Predictors of treatment error for children with uncomplicated malaria seen as outpatients in Blantyre district, Malawi

BACKGROUND: Past studies have shown that health workers in developing countries often do not follow clinical guidelines, though few studies have explored with appropriate methods why errors occur. To develop interventions that improve health worker performance, factors affecting treatment practices must be better understood. METHODS: We analysed data from a health facility survey in Blantyre District, Malawi, in which health workers were observed treating ill children, and then children were independently re-examined by 'gold-standard' study clinicians. The analysis was limited to children with uncomplicated malaria (defined according to Malawi's guidelines as fever or anaemia without signs of severe illness), and a treatment error was defined as failure to treat with an effective antimalarial. RESULTS: Twenty-eight health workers and 349 ill-child consultations were evaluated; 247 (70.8%) children were treated with an effective antimalarial, and 102 (29.2%) were subject to treatment error. Logistic regression analysis revealed that in-service malaria training was not associated with treatment quality (univariate odds ratio (OR) = 1.16, 95% confidence interval (CI): 0.46-2.93); whereas acute respiratory infections training was associated with making an error (adjusted OR (aOR) = 2.42, 95% CI: 1.23-4.76). High fever and chief complaint of fever were associated with fewer errors (aOR = 0.25, 95% CI: 0.10-0.60 and aOR = 0.25, 95% CI: 0.13-0.48, respectively). Errors were more likely to occur in consultations starting before 1 p.m. (aOR = 1.88, 95% CI: 1.07-3.31). Supervision was not associated with better treatment quality. CONCLUSIONS: These results suggest that the disease-specific training and supervision, performed before the survey, did not lead to long-term improvements in health care quality. Furthermore, case management training for one specific disease may have worsened quality of care for another disease. These results support integration of guidelines for multiple conditions. Interventions should be evaluated for unintended negative effects on overall quality of care.