Systematic review on chronic non-communicable disease in disaster settings

Abstract

Background

Non-communicable diseases (NCDs) constitute the leading cause of mortality globally. Low and middle-income countries (LMICs) not only experience the largest burden of humanitarian emergencies but are also disproportionately affected by NCDs, yet primary focus on the topic is lagging. We conducted a systematic review on the effect of humanitarian disasters on NCDs in LMICs assessing epidemiology, interventions, and treatment.

Methods

A systematic search in MEDLINE, MEDLINE (PubMed, for in-process and non-indexed citations), Social Science Citation Index, and Global Health (EBSCO) for indexed articles published before December 11, 2017 was conducted, and publications reporting on NCDs and humanitarian emergencies in LMICs were included. We extracted and synthesized results using a thematic analysis approach and present the results by disease type. The study is registered at PROSPERO (CRD42018088769).

Results

Of the 85 included publications, most reported on observational research studies and almost half (48.9%) reported on studies in the Eastern Mediterranean Region (EMRO), with scant studies reporting on the African and Americas regions. NCDs represented a significant burden for populations affected by humanitarian crises in our findings, despite a dearth of data from particular regions and disease categories. The majority of studies included in our review presented epidemiologic evidence for the burden of disease, while few studies addressed clinical management or intervention delivery. Commonly cited barriers to healthcare access in all phases of disaster and major disease diagnoses studied included: low levels of education, financial difficulties, displacement, illiteracy, lack of access to medications, affordability of treatment and monitoring devices, and centralized healthcare infrastructure for NCDs. Screening and prevention for NCDs in disaster-prone settings was supported. Refugee status was independently identified both as a risk factor for diagnosis with an NCD and conferring worse morbidity.

Conclusions

An increased focus on the effects of, and mitigating factors for, NCDs occurring in disaster-afflicted LMICs is needed. While the majority of studies included in our review presented epidemiologic evidence for the burden of disease, research is needed to address contributing factors, interventions, and means of managing disease during humanitarian emergencies in LMICs.

Background

Non-communicable diseases (NCDs) constitute the leading cause of mortality globally, accounting for 70% of deaths worldwide [1]. This percentage is projected to rise in the next fifteen years, with the steepest increase in morbidity and mortality from NCDs projected to occur in Low and Middle-Income Countries (LMICs). The World Health Organization (WHO) projects a 10% rise in mortality in Africa from NCDs in from 2015 to 2030 [2]. This rise in NCDs in LMICs coincides with an increasing burden of humanitarian disasters [3].

The International Red Cross defines a disaster as: “a sudden, calamitous event that seriously disrupts the functioning of a community or society and causes human, material, and economic or environmental losses that exceed the community’s or society’s ability to cope using its own resources” [4], and can be divided into: mitigation, preparedness, response, and recovery phases [5]. The United Nations Office for Disaster Risk Reduction (UNISDR) recorded over 1.35 million people killed by natural hazards between 1997–2017, with disproportionate mortality in LMICs [6]. Poverty, rapid urbanization, inadequate infrastructure, and underdeveloped disaster warning and health systems are all contributors to morbidity and mortality in disasters [6, 7].

According to the UNHCR Global Trends Report, an unprecedented 79.5 million people are estimated to have been displaced from their homes as internally displaced persons (IDPs) or refugees in 2019—the largest figure ever recorded [8]. The scale of humanitarian disasters has increased in recent decades for two primary reasons. Firstly, the frequency and ferocity of natural disasters are increasing due to climate change [9]. Secondly, the number of refugees, displaced persons, and migrants are at an all-time high due to the unprecedented refugee crises in Syria, Iraq, and the Democratic Republic of Congo [10]. Disasters may directly exacerbate NCDs through effects such as increased stress levels [11], exposures such as inhalation of substances that trigger worsening of pulmonary disease [12], and exacerbation of underlying disease secondary to limited access to care [13].

Despite the growing burden of humanitarian crises with increasing populations at risk for morbidity and mortality from NCDs, primary focus on the topic is lagging. It is essential to better understand the effect of disasters on NCDs in LMICs as the mortality and morbidity are projected only to increase given climate change and population growth in vulnerable areas [14]. In this context, we conducted a systematic review on the effect of humanitarian disasters on NCDs in LMICs assessing epidemiology, interventions, and treatment. While a limited number of articles have reviewed interventions for NCD management [15, 16], a single NCD disease type [17, 18], or a single geographic region in disaster settings [18,19,20,21], to our knowledge, this is the first systematic review of its kind cross-cutting both regions and disease type. Our aims are to guide allocation of resources, future research, and policy development.

Methods

An experienced medical librarian performed a comprehensive search of multiple databases after consultation with the lead authors and a Medical Subject Heading (MeSH) analysis of key articles provided by the research team.

Eligibility criteria

In each database, we used an iterative process to translate and refine the searches. English, Arabic and French language articles were eligible based on these languages being spoken frequently in LMICs, our team’s language capabilities, and so as not to limit solely to English language articles and potential reporting bias as a result [22]. The formal search strategies used relevant controlled vocabulary terms and synonymous free text words and phrases to capture the concepts of noncommunicable, chronic and noninfectious diseases, and different types of humanitarian emergencies including natural disasters, armed conflicts, terrorism, and failed states (see Additional file 1).

Information sources

The databases searched were MEDLINE (OvidSP 1946-August Week 2 2015), MEDLINE (PubMed, for in-process and non-indexed citations), Social Science Citation Index, and Global Health (EBSCO).

Search strategy

We included studies conducted in LMICs investigating non-communicable diseases in the context of humanitarian emergencies; LMICs were categorized as outlined by The World Bank [23]. Studies conducted in high income countries (HICs) and review articles were excluded. Mental health and associated terms were not included in this review given evidence on the disease burden in existing literature [24,25,26,27,28] and our own research question which sought to address the leading four NCDs (cardiovascular disease, diabetes, cancer and chronic respiratory disease) as outlined by the WHO [29]. No other restrictions on study type were applied. The original searches were run August 10, 2015 and were rerun on December 11, 2017. No date restrictions were applied such that any publication prior to this date was potentially eligible for inclusion. The full strategy for PubMed is available in the Additional file 1. The study is registered at PROSPERO (CRD42018088769).

Selection process

Retrieved references were pooled in EndNote and de-duplicated to 4,430 citations. Two separate screeners independently evaluated the titles, abstracts and full text of the eligible articles (RB and LW), with vetting by a third reviewer (CN). The flowchart per PRISMA is presented in Fig. 1. An assessment of the risk of bias of included studies is provided in tabular format in the Additional file 1.

Fig. 1

PRISMA Flow Diagram

Study risk of bias assessment

Bias was evaluated using the Newcastle–Ottawa scale for assessing risk of bias given majority observational studies in our findings [30].

Results

We retrieved a total of 4,430 references. Four thousand three hundred forty-two studies were excluded by title or abstract, and 158 articles were read in full. Out of the studies screened by full text, 85 studies are included in the final thematic analysis (Tables 1, 2, 3, 4, 5, 6, 7, 8, 9 and 10; Fig. 2), with increasing publications on the topic over time (Fig. 2). For ease of review, we have presented the results by disease type (Tables 1, 2, 3, 4 and 5; Fig. 3) including summaries on study type as well as epidemiology of disease addressed. We felt that the study design would be relevant, in addition to the disease focus, in order to elucidate opportunities for future research based on study approaches that were lacking. The diseases types are split into five categories, which consist of the lead four NCDs in order of burden [29]: cardiovascular disease (CVD), cancer, chronic respiratory disease, diabetes, and a section on other NCDs (defined as those identified in our results that assessed NCDs not fitting into one of the lead four categories). We have also grouped the articles by region, and those results have been presented in tabular format and graphically (Tables 6, 7, 8, 9 and 10; Fig. 4). We present the results on interventions in detail elsewhere [31].

Table 1 Characteristics of included publications by disease type: Cardiovascular Disease

Table 2 Characteristics of included publications by disease type: Cancer

Table 3 Characteristics of included publications by disease type: Chronic Respiratory Disease

Table 4 Characteristics of included publications: Diabetes Mellitus

Table 5 Characteristics of included publications by disease type: Other Non-Communicable Diseases

Table 6 Characteristics of included publications by region: Africa

Table 7 Characteristics of included publications by region: Region of the Americas

Table 8 Characteristics of included publications by region: Eastern Mediterranean Region

Table 9 Characteristics of included publications by region: Western Pacific

Table 10 Characteristics of included publications by region: Southeast Asia

Fig. 2

Number of Publications by Year

Fig. 3

Number of Publications by NCD

Fig. 4

NCD Heat Map

Cardiovascular disease

Regarding overall number of publications, cardiovascular disease was the most commonly studied NCD after diabetes, and 29 studies addressed this (Table 1, Fig. 1). Syrian refugees were the most commonly studied population among studies addressing CVD (Table 11) [44, 52, 53, 83].

Table 11 CVD Key Findings 

Prevalence of disease was high [32, 40, 52], as demonstrated by Sibai et al. in a community-based cross-sectional study of residents of Beirut, Lebanon with circulatory diseases accounting for nearly 60% of diagnoses, and ischemic heart disease was the leading diagnosis [52]. They also demonstrated that strokes had the second highest case fatality rate (54%), which was second only to sepsis (60%). However, most studies assessed cardiovascular disease risk factors, or intermediate risk factors [68], as opposed to actual diseases such as heart attack or stroke. Among cardiovascular disease studies, only two studies primarily addressed strokes [43, 46].

As highlighted, only two studies primarily addressed strokes [43, 46]. In one of the studies, which was conducted two years after the 1991–97 Croatia War in the Baranya region of Croatia, they found 513 stroke cases in a single-site emergency department study [46]. The patients had an average age of 68.4y, with an age range from 25-91y, and a near equal distribution of the cases between men and women (51.7% male). Only 50.6% of patients presented within 6 h, another 16.2% presented after 24 h [46], paresis, speech impairment and vision impairment were the most common presenting symptoms. 38.4% died in hospital. 85.8% of patients had hypertension, 27% had diabetes, 44.6% had hyperlipidemia and 46% also had cardiac disease.

As far as risk factors, hypertension was the lead CVD risk factor in several studies [42,43,44, 47, 50, 53, 54, 112], and reason for presentation for care among refugees. This was evident for Iraqi refugees in Jordan where, for adults 18 and older, primary hypertension was the top diagnosis [48]. However, blood pressure control remains a problem, as demonstrated among victims staying in temporary shelter more than 1 year after a 2008 earthquake in the Sichuan province of China, where only half of those diagnosed had medications (53.4%) [54] and less than one in five (17.8%) demonstrated control.

Regarding CVD risk factors along gender lines, generally men tend to have a higher prevalence of hypertension as compared to women, and associated CVD (myocardial infarction, congestive heart failure, and stroke) [45, 51, 52]. However, in several studies we found a trend of disproportionate prevalence and worse outcomes for women for a variety of CVD outcomes [42, 45, 50, 83, 87]. Those citations with observed gender differences are described in detail in a separate review [116].

In turn, the effect of being exposed to a disaster was demonstrated to be a primary contributor to developing CVD in several studies [32, 34, 37, 39, 49, 55, 56]. In a Croatian study assessing the patterns of presentations for acute myocardial infarction (AMI) in 3,454 patients, they found a 23% increase in presentations during the war (1,254 vs 1,024 hospitalized patients) as compared to the 3-year period preceding the war of 1989–1991, and a 15% increase (1,173 hospitalized patients) as compared to the 3-year period even after the war (1995–1998) [49]. The incidence of hypertension and heart disease was also affected by those with death or injury in their family from disaster, with disease occurring most commonly within the first six months after an event [34]. In another study on residents affected by the Bosnian war, they assessed incidence of AMI and unstable angina (UA) 5 years prior to, during, and 5 years after the war [36]. The overall incidence of both AMI (n = 428 vs 365, p = 0.025) and UA (n = 185 vs 125, p = 0.001) was found to be higher during the war as compared to the period prior. In a Kuwait missile attack, Zubaid et al. found that the incidence rate of AMI hospital presentations more than doubled (incidence rate ratio = 2.43; 95% CI: 1.23 – 4.26, p < 0.01) for one year after the event [57]. Another study assessing the effects of dust storms in western Iran, showed there was an increase in cardiovascular events with a 1.35% increase in incidence of events for every 100 μg/m3 increase in the PM10 concentration (particulate matter greater than 10 μm) (p < 0.05) [39]. Finally, a retrospective cohort study assessing the effects of famine during the Biafran war (1967–1970) demonstrated association between undernutrition and the presence of hypertension, glucose intolerance, and overweight in Nigerian adults affected [41].

Furthermore, refugee status was associated with higher prevalence of CVD as compared to non-refugee counterparts in several studies. Abukhdeir et al. demonstrated a lower prevalence of CVD among those reporting non-refugee status in a representative sample of Palestinian households within the West Bank and the Gaza Strip (OR 0.539, p < 0.001), as compared to their refugee counterparts [32]. Yusef et al. highlighted an alarming predominance of late presentations for CVD, and other NCD risk factors, at United Nations Relief and Works Agency (UNRWA) primary health care facilities in Lebanon with 42% of respondents having at least one complication (such as retinopathy, nephropathy, and neuropathy) [83]. Similarly, Kadojic demonstrated that displaced persons in Croatia residing in camps had higher prevalence of hypertension, hyperlipidemia and obesity when compared to age-matched controls in settlements not impacted by the war [43].

Only one study assessed management of disease. This was a descriptive analysis by Yusef et al., showing that among refugees accessing care at UNRWA facilities in Lebanon, only 3% were on first-line anti-hypertensive therapy, up to 14.2% were on third line treatment, and 10% reported lifestyle modifications [83]. Another study discussed a complex intervention that included capacity-building of staff, provision of key diagnostic tools such as blood pressure cuffs, stethoscopes and glucometers), and advocacy on providing NCD care. The intervention took place in Lebanon [44], and they implemented screening for DM and hypertension in those 40y and older attending any of the clinics (five health centers and three mobile units), with the potential for referral to a specialist, such as cardiology, in case of need. This and the scant other interventions found in our study [38, 44, 61, 62, 67, 86, 96] are further described in a separate publication [31].

Cancer

Multiple studies demonstrated that cancer and oncological emergencies affect populations in conflict (see Table 2). Of the ten articles included, there was a predominant geographic focus on the EMRO region. In Lebanon, Sibai et al. [52] observed that cancer was second only to cardiac disease as a cause of death. Cancer represented 15% of all causes of deaths in their retrospective cohort study of 1,567 Lebanese aged 50 years and over residing in Beirut during the Lebanese Civil War (1975–1990). This was followed post-war by Shamseddine et al. [64] who identified an overall crude incidence rate for all cancers combined of 141.4 per 100,000 among males and 126.8 among females, a sharp contrast to earlier estimates made in 1966, of 102.8 and 104.1, respectively [64]. Of note, few studies addressed refugees, Internally Displaced Persons (IDPs) or noncombatants, in particular [52, 64]. We identified no articles relating to cancer prevalence among refugees in Africa, Asia, or the Americas. No studies addressed palliative care for oncology patients in the disaster setting (Table 12).

Table 12 Cancer Key Findings

Multiple studies indicate a high prevalence of modifiable cancer risk factors [60, 63,64,65] in conflict-affected populations that could be targets for future intervention such as Human papillomavirus (HPV) vaccination, anti-tobacco smoking campaigns, and access to adequate nutrient-rich food. Cervical cancer, in particular, was identified as being related to or affected by war [58, 63]. For example, in the study by Huynh et al. [58], they demonstrate that southern Vietnamese women whose husbands served in the armed forces experienced a more than 160%-290% increase in cervical cancer risk, relative to women whose husbands had not served in the armed forces. The authors attribute the association between male combat activity and cervical cancer as men become reservoirs of high risk subtypes of HPV which cause cervical cancer, acquired during wartime movement patterns [58, 117].

We also found a variety of tobacco-related cancers. Shamseddine et al. [64] found in reviewing 4,388 new cancer cases in post-civil war Lebanon, that lung cancer was the third most prevalent cancer type. In addition, they highlight that bladder cancer incidence rates are disproportionately higher in Lebanon than in the region, and globally. Breast cancer was listed by multiple studies as the most significant cancer burden amongst women in conflict affected LMICs—including studies relating to Lebanon [64], Afghanistan [59], and Pakistan [59]. Tobacco associated cancers were noted as prominent in multiple conflict affected nations and as amenable to prevention efforts through anti-smoking campaigns [59, 64].

Malnutrition in early life had demonstrated association with stomach cancer mortality for survivors of the 1959–1961 Chinese famine [60]. Birth cohorts of Zhaoyuan County, China who were exposed to famine or experienced malnutrition had stomach cancer mortality rates around twice as high as birth cohorts not exposed to malnutrition 15 to 20 years post-famine [60]. Proposed mechanisms by the authors for this relationship include a correlation between nutritional deficiency and H. Pylori infection, consumption of foods associated with development of gastric carcinoma in times of famine such as salted meat containing N-nitrosamines or nitrite, vitamin deficiencies, and heavy alcohol use [60].

Relating specifically to refugees, Otoukesh et al. [51], provided cancer prevalence data for refugees in a 2012 retrospective cross-sectional study of Afghani refugees residing in Iran. Using demographic and medical data collected between 2005 and 2010 from referrals to the United Nations High Commissioner for Refugees (UNHCR) offices in Iran for Afghani refugees, they found that neoplasms represented 13.3% of all referrals second only to ophthalmic diseases. Likewise, McKenzie et al. [62] found that amongst UNHCR registered Iraqi and Syrian refugees in Jordan, brain tumors accounted for 13% of all neuropsychiatric applications. Furthermore, Khan et al. found a divergence in the epidemiology of cancer diagnosis from the host population when compared to refugees, with esophageal cancer representing 16.6% of oncological cases amongst male Afghan refugees compared to only 4.6% of cases amongst Pakistani residents [59], and further evidence shows a difference in breakdown by ethnicity exemplified by Pashtun refugees who experienced a disproportionate frequency of referrals for oncologic disease (17%) amongst Afghani refugees residing in Iran despite receiving only two percent of all referrals [51].

Further studies identified challenges specific to refugee populations or subgroups of refugee populations [51, 58, 59, 61, 63]. Marom et al. [61] described clinical and ethical dilemmas in patients with head and neck cancers presenting to a joint Israeli-Filipino field hospital during the subacute period following a 2013 typhoon in the Philippines. They highlight the importance of awareness of cancer epidemiology in the target country prior to deployment. In this case, it guided the Israeli team’s clinical management such as prioritizing physical examination for cervical nodal metastases based on known prevalence of regional lymph node involvement at presentation in 70% of Filipinos with head and neck cancers [61].

Cost of care as a barrier for refugees with cancers was studied by McKenzie et al. [62] who aimed to assess the prevalence and cost of neuropsychiatric disorders among Syrian and Iraqi refugees requiring advanced specialty care in Jordan. The UNHCR funds tertiary level medical care for refugees based on the cost and acuity of required care by means of application to an Exceptional Care Committee (ECC). In reviewing refugee applications for tertiary care to the ECC, McKenzie et al. [62] found that brain tumors represented the most expensive neuropsychiatric diagnosis overall ($181,815 USD, $7,905 USD/ applicant). Other referral diagnoses were stroke, psychiatric diagnoses, trauma, infectious diseases, multiple sclerosis, neurodevelopmental abnormalities, and epilepsy.

Chronic respiratory disease

Of the fourteen articles that addressed chronic respiratory disease, six were related to war, and most addressed health hazards faced by refugees or victims of chemical weaponry (see Table 3). The geographic focus of most of these studies was the Middle East, with six studies from Iran alone (Table 13).

Table 13 Chronic respiratory disease Key Findings 

Two studies conducted in Kuwaiti patients affected by the Gulf War demonstrated the association between war trauma and increased incidence of asthma exacerbations. However, despite the increase in frequency, there was no change in severity of exacerbations [66, 78]. One study found increasing levels of self-reported stress exposure were correlated with reports of asthma [78]. In contrast, a chart review on patients admitted with asthma in Kuwait found no difference in admission or mortality rates from asthma when comparing the pre-war and post-war periods [66].

Chemical agents used during warfare, such as sulfur mustard gas, confer an additional risk for chronic respiratory disease [105]. In one study assessing incidence of asthma among children of individuals exposed to chemical warfare, a similar incidence of disease was found to that of individuals born to parents with asthma [73]. The comparable incidence is concerning for chemical warfare as an independent contributor to the development of asthma. Additionally, a cross-sectional study of a Chronic Obstructive Pulmonary Disease (COPD) cohort demonstrated increased morbidity of patients exposed to sulfur mustard gas also conducted in Iran, and validated use of the COPD Assessment Tool (CAT) for quality of life in this population [72].

The effect of storms on respiratory illness was also studied [39, 70]. The only prospective observational study within our review on chronic respiratory disease was on this topic, evaluating asthma exacerbations and bronchospasm associated with thunderstorms in southwestern part of Iran, Ahvaz [70]. Two thousand patients who presented with these complaints within three weeks of a thunderstorm were surveyed. This represented an abnormal surge in such complaints for emergency departments there. 30% of patients reported developing their symptoms on the day of the thunderstorm, although only 2% presented within 24 h. At 3 weeks follow-up, more than two thirds were still using medications, with beta-agonists being the most likely prescriptions, and corticosteroids following. More than half (51.7%) had no prior history of respiratory disease or complaints of shortness of breath. A retrospective chart review similarly looked at respiratory illness and evaluated correlation with dust storms [39]. In contrast, this study concluded that cardiac (P < 0.05), but not respiratory, disease was associated with occurrence of dust storms.

Beyond storms, a variety of studies looked at the health effects of different types of natural disasters via chart review of patients who presented after the disaster. A large forest fire in Indonesia caused a “haze disaster” in 1997 resulting in increased respiratory complaints [71]. Among 543 respondents, while only 7.4% had a history of chronic respiratory illness (asthma), 98.7% presented with respiratory complaints [71]. 49.2% of all respondents reported symptoms which disturbed their daily life [71]. In Ecuador, researchers looked at pediatric emergency department visits and found that there was an increase in frequency of visits associated with volcanic eruptions. Visits for asthma and asthma-related conditions doubled (RR 1.97, 95% CI 1.19, 3.24) during the three weeks following volcanic activity [75]. Among NCD presentations to an International Committee of the Red Cross (ICRC) Hospital in Banda Aceh, Indonesia post-tsunami respiratory diseases were one of the most commonly recorded conditions (21%), which included acute asthma exacerbations [76]. Similarly, Redwood-Campbell et al. [77] cited respiratory complaints as constituting 12% of presentations in the outpatient/ emergency department at the same Indonesian ICRC facility, with asthma making up 29% of those cases.

Studies looking at populations in refugee camps were epidemiologic in nature. In the Palestinian West Bank, children from refugee camps were at higher risk of asthma than children from neighboring villages or cities [69]. Having a history of wheezing was reported for 22.1% of children in refugee camps versus 16.5% in cities, and 15.5% in villages. Overall, 8.8% (n = 298) of children reported wheezing in the previous year, with a 17.1% lifetime prevalence of wheezing [69]. Similarly, in the slums of Dhaka, Bangladesh, children under 5 who were part of a “climate refugee” community were studied and compared to a non-refugee group. Asthma caused a 1069-fold higher number of disability adjusted life years (DALYs) lost in the group displaced due to climate change in comparison to non-affected populations [74].

Diabetes

We found that studies addressing diabetes were predominantly conducted in the EMRO Region (see Table 4). Specifically, 20 studies were conducted in the Eastern Mediterranean Region, two studies were conducted in the Caucasus region, three studies occurred in Sub-Saharan Africa, six studies occurred in Asia including South and Southeast Asia, and two studies were conducted in Eastern Europe (Table 14).

Table 14 Diabetes Key Findings

Multiple studies point to the relationship between stress and personal loss incurred in natural disasters and conflict, and a subsequent rise in occurrences of impaired fasting glucose (IFG) and diabetes mellitus (DM) among survivors [34, 38, 82, 91, 94]. One such retrospective cohort study by An et al. [82] investigated the long-term impact of stresses from the 1976 Tangshan earthquake on the occurrences of impaired IFG and DM among survivors and found that the incidences of IFG and DM for the exposure groups were significantly higher than that for the control group (P = 0.043 for IFG; P = 0.042 for diabetes), with those who had lost relatives exhibiting a higher diabetes incidence than those who had not lost relatives. This effect was only statistically significant in women earthquake survivors (p = 0.009) [82]. In addition, refugees with diabetes were found to have strongly reduced quality of life (HRQOL) as compared to age-matched non-diabetic controls as identified by Eljedi et al. using the World Health Organization Quality of Life questionnaire (WHOQOL-BREF), with particularly severe effects noted among females (p < 0.05 in all four domains) [88].

Additionally, several studies addressed food insecurity, and identified it as a primary contributing factor affecting diabetes management [38, 50, 53, 95]. A study focusing on older Palestinian refugees [53] found that participants practiced reduced meal portion sizes, skipping a meal, or foregoing a full day’s meals due to food shortage at a significantly higher rate than an age matched host population in Syria (reducing portion sizes p < 0.001; skipping a meal p < 0.001; not eating at all p < 0.001). Factors associated with skipped meals or reduced portion sizes included low economic status, larger household size, and type of residence (financial status p = 0.009; household size p < 0.001; type of residence p < 0.001). The number of days older refugees reported eating only bread and nothing else corresponded to reported financial status (p = 0.036). The authors theorized that food insecurity may result in challenges in the management of diabetes [53].

Further studies specifically addressed effects of fetal exposure to malnutrition and impaired glucose tolerance or diabetes later in life [41, 85, 90, 92]. Hult et al. [41] examined the accumulated risk for glucose intolerance 40 years following fetal exposure to famine in Biafra, Nigeria during the Nigerian civil war. The crude odds ratios for both impaired glucose tolerance and diabetes diagnoses were significantly higher for the group exposed to fetal or infant famine in comparison to controls [41]. Consistent findings were identified by a retrospective cohort study from China by Li et al. [85], who also identified a relationship between the severity of famine for fetal exposed subjects and risk of hyperglycemia later in life (OR = 3.92; 95% CI: 1.64–9.39; P = 0.002). Similarly, in a region of Northern Ethiopia recently affected by severe famine, clinical features of 100 insulin-treated diabetic patients were consistent with previous descriptions of malnutrition-related diabetes mellitus (MRDM): young age of onset (70% < 30 yrs), low BMI (mean 15.8), and resistance to ketosis (only 4% admitted with diabetic ketoacidosis despite 48% reporting insulin treatment interruption) [90].

Additional barriers to glycemic control in patients affected by conflict were: migration after war, lack of self-monitoring glucose strips, lack of access and cost of medications, failure to adequately screen for diabetes, inability to travel to a heath facility, lack of education regarding diabetes complications and management, food availability, and difficulty following patients over time [38, 45, 50, 53, 79, 83, 84, 86, 89, 90, 93, 104, 109,110,111]. One cross sectional study [93] which aimed to identify barriers to glycemic control from the patient perspective in a diabetic clinic in the south of Iraq, found that lack of drug supply from a primary health care center or drug shortage is a barrier for 50.8% of patients, while drug and/or laboratory expenses were a barrier for 50.2% of patients. 30.7% of patients said that they were not aware of possible diabetic complications and 30% thought that their failure to control their diabetes was due to migration after the war. Lack of electricity, lack of access to blood glucose monitoring devices, and illiteracy as a cause were cited by 15%, 10.8% and 9.9% respectively [93]. In Mali and Ethiopia, insulin was not widely available and access was limited by cost (US$ 11 per vial in Mali) [86, 90]. Multiple studies noted that syringes and self-monitoring blood glucose devices were not readily available and posed a financial burden to those who required access to them [84, 86, 90].

Diabetic limb amputations were also found to be highly prevalent amongst populations in disaster affected LMICs [67, 83, 97, 110] corresponding to low rates of diabetic foot examinations in refugee settings (e.g., Palestinian refugee diabetic patients’ feet were examined in only 8% of encounters at a UNRWA clinic) [79, 109]. In Lebanon, during the 2006 Lebanese–Israeli conflict, diabetes was the main indication for limb amputation (59%), followed by vascular disease (18%), and trauma (12%), with the highest amputation rates reported in the region experiencing the greatest conflict burden (3.82 per 10, 000 persons) [97]. Diabetic patients were older (mean age 73 years versus 30 years), more likely to have major surgery (OR = 7.87; 95% CI: 2.83–21.9), and stay in hospital longer (RR = 4.56, 95% CI: 2.41–8.64) than patients with trauma-related amputations [97]. Other complications of late stage disease were also prevalent, as demonstrated by Khader et al. in a community-based cross-sectional study of Palestinian refugees in Jordan with 10–20% of diabetic patients presenting with late stage complications of diabetes including blindness, cardiovascular disease, and limb amputations [109].

One study investigated complementary and alternative medicine (CAM) use among Palestinian diabetic patients and found the use of CAM differed significantly between residents of refugee camps as compared to residents of urban or rural areas (p = 0.034) [80]. Those who were on CAM reported they were using it to slow down the progression of disease or relieve symptoms and 68% of patients interviewed reported not disclosing CAM use to their physician or pharmacist.

While no study specifically aimed to focus on gender in their primary research objectives, we found a relationship between gender and prevalence or access to resources for diabetes, emerged as a recurring theme [32, 82, 83, 88,89,90, 94, 113, 118]. These findings will be presented in a separate publication [116]. Other common risk factors associated with diabetes type 2 included age, having a higher BMI, being divorced/widowed/separated, having never attended school, illiteracy, comorbid hypertension, hyperlipidemia, family history, sedentary lifestyle, history of traumatic exposure, and refugee status [32, 50, 53, 81, 83, 91, 97, 113].

Several studies also took a health systems approach and found that reliance on tertiary care for diabetes management fostered unequal access by socioeconomic status, geographic location, and escalating healthcare costs overall [33, 35, 84, 86]. One study from Georgia [84], which sought to identify the extent to which the Georgian health system provides for effective diabetes control post-independence, identified a systems level concern that only tertiary-level endocrinologists were able to modify treatment regimens and prescribe insulin whereas even endocrinologists who worked in polyclinics were unable to determine insulin regimens or prescribe insulin. Three studies from Syria [33], Tunisia [35], and Mali [86] identified a similar shift of diabetes care to the tertiary level prior to the emergence of conflict in these countries due to an emerging private sector [33, 35] and lack of specialists [86], respectively. In Mali, the lack of specialists was augmented by a lack of available guidelines, treatment protocols, and training for primary care level providers which prevented a transition of care to primary or general practitioners [86]. The authors theorized that this shift of diabetes care to the tertiary level contributed to reduced care access during active conflict in these countries [33, 35, 86].

Other NCDs

Studies investigating other NCDs centered on musculoskeletal and joint disorders [34, 42, 48, 77, 119] epilepsy and other neuropsychiatric disorders [51, 62, 103], ophthalmic diseases [48, 51, 100], nephropathies and urologic complaints [48, 51, 77] (see Table 5). Two studies measured mortality rates [34, 52] and two also studied quality of life [98, 101]. The effects of disability were briefly touched on by Leeuw et al. [101], with Amini et al. [98] further identifying hearing loss, and tinnitus as having negative impacts on quality of life among blind survivors from the Iranian War (p = 0.005, p < 0.0001) as compared to non-afflicted counterparts. We found that the majority of the studies on other NCDs did not refer to specific diseases or illnesses [42, 48, 77, 99], but rather represented epidemiological studies referring to conditions more broadly such as in the case of Mateen et al. [48] referring to “joint disorders”, and Hung and Redwood-Campbell describing “musculoskeletal”, “respiratory complaints,” and “gastrointestinal complaints” of unclear etiology [42, 77].

Hung et al. described musculoskeletal complaints constituting 30.4% of presentations among those visiting a Hong Kong Red Cross clinic in rural China following the 2008 Sichuan earthquake [42]. Mateen et al. conducted a far-reaching study of refugees in 127 camp settings across 19 countries and found that reportable neurologic diseases accounted for 59,598 visits over a 4-year period [103]. Nearly 90% of these cases were for epilepsy, which they highlight far outweighed the prevalence of neurological diagnoses of an infectious nature. Another study investigated neuropsychiatric disorders among Syrian and Iraqi refugees in Jordan via retrospective review of applications to the Jordanian Exceptional Care Committee, and found stroke to be the most common neuropsychiatric diagnosis (n = 41 applications, 16% of neuropsychiatric applications; median age 64 years) [62].

Specific ophthalmic diseases identified by Mateen et al. include cataracts (1.44 visits per refugee) and glaucoma (1.46 visits per refugee), which were exceeded only by cerebrovascular disease (1.46 visits per refugee) among Iraqi refugees in Jordan [48]. Of note, more than half of the refugees received concomitant diagnoses in one visit. Otoukesh describes ophthalmic disease as the most common health referral (13.65%) for those aged 15–59 among Afghan refugees in Iran [51]. Amini et al. [98] measured Quality of Life (QOL) scores in Iranian survivors totally blinded during the Iran-Iraq War, the effects of which were mitigated among those with higher levels of education (p = 0.006). Urologic complaints were identified as predominant in the ICRC hospital in Banda Aceh, Indonesia with 19% of complaints [77]; specific examples of urologic disorders from Mateen et al. among Iraqi refugees constituting a significant amount of morbidity were prostatic hypertrophy and nephrolithiasis [48]. Hematologic disorders were described by Otoukesh, and the type of disorder varied by ethnicity, with referrals for the Baluch being the highest at 25% [51].

Concomitant affliction with NCDs

Finally, co-affliction with multiple NCDs was a recurrent issue in our findings. This was demonstrated by Strong et al. among Palestinian refugees in Lebanon, with an average of 4 NCDs per person; Syrian refugees in the same study had an average of 2.5 NCDs per person [53]. Three or more risk factors were also seen in displaced persons in Croatia, a statistically significant difference in prevalence when compared to age-matched controls who were not displaced [43]. Clustering of risk factors was also evident in a populations being served by UNRWA in Jordan, Syria, Lebanon, West Bank, and the Gaza Strip, and the risk of having CVD was 2.7 times higher in individuals with 4 risk factors as compared to those with only 1 risk factor [50]. Concomitant affliction also conferred worse outcomes among Palestine refugees in Jordan with CVD (myocardial infarction, congestive heart failure, stroke and blindness) among those with hypertension and diabetes, when compared to those with hypertension alone in the same cohort (p < 0.01) [45]. Yusef et al. also demonstrated that having concomitant risk factors (such as diabetes and hypertension) resulted in a higher likelihood of presentation with late complications of NCDs at a UNRWA primary healthcare field site in Lebanon [83].

Discussion

NCDs represented a significant burden for populations affected by humanitarian crises and natural disasters for all regions [40, 51, 52, 57, 58, 61, 62, 64, 74, 86, 89, 97, 106, 120] and even conferred increased mortality and morbidity when compared to infectious diseases in one study [40]. Diabetes was the most commonly studied disease, even exceeding cardiovascular disease despite a higher global epidemiologic burden of the latter [29]. Late stage complications of cardiovascular diseases and diabetes including stroke [40, 46, 62, 103, 110], diabetic foot amputations [67, 83, 97, 110], and myocardial infarctions [55, 57] were described in all regions [83, 110]. However, studies addressing chronic respiratory diseases and cancer were noticeably lacking. All the same, studies on cancer highlighted tobacco as a key underlying factor, which contributes to predominant cancer etiologies in these populations such as breast, lung and bladder cancer [59, 64]. Musculoskeletal and joint disorders [34, 42, 48, 77, 119] epilepsy and other neuropsychiatric disorders [51, 62, 103], ophthalmic diseases [48, 51, 100], nephropathies and urologic complaints [48, 51, 77] constituted additional commonly encountered NCDs, and co-affliction with multiple NCDs or NCD risk factors should be expected in care [43, 45, 50, 53, 83]. As far as regional focus of studies assessed, both Sub-Saharan Africa and the Americas were poorly represented in the literature on NCDs in humanitarian crises [40, 41, 86, 90] (Tables 6 and 7). This is in spite of the fact that these regions experience a marked dual burden of armed conflict [115, 121] and natural disasters [122, 123], and represent a significant portion of the global NCD burden [107, 108]. Several studies demonstrated that NCDs adversely affected morbidity of populations in humanitarian crises with women and older populations disproportionately affected [72, 88, 98, 124]. Meanwhile, scant studies focused on pediatric populations, another vulnerable population [69, 73,74,75]. Strengthening and broadening the spectrum of NCD diagnoses included in disaster management planning is key, and particular focus on children and adolescents is critical as these age groups present key opportunities for interventions to mitigate future NCD morbidity [114, 125]. Several studies identified challenges and epidemiologic factors specific to refugee populations or subgroups of refugee populations [51, 58, 59, 61, 63], and they also highlight that refugee populations are heterogenous in their disease burden as compared to the host population. Understanding these sub-populations is key in guiding the medical equipment, personnel including specialists, and potential screening programs that should be considered in future humanitarian efforts. Assessing contributors to disease development and progression to complications, in prospective studies, would also be beneficial. Additionally, studies are needed on effective management of NCDs in these settings including for intervention implementation, which were notably lacking but critical in these uniquely resource-constrained contexts. Finally, policies that ensure established host country capacity for NCD care in addition to those for disaster response agencies that incorporate guidance for NCD care are key.

Further research, policies and interventions needed for lead four NCDs among diverse populations in disaster settings

There was a predominant focus on diabetes, including among studies from the EMRO region which constituted the region with the highest number of publications. For studies on DM, 57.6% of studies were conducted in that region alone. Furthermore, 32.4% of DM studies focused on the Palestinian population alone [32, 50, 79,80,81, 83, 88, 104, 109,110,111], higher than Africa, the Americas, Western Pacific (WP), and South East Asia (SEA) combined. The high prevalence of articles conducted in the EMRO may reflect the higher prevalence of diabetes there [126]. However, with a rise of diabetes in all regions including Sub-Saharan Africa, where the largest percentage increase in the incidence of diabetes is projected in the coming decade, this represents a significant gap in the available literature [127]. Increased research on diabetes in these understudied regions is particularly needed on interventions targeting screening and early disease recognition in order to forego complications, highlighted in several articles [67, 83, 97, 109, 110]. Additionally, best practices on management of disease such as through controlled trial designs would be ideal. Innovative interventions are needed given limited access to self-monitoring devices, insulin as well as potentially limited health literacy [84, 86, 90, 93]. Implementation of clinical policies as well as education for providers and patients are important [128]. Additionally leveraging of community leaders [129], as well as more novel interventions such as the poly-pill [130] and mHealth [131] may be potential opportunities for treatment in these highly limited resource-variable settings.

Further studies are also needed on the additional leading NCDs, particularly cancer and chronic respiratory disease; this includes on epidemiology to guide policy and further research. For example, the number of studies on CVD were surpassed by those on DM, despite the significant global public health burden as the leading cause of death [29]. Understandably, CVD may be more challenging to diagnose, screen and test for, but that does not mean that efforts should not be made to do so in these contexts. Moreover, disease focus lacked as compared to the primary focus on risk factors [42,43,44, 47, 50, 53, 54, 112]. To that end, a focus on risk factors is laudable given potential for guided interventions that target prevention, however, understanding the epidemiology of disease is also important in order to effect policy and practice change. Evidence in these settings including but not limited to disease presentation patterns, socioeconomic characteristics of patients, responsiveness to medications, and overall outcomes would be ideal. Surveillance and registries that are set up prior to conflict settings, or early in response, would be ideal in tackling these and other key research questions [132].

Furthermore, the gap in publications on palliative care was alarming given its particular relevance and importance in settings with limited access to care for later stages of disease. This highlights the importance of developing interventions with palliative care implications as well as policies that ensure access to palliative care management such as medications to treat pain, mental health symptoms and gastrointestinal symptoms [133, 134].

Further prioritization by policy-makers and other stakeholders on NCDs in diverse disaster settings needed

Increased understanding of the effects of diverse crises, rather than just armed conflict is also key. In Asia, the Western Pacific, and the Americas there was a specific focus on natural disasters [67, 75, 76], whereas the Eastern Mediterranean Region (EMRO) and Africa regions focused primarily on armed conflict [40, 57, 86, 97]. The effects of climate change [9], and subsequent increasing natural disasters, highlight our need to identify NCD burden in order to guide appropriate responses during these events. These feats can be achieved by greater prioritization by stakeholders that are already based in the underrepresented regions, such as development agencies and non-profits, increased political will, funding mechanism opportunities such as from the World Bank and the UNHCR, as well as through consideration by new partners establishing work in these settings.

Further research in diverse disaster phases needed

Of note, most studies either reported on the consequences of conflict after the fact [40] or when the population of interest had relocated to a refugee camp or host country [51, 59, 62, 109]. Meanwhile, there were few studies on NCDs [44, 86, 97] affecting populations during acute crises, active conflict, or for internally displaced persons (IDPs) [51, 58, 59, 61, 62] (see Tables 6, 7, 8, 9 and 10). These findings indicate the importance of research during active crisis, including by organizations doing relief efforts such as through tracking reporting on NCD diagnoses and outcomes. While this can be challenging in such settings, it is necessary. Kohrt et al. suggest increased focuses on conducting research ethically among these vulnerable populations, as well as increased community engagement and facilitation of improved research capacity with LMIC-based partners as potential initial solutions [3]. They also encourage flexible research methods with sensitivity to these unique needs by researchers when being reviewed by funding bodies and ethics review boards [3]. Ultimately, clearly outlined policies that guide agencies responding during crises on establishing research protocols even as they provide clinical care, which are developed a priori, would be beneficial to increase the quality, rigor and ethical nature with which research can be conducted among these populations.

Concomitant affliction with NCDs and NCD risk factors

We found that the populations studied were commonly afflicted with multiple NCD risk factors [32, 50, 53] and multiple NCDs [45, 50, 53, 83], which supports the need for consolidated care for NCDs as co-affliction confers higher risk of complications [83]. Many commonly cited risk factors in HICs such as age, family history, higher BMI, comorbid hypertension, smoking, hyperlipidemia, family history, sedentary lifestyle for DM, cancer, cardiovascular disease were cited [32, 50, 53, 64, 81, 83, 113]. However, a lack of association between NCDs and family history as well as other traditional risk factors was also found, and this may result in under-recognition and subsequent under-diagnosis in these settings [89, 135]. Packages, such as the WHO PEN, which provide a comprehensive approach to NCDs and NCD risk factors may be worthwhile when considering establishing care in these settings to ensure that NCDs are considered routinely [136]. Furthermore, an adaptation to PEN for humanitarian settings called the “PEN-H” should be considered for dissemination during crisis relief efforts [137].

Disaster related exposures as unique contributors to NCD development and morbidity

Multiple studies identified disaster-related psychologic and physical stressors as significant risk factors for NCDs [37, 38, 41, 43, 63, 65, 78, 82, 85, 91, 92, 94], as well as described subsequent increased NCD related morbidity as a result of disaster stressors [34, 36, 43, 56, 57, 78, 95, 97]. Bereavement, injuries in the family [34, 36, 82], displacement [120], temporal/ geographical proximity [32, 56, 57, 95], and war-related physical and psychological trauma [34, 78, 94] were some of the independent predictors of diagnosis, and increased NCD morbidity [34, 36, 37, 57, 78, 82]. Refugee status was independently identified both as a risk factor for diagnosis with an NCD [32, 38, 43, 50, 69], and conferring worse morbidity as indicated by Disability Adjusted Life Years (DALYs) lost [74].

Malnutrition and food insecurity during disaster were commonly cited risk factors for increased NCD morbidity. Notably, fetal exposure to severe famine was associated with an increased risk of cancer [60], DM/impaired glucose tolerance [41, 85, 90, 92], metabolic syndrome later in life [102], and the unique phenomenon of Malnutrition Related Diabetes Mellitus (MRDM) [90]. The risk of MRDM was exacerbated by a nutritionally rich environment later in life [85, 92]. Another hypothesis for the higher prevalence of DM was lack of ability to monitor and control dietary intake and blood sugar during a crisis [38, 50, 53, 95].

Finally, environmental exposures from natural disasters [39, 42, 70, 71, 75] and war related toxins [73, 100, 105] contribute to NCD burden for these populations particularly for respiratory and cardiovascular diseases. Natural disasters impacting chronic respiratory illness include thunderstorms, earthquakes, forest fires, volcanic eruptions, and tsunami. Dust storms were a notable exception in one study [39], for which a link to increased pulmonary illness was not shown, while in contrast there was evidence of an effect on cardiovascular disease.

In sum, disaster settings confer higher incidence of NCDs and associated comorbidity. Furthermore, attention to refugee status in disaster settings is key given a disparate disease burden. Refugee populations have greater burden of disease and worsened outcomes when compared to host populations. Distribution of disease within a refugee population may be unique, and further divergent by ethnic group even among refugee populations. This is critical information to guide future humanitarian intervention design and implementation that should be sensitive to the need for tailored interventions for these sub-populations affected in that context. Furthermore, increased research is needed on the magnitude of the effect of disasters on NCD development, the development of complications of NCDs, as well as the timeliness of development of NCDs associated with exposure to disaster settings and the duration of effect.

Overcoming barriers to management of NCD care in humanitarian crisis settings through increased health system preparedness and responsiveness

The most commonly cited barriers to healthcare access in all phases of disasters and major disease diagnoses studied, included personal attributes: low levels of education [90, 93, 138], financial difficulties [53, 93, 120], displacement [86, 93], and illiteracy [32, 50, 93]. The most commonly cited systems level concerns were lack of access to medications, and affordability of medications [53, 83, 86, 90, 93, 139]. Multiple DM specific studies noted that syringes and self-monitoring blood glucose devices were not readily available and posed a financial burden to those who required access to them [84, 86, 90]. Several studies also noted shifting of medications from the clinically indicated medication to cheaper or more available options, which may lead to worse outcomes [83, 84]. Such challenges may be magnified more for migratory refugees as compared to those who are more established in refugee camps, as demonstrated by Yusef et al. [83].

Furthermore, in many countries affected by humanitarian emergencies, there is scarce data on NCD surveillance, epidemiology, and outcomes in populations at risk in the pre-disaster setting, which creates challenges for disaster mitigation efforts [89, 99]. As a result, poorly functioning systems for delivery of NCD care [33, 86], and underdiagnosis of NCDs [89, 135] in the pre-disaster setting are compounded by new challenges resulting from widespread destruction of the health system [33, 140].

Greater attention to screening and allocation of resources to treat NCDs including acute cardiovascular events such as acute myocardial infarction and stroke are needed in disaster-prone settings, outside of other medical relief efforts. A health system situational analysis in Tunisia demonstrates the effectiveness of a robustly developing primary health care system, which falls short in the humanitarian crisis setting without established human resources, reimbursement for public sector, consensus around guidelines for management, and the absence of ancillary providers such as nutritionists or specialists for referral, when needed [35]. In post-war Liberia, with majority of CVD deaths occurring within 24 h of admission, optimization of emergency care which is the first point of contact, was also highlighted [40]. Hung et al., the only researchers focusing on the pre-hospital setting, also enforce the importance of raising awareness among first responders of the associated increased burden of NCDs during crisis and propose guidelines adapted to this [42].

Multiple studies also enforced the importance of decentralized care of NCDs from tertiary health facilities pre-disaster as this was commonly noted to hinder NCD care access during the relief phase [33, 35, 84, 86]. Several diseases including leading cancer diagnoses are amenable to prevention, screening, and early detection such as breast cancer [59, 64], cervical cancer [58, 63], and other cancers associated with tobacco use [59, 64]. Cervical cancer, for example, is amenable both to primary prevention strategies (HPV immunization and barrier protection during sexual intercourse) as well as secondary prevention (pap smears), and was identified as an opportunity for targeting by several studies with high prevalence including in Vietnam [58] and Croatia [63]. Decentralization of primary care provision to community-based settings, such as for eye care, was advocated to address the loss of healthcare infrastructure [77], and may reduce stress on facilities providing emergent care [48, 77]. Reinforcement of the public health sector’s capacity for NCD management benefits both the relief phase of disaster response as well as post-disaster rehabilitation and reconstruction [140].

Overall, increased preparedness [33, 67, 84, 86] and responsiveness by aid providers, health providers, and local governments to NCDs in disasters [67, 86] would help improve disaster mitigation assessments. Validated tools such as the WHO Stepwise approach to Surveillance [141] or Demographic Health Surveys [142] could be used for surveillance or to develop registries in countries to allow for increased pre-disaster preparedness. Finally, healthcare systems can address the imminent need for palliative interventions that aim to reduce excess morbidity and suffering from NCDs [133].

Methodology and research infrastructure also key

As further guidance, we wanted to comment on study design as a key focus for future studies in this setting. The predominant study design consisted of retrospective chart reviews [36, 40, 42, 46, 49] with a minority of cross-sectional studies [39, 47, 51, 59, 64]. We observed that several studies either did not include a comparison group in their study design, or they used a time period across which the comparison was made that was arbitrary in nature. While challenging to conduct given the context of the studies, this limits validity of findings in many studies [52, 63,64,65]. These predominant research designs limit the ability to draw conclusions for causation, or to accurately measure the effects of disaster itself, although associations have been noted as previously stated under the discussion heading on disaster related exposures. In future, cohort study designs, as well as potential registries [132] alongside other prospective studies would enrich current knowledge on NCD determinants in disaster settings. Additionally, these study designs would enable better assessment of the long-term effects and complications of NCDs in these settings that are potentially exacerbated by the disaster context.

In addition, it was noted that publications were clustered by research group or author [45, 55, 85, 102, 104, 109,110,111, 143], which speaks to the need for increased academic outputs in LMICs, and Africa in particular. Increased infrastructure and capacity on research development is needed, including support for agencies at the front-lines implementing and delivering clinical programs, and who have the potential to concomitantly implement effective research on target populations or their programs.

Finally, several articles that were included in our results included NCDs as a peripheral focus, rather than as primary outcomes [77, 135]. Consideration should be given to include comparison or control groups in study design, for example individuals in neighboring regions, non-refugee counterparts, or matched sample populations not afflicted by the disease [34, 58] to be able to better assess and thus delineate the effects of the humanitarian crises itself on disease outcomes. Additionally, long-term cohorts and registries [109, 111] would be ideal to better understand the diversity of diseases and contributory factors in even greater depth. Of all the studies included, none referred to the Sphere guidelines [144], WHO Noncommunicable Diseases in Emergencies brief [145], or WHO PEN package of essential NCD interventions [136] as markers for study design, which we propose be included in future research.

Limitations

The last date of publication submission included in our findings is 2017, which limits conclusions based on findings published thereafter. With that said, we have demonstrated inciting evidence to guide future research, including through summarizing trends over the past two decades (see Fig. 2), and highlighting key findings from existing literature during that window that includes a predominant focus on cancer, and less focus on cardiovascular disease which carries a predominant NCD burden globally, a relative lack of focus on chronic respiratory disease and pediatric populations among other key findings. In turn, we have followed the protocol outlined and pre-published in PROSPERO demonstrating the novelty of this comprehensive review on NCDs in humanitarian settings that has not been undertaken as yet. In addition, although we were inclusive of major languages spoken in our search strategy (specifically English, French and Arabic) if the articles were not indexed using English words, or in the databases utilized, they would not have been included in our results. We are cognizant of the limitations of publication bias, and support continued advocacy for representation of various languages in primary research, journals and popular databases. All the same, we trust that despite these limitations the findings will contribute to increasing prioritization of NCDs in humanitarian settings, stimulate research ideas, and engage policy-makers at the country, national and international level.

Conclusion

An increased focus on the effects of, and mitigating factors for, NCDs occurring in disaster-afflicted LMICs is direly needed. While majority of studies included in our review presented epidemiologic evidence for the burden of disease, research is needed to address contributing factors, and means of managing disease in these extremely resource-variable settings. Regions particularly lacking evidence on LMICs in our study were Africa and the Americas; majority of evidence was from the EMRO region. Among the four lead NCDs, chronic respiratory disease was under-addressed despite evidence that it contributes to high morbidity in crisis. Furthermore, increased evidence on actual diseases such as myocardial infarction and diabetes, rather than simply focusing on risk factors such as hypertension is also needed with greater understanding of NCD epidemiology to guide allocation of resources and policy-makers. Attention to vulnerable populations including women and refugees is also a priority. Refugees have unique exposures that may predispose them to certain illnesses, such as MRDM, and management needs that warrant separate attention from host populations. Given this, we propose that refugee status be considered as an independent risk factor for future studies and interventions. All in all, screening and prevention for NCDs should be a priority alongside communicable disease programs, such as counseling for smoking cessation, counseling on diet, HPV vaccination, and screening for common cancers like breast and cervical cancer. Studies on implementation for these and other interventions will be key, and the use of implementation science to guide design and assess feasibility could be useful in these challenging settings. Additionally, policies allocating resources to equip health systems to address NCDs both pre-disaster and during crisis will enhance these efforts, such as through decentralization of care from tertiary settings that are already overextended during crisis. Finally, the need to address disease in disaster settings in collaboration with LMIC-based partners, community members, as well as other sectors outside of health silos such as agriculture and urban policy-makers, was also supported.