key: cord-0952088-1s47tga2
authors: Sonenthal, Paul D.; Nyirenda, Mulinda; Kasomekera, Noel; Marsh, Regan H.; Wroe, Emily B.; Scott, Kirstin W.; Bukhman, Alice; Connolly, Emilia; Minyaliwa, Tadala; Katete, Martha; Banda-Katha, Grace; Mukherjee, Joia S.; Rouhani, Shada A.
title: The Malawi emergency and critical care survey: A cross-sectional national facility assessment
date: 2022-01-13
journal: EClinicalMedicine
DOI: 10.1016/j.eclinm.2021.101245
sha: f98c1d947bdee165256164fd55f7993e85a62110
doc_id: 952088
cord_uid: 1s47tga2

BACKGROUND: Data on emergency and critical care (ECC) capacity in low-income countries (LICs) are needed to improve outcomes and make progress towards realizing the goal of Universal Health Coverage. METHODS: We developed a novel research instrument to assess public sector ECC capacity and service readiness in LICs. From January 20th to February 18th, 2020 we administered the instrument at all four central hospitals and a simple random sample of nine of 24 district hospitals in Malawi, a landlocked and predominantly rural LIC of 19·1 million people in Southern Africa. The instrument contained questions on the availability of key resources across three domains and was administered to hospital administrators and clinicians from outpatient departments, emergency departments, and inpatient units. Results were used to generate an ECC Readiness Score, with a possible range of 0 to 1, for each facility. FINDINGS: A total of 114 staff members across 13 hospitals completed interviews for this study. Three (33%) district hospitals and all four central hospitals had ECC Readiness Scores above 0·5 (p-value 0·070). Absent equipment was identified as the most common barrier to ECC Readiness. Central hospitals had higher median ECC Readiness Scores with less variability 0·82 (interquartile range: 0·80–0·89) than district hospitals (0·33, 0·23 to 0·50, p-value 0·021). INTERPRETATION: This is the first study to employ a systematic approach to assessing ECC capacity and service readiness at both district and central hospitals in Malawi and provides a framework for measuring ECC capacity in other LICs. Prior ECC assessments potentially overestimated equipment availability and our methodology may provide a more accurate approach. There is an urgent need for investments in ECC services, particularly at district hospitals which are more accessible to Malawi's predominantly rural population. These findings highlight the need for long-term investments in health systems strengthening and underscore the importance of understanding capacity in LIC settings to inform these efforts. FUNDING: Division of Pulmonary and Critical Care Medicine, Brigham and Women's Hospital and Department of Emergency Medicine, Brigham and Women's Hospital.

In the 40 years since the Alma Ata Declaration, there have been significant gains worldwide in the provision of primary care services as a cornerstone of Universal Health Coverage (UHC). 1 In contrast, the development of hospital-based services has remained limited, particularly in low-income countries (LICs). 2 Emergency and critical care (ECC) is the care of patients "who are critically-ill at arrival, or who were stable and subsequently deteriorated, and can be provided anywhere in the hospital: in the emergency department [ED] , the intensive care unit (ICU), medical wards, postoperative recovery units, and high-dependency units [HDUs] .". 3 Recent efforts have been directed at defining the most basic, core ECC capabilities that should be provided in all hospitals. 4 These services have potential to provide significant benefit in LICs where critically ill patients tend to be younger with fewer comorbidities. 5 In lower-middle income countries (LMICs), an estimated 54% of annual deaths are from conditions treatable by prehospital and facility-based emergency care. 6 Despite this high burden of illness, information on the status of ECC services in LICs is lacking, limiting development of interventions to improve care 7 as well as efforts to progress towards the objectives of UHC.

The World Health Organization (WHO) defines health service-specific readiness as "the capacity of health facilities to provide a specific service, measured through the presence of tracer items" 8 organized into domains such as staff and training, equipment, and diagnostics. Standardized sets of tracer items exist for measuring readiness to deliver a number of specific services including emergency obstetric care, HIV counselling and testing, malaria diagnosis and treatment, and chronic disease management. 8 However, there is no established methodology for measuring and reporting ECC service readiness.

Data from standardized assessments of ECC service readiness from LICs are needed to improve outcomes and advance towards the goal of UHC. The Malawi Emergency and Critical Care (MECC) survey is a crosssectional study that aims to measure ECC service readiness in district and central hospitals in Malawi.

Malawi is a landlocked LIC of 19¢1 million people 9 with 83% of the population living in rural areas. 10 The national health system is organized into four tiers. 11 The first two tiers are the community and primary levels, providing health services at health posts, dispensaries, village clinics, health centres, and community hospitals. The next tier comprises district hospitals, located in predominantly rural settings. The top tier consists of four central hospitals and one central psychiatric hospital. Malawi's health system faces many challenges including inadequate funding and shortages of

Evidence before this study

We identified seven studies reporting facility-level data on emergency and critical care (ECC) services from 244 low-income country (LIC) hospitals by searching PubMed for English language publications between Jan 1, 2000 and May 24, 2021 using terms related to ECC service delivery ("emergency care", "critical care", "intensive care unit", and "high dependency unit"), critical illness ("diabetic ketoacidosis", "respiratory failure", "shock", and "sepsis"), and the country setting (full list of LIC names and standard terms for LIC settings). All studies employed convenience samples and tertiary referral centres accounted for 92% of facilities in the four studies with hospital referral level information. Studies assessing critical care capacity outside of ICUs reported that critically ill patients were often managed in emergency departments and general medical wards.

To our knowledge this is the first systematic national assessment of facility-level ECC capacity in an LIC. Our inclusion of secondary hospitals expands the evidence base, with increased relevance to predominantly rural LIC populations. Additionally, this is one of the few studies to assess ECC across multiple locations in the hospital, including the emergency department (or outpatient department if there was no emergency department) and inpatient wards.

The available evidence shows that LICs hospitals provide treatment for critically ill patients in multiple areas, not just ICUs. There are significant gaps in ECC services, particularly at secondary hospitals, including lack of designated space, inadequate availability of equipment, and absence of clinical protocols. These findings have the potential to guide action by national governments as well as the international community to improve the availability and quality of ECC services in LICs.

staff, essential medicines, and equipment. 11 While there are no prospective nationwide evaluations of ECC, one study showed that only 3¢5% of the country's facilities are fully equipped to provide basic pediatric emergency care. 12 These challenges translate to poor outcomes with inpatient pediatric mortality above 7% 13 and adult sepsis mortality as high as 75%. 14 The MECC survey measured cross-sectional ECC service readiness at a sample of Malawian public sector hospitals. From January 20th to February 18th, 2020, the research team visited all four full service public sector central (tertiary) hospitals and a simple random sample of nine of 24 public sector district (secondary) hospitals in Malawi. All hospitals included in the study sample granted permission for an on-site visit. Ethics approval was obtained from the Partners Healthcare Institutional Review Board in Boston, USA (2019P003457) and the National Health Science Research Committee in Malawi (Protocol #19/05/2346, approval number 2346). The Malawi Ministry of Health also approved the study. Recognizing the urgent nature of the COVID-19 pandemic, a post-hoc analysis of relevant unit-level data from the MECC survey was rapidly published in early 2020. 15 

The MECC survey instrument measures unit-and facility-level ECC capacity and service readiness in LICs. The instrument contains questions drawn from three sources: the WHO Hospital Emergency Unit Assessment Tool (HEAT), 16 novel questions developed and piloted by the research team to assess ECC capacity in LICs, and three expert-developed questions specific to the Malawian context.

Novel questions included in the MECC survey instrument were developed by first generating a list of concepts and topics from review of previously published instruments, 16−20 standards established by international organizations, 21 and expert opinion. Questions were then formulated within the following three primary domains: (1) systems and space; (2) essential equipment, diagnostic tests, and medications; and (3) staff. Question response structures were selected to match the WHO HEAT survey tool. Questions were iteratively refined and reduced through a modified Nominal Group Technique including content experts in emergency medicine, pulmonary and critical care medicine, global health implementation, and survey methodology. Questions were then pre-tested by outside clinical experts from LICs. In November 2018, the questions were piloted using a convenience sample of staff at an LIC district hospital. We also administered an established clinical sensibility tool for survey development to gather information on comprehensiveness, clarity, and face validity. 22 The questions were further refined based on feedback from the pilot and clinical sensibility testing.

The finalized instrument collected participant-level data using WHO HEAT response structures. 16 Signal function questions asked participants to describe the availability of a given resource or ability to perform an intervention on a scale of 1 to 3, with 1 indicating generally unavailable, 2 some availability, and 3 adequate availability. For signal function questions we instructed participants to consider how often all patients in their unit who need the service or resource are able get it within the timeframe needed for emergency care or critical care. Questions asking the frequency of an activity were reported on a scale of 1 to 5, with 1 indicating almost never, 2 infrequently, 3 sometimes, 4 frequently, and 5 almost always.

For each signal function rated generally unavailable or some availability, participants were asked to identify all relevant barriers. There was no limit to the number of barriers participants could identify per signal function. Barriers were coded by participants and data collectors into WHO HEAT categories of infrastructure, absent equipment, broken equipment, stockout, personnel, training, user fees, and opening hours.

Eligible participants were 18 years or older and selfreported working as a nurse or clinician (ie, clinical officer, medical assistant, or physician) for at least one month in the selected hospital area (listed below). Potential participants were identified through discussions with local hospital leadership, personal introduction during facility visits, and announcements during hospital lectures and meetings. All participants provided written informed consent.

Using the instrument, we interviewed staff working in the following hospital areas: administration, EDs, outpatient departments (OPDs), medical wards, ICUs, and HDUs. Survey content for each unit was similar with adjustments based on varying anticipated care activities at each location. At each facility, the research team administered the survey in person to a hospital administrator and three clinicians in each of the ED, medical ward, and ICU and/or HDU, if present. For hospitals without a designated ED, the OPD was substituted as the site most likely to receive new patients arriving to the facility. Data were collected from three clinicians per unit to address discrepancies among participant responses when generating unit-level data. The survey took approximately 45 min to complete. To reduce question burden, questions on availability of protocols, electronic cardiac monitoring, crash trolley, social work, security, dietician, physiotherapist, and spiritual support were only asked to one designated clinical lead out of the three respondents from each unit.

The primary outcome was the ECC Readiness Score, which is a measure of service readiness to provide ECC. The research team developed the ECC Readiness Score using the WHO's recommended methodology for measuring service-specific readiness as "the cumulative availability of components required in health facilities to deliver specific services, expressed as percentage.". 8 Tracer items for the ECC Readiness Score were divided into three domains: (1) systems and space; (2) essential equipment, diagnostic tests, and medications; and (3) staff. The research team selected tracer items essential to basic ECC practice and analysed their availability at a facility-level by combining unit-level data from the ED/ OPD and inpatient units (ie, medical ward, ICU, and HDU). To reflect the entire care continuum, a tracer item was considered present (or adequately available) at the facility-level if it was reported as available in the ED/ OPD and at least one inpatient unit, with the exceptions of tracer items only applicable to specific units (ie, triage, method of identifying critically ill patients on medical wards, and twice daily reassessment of critically ill inpatients). This facility level approach was selected because ECC, by definition, occurs across multiple areas of the hospital, 3 and is consistent with prior assessments of ECC capacity. 19, 23 To generate facility-level tracer item data, unit-level data were first calculated by averaging the three participant responses within each unit. For signal function tracer items, we considered there to be "adequate availability" within a unit if the average score was >2¢5 (out of 3). For yes/no questions, the tracer item was considered present if at least two staff members answered affirmatively. For tracer items reporting frequencies, an item was considered present if the average unit score was >4 out of 5. Using emergency obstetric care service readiness as a model, 17 we first calculated domain subscores as the unweighted proportion of adequately available tracer items and then calculated ECC Readiness Score as the unweighted mean of the three domain subscores so that each domain contributed 33¢3% to the ECC Readiness Score.

Facility-level barrier data are reported as a percentage calculated using the frequency with which each barrier category was identified relative to the total number of times participants at the facility were asked to identify barriers (ie, frequency with which participants rated a signal function as generally unavailable or some availability).

It is unlikely a service or resource is rapidly and reliably accessible if a clinician is unaware of its availability. Therefore, for the purposes of data analysis, a response of "don't know" was considered equivalent to "generally unavailable" or "no" for signal functions and yes/no questions, respectively. For tracer items reporting frequencies, we treated an answer of "don't know" as missing/incomplete data. Frequency variables were treated differently than other tracer items because an event may still occur even if a respondent is unaware of how often it occurs. If data were only available from two respondents at a unit, we used the same methods as above. If data were available for less than two respondents in a unit, we considered the tracer item to be missing/incomplete for that unit.

Study data were collected and managed using REDCap electronic data capture tools. 24 Data were analysed in Stata (Release 16). Continuous and ordinal variables were summarized by medians and interquartile ranges. Categorical variables were described using frequencies, proportions, and 95% confidence intervals. Individual hospital data were not reported to maintain anonymity.

We compared proportions of central and district hospitals with ECC scores above 0¢5. The use of 0¢5 for dichotomization was selected for consistency with prior studies employing service readiness methodology. 25−27 In additional analyses we compared median ECC Readiness Scores, domain sub-scores, and barrier frequencies between district and central hospitals using Fisher's Exact test. A nominal level of 5% for statistical significance (two-tailed) was used for comparisons.

We also assessed instrument reliability (see Supplement). Interrater reliability of newly developed questions was assessed by calculating the intraclass correlation coefficients for one-way random-effects models measuring average absolute agreement of participants within each hospital unit. Internal consistency was assessed by calculating Kuder-Richardson coefficients for groups of questions tapping into the same construct.

We hypothesized that less than 20% of district hospitals surveyed would have ECC Readiness Scores above 0¢5. This determination was made based on prior studies employing readiness score methodology, literature on availability of critical care resources in similar settings, [17] [18] [19] [20] 23, 28, 29 review of pilot data, and local expert opinion. The district hospital sample size of 9 was calculated using methodology recommended by the WHO Service Availability and Readiness Assessment (SARA), 17 assuming a confidence level of 95 and a 15% margin of error (Supplement). A simple random sample of the 9 district hospitals included in the MECC survey was generated from a list of all 24 Malawian public sector district hospitals using a random number generator.

The research presented in this manuscript adheres to the reporting standards of the Strengthening the Reporting of Observational studies in Epidemiology (STROBE) Statement and STROBE checklist for crosssectional studies.

The funders had no role in study design; collection, analysis, and interpretation of data; decision to publish; or preparation of the manuscript.

A total of 114 staff members across nine district and four central hospitals participated (Table 1 ); 54% were nurses. All staff members who were approached by the research team agreed to participate. We interviewed three staff members from 33 (97%) units. At the one remaining unit, we interviewed the only two staff members present during data collection and data were treated as missing for one staff member. The median facility catchment area populations were 601,000 (Interquartile range (IQR): 451,000 to 681,000) for district hospitals and 4,492,000 (3,351,000 to 6,048,000) for central hospitals, with median annual inpatient admissions of 13,300 (10,000 to 21,000) and 35,100 (27,900 to 51,600), respectively.

Care of critically ill patients occurred across multiple hospital areas. Emergency units (defined as EDs and designated areas within OPDs) and operating theatres were the most common sites of critical care, used by 77% of all hospitals. Three (33%) district hospitals and all four central hospitals had a designated ICU and/or HDU. In addition to designated units, 3 (75%) central and 6 (67%) district hospitals also cohorted critically ill patients in general medical wards.

Only 3 (33%) district hospitals, compared to all 4 central hospitals (Table 2) , had ECC Readiness Scores above 0¢5 (p-value 0¢070). Of the 3 district hospitals with HDUs, only 1 had an ECC Readiness Score above 0¢5. Figure 1 shows the distribution of ECC Readiness Post-operative recovery area n (%) 2 (22%) 3 (75%)

Emergency unit^n (%) 6 (67%) 4 (100%)

Cohorted areas within general inpatient wards n (%) 6 (67%) 3 (75%)

Interspersed throughout general inpatient wards n (%) 1 (11%) 3 (75%)

Operating theater n (%) 7 (78%) 3 (75%) Increased nursing to patient ratios for critically ill patients* n (%)

Staff domain sub-score median (IQR) 0¢67 (0¢33 to 0¢67) 0¢67 (0¢33 to 0¢67) 0¢67 (0¢67 to 1¢0) 0¢20 +

Available critical care protocol y n (%) 9 (69%) 5 (56%) 4 (100%) 0¢50 (0¢30 to 0¢80) 0¢40 (0¢30 to 0¢50) 0¢85 (0¢75 to 0¢90) 0¢007 + (Figure 2 ). There was no clear evidence of differences in reported barriers between district and central hospitals.

Radiologists were available to interpret imaging results at 3 (75%) central and 2 (22%) district hospitals (Table 3) . Social work staff were adequately available at half of central and a third of district hospitals. Dedicated spiritual support staff were adequately available at 1 (25%) central and no district hospitals. The most commonly available clinical protocol was for diabetic ketoacidosis, at all 4 central and 7 (78%) district hospitals. At least half of central and a third of district hospitals had clinical protocols for volume resuscitation, asthma, pneumonia, and sepsis. No hospitals had protocols for end-of-life care.

The ability to perform non-portable xray was adequately available at 8 (89%) district and 3 (75%) central hospitals (Table 4 ). Neither electrocardiograms nor cardiac markers were available at any hospitals. Most facilities were able to perform bag-valve-mask ventilation. All 4 central hospitals were able to administer insulin and treat hypoglycemia as opposed to 5 (56%) and 3 (33%) district hospitals, respectively. Only 1 (11%) district and 3 (75%) central hospitals reported adequate ability to communicate poor prognoses with patients and their families and no hospitals had the ability to de-escalate care (ie, transition to comfort measures). 

ECC is an essential component of high-quality health systems but there is limited information about facilitylevel ECC capacity from LICs. Applying the WHO methodology for assessing service readiness, we have provided a detailed description of ECC capacity and service readiness across district and central hospitals in an LIC health system. We found that critically ill patients were treated across multiple hospital units, even in facilities with designated critical care areas, highlighting that the provision of ECC in Malawi is not limited to ICUs or HDUs, supporting a facility-wide approach to understanding these services. Although most hospitals had ECC Readiness Scores above 0¢5, significant gaps exist. Effective care of critically ill patients requires timely collaboration across multiple areas of the hospital from initial triage, recognition, and stabilization to ongoing monitoring, treatment, and supportive care. An alternative interpretation of ECC Readiness Score is as a measure of the extent of improvement and increased resources required for optimal care of critically ill patients. In other words, what inputs are needed to achieve an ECC Readiness Score of 100%.

Protocols and standardized procedures for ECC were notably absent at many hospitals. Only 62% of facilities used triage systems, which ensure rapid care for the sickest patients. 30 On the inpatient wards, less than half of facilities regularly reassessed critically ill patients and only 15% had a method of identifying critically ill patients, a known practical and effective intervention in LIC hospitals. 31 Of particular concern is the large disparity in ECC readiness between levels of Malawi's health systemdistrict hospitals had lower and more variable capacity to provide ECC services compared to central hospitals. The population of Malawi, like most LICs, is predominantly rural with only 17% residing in the urban areas 10 where central hospitals are located. Due to their geographic distribution throughout the country, most critically ill patients are likely to receive at least initial care at district hospitals 32 . Although district and central hospitals are designed and resourced to provide different levels of care, the ECC Readiness Score measures only Upper adjacent values, n u , defined as n u < 75th percentile + 1.5*interquartile range and n u+1 >75th percentile + 1.5*interquartile range. Lower adjacent values, n l , defined as n l > 25th percentile -1.5*interquartile range and n l-1 < 25th percentile -1.5*interquartile range. The numerator for a given facility level barrier category is calculated as the number of times the barrier category was identified by a participant at the facility for signal functions included in the ECC Readiness Score. The denominator is the number of times participants at the facility were asked to identify barriers for those signal functions (ie, the number of times participants rated any of the signal functions as "generally unavailable" or "somewhat available"). (Signal functions included in the ECC Readiness Score: Administration of therapies for reactive airway disease, place peripheral intravenous access, place urinary catheter, administration of adrenaline, diagnose and treat hypoglycemia, perform portable xray, perform non-portable xray, availability of ultrasound machine, check electrolytes, check blood urea nitrogen/creatinine, check hemoglobin, perform transfusion, administration of intravenous fluids, administration of intravenous or intramuscular antibiotics, and administration of intravenous vasopressors.) basic essential elements of ECC which should be available at all secondary and tertiary facilities. Accordingly, there is significant overlap between the tracer items included in the ECC Readiness Score and a recent consensus statement on ECC processes that should be available to critically ill patients in all hospitals. 4 When probed, participants identified absent equipment as the most common barrier to performing signal functions. This includes situations where a single item of equipment is shared across multiple units, limiting timely access. This suggests that even when an item is technically "present" it may not be immediately available for patients when needed, thus limiting care delivery. This is particularly relevant for critically ill patients who require interventions in a timely manner. Interestingly, a different instrument used in two prior assessments of ECC in Tanzania and Sierra Leone reported availability of equipment and commodities among the highest scoring domains. 19, 23 This discrepancy between prior work and this study may be attributable to methodology: our survey asked about the ability to deliver an intervention or perform a test when needed, considering equipment as just one component of a clinical process. In other words, our question-response structure accounted for the possibility that the mere presence of equipment within a hospital does not guarantee it is always available when needed. Using signal function questions with follow-up barrier probes and interviewing multiple participants from each unit may have improved the accuracy of our estimates.

To our knowledge, this is one of the first assessments to analyze ECC at a facility level in an LIC setting by considering capabilities across multiple locations in the hospital, including the ED (or OPD if there was no ED) and inpatient units. By including district hospitals, these findings expand existing evidence on ECC in LICs, which is biased towards academic and central referral hospitals in large cities. 33 The Malawi Ministry of Health, recognizing the importance of long-term health system investment for ECC services, recently developed a 10-year ECC national strategy. These results provide the first national baseline for ECC capacity at district and central hospitals in Malawi and have the potential to influence policy interventions across the health system. The MECC survey also affords an opportunity to measure impact of these interventions through re-administration. More broadly, the MECC survey provides a standard approach for harmonized multi-country assessments of ECC capacity in LICs, emphasizing inclusion of district hospitals and multiple care areas within hospitals. These findings highlight the need for interventions to improve ECC service readiness in LICs. Although ECC health systems strengthening requires a combination of equipment, commodities, clinical protocols, dedicated space, and trained staff, these results provide valuable insight into specific areas of immediate need. To ensure universal access to quality ECC, particular attention should be paid to district hospitals, where most patients initially seek care and ECC Readiness is low. These data also suggest a need for designated critical care areas and proven interventions such as early warning systems, 31 critical care training, 34 and triage. 30 Adequate access to proper equipment when caring for critically ill patients is essential. Our findings support a more nuanced approach to understanding equipment availability in the context of ECC. When planning health system strengthening interventions, it should not be assumed that the mere presence of equipment equates with adequate availability and usability. 

Procurement should be linked to burden of disease by adapting tools such as the Partners In Health UHC Monitoring and Planning Tool 35 to include ECC services. Finally, the capacity to care for patients at the end of life, provide spiritual support, and transition to comfort-focused care when appropriate must not be overlooked as an essential aspect of ECC.

Although this study provides important data from an LIC context, there are several limitations. Data were collected from a single country. Although similar gaps are likely present in other LICs, the degree of generalisability is unclear. Although the facility sample was relatively small, it included nearly half of all public sector secondary and tertiary hospitals in Malawi. The sample size was calculated using a formula recommended by the WHO SARA. However, the WHO SARA advises oversampling when variation is likely (ie, smaller strata), 17 suggesting that the study was underpowered. Despite this limitation, we believe the MECC survey still provides valuable findings. Furthermore, the random selection of district hospitals helped reduce the likelihood that sampled hospitals differ significantly from nonsampled ones. Administration of the survey in person may have introduced some reporting bias which could result in overestimation of resource availability and capacity. Though our primary outcome, ECC Readiness Score, has not been validated, we followed the approach to quantifying service specific readiness used by the WHO SARA. We also ensured expert consultation and piloted the instrument prior to implementation. Future studies should explore the relationship between ECC Readiness Scores and patient outcomes.

In conclusion, this study provides a detailed description of ECC capacity and service readiness across district and central hospitals in an LIC health system and establishes a framework for measuring ECC in other LICs. Using signal function questions may provide a more accurate method of assessing availability of equipment and commodities. Future efforts to improve ECC services should include district hospitals where the need is greatest. This study underscores the need for long-term investments in health systems strengthening through proven interventions as well as the importance of assessing capacity and service-specific readiness in LIC settings to inform these efforts. 

De-identified data collected for this study will be made available via email request to the first author for noncommercial research purposes. Data collection tools will be made available upon email request to the first author with the exception of questions from the WHO HEAT, which can be requested from the WHO Emergency Care team via email (emergencycare@who.int).

PDS received support for this study from Brigham and Women's Hospital, Division of Pulmonary and Critical Care Medicine in the form of a faculty research fund and has received consulting fees from the University of California-San Francisco/Sustaining Technical and Analytic Resources as lead technical advisor for ventilator technical assistance in Haiti. SAR received support for this study in the form of a seed grant from Brigham and Women's Hospital Department of Emergency Medicine. JSM is the Chief Medical Officer at Partners In Health and sits on the boards of Village Health Works (Burundi/Muso and Mali), The Institute for Justice and Democracy in Haiti, and Free Speech for People.

A vision for primary health care in the 21st century: towards universal health coverage and the sustainable development goals

Intensive care in low-income countries−a critical need

The global need for essential emergency and critical care

Essential emergency and critical care: a consensus among global clinical experts

Global health care of the critically ill in lowresource settings

The International Bank for Reconstruction and Development /The World Bank

Critical care in resourcerestricted settings

Monitoring the building blocks of health systems: a handbook of indicators and their measurement strategies. Geneva: World Health Organization

World Bank Country and Lending Groups

World Bank Group

World Urbanization Prospects

Lilongwe: Government of the Republic of Malawi

Accessibility of basic paediatric emergency care in Malawi: analysis of a national facility census

Decreased pediatric hospital mortality after an intervention to improve emergency care in Lilongwe, Malawi

Prevalence Charles A. Etiology, and outcome of sepsis among critically Ill patients in Malawi

COVID-19 preparedness in Malawi

World Health Organization. Emergency care [Internet]. Geneva: World Health Organization

World Health Organization. Service availability and readiness assessment (SARA) an annual monitoring system for service delivery: implementation guide, version 2.2. Geneva: World Health Organization

A survey on critical care resources and practices in low-and middle-income countries

Emergency and critical care services in Tanzania: a survey of ten hospitals

Development of an intensive care unit resource assessment survey for the care of critically ill patients in resource-limited settings

What is an intensive care unit? A report of the task force of the world federation of societies of intensive and critical care medicine

A guide for the design and conduct of self-administered surveys of clinicians

Emergency care capacity in Freetown, sierra Leone: a service evaluation

Research electronic data capture (REDCap)-A metadata-driven methodology and workflow process for providing translational research informatics support

Health facility service availability and readiness to provide basic emergency obstetric and newborn care in a low-resource setting: evidence from a Tanzania national survey

Preparedness of lower-level health facilities and the associated factors for the outpatient primary care of hypertension: evidence from Tanzanian national survey

Determinants of facility readiness for integration of family planning with HIV testing and counseling services: evidence from the Tanzania service provision assessment survey

Critical care capacity in Haiti: a nationwide cross-sectional survey

Availability of critical care resources to treat patients with severe sepsis or septic shock in Africa: a self-reported, continent-wide survey of anaesthesia providers

Emergency department design in low-and middle-income settings: lessons from a university hospital in Haiti

Modified early warning score (MEWS) identifies critical illness among ward patients in a resource restricted setting in Kampala, Uganda: a prospective observational study

Gaps in physical access to emergency care in sub-Saharan Africa

Intensive care unit capacity in low-income countries: a systematic review

Impact of a structured ICU training programme in resource-limited settings in Asia

A practical approach to universal health coverage

We thank the Ministry of Health and Population of the Republic of Malawi, including district medical officers and the office of non-communicable diseases for their support and dedication to these important topics. We thank the staff at Abwenzi Pa Za Umoyo/Partners In Health for their support of survey implementation. Most of all, we wish to thank all the study participants.

Supplementary material associated with this article can be found in the online version at doi:10.1016/j. eclinm.2021.101245.