Aetiologies, neuroradiological features, and risk factors for mortality and long-term neurosequelae of febrile coma in Malawian children: a prospective cohort study

Abstract

Background: Children in febrile coma in Africa are frequently hospitalised, with poorer outcomes than those in high-income settings. Cerebral malaria is historically the most common cause of febrile coma. Due to limited diagnostic and radiological resources and a decrease in malaria prevalence, there might be under-recognition of non-malarial coma. However, prospective data are scarce. We aimed to determine causes, neuroradiological features, risk factors for mortality, and neurosequelae of children in febrile coma in Malawi.

Methods: In this prospective cohort study, we enrolled children in a coma (Blantyre Coma Scale score ≤2) who were aged between 3 months and 15 years at Queen Elizabeth Central Hospital, Blantyre, Malawi. We used pathogen-specific PCR analysis of blood and cerebrospinal fluid for 15 pathogens including Streptococcus pneumoniae, Neisseria meningitidis, Haemophilus influenzae, Salmonella spp, non-typhoidal Salmonella, Salmonella enterica serotype Typhi (S Typhi), Klebsiella spp, Escherichia coli, Mycobacterium tuberculosis (also using GeneXpert), Streptococcus agalactiae, herpes simplex virus (types 1 and 2), varicella zoster virus, cytomegalovirus, enteroviruses, and SARS-CoV-2; microscopy for malaria; admission brain MRI to enhance the diagnosis of cause and identify brain injury, swelling, and any other complications; and electroencephalography tracings were used identify subclinical seizures or non-convulsive status epilepticus. Assessment of malarial retinopathy was performed by a trained ophthalmologist. We used regression models to estimate risk factors for (and the difference in) 30-day mortality and 180-day neurosequelae (outcome assessed in-person) between children with non-malarial coma and cerebral malaria.

Findings: Between Jan 31, 2018, and June 30, 2021, we recruited 352 children with febrile coma. Cerebral malaria was the most common cause (in 231 [66%] of 352 children). Pathogenic diagnosis was possible in 289 (82%) of 352 children. Co-infection was identified in 63 (27%) of 231 children with cerebral malaria, of which 49 (78%) were bacterial. The most common non-malarial causes of coma were meningitis (48 [14%] of 352 children) and encephalitis (24 [7%] of 352); 32 (9%) cases had an unknown cause. Compared with standard cultures, PCR significantly increased pathogen diagnosis (p<0·0001), with the highest yield in patients with meningitis (seven [15%] of 48 vs 30 [63%] of 48). S pneumoniae (n=44) and non-typhoidal salmonella or S Typhi (n=24) were the most frequently detected bacterial pathogens. Brain parenchymal abnormalities were identified on MRI in most children with febrile coma (165 [92%] of 178), and were significantly more common in children with non-malarial coma (68 [100%] of 68]) than cerebral malaria (98 [89%] of 110; p<0·0001). Overall, at 30 days after discharge, death (69 [21%] of 323) or any neurological impairment (163 [50%] of 323) were common, but poorer long-term outcomes were more frequent following non-malarial coma than cerebral malaria (death at 30 days: 32 [28%] of 114 vs 37 [18%] of 209, p=0·029; severe neurological impairment at 180 days: 19 [17%] of 114 vs 15 [7%] of 209, p=0·0079). Children who had cerebral malaria with CNS co-infection had higher mortality (ten [37%] of 27) than those with cerebral malaria alone (19 [12%] of 154, p=0·0033).

Interpretation: Despite malaria control efforts, cerebral malaria remains the most common cause of febrile coma in Malawi. However, non-malarial coma causes a greater disease burden (death and disability), and a higher case-fatality rate was observed in non-malarial coma and cerebral malaria with non-malarial co-infection than cerebral malaria alone. To adequately treat severe invasive bacterial infections, that are frequently not detected in routine clinical practice, commencing empirical antimicrobials in all children in febrile coma, including those with cerebral malaria, could and should be rapidly implemented across Africa and must be considered. The study highlights the value of molecular diagnostics and imaging to guide diagnosis. The frequent findings of brain abnormalities from imaging at admission emphasises the need for earlier escalation of children with febrile coma to specialist care. Further work is needed to develop feasible molecular and radiological diagnostics for their successful deployment across the continent. Implementation of these methods could improve diagnosis and outcomes for children with febrile coma in Africa.

Funding: Wellcome Trust TRANSLATIONS: For the Chichewa, French and Portuguese translations of the abstract see Supplementary Materials section.