income countries, which is where humanitarian crises are more likely to occur and where the humanitarian impact is greater.
An estimated 172 million people (all ages) are currently affected by armed conflict worldwide, [2] including over 59 million people forcefully displaced from their homes as either internally displaced persons (IDPs) or as refugees. Natural disasters are also estimated to affect 175 million people annually [3]. The combination of global demographic changes and a growing number of humanitarian crises in middle-income countries with higher life expectancy has led to an increase in the number of older populations affected by humanitarian crises [4][5][6].
Older populations are more likely to be disproportionately affected by humanitarian crises [7]. Older age is associated with increased likelihood of disability and ill health which can limit functioning and physical mobility, and impede access to health services. Ageing also increases dependency on others for financial and social support. These collective vulnerabilities put older populations at a higher risk during humanitarian crises when health risks are increased and support networks and existing social infrastructure compromised [8]. Specific health risks for older populations in humanitarian crises include: greater susceptibility to ill health, malnutrition, disability and injury; difficulties in accessing health services (including psychological services); inappropriate health services such as services not addressing noncommunicable diseases which older people are more likely to suffer from; failure to collect data on health needs of older people; and broader social and economic marginalisation [9,10].
While older populations are recognised as a vulnerable group in humanitarian crises, [11,12] the particular needs of older populations in humanitarian crises appear poorly understood [13]. Reviews have been conducted on crisis-affected older populations, [14][15][16] but these have not been systematic, have focused on natural disasters only, and on high-income countries where the health needs and health sector resources and responses are likely very different compared to low-and middleincome countries (LMICs) where the vast majority of crisis-affected populations live.
The aim of this review was to systematically examine evidence on the health needs of older populations in humanitarian crises in LMICs. The specific objectives were to: identify the vulnerabilities of older populations in humanitarian crises; assess health service access and responsiveness for older populations in humanitarian crises; and evaluate the quality of the evidence.

Eligibility criteria
The population of interest were older populations affected by humanitarian crises in LMICs (with LMICs classified according to Word Bank listings [17]). No age limit was set as the definition of 'older' varies across country contexts. The study population included refugees, returnees, IDPs, and non-displaced crisis-affected people. Humanitarian crises were defined as a serious disruption of the functioning of a community or a society causing widespread human, material, economic or environmental losses which exceed the ability of the affected community or society to cope using its own resources, necessitating a request to national or international level for external assistance [18]. Humanitarian crises included both armed conflict and natural disasters [19]. Natural disaster events included earthquakes, tsunamis, floods, hurricanes, landslides, and volcanic eruptions (see Additional file 1 for the full list of events). All health outcomes were included. Research on military or veteran military populations was excluded, as were studies of an older population that had experienced a crisis at a younger age. Studies of all-age populations showing age as a risk factor but not focusing specifically on older populations were excluded.
Primary published and grey literature using quantitative and qualitative methods were included. All languages were included. No date restrictions were set (the end date was 18 July 2016).

Search strategy
The following bibliographic databases were used: Medline, Embase, Global Health, Psychinfo, and IBSS. The search methodology consisted of three strings, with terms related to LMICs, humanitarian crises, and older populations. Free-text searching was used, and subject heading (MeSH) were also used for Medline. The search terms are listed in Additional file 1. Broad search terms such as 'elderly' and 'humanitarian' were applied to the Desastres database (mixed published and grey literature) and also to the following grey literature sources: UNHCR, MSF Field Research, HelpAge International, Handicap International, International Committee of the Red Cross (ICRC), WHO Institutional Repository for Information Sharing (IRIS), Open Grey, ReliefWeb, PsycEXTRA, ALNAP, and Google (first ten pages only).

Study selection and data extraction
Study selection involved a four stage process: removal of duplicates (stage 1); screening by title (stage 2a) and abstract (stage 2b) and then full text (stage 2c); grey literature screening and review of the reference lists of the final selected studies (stage 3); and final review and analysis of the selected studies (stage 4).
The information extracted from the final selected studies included: author/date, location, crisis/population type, older age definition, methods, health outcomes/ measurement, and findings that related to the three study objectives. Where both bivariate and multivariate analyses were performed, only multivariate results were extracted. In relation to objective one, where statistical significance tests were used, only results that were considered statistically significant (p < 0.05) were extracted. The study screening, data extraction, and quality assessment was conducted separately by EM and JS and any differences discussed and reconciled.

Analysis and quality assessment
Descriptive analysis was used given the heterogeneous nature of study context, population exposure, health outcomes, and study methodologies. Findings were organised by the three study objectives, and then into commonly recurring themes. For quality appraisal, quantitative studies were appraised using the Newcastle-Ottawa Scale (NOS), [20]. with cohort studies given a score of 1-9, and cross-sectional studies given a score of 1-10 (using a modified NOS version for cross-sectional studies) [21]. For qualitative studies, the Critical Appraisal Skills Program (CASP) checklist was used, [22]. with studies given a score of 1-10. Higher scores in the quality appraisals indicate better quality. The quality appraisal process sought to identify common strengths/weaknesses, rather than to exclude studies. This review follows the PRISMA Statement for reporting systematic reviews (see Additional file 2 for the completed PRISMA checklist) [23] Results

Study selection and characteristics
Thirty-six studies met the eligibility criteria,  of which two were from the grey literature [37,40] (Fig. 1). The most common reasons for exclusion at stage two were studies not reporting: primary research, populations in LMICs, specifically on older populations.
Being widowed, unmarried, single or separated were associated with PTSD, [33] depression, [48] and worse psychological quality of life [26,56,59]. Reduced social support was associated with PTSD, [33] depressive symptoms, [42] quality of life, [59] (all in China) and anxiety (refugees in Lebanon), [52] as was a reduced sense of community with depressive symptoms (China) [42]. Regular religious attendance was associated with reduced probability of depression among refugees in Lebanon [30].
One study following the 2003 Bam earthquake in Iran observed that rural residents scored a higher psychological quality of life than affected urban residents [26]. However, rural residents were more likely than urban residents to report sleeplessness and a feeling of depression or helplessness after the 2005 Kashmir earthquake in Pakistan [31].
Exposure to crises, traumatic events and forced displacement Five studies observed greater intensity of exposure to crises increased the risk of PTSD, [36,41,58] depression, [42,58] anxiety, [58] worse psychological quality of life, [26] and psychological distress [41]. Three studies showed an association between bodily injury from a crisis exposure (most commonly in the context of a natural disaster) with PTSD, [33] anxiety [58] and worse psychological quality of life [26]. Three studies with earthquake and Tsunami survivors in Sri Lanka, Thailand, and China reported the effects of loss, disability or injury of a family member on PTSD, [44]. depression, [48] psychological distress, [29] and quality of life [59].
Three studies assessed PTSD levels at 1 year, [58] 15 months, [39] and 3 years [33] after the Wenchuan earthquake in China, and observed PTSD remained high many months and years after the earthquake. The study in Lebanon of long-term Palestinian refugees and shorter-term Syrian refugees found that Palestinian refugees had higher levels of depression and experiencing fear than Syrian refugees (the time period of displacement was not recorded in the study but we have assumed that Palestinian refugees had been displaced for a longer time than Syrian refugees given their histories of forced displacement) [52]. A qualitative study in Iran found that older populations experienced a significant amount of emotional distress years after the Bam Earthquake, and they found it difficult to move on from the earlier crisis events [25]. Conversely, a study in Georgia found that IDPs displaced for a shorter period of time were more susceptible to depression [40].
Forced displacement and dissatisfaction with current living conditions after a crisis was related to worse psychological quality of life [26] and psychological distress [29] among earthquake survivors in Iran and China respectively, [26,29] and anxiety disorder among Tsunami survivors in India [53]. However, the 2004 Andaman and Nicobar study reported that remaining in the crisis-affected area increased the likelihood of suffering from adjustment disorder [53].
Demographic and socio-economic factors Older age was associated with lower physical quality of life among earthquake and flood survivors in Iran and China, [26,56] lower physical functioning in Iran, Rwanda, and Syrian refugees in Lebanon, [27,45,52] higher mortality risk among Ethiopian refugees in Sudan and earthquake survivors in China, [35,54] worse nutritional outcomes, [24,28,45] worse clinical outcomes (except for oliguria) among patients with traumatic injuries following the 2008 Sichuan Earthquake in China, [57] and higher intra-operative mortality in 21 countries [55].
Female gender was associated with lower physical quality of life [26] and physical functioning among earthquake survivors in Iran, [27] and with cardiovascular and all-cause mortality among war-affected persons in Lebanon [50] Low education was associated with worse physical functioning among earthquake survivors in Iran [27] and self-reported health status among refugees in Lebanon [52]. Lower socio-economic status was associated with a higher prevalence of chronic energy deficiency following a period of severe drought in India [28]. The loss of property had a greater mortality risk for war-affected men in Lebanon [50].
Being single, divorced, widowed or separated increased the risk of death from cardiovascular disease and all-cause mortality in Lebanon [51] and worse self-reported health among flood survivors in China [56]. A qualitative study with Rwandan refugees in Tanzania found that older populations perceived that those who lived alone and had no family or spouse to care for them were at the greatest risk of poor nutrition, citing reduced income and inadequate support networks [47]. Conversely, living with others was associated with a worse physical functioning score in the older Bam earthquake survivors in Iran [27]. The study of Palestinian refugees in Lebanon observed living in a larger household size was associated with worse functional status [52].
A study of survivors of the 2005 Kashmir earthquake in Pakistan observed a higher prevalence in rural areas than urban areas of dental, visual, eating, and hearing problems, headache, dizziness, muscle and joint pains, and of established yet untreated medical problems [31].
Trauma exposure and forced displacement The study from Lebanon observed excess risk of both cardiovascular and total mortality following human loss (deaths of close relatives/friends, injuries, kidnappings, and serious threats) among women, and of cumulative exposure to war events among men and women [50]. Flood-affected populations in China reported lower health related quality of life compared to non-flood affected populations [56]. The study of Ethiopian refugees in Sudan observed mortality rates markedly increased one year after migration compared to the pre-migration period [35]. The study of displaced Palestinian and Syrian refugees in Lebanon observed longer-term Palestinian refugees were more likely to suffer from NCD's, poor physical functioning, physical limitations and impaired vision and hearing when compared to shorter-term Syrian refugees [52]. However, a study of Mozambican refugees reported lower poorer nutritional status among those who had been displaced for a shorter period of time [49].
Health-related factors One study in the Democratic Republic of Congo observed that taking no physical exercise, taking multiple prescription drugs, and limited mobility and functioning were associated with malnutrition [24]. The study with Rwandan refugees in Tanzania reported that malnutrition had a negative effect on physical functioning in terms of handgrip strength [46]. The related qualitative study reported perceptions that older people who were physically impaired were at greater risk of poor nutrition due to reduced income [47]. The study of flood survivors in China reported poor sleep patterns, diagnosed chronic disease, and hospitalisation in the preceding year were all associated with poor physical health [56]. The study of refugees in Lebanon noted dementia, poor vision, difficulty walking, poor self-reported health status were associated with lower functional status [52].

Health service access and responsiveness
Ten studies examined aspects of health service access and responsiveness for older populations (Table 3) [25,27,31,32,34,39,40,43,52,56]. The majority of these studies were based on descriptive self-reporting, with no statistical tests.
Five studies reported how older populations had difficulty accessing medical services [27,31,34,40,52]. Reasons included: a lack of financial resources for treatment and transport; [34,40,52] the systematic exclusion of older populations from programmes targeting other groups; [31,34] limited knowledge about appropriate facilities; [52] an absence of outreach programmes; [34] and inability to travel to clinics [52]. The study of Syrian and Palestinian refugees in Lebanon reported over 97% of older populations experienced difficulties accessing medical services and medicines [52].
Four studies assessed health service utilisation [31,39,43,56]. Rural residents in post-earthquake Kashmir were less likely to utilise health services than urban residents, particularly womenwith clinician gender playing an important role [31]. A study in eastern Democratic Republic of Congo found a very small proportion (3.3%) of older populations utilised health services when they were sick [43]. A study of flood-affected residents in Bazhong in China found the two-week healthcareseeking rate was significantly higher than non-flood affected older populations in Sichuan province [56]. The study of surgical outcomes in 21 countries observed a lower proportion of urgent surgical cases when compared to younger age groups (<50 years); and the most commonly performed surgical procedures for older people included herniorrhaphies, simple and extensive wound debridement, abscess incision and drainages, minor tumorectomies, and urological procedures [55].
Four studies addressed the responsiveness of health services to the needs of older populations [25,31,32,34]. The qualitative study of Bam earthquake survivors in Iran revealed they perceived services to be inappropriate, with a lack of respect paid to the needs and dignity of older people [25]. Another qualitative study of survivors of the 2004 Tsunami in Sri Lanka observed that older populations felt that they were not adequately consulted about their specific needs [34]. The two remaining studies assessed service responsiveness from the perspective of health service providers after the 2005 Kashmir earthquake in Pakistan, and found that many of their medical problems were undertreated, [31] and the level of awareness of the special needs of older populations was inadequate among all types of healthcare providers [32].

Quality of the evidence
A commonly recurring issue with the quantitative studies was the limited statistical analysis, including a substantial proportion of the studies only performed descriptive bivariate analysis and so could not control for potential confounding [24, 28, 29, 31, 35-38, 40, 42, 45, 48, 49, 52-56]. Only four studies reported descriptive prevalence without calculating confidence intervals or statistical significance tests where it would have been appropriate [31,32,35,43]. Many of the studies did not justify their sample size, and non-response rates were rarely reported. Furthermore, the representativeness of some study populations was negatively affected by suboptimal sampling strategies [24,29,36,46,48,49,52,53,59]. Many of the included studies did not employ comparison groups, making it difficult to interpret whether a particular factor was more likely to influence an outcome in older populations than in the general population. Of the studies that did include a comparison group, the selection process was often poorly justified [36,52]. Inadequate justification was also given for the selection of particular outcome measures. Among the qualitative studies, a superficial engagement with the role of the researcher and their subjectivity was a common weakness. The scores for the quality appraisal of individual studies are given in Tables 2 and the detailed results provided in Additional file 3.

Discussion
This is the first systematic review to examine the evidence related to the specific health needs and vulnerabilities of older populations affected by humanitarian crisis in LMICs. The majority of the 36 studies meeting eligibility criteria were cross-sectional in design, restricting our ability to imply causation between vulnerability factors and health outcomes. In light of the limited breadth and quality of evidence, the following findings should be treated with caution. This review identified that older age, female gender, socio-economic deprivation and rural residency were frequently associated with adverse health outcomes, reflecting findings from elsewhere for mental health, [60,61] and nutrition [62]. The influence of female gender with worse health outcomes is consistent with existing research in stable settings and highlights the importance of genderdisaggregated data and further research on older women's health needs in humanitarian crises [61,63,64]. The discrepancy in health outcomes between urban and rural areas is particularly concerning given that the majority of older populations in low-income countries live in rural areas [65]. Many of these risk-factors, particularly for mental health outcomes, are similar to those in all-age adult populations affected by humanitarian crises [61,66,67]. The limited number of studies on non-communicable diseases is also surprising given their higher burden among older people and increasing concern about noncommunicable diseases in humanitarian crises [68].
The limited quantity and quality of research can be partly attributable to the inherent complexity of providing services and conducting research during humanitarian crises, but such research has been successfully undertaken with other population in humanitarian crises [69]. We identified no intervention studies on the effectiveness of existing health interventions specifically with older populations. As the context of humanitarian crises can make randomized control trials difficult to carry out (though by no means impossible [69]), quasiexperimental methods and variants such a stepped-wedge approaches could be used to gain a fuller understanding of the effectiveness (and cost-effectiveness) of health programs in meeting the needs of older populations in humanitarian crises. Routine service data could also be more effectively utilised, but this is currently hampered by the common absence of routine age-disaggregated data for older populations [70,71]. There also needs to be considerably more qualitative research to better understand the perspectives of older populations and health care providers.
In addition to the above research recommendations, humanitarian agencies should consider ways to strengthen their work and capacity to better understand and address the health needs of older people. This includes strengthening and adhering to best practice guidelines for older people in humanitarian crises [9]. UNHCR's Accountability Framework for Age, Gender and Diversity Mainstreaming [72] provides some information on activities for older people but much more detailed and rigorous data reporting is required. This necessitates the collection of age disaggregated routine data (as done by the Office of U.S. Foreign Disaster Assistance (OFDA) which requests disaggregated data for older age groups of 50-59 and then 60+) as well as specific data on the health needs of older people. Other activities include more training and sensitisation for humanitarian health workers on detecting and reporting the health needs of older people. This all requires substantially greater financial investment given the negligible number of funded aid projects specifically for older people in humanitarian crises. For example, of 16,221 humanitarian projects implemented between 2010 and 2014, only 74 projects were funded which included at least one activity specifically targeting older people [7]. This review has highlighted considerable weaknesses in the quantity and quality of research on the health needs of older people in humanitarian crises. While recognising the inherent constraints of humanitarian settings, the lack of research does suggest low levels of awareness and prioritisation of the needs of older populations among the heath care actors and researchers in humanitarian crises.

Limitations
For the quality review, the NOS does not employ weighted scores for different categories and so studies can receive a strong score while still failing to consider important factors such as the representativeness of the study sample. Bivariate results were extracted where multivariate analysis was not conducted, and so these do not adjust for potential confounders.