Piloting a pragmatic clinical audit tool for quality improvement in rural paediatric care in northern Sierra Leone

Hospital admissions and their clinical outcomes can reflect the disease burden in a population and can be utilised as effective surveillance and impact monitoring tools. Inadequate documentation of admissions and their outcomes have contributed to the poor quality of paediatric care in many health care settings in sub-Saharan Africa. We have developed and piloted a simple tool for documentation of basic, standardised patient-level information on causes of admissions, diagnoses, treatments and outcomes in patients admitted to the paediatric ward of a district hospital in a rural community in Sierra Leone. From 1 August 2019 to 31 July 2021, we used this tool to document the admissions, treatments and clinical outcomes of 1,663 children admitted to this hospital. The majority of the children (1015, 62%) were aged between 12-59 months, were boys (942, 57%), were wasted (516, 31%), stunted (238 14%) or underweight (537, 32%). More than half of the children lived more than 1 km distance from the hospital (876/1410, 62%). Most were admitted before 4pm (1171/1626, 72%) and during weekdays (1231/1662, 74%). The highest number of paediatric admissions occurred in November 2019 and the lowest in April 2020. Severe malaria was the leading cause of admission. More than 80% of the children were successfully treated and discharged home (1356/1663, 81.5%) while 122/1663 (7.3%) died. Children aged under-five years who were underweight, and those who presented with danger signs (e.g. signs of breathing difficulty, dehydration, head injury or severe infections) had a higher risk of death than children without these features (p<0.01; p=0.03; p=0.011 and p= 0.009, respectively). Lack of systematic documentation of medical histories and poor record keeping of hospital admissions and outcomes can be overcome by using a simple tool. Continuous use of the tool with regular audits could improve delivery of paediatric care in resource-limited settings.