Article Text

Original research
Comprehensive assessment of pediatric acute and inpatient care at a tertiary referral hospital in Malawi: opportunities for quality improvement
  1. Elizabeth Fitzgerald1,
  2. Emily Jane Ciccone2,
  3. Tisungane Mvalo3,4,
  4. Msandeni Chiume5,
  5. Yamikani Mgusha5,
  6. Treasure B Mkaliainga5,
  7. Alyssa Evelyn Tilly6,
  8. Jane Chen7,
  9. Griffin Bell8,
  10. Heather Crouse9,
  11. Jeff A Robison10,
  12. Michelle Eckerle11,12
  1. 1Department of Pediatrics, Division of Emergency Medicine, The University of North Carolina at Chapel Hill School of Medicine, Chapel Hill, North Carolina, USA
  2. 2Division of Infectious Diseases, University of North Carolina at Chapel Hill School of Medicine, Chapel Hill, North Carolina, USA
  3. 3Department of Pediatrics, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA
  4. 4University of North Carolina Project Malawi, Lilongwe, Malawi
  5. 5Department of Pediatrics, Kamuzu Central Hospital, Lilongwe, Malawi
  6. 6Divisions of General Medicine and Clinical Epidemiology, University of North Carolina at Chapel Hill School of Medicine, Chapel Hill, North Carolina, USA
  7. 7Institute for Global Health and Infectious Disease, The University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA
  8. 8Epidemiology, The University of North Carolina at Chapel Hill Gillings School of Global Public Health, Chapel Hill, North Carolina, USA
  9. 9Department of Pediatrics, Division of Emergency Medicine, Baylor College of Medicine, Houston, Texas, USA
  10. 10Division of Pediatric Emergency Medicine, The University of Utah School of Medicine, Salt Lake City, Utah, USA
  11. 11Division of Pediatric Emergency Medicine, Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio, USA
  12. 12University of Cincinnati College of Medicine, Cincinnati, Ohio, USA
  1. Correspondence to Dr Elizabeth Fitzgerald; elizabeth.fitzgerald{at}


Background Despite the reduction in global under-5 mortality over the last decade, childhood deaths remain high. To combat this, there has been a shift in focus from disease-specific interventions to use of healthcare data for resource allocation, evaluation of performance and impact, and accountability. This is a descriptive analysis of data derived from a prospective cohort study describing paediatric admissions to a tertiary referral hospital in Malawi for the purpose of process evaluation and quality improvement.

Methods Using a REDCap database, we collected data for patients admitted acutely to Kamuzu Central Hospital, a tertiary referral centre in the central region. Data were collected from 17 123 paediatric inpatients from 2017 to 2020.

Results Approximately 6% of patients presented with either two or more danger signs or severely abnormal vital signs. Infants less than 6 months, who had the highest mortality rate, were also the most critically ill on arrival to the hospital. Sepsis was diagnosed in about 20% of children across all age groups. Protocols for the management of high-volume, lower-acuity conditions such as uncomplicated malaria and pneumonia were generally well adhered to, but there was a low rate of completion for labs, radiology studies and subspecialty consultations required to provide care for high acuity or complex conditions. The overall mortality rate was 4%, and 60% of deaths occurred within the first 48 hours of admission.

Conclusion Our data highlight the need to improve the quality of care provided at this tertiary-level centre by focusing on the initial stabilisation of high-acuity patients and augmenting resources to provide comprehensive care. This may include capacity building through the training of specialists, implementation of clinical processes, provision of specialised equipment and increasing access to and reliability of ancillary services. Data collection, analysis and routine use in policy and decision-making must be a pillar on which improvement is built.

  • Data Collection
  • Infant
  • Mortality
  • Epidemiology

Data availability statement

Data are available upon reasonable request. For further information email the corresponding author.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • Worldwide, under-5 mortality remains unacceptably high, and improving the quality of acute care in low-and-middle-income countries can have a significant impact on decreasing global paediatric mortality.

  • Routine use of healthcare data is essential for resource allocation, evaluation of impact and accountability.


  • We implemented a database describing acute paediatric admissions to a tertiary referral hospital in Malawi for the purpose of process evaluation and quality improvement.

  • Protocols for common conditions were generally well adhered to, yet overall mortality remained high, with most deaths occurring in the first 48 hours.

  • There was a low rate of completion for laboratory tests, radiology studies and subspecialty consultations.


  • There needs to be a focus on improving care of high-acuity patients through investments in capacity building for paediatric medical and nursing specialists, specialised equipment and strengthening of ancillary services.

  • This improvement needs to be based on the routine use of reliable healthcare data.


Despite significant reductions in global under-5 (U5) mortality, childhood deaths remain unacceptably high. The United Nations Millennium Development Goals (MDGs), put forth in 2000, sought to reduce child mortality by 2015 by improving access to healthcare, mainly through global efforts focused on preventative or disease-specific interventions.1 While the global U5 mortality rate fell by 53% from 1990 to 2015, there were still nearly 6 million annual child deaths at the end of that period.2 In response, the United Nations issued the sustainable development goals which call for universal health coverage (UHC) and a focus on ensuring access to high-quality healthcare.3

Although access to basic healthcare has improved dramatically in the last few decades, there are significant deficits in the quality of care provided to patients in low-income and middle-income countries (LMICs). It is estimated that lack of high-quality care resulted in 224 million years of life lost and 5 million deaths of children and adults in 2016.4 5 It is critical to ensure that, as countries strive to achieve UHC, the quality of care is also improved.

The acquisition and availability of reliable healthcare data are a priority for ensuring a high standard of care as healthcare systems are developed or strengthened. Quality healthcare data are essential for resource allocation, evaluation of performance and impact, and accountability.6 While most health information systems in LMICs measure some basic indicators of health and processes at the facility, district and national levels, the quality of data and its use for systems improvement are lacking.6–10 Global health leaders have called for countries to strengthen systems-level and facility-level data collection systems so that care delivery can be routinely measured and improved.6

The discrepancy between access to care and access to high-quality care, and the limitations of current data systems are highlighted in Malawi. Although Malawi notably met MDG 4 and reduced its under-5 child mortality by two-thirds before 2015,11 a recent data quality analysis in Malawi showed challenges with the availability, completeness and accuracy of data generated by the national Health Management Information Systems (HMIS) used in all health centres and hospitals.12 Despite paediatric inpatient mortality rates at tertiary government hospitals ranging from 3.3% to 8.7%,13–16 the data collection system currently in place in these facilities is unable to provide timely and meaningful aggregated information to clinical teams and policy-makers, hampering the ability to monitor or improve care on a local and national level.

Through a collaborative, multi-institutional partnership,17 we implemented a comprehensive database describing acute paediatric admissions to one tertiary referral hospital in Malawi.18 This database was intended to augment the existing HMIS system by acting as a more comprehensive electronic medical record, to provide a repository of data for process evaluation and improvement interventions at the facility level and to contribute to the global body of literature about paediatric inpatient care in LMICs. Here, we describe the characteristics, diagnoses and outcomes of the patients included in the database from 1 January 2018 through 31 December 2019. We further describe patients’ access to quality care, as measured by adherence to relevant local protocols and completion of diagnostic tests and studies.


Study site

The study was conducted at Kamuzu Central Hospital (KCH), a 900-bed tertiary referral centre in Lilongwe, Malawi. KCH serves a population of about 7 million people in the central region of the country, an estimated 43% of whom are between the ages of 0 and 14 years.19 The paediatric department has 199 beds and admits approximately 20 000 patients each year.

Study population

A detailed description of the methods for the design and implementation of this database, including the collaborative process of creating the data collection tools, the assembly and training of the team, and the evolution of the data collection process, is provided elsewhere.18 In brief, from 1 January 2018 through 31 December 2019, we collected data at the point of admission for all patients ages 6 months to 36 months admitted acutely through the outpatient (under 5) department to the KCH paediatric ward; in January 2019, the data collection cohort was expanded to include all admitted patients ages 2 weeks to 60 months. Patients were included if they were admitted for the evaluation and treatment of an acute medical illness and were excluded if they were admitted for preventive care services (such as vaccine administration), follow-up of a previously diagnosed condition (cancer or HIV), if they fell outside of the designated age range, or if they presented acutely but were discharged home without being admitted. In addition, neonates in kangaroo mother care and the neonatal high acuity unit, known oncology patients and patients admitted for trauma or injury through the casualty department were not included in data collection, as they are cared for in separate wards. Written consent was not obtained from guardians as the data collected were part of routine clinical documentation.

Patient involvement

Data were collected as part of routine documentation of medical records for patients, so they were not involved in the design of the study. The project was prioritised and led by the local healthcare team, as documented in previous publications.18

Data collection

The flow of admitted patients through the hospital from initial presentation to discharge has been previously described.18 Briefly, patients were triaged, evaluated and stabilised in the ‘Under 5’ unit (U5), then moved to the inpatient wards. Data were collected from guardians’ reports, the patient’s health passport and the written hospital record at admission on each day of the hospitalisation (up to day 10), and at time of discharge or death. Data were entered into a REDCap database at the point of collection.20 Vital signs were collected by the clinician or data clerk, and admission and discharge diagnoses were collected from the written medical record as documented by the clinician.


Vital signs were categorised as severely abnormal based on a previous study performed in Malawi14 (see footnotes for table 1 for exact values). Danger signs were based on the WHO’s Emergency Triage Assessment and Treatment (ETAT) system.21–23

Table 1

Patient clinical characteristics, 2018–2019

Pneumonia and bronchiolitis were consolidated into one category representing clinically diagnosed lower respiratory tract infection (LRTI). Malaria included the following admission diagnoses: uncomplicated malaria, malaria with severe anaemia and malaria with convulsions. In the analysis of guideline-concordant care (figure 2), pneumonia was kept separate as its recommended management is different than that of bronchiolitis. The WHO definition for severe acute malnutrition (SAM) was used (detailed in the footnotes for table 2).

Figure 2

Proportion of children for whom key processes of guideline-adherent care were completed in the under-5 unit by admission diagnosis. †Antibiotics include ampicillin, gentamicin, amoxicillin, ceftriaxone and benzyl penicillin. ‡Considered completed if weight and height measured, or mid-upper arm circumference measured, or oedema assessed. §Transfusions ordered in U5 or on inpatient wards, and resulted before discharge or death. Hgb, haemoglobin; MRDT, malaria rapid diagnostic test; NRU, nutritional rehabilitation unit; RBS, random blood sugar; U5, under-5 unit.

Table 2

Patient demographics, 2018–2019

Data analysis

We used descriptive statistics to characterise the patient population and process measures. To assess admission diagnoses over time, we selected the five most common diagnoses by age group (2 weeks to 5 months, 6–36 months, 37–60 months) and presented them graphically over time. All analyses were conducted using R V.4.0.2 (Vienna, Austria).

Data dissemination

Members of the clinical and nurse leadership teams from the KCH Department of Pediatrics were provided unlimited access to the data as well as training in the use of REDCap to generate reports. The paediatric department was provided with focused summarised data monthly in the form of digital slide decks disseminated by email and WhatsApp and formal presentations at the paediatric department morning handover meetings.


Between 1 January 2018 and 31 December 2019, data were collected from 17 123 paediatric inpatients (table 2). There was a slight male predominance (n=9819, 57%). Most children had no chronic medical conditions (n=16 305, 95%), were referred from primary health centres (n=12 667, 74%), were up to date on their immunisations (n=12 857, 75%) and had an unknown HIV status (n=12 873, 75%) at admission. Approximately 10% (n=1790) met the WHO criteria for SAM at presentation based on their anthropometric measurements (table 2).

Most of the patients had at least two vital signs (temperature, heart rate, respiratory rate, oxygen saturation and/or blood pressure) measured by data clerks or hospital staff (n=13 327, 78%, data not shown), with temperature most frequently measured. In 6% of patients (n=974), 2 or more of the vital signs recorded were severely abnormal (see table 1 footnotes for definitions). The most common danger sign at presentation (respiratory distress, capillary refill >3 s or convulsions) was respiratory distress (n=5899, 34%), and about 6% of patients (n=1002) had 2 or more danger signs documented (see table 1 footnotes for definitions). Older children were almost seven times more likely to present with convulsions than infants, and infants less than 6 months were 1.7 times more likely to present in respiratory distress than older children (table 1).

Admission diagnoses

Admission diagnoses varied with age (table 3). In older children, LRTIs were less frequent, and malaria was more frequent; only 27% of older children presented with LRTIs (n=334), but almost half presented with some form of malaria (n=616). In the less than 6-month age group, 66% (n=1273) presented with LRTIs and only 6% (n=121) with malaria. Sepsis was diagnosed in about 20% of children across all age groups. We also observed seasonal patterns by age group for diseases such as LRTIs, which peaked in February, and malaria, which peaked in May (figure 1).

Table 3

Five most prevalent admissions diagnoses by age group

Figure 1

The five most prevalent admissions diagnoses by age group and month of presentation. LRTI, lower respiratory tract infection.

Process indicators

Figure 2 describes initial patient management for common admission diagnoses; adherence to local protocols varied. Evaluation and management of malaria were largely consistent with national protocols24; most patients diagnosed with malaria at admission had a point-of-care malaria rapid diagnostic test (MRDT) completed on initial evaluation (n=5421, 92%), and most who were positive were treated with artesunate (n=3984, 92%). Similarly, 99% of patients diagnosed with pneumonia received antibiotics on admission, although only 80% had oxygen saturation documented, and only 78% of those with oxygen saturation below 90% were placed on O2 therapy. Nearly all patients with an admission diagnosis of SAM had nutritional assessments completed (n=682, 100%); however, only 27% (n=181) had a random blood sugar performed, which is one of the first steps in the inpatient management of SAM in Malawi.25 A consultation at the time of hospitalisation with the nutritional rehabilitation unit team that manages malnourished inpatients occurred about half of the time (n=366, 54%). Dehydration was addressed with fluid resuscitation in most patients (n=1115, 86%), but only 66% of children with a haemoglobin less than 4 g/dL (n=336) received a blood transfusion while hospitalised. Ninety-one per cent of patients diagnosed with sepsis received antibiotics; however, only 24% of those with septic shock (as diagnosed by capillary refill greater than 3 s) received any fluid resuscitation and 14% were placed on oxygen therapy.

Diagnostic testing

Most inpatients had some laboratory evaluation done on arrival (n=14 277, 83%), with MRDT being the most common test performed (n=13 563, 79%, data not shown). Overall, the rates of completion for diagnostic investigations at KCH were low, ranging from 21% to 49% for laboratory tests, 32% to 42% for radiology studies and 13% to 70% for subspecialty consultations (figure 3).

Figure 3

Laboratory tests, radiology studies and consultations ordered at presentation and per cent of orders completed during hospitalisation. Percent indicates the percent of tests, studies or consults that were conducted during hospitalisation. CM, crossmatch; ENT, ear nose and throat; FBC, full blood count; Hem/Onc, haematology/oncology; LP, lumbar puncture; MPS/PBF, malaria parasite smear/peripheral blood film; NRU, nutritional rehabilitation unit; U&E, urine and electrolytes.


The overall mortality rate for our entire study population was 4%, and 7% of patients either absconded or had missing data so an outcome could not be determined (table 4). Infants under 6 months had the highest overall mortality rate compared with children in other age groups. Most deaths occurred within the first 24 (44%) and 48 (60%) hours of hospitalisation (data not shown). The diagnoses associated with the highest overall number of deaths align with the diagnoses with the highest number of admissions for each age group (LRTI in infants less than 6 months, and malaria in children greater than 6 months). The diagnosis of sepsis was common among patients who died across all ages. Congenital heart disease accounted for 10% (n=11) of deaths in infants aged 2 weeks to 5 months and malnutrition accounted for 15% (n=80) of deaths in the 6–36 months age group.

Table 4

Patient outcomes and five most prevalent diagnoses among those who died, by age category


Here, we report a descriptive analysis of more than 17 000 paediatric admissions to a tertiary-level hospital in Malawi over a 2-year period, providing a unique level of detail about patient characteristics and hospital care practices in a low-resource setting. While there have been multiple studies done in Malawi looking at the care of hospitalised children related to specific diseases or care processes, describing specific cohorts, or focusing on potential determinants of mortality, to our knowledge, this is the first study to provide a comprehensive description of all children admitted for care to the medical wards.

Paediatric inpatients were largely referred for care for an acute illness. Infants less than 6 months, who had the highest mortality rate overall, were also the most critically ill on arrival; they were twice as likely to have two or more abnormal vital signs on admission as older children. Vital signs were considered severely abnormal based on a previous study performed in Malawi that associated abnormal vital signs with a significant increase in mortality.14 Unfortunately, hospitals in low-resource settings such as this are challenged by both equipment and staffing shortages that frequently limit the collection of vital sign data. A mortality audit completed at KCH in 2018 revealed that 44% of patients who died had no vital signs done during the first 24 hours of admission.15 To address this potential gap, we trained data clerks to support clinicians by obtaining vital signs, therefore improving their overall collection. However, patients who were the most critically ill frequently bypassed the under 5 triage area and went straight to the wards, so that information about the initial vital signs of the sickest children may be underrepresented. We attempted to mitigate this challenge by having data clerks travel bed to bed through the wards several times per day to obtain information about patients who may not have come in through the under 5 area.

ETAT is a triage system used widely in low-resource settings, including KCH, that has proven successful in reducing reliance on vital signs equipment by providing healthcare workers of all cadres with the tools to rapidly complete a clinical assessment of patients and identify those most in need of immediate intervention.16 22 23 The danger signs included on our admission form, taken from the KCH admission form, are modelled on ETAT danger signs.22 23 Although ETAT has been shown to be an effective triage and stabilisation tool in LMICs and in Malawi, hospitals in low-resource settings often struggle with staff shortages and turnover which hamper its consistent use.16 In our study, clinician documentation of danger signs and presumed use of ETAT was generally poor. It is possible that, as with vital signs, the sickest children did not have documentation completed as they were immediately taken to the wards for further management.

Our study showed that admission diagnoses varied by age and season, with infants under 6 months largely presenting in the month of January with LRTIs, and older children presenting in May with malaria and its complications. Although 1790 (10.5%) of patients met the criteria for SAM by anthropometric measurement, only 683 (4%) of children were given the diagnosis, indicating that it is underdiagnosed and likely under-recognised, and perhaps explaining why it was a leading cause of death in children older than 6 months. Almost half of patients had more than one admission diagnosis, a finding common in LMICs and likely a reflection of the diagnostic uncertainty related to limited available testing.

Protocols for the management of high-volume, lower-acuity conditions such as uncomplicated malaria and pneumonia were generally well adhered to in our study, but our data reveal a low rate of return for the laboratory results, radiology studies and subspecialty consultations required to provide the standard of care for critically ill children presenting with common diagnoses like sepsis and severe malaria or complex conditions such as SAM. While there remains some controversy over the optimal fluid resuscitation for patients with septic shock in this setting,26–31 the least invasive management includes maintenance fluids, administration of oxygen and antibiotics.21 It is likely that limitations in staffing and the equipment required for patient monitoring impeded adherence to septic shock protocols. It is notable that 79% of children admitted to KCH were tested for HIV in accordance with guidelines promoting universal testing.

The overall mortality rate for our population was 4%, an improvement from previous studies in this hospital,15 16 with the highest mortality rate in the group of infants under 6 months of age. Most inpatient paediatric deaths in LMICs occur in the first 24–48 hours,13–15 32 33 which is consistent with our findings of 44% and 60%, respectively (online supplemental figure 1), highlighting the need for improved early recognition and management of high-acuity patients to decrease inpatient mortality. There has been an international focus in the post-COVID era on defining the care that should be provided to critically ill patients in hospitals around the world, and a 2021 consensus statement by global experts provides an outline for ensuring preparedness.34 Recently published assessments of the Malawi healthcare system’s emergency and critical care capabilities have highlighted opportunities for improvement, addressed in the Malawi Ministry of Health Emergency & Critical Care Strategy: Framework for Implementing Emergency & Critical Care Services in Malawi.35–37 This document provides a comprehensive structural outline based on the WHO Healthcare System Building Blocks to facilitate development of robust local and national emergency and critical care systems. Our setting does not have an Accident and Emergency Department, nor a dedicated space for paediatric patients in its Intensive Care Unit (ICU). Critically ill patients are managed in a high-density unit or on the general wards.

Supplemental material

Our study has several important limitations. We were limited by our reliance on review of paper charts and interviews with families. Our results may therefore be impacted by recall bias or be underestimated due to incomplete documentation by clinicians. Admission and discharge diagnoses were recorded in the database as documented by clinicians; no independent assessment of the accuracy of the diagnoses was made. A significant number of our patients absconded, which may have affected our mortality rates, although it is unlikely that parents absconded with deceased patients. During the study period, the clinicians at KCH adopted the use of WhatsApp groups for communication between consultants and staff, so there have likely been consultations ordered or completed that we did not capture, and we therefore likely underestimate the availability of consultant review. Lastly, while we provide a comprehensive description of the care of children admitted through the paediatric ward, we did not include children who presented with traumas or through the surgical service or oncology wards, who may account for a significant number of the overall admissions to the department.

While progress has been made in improving the survival of children under 5 years, the global equity gap remains. Our data highlight the need to focus on improving the initial stabilisation of critically ill patients, increasing patient access to ancillary services and providing the subspecialty nursing and medical care required for the ongoing management of the sickest hospitalised children or those with complex disease processes. Improving the overall and 24-hour mortality rate will require training clinicians and nurses to identify and stabilise critically ill patients, developing the infrastructure necessary for consistent diagnostic testing, use of standardised clinical processes and provision of specialised paediatric equipment. These are essential components of paediatric emergency and critical care services advocated for by national and international experts. Additionally, as the global health community commits to addressing the world’s remaining health inequities by advocating for UHC, it is imperative we build on a sound foundation. Data collection, analysis and routine use in policy and decision-making must be a pillar on which UHC is built, with an emphasis on the development of a culture of data use for feedback, evaluation and accountability.

Data availability statement

Data are available upon reasonable request. For further information email the corresponding author.

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants and was approved by University of North Carolina Institutional Review Board (#17-1942) and Malawi National Health Sciences Research Committee (17/07/1870). Routinely obtained health data that did not require consent to be documented.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Contributors All authors were involved in the study conception and design. Data collection was performed by EF, EJC and TM. Analysis and interpretation of results were done by EF, EJC, JC, GB and ME. The draft manuscript was prepared by EF, EJC, JC, GB and ME. All authors reviewed the results and approved the final version of the manuscript. EF is the guarantor.

  • Funding University of North Carolina Development Fund. Grant number N/A.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.