Background: In developing countries, health workers often do not follow clinical practice guidelines. However, few studies have examined why different types of errors occur.
Methods: We analysed a sample of consultations of children with non-severe malaria (defined as fever without signs of severe illness) from a health facility survey conducted in Ouémé Département, Benin. Treatment was defined as correct (recommended antimalarial), a minor error (non-recommended antimalarial), or a major error (no antimalarial).
Results: In all, 85 health workers and 289 children were studied. In a multivariate logistic regression analysis, the following factors were significantly associated with major errors: treatment by a physician (adjusted odds ratio [aOR] = 13.57, 95% CI: 1.45-126.75), child's age <12 months (aOR = 3.41, 95% CI: 1.15-10.07), and child's temperature (aOR = 0.58 per degrees C, 95% CI: 0.34-0.97). Factors significantly associated with minor errors were: child's temperature (aOR = 1.43 per degrees C, 95% CI: 1.07-1.92), electricity at the health facility (aOR = 3.10, 95% CI: 1.05-9.17), >/=1 supervision visit in the past 6 months (aOR = 0.33, 95% CI: 0.14-0.77), fever treatment wall chart in the consultation room (aOR = 0.29, 95% CI: 0.12-0.73), and number of non-fever chief complaints (aOR = 0.67 per complaint, 95% CI: 0.48-0.93). In-service training in malaria treatment was not significantly associated with either error type.
Conclusions: Many factors may influence health worker performance, and factors such as pre-service training may influence performance in unexpected ways. Identifying different errors and analysing them separately can help reveal potential causes that may be masked by combining errors into a single category.