Information on the causes of deaths from diarrhoea in children younger than 5 years is needed to design improved preventive and therapeutic approaches. We aimed to conduct a systematic analysis of studies to report estimates of the causes of deaths from diarrhoea in children younger than 5 years at global and regional levels during 2000-21. For this systematic review and Bayesian multinomial analysis, we included 12 pathogens with the highest attributable incidence in the Global Enteric Multicenter Study. We searched PubMed, Scopus, Embase, Web of Science, Global Health Index Medicus, Global Health OVID, IndMed, Health Information Platform for the Americas (PLISA), Africa-Wide Information, and Cochrane Collaboration for articles published between Jan 1, 2000, and Dec 31, 2020, using the search terms "child", "hospital", "diarrhea", "diarrhoea", "dysentery", "rotavirus", "Escherichia coli", "salmonella", "shigella", "campylobacter", "Vibrio cholerae", "cryptosporidium", "norovirus", "astrovirus", "sapovirus", and "adenovirus". To be included, studies had to have a patient population of children younger than 5 years who were hospitalised for diarrhoea (at least 90% of study participants), at least a 12-month duration, reported prevalence in diarrhoeal stools of at least two of the 12 pathogens, all patients with diarrhoea being included at the study site or a systematic sample, at least 100 patients with diarrhoea, laboratory tests done on rectal swabs or stool samples, and standard laboratory methods (ie, quantitative PCR [qPCR] or non-qPCR). Studies published in any language were included. Studies were excluded if they were limited to nosocomial, chronic, antibiotic-associated, or outbreak diarrhoea or to a specific population (eg, only children with HIV or AIDS). Each article was independently reviewed by two researchers; a third arbitrated in case of disagreement. If both reviewers identified an exclusion criterion, the study was excluded. Data sought were summary estimates. Data on causes from published studies were adjusted when necessary to account for the poor sensitivity of non-qPCR methods and for attributable fraction based on quantification of pathogens in children who are ill or non-ill. The causes of deaths from diarrhoea were modelled on the causes of hospitalisations for diarrhoea. We separately modelled studies reporting causes of diarrhoea in children who were hospitalised in low-income and middle-income countries (LMICs) and in high-income countries (HICs). Of 74 282 papers identified in the initial database search, we included 138 studies (91 included data from LMICs and 47 included data from HICs) from 73 countries. We modelled estimates for 194 WHO member states (hereafter referred to as countries), including 42 HICs and 152 LMICs. We could attribute a cause to 1 003 448 (83·8%) of the estimated 1 197 044 global deaths from diarrhoea in children younger than 5 years in 2000 and 360 730 (81·3%) of the estimated 443 833 global deaths from diarrhoea in children younger than 5 years in 2021. The cause with the largest estimated global attribution was rotavirus; in LMICs, the proportion of deaths from diarrhoea due to rotavirus in children younger than 5 years appeared lower in 2021 (108 322 [24·4%] of 443 342, 95% uncertainty interval 21·6-29·5) than in 2000 (316 382 [26·5%] of 1 196 134, 25·7-28·5), but the 95% CIs overlapped. In 2000, the second largest estimated attribution was norovirus GII (95 817 [8·0%] of 1 196 134 in LMICs and 225 [24·7%] of 910 in HICs); in 2021, Shigella sp had the second largest estimated attribution in LMICs (36 082 [8·1%] of 443 342), but norovirus remained with the second largest estimated attribution in HICs (84 [17·1%] of 490). Our results indicate progress in the reduction of deaths from diarrhoea caused by 12 pathogens in children younger than 5 years in the past two decades. There is a need to increase efforts for prevention, including with rotavirus vaccine, and treatment to eliminate further deaths. Bill & Melinda Gates Foundation via Johns Hopkins University and the University of Virginia.
Read full abstract