Co-diagnoses in hospitalised children – revealing the cause of admission in a malaria endemic area

In tropical settings many children attending a hospital suffer from multiple diseases. Especially in areas with a high malaria prevalence co-diagnoses with malaria are common. In these children it is unclear whether they suffer from two independent illnesses or whether they have a symptomatic disease along with an uncomplicated co-diagnosis. In hospital-based case-control studies negative associations between symptomatic diseases are observed, when both diseases are associated with hospitalisation (i.e., Berkson's Bias). Utilising this effect, we estimated which groups of diseases are a likely cause for hospital admission.

Febrile (≥38 °C) children (≤15 years) admitted to a rural Hospital in Ghana were included into the study. Extended clinical and microbiological diagnostics were applied according to a clinical algorithm to establish diagnosis. Using a case-control approach, associations between all possible diagnoses pairs were estimated, so that diagnosis A was the exposure and diagnosis B the outcome. An odds ratio (OR) below one points towards two symptomatic diseases.

In total, 1,238 children were included into the study, of these 163 (13%) children had no diagnosis, 691 (56%) had a mono-diagnosis and 384 (31%) children had co-diagnoses. Most diagnoses occurred independently and patients with or without a diagnosis were as likely to have a particular co-diagnosis. Diagnoses less often observed in the presence of another condition were malaria and bacteraemia (OR = 0.1; 95%-CI: 0.1 – 0.2), CNS infection and LRTI (OR = 0.5; 95%-CI: 0.3 – 0.8), and malaria and LRTI (OR = 0.3; 95%-CI: 0.2 – 0.4).

Negative associations may point to diagnoses, which are independent reasons for hospital admission. Within our study malaria, CNS infection, bacteraemia and LRTI could be revealed as independent causes for hospitalisation. Nevertheless, the data also highlight that co-diagnoses occur frequently, which must be considered when making individual treatment decisions.