Disasters and humanitarian crises offer an extensive range of challenges for disaster responders and humanitarian actors. These settings frequently feature the necessity to operate despite political tensions, active armed conflict and violence, significant suffering and loss of life, and profound resource scarcity.1 Resource scarcity can develop due to several key processes seen in the aftermath of disasters or during humanitarian crises, primarily through shifting the balance between demand for care and the availability of resources.
Resources essential for health-care delivery, according to surge capacity constructs, can be grouped into the following: staff, stuff, space, and systems.Reference Adams2–Reference Paganini, Conti and Weinstein4 Significant threats to or shortages within these domains similarly threaten the capacity of a health service to deliver care, thus precipitating or exacerbating disparities. For example, the surge capacity domain of space includes appropriate physical space equipped with the required infrastructure and equipment that can be directly damaged and threatened by events such as floods or earthquakes.Reference Ardagh, Richardson and Robinson5–7 Additionally, the staff domain includes the number and skill sets of available personnel which can be directly and indirectly threatened by conflict, violence, and the brain-drain phenomenon, in which an exodus of experts seeking safety leaves a knowledge and skill gap while demand continues to grow.7, Reference Dodge8 It is, therefore, important to consider these domains and their relevance in the allocation of health resources.
The principles and practice of triage have evolved, but the primary function remains to provide mechanisms for distributing finite health-care resources.Reference Gupta9–Reference Robertson-Steel10 Operationally, triage is the application of systems which rank individuals according to the urgency of care required, providing clinicians with a means of patient prioritization.Reference Robertson-Steel10–Reference FitzGerald, Jelinek and Scott12 While deployed in the routine delivery of health-care services and in mass casualty incidents, traditional applications of triage can become saturated by an overwhelming number of patients in disaster and humanitarian crisis settings. Patient prioritization and health resource allocation in these settings therefore requires different approaches to clinical decision-making, patient prioritization, and potentially even the categorical denial of clinical care in response to disparity between demand and availability.
Emerging resource allocation principles include, firstly, the level of allocation. In terms of level of allocation, there are 2 primary levels: (1) macro-allocation, representing determinations affecting the overall availability of resources for health service delivery,Reference Gallagher, Little and Hooker13, Reference Scheunemann and White14 and (2) micro-allocation, representing determinations affecting the allocation of a specific resource to an individual.Reference Gallagher, Little and Hooker13, Reference Scheunemann and White14 The second principle is the basis of triage, which can be grouped according to individual-based, where triage is concerned only with prioritizing or allocating resources to individuals, or population-based, in which the population context is considered in determining resource allocation.Reference Burkle15–Reference Powell, Christ and Birkhead17 The final principle considered is transparency, with subcategories of explicit determinations with formal and transparent means of allocation often underpinned by policy or procedures, or implicit decisions made by individuals on an ad hoc basis.Reference Oei18, Reference Spector-Bagdady, Laventhal and Applewhite19
Health resource allocation in disasters and humanitarian settings has been the subject of theoretical and conceptual debate and analysis, and there has been some exploration of the ethical challenges faced by health-care workers in these contexts; however, there are several significant barriers that exacerbate the limitations of established mechanisms of health resource allocation, as well as the translation of knowledge from hospital settings (even during disasters) and the humanitarian crisis context. This gap is further exacerbated by the present lack of a comprehensive exploration of the state of evidence derived from direct experience with negotiating triage practices and decision-making mechanisms in these settings.
Aim and Research Question
The aim of this review is to identify, evaluate, and display graphically relevant studies to clearly identify gaps in the existing literature related to scarce health resource allocation in humanitarian aid settings. In doing so, this review will answer the question “what evidence has been derived from experiences of managing health resource allocation in real-world settings requiring humanitarian aid?”.
Methods
A mapping review utilizes a transparent and systematic approach to identifying, extracting, and mapping relevant literature according to an adopted framework.Reference Campbell, Tricco and Munn20, Reference Snilstveit, Vojtkova and Bhavsar21 A mapping review can be supplemented by an evidence gap map (EGM), which visually presents evidence according to the framework utilized in the mapping review, making evidence distribution and critical gaps rapidly and visually accessible to users.Reference Campbell, Tricco and Munn20–Reference Albers and Gaarder23 The methodological approach to performing a mapping review and EGM is provided in the literature as follows: (1) develop the framework to underpin the search and resulting EGM, (2) establish inclusion criteria, (3) conduct search for literature, (4) screen and assess evidence for inclusion, (5) perform data coding, extraction, and appraisal, and (6) perform analysis and produce visual representation of data.Reference Snilstveit, Vojtkova and Bhavsar21–Reference Albers and Gaarder23
Develop Framework
Developing a mapping review framework requires identification of categories, domains, and filters relevant to the central phenomenon.Reference Snilstveit, Vojtkova and Bhavsar21–Reference Albers and Gaarder23 The categories were determined a priori to be (1) key resource allocation principles and (2) the surge capacity construct. The resource allocation category informed row headings, with principle subcategories determining the domains: (1) level of allocation produced the domains of macro-allocation and micro-allocation, (2) basis of triage produced the domains of population-based and individual-based, and (3) transparency produced the domains of explicit and implicit. The surge capacity category informed the column headings, and these domains were determined to be staff, stuff, space, systems, and other/unspecified (added to ensure capture of relevant data not captured by these domains). This configuration served both as a theoretical framework underpinning this review and the framework informing the resulting EGM matrix. Finally, the primary filter for this mapping review was determined to be the “level of evidence.”
Establish Inclusion Criteria
Inclusion criteria were determined to capture evidence derived from experience and pertaining to health resource allocation or surge capacity occurring in the disaster and/or humanitarian crisis setting. Although acceptable for systematic reviews to be included in mapping reviews and EGMs, the scarcity of focused research on this topic underpinned on a focus on structured primary research and unstructured experiential accounts (such as discussion papers, editorials, narrative accounts). Inclusion and exclusion criteria for this literature review are presented in Table 1. Of note, papers were still considered for inclusion if the experience was considered or analyzed according to specific ethical challenges experienced rather than presenting ethical commentary or theoretical ethical recommendations.
Inclusion and exclusion criteria for article screening
Qual, qualitative methods; Quan, quantitative methods; MM, mixed methods.
Conduct Search for Literature
A systematic search was undertaken, guided by the Preferred Reporting Items of Systematic review and Meta-Analysis (PRISMA) guidelines.Reference Liberati, Altman and Tetzlaff24 Databases and search engines searched included MEDLINE (Medline Industries, Inc; Mundelein, Illinois, USA) via OvidSP (Ovid Technologies; New York, New York, USA), Scopus (Elsevier; Amsterdam, Netherlands), Embase (Elsevier; Amsterdam, Netherlands), CINAHL Complete (EBSCO Information Services; Ipswich, Massachusetts, USA), and ProQuest Central (Clarivate; Ann Arbor, Michigan, USA). The search strategy included combinations of Medical Subject Headings (MeSH) terms and keywords, as outlined in Table 2. Terms and keywords within cells were combined using the OR Boolean operator, and cells within columns were combined using the AND Boolean operator. The search was run on November 22, 2022.
Search terms used during systematic database search
Search terms in bold are indexing terms relevant to the database (MeSH terms in Medline, and Emtree Controlled Vocabulary in Embase); *designates actual use of the wildcard operator.
Data extraction – description of context, event, and response
Data extraction – research focus, design, limitations, and level of evidence
Article screening
Identified articles were imported into Covidence (Covidence; Melbourne, Australia) to facilitate screening. Each article was screened by title and abstract by 2 authors, with disagreements resolved by the third author. Those papers included by title and abstract had their full text reviewed by 2 authors with any disagreements during full review resolved by consultation between all authors.
Data Coding, Extraction, and Appraisal
Data were extracted from the reviewed articles into data extraction tables. Key information extracted included author(s), event description (external references were sourced to ensure an adequate description was provided if not described in sufficient detail within the article), response description, research focus and design, limitations, and an assessment of the level of evidence. The level of evidence was assessed according to the 7-tier hierarchy of evidence provided by Polit and Beck.Reference Polit and Beck25 This hierarchy was utilized, as it goes beyond the scope of evidence typical of biomedical experimentation to include qualitative and descriptive methodologies (level VI evidence) and evidence derived from opinion and committee (level VII evidence).Reference Polit and Beck25
Data in the form of direct excerpts from articles which offered a meaningful contribution were also extracted into data tables. Each extract was then considered and classified according to its relevance within the framework in a binary manner in that each statement either was or was not relevant to each domain. Importantly, data were not extracted for congruence with these domains but rather were only considered according to the domains once already extracted and determined to be relevant.
Analysis and Evidence Gap Map Production
The EGM matrix was produced in direct response to the developed framework. Extracted data were aggregated to produce a binary result for each domain intersection according to resource (presence or absence). Data was plotted on the matrix according to the level of evidence filter and scaled according to the prominence of the data within each domain intersection.
Although not typical of a mapping review and EGM, data analysis in this review was supplemented by high-level content analysis of data within domains; however, a comprehensive synthesis remains outside the scope of this review. Themes were taken directly from the developed framework, so overarching themes are identical to the “categories” included in the mapping review framework.
Results
In total, 12 papers met the criteria for inclusion (Figure 1). Data extracted to inform this literature review are displayed in Table 3 and Table 4. The produced EGM (Figure 2) presents the included sources mapped according to how data extracted from each source related to surge capacity domains and the identified principles.
Modified PRISMA flow diagram.
Evidence Gap Map.
The papers vary in depth of detail relating to health resource allocation in humanitarian settings. There are several key points related to the nature of the data gathered. The reviewed papers did not utilize approaches to structuring or reporting their findings that supported standardizing this type of disaster and humanitarian research. Much of the extracted data represented only superficial consideration and lacked purposeful in-depth exploration of resource allocation. Finally, much of the extracted data were not produced with the intent to report on how decision-making occurred in these settings; for example, data extracted from numerous papers were derived from explorations of ethical, or even broader, challenges in humanitarian contexts.
All 4 surge capacity domains were represented by extracted data. The most represented of the 4 surge capacity domains was the staff domain, with data derived from 8 (67%) of the included papers. The least represented was the space domain, with data derived from only 3 (25%) papers. Both the stuff and systems domains were informed by data extracted from 6 (50%) sources each. Data from 10 (83%) papers were assigned to the unspecified category, particularly prominent across the basis of triage and transparency themes, after the extract could not be otherwise classified.
Level of Allocation
Data extracted from 11 (92%) sources related to the level of allocation. Of all included sources, data from 9 (75%) papers related to macro-allocation and data from 10 (83%) papers related to micro-allocation. Across the surge capacity domains, the staff domain was most prominent within this theme, with equal distribution of sources across the macro-allocation and micro-allocation sub-themes. Of note, level of allocation is the only theme in which the space domain was addressed across both subthemes.
Extracted data related to macro-allocation or decisions determining the overall availability of resourcesReference Gallagher, Little and Hooker13, Reference Scheunemann and White14 noted high-level directives and mandates,Reference Akik, Semaan and Shaker-Berbari26, Reference Durocher, Chung and Rochon27 donor fatigue and donor influence,Reference Akik, Semaan and Shaker-Berbari26–Reference Hunt, Nouvet and Chenier29 tensions felt by in-field operators due to external determinations,Reference Asgary and Lawrence28, Reference Civaner, Vatansever and Pala30 decisions prioritizing risk mitigation,Reference Durocher, Chung and Rochon27 and decisions prioritizing impact maximisationReference Civaner, Vatansever and Pala30–Reference Lamblin, Derkenne and Trousselard32. Data related to micro-allocation, or decisions determining allocation of resources to individuals,Reference Gallagher, Little and Hooker13, Reference Scheunemann and White14 noted the influence of external factors on resource allocation, such as organizational priorities and policiesReference Hunt, Nouvet and Chenier29, Reference Civaner, Vatansever and Pala30, Reference Cereste33, Reference Daniel34 and the balance of risk,Reference Drevin, Alvesson and van Duinen35 and the prioritization of survival or at least greatest impact.Reference Durocher, Chung and Rochon27, Reference Kreiss, Merin and Peleg31, Reference Lamblin, Derkenne and Trousselard32, Reference Fardousi, Douedari and Howard36, Reference Sloand, Ho and Kub37
Basis of Triage
Data extracted from all 12 sources related to the level of allocation, with data from 12 (100%) papers related to population-based triage and data from 5 (42%) papers related to individual-based triage; however, when considering only data assigned to surge capacity domains, this reduced to 10 (83%) and 3 (25%) papers, respectively. Despite all papers contributing data to this theme, it remained grossly underrepresented compared to the other 2 themes. Of additional note is that the space domain consisted of data from only 1 (8%) paper which, notably, reported only level VII evidence.
The basis of triage was considered according to whether decisions are concerned with allocating resources to individuals according to individual characteristics (individual) or according to the broader population context (population-based).Reference Burkle15–Reference Powell, Christ and Birkhead17 When considering the population context, the included sources noted defining the “population” by need or vulnerability,Reference Akik, Semaan and Shaker-Berbari26, Reference Asgary and Lawrence28 military affiliation,Reference Lamblin, Derkenne and Trousselard32, Reference Cereste33 or infection status.Reference Civaner, Vatansever and Pala30 Where individual clinical status alone was not sufficient to drive allocation decisions, it was noted that broader population perspectives became relied upon,Reference Daniel34 such as the use of the expectant category to clearly denote assessment and acceptance of clinical futility.Reference Civaner, Vatansever and Pala30, Reference Cereste33
Transparency
Data extracted from 11 (92%) sources related to the theme of transparency. Both explicit and implicit decision-making subthemes contained data extracts from 9 (75%) of the included sources. Including only data which related to the 4 surge capacity domains, both the explicit and implicit subthemes contained data from 7 (58%) papers. The most prominent domains in the explicit subtheme were staff and systems, while in the implicit subtheme, staff and stuff were the most prominent. Notably, no extracted data related to the space domain within this theme.
Explicit decision-making occurs particularly when the interest of donors must be consideredReference Akik, Semaan and Shaker-Berbari26 and when the outcomes must align with organizational mandates.Reference Durocher, Chung and Rochon27 By comparison, implicit decision-making was noted to occur when balancing patient-versus-patient decisions.Reference Cereste33 Explicit decision-making was noted to be informed by rules and professional standards,Reference Asgary and Lawrence28, Reference Civaner, Vatansever and Pala30, Reference Cereste33 while implicit decision-making without these inputs is more prone to individual critique.Reference Hunt, Nouvet and Chenier29, Reference Civaner, Vatansever and Pala30 Data hinted at a point where the transition from explicit to implicit rationing becomes necessitated, particularly with implicit decision-making becoming increasingly dominant in the absence of operationalizable guidelinesReference Asgary and Lawrence28, Reference Hunt, Nouvet and Chenier29 or when decision-making is less straightforward.Reference Lamblin, Derkenne and Trousselard32, Reference Daniel34, Reference Drevin, Alvesson and van Duinen35
Security
During analysis, “security” emerged from within already extracted and otherwise classified data as a potential theme. Extracts related to security contained references to the selection of health-care facilities, the adoption of dispersed models of health delivery, decisions made in light of security threats to personnel, and the inability to treat patients in settings where security had become questionable.Reference Akik, Semaan and Shaker-Berbari26, Reference Lamblin, Derkenne and Trousselard32, Reference Fardousi, Douedari and Howard36 Specifically, security emerged from data classified according to the following: staff (micro-allocation), space (macro-allocation), system (macro-allocation), and unspecified (micro-allocation, explicit).
Level of Evidence
Level of evidence is assessed and reported in accordance with the hierarchy of evidence provided by Polit and Beck.Reference Polit and Beck25 Of the included papers, 9 (75%) reported level VI evidence or data gained by a single descriptive or qualitative research approach. The remaining 3 (25%) contained level VII evidence, consisting of data derived from unstructured reflection or opinion. Among the level VII papers, 2 (66%) contributed to the stuff and systems domains and 1 (33%) contributed to the staff and space domains.
Limitations
Firstly, many humanitarian crises occur in non-English-speaking regions, and thus, accounts and experience-based research may not be published or identifiable in English. Secondly, while a systematic search approach was deployed, it cannot be guaranteed that all potentially relevant sources have been captured. Additionally, the scarcity of data in this area meant there was insufficient data to perform a rigorous meta-synthesis, compounded further by the overall low level of evidence identified; however, the level of evidence must be considered in the context of humanitarian health-care settings and expectations around higher levels of evidence scaled by methodological and ethical feasibility.
Discussion
The aim of this review was to identify and produce an evidence gap map of the existing literature related to scarce health resource allocation in humanitarian aid settings. Overall, a limited and superficial body of experience-based evidence was available to inform this complex phenomenon in humanitarian settings.
There are several key frameworks available to underpin and guide disaster health research; however, there is no widely accepted framework or nomenclature system promoted specifically in relation to health resource allocation in these settings. The conceptual framework provided by Birnbaum et al.,Reference Birnbaum, Daily and O’Rourke38 as a leading framework in disaster health research, details the progression from a hazard through to the requirement for a relief phase, providing a backbone for conceptualizing the role of disaster health interventions. However, in its current form, this collection of frameworks does not extend to address the nature and impacts of resource scarcity during health responses to disasters. This review, particularly with its focus on the intersection of surge capacity models and health resource allocation, creates and occupies a space not currently captured or explored conceptually within existing literature.
Surge Capacity Models: Hospital Versus Humanitarian Settings
Despite recent advancements in surge capacity models for health services, such models are yet to be formalized within the humanitarian landscape. Developed primarily as conceptual models addressing surges in hospital-based care settings, existing surge capacity models consist of 4 domains: staff, stuff, space, and systems.Reference Adams2, Reference Bonnett, Peery and Cantrill3, Reference Kaji, Koenig and Bey39 The COVID-19 pandemic saw surge capacity models operationalized and clinicians reporting strategies employed to enhance the capacity of their health services. For example, Cammarota et al.,Reference Cammarota, Ragazzoni and Capuzzi40 Al Mutair et al.,Reference Al Mutair, Amr and Ambani41 Gauss et al.,Reference Gauss, Pasquier and Joannes-Boyau42 and RosenbaumReference Rosenbaum43 each contributed to either the conceptual or pragmatic advancement of surge capacity operationalization during the COVID-19 pandemic; however, surge capacity models seemingly remain the exclusive realm of hospital-based services aiming to enhance capacity from a functional baseline to accommodate further demand.
Despite contextual differences between the hospital-based settings within which existing surge capacity domains have been developed and humanitarian response settings, this review demonstrates the applicability of surge capacity domains within previous humanitarian health-care operations. Extracted data were amenable to inductive coding according to the surge capacity domains, with data consisting of experiences or strategies relevant to either individual or multiple domains. In this review, we therefore not only map existing literature according to this conceptualization but also establish the relevance of the framework developed to underpin the resulting EGM matrix. Additionally, the demonstrated utility of surge capacity models in response to the COVID-19 pandemic adds further weight to the potential contribution that further development and refinement of surge capacity models may have for humanitarian health responses.
Security as an Emerging Theme
Although the aim and nature of a mapping review and EGM is not to provide a meta-synthesis of available data, the theme of “security” emerged from data already inductively coded according to preexisting surge capacity domains. As already highlighted, the existing surge capacity model has been developed in hospital-based settings with a focus on the continuation of services above a secured operational baseline, a characteristic often incongruent with humanitarian health-care responses. When considering the 4 S’s of surge capacity, each domain is traditionally considered according to whether there is a sufficient supply within each domain; however, the way in which “security” seems to cut across domains suggests additional factors and complexities relevant to humanitarian contexts not captured by the existing 4-domain model. Therefore, in addition to establishing the applicability of such models in the humanitarian space, this review also rapidly identifies that further work is required to conceptually and pragmatically refine a model fit for purpose in these settings.
Resource Allocation
The COVID-19 pandemic casts a spotlight on the concept of population-based triage, but developments continue to fall short of facilitating the translation of constructs to humanitarian settings. Despite its potential, population-based triage remains seemingly limited to infectious disease outbreaks, as demonstrated by formal attempts to operationalize population-based triage focusing primarily on ventilator scarcity during outbreaks of respiratory viruses.Reference Adeniji and Cusack44–Reference Daugherty Biddison, Faden and Gwon48 For many international health systems, the COVID-19 pandemic highlighted the shortcomings of standard triage practices, sparking discussion around population-based triage; however, contributions in the form of protocols or guidelinesReference Aziz, Arabi and Alhazzani49–Reference Sprung, Joynt and Christian53 remain specific to the scope of pandemic responses alone.
Through this review, we confirm the relevance of population-based triage approaches beyond the scope of pandemics through inductive coding, and a scarcity of available literature specific to the humanitarian landscape is also demonstrated. The ongoing significance of this scarcity rests alongside the inability to translate existing constructs from pandemic to non-pandemic settings and, thus, capitalize on the surge in pandemic-related outputs to address this gap. Therefore, we recommend that further research into the nature and operationalization of population-based triage and, more broadly, resource allocation principles is required specifically in the humanitarian health response.
Conclusion
Through undertaking this systematic review and EGM, we have identified and graphically displayed existing experience-based research in relation to the allocation of scarce health resources in humanitarian settings. Among the identified sources, data frequently related sufficiently to justify extraction and analysis but were not derived with explicit intent to contribute to this research area. Data extracts could be mapped across all surge capacity domains, although a large proportion of data could not be classified accordingly. In addition to identifying areas of scarcity, data analysis and mapping identified an emerging theme of “security” as a potentially necessary addition to existing models of surge capacity when translated to humanitarian settings. Adopting surge capacity models in humanitarian health-care research may provide a potential way forward in terms of reporting, collating, and maximizing the translation experiences and findings.
Author contribution
Zachary Horn – systematic search of literature, screening of articles (title and abstract, and full-text review), data extraction, data analysis, production of evidence gap map graphic, and manuscript drafting.
Jamie Ranse – screening of articles (title and abstract, and full-text review), manuscript editing and review, and supervision of doctoral candidate.
Andrea Marshall – screening of articles (title and abstract, and full-text review), manuscript editing and review, and supervision of doctoral candidate.
Funding statement
This research was supported by an Australian Government Research Training Program (RTP) Scholarship.
Competing interest
None.
Open access
