Electronic Integrated Management of Childhood Illness (eIMCI): a randomized controlled trial to evaluate an electronic clinical decision-making support system for management of sick children in primary health care facilities in South Africa

Background: Electronic clinical decision-making support systems (eCDSS) aim to assist clinicians making complex patient management decisions and improve adherence to evidence-based guidelines. Integrated management of Childhood Illness (IMCI) provides guidelines for management of sick children attending primary health care clinics and is widely implemented globally. An electronic version of IMCI (eIMCI) was developed in South Africa. Methods We conducted a randomized controlled trial comparing management of sick children with eIMCI to the management when using paper-based IMCI (pIMCI) in one district in KwaZulu-Natal. From 31 clinics in the district, 15 were randomly assigned to intervention (eIMCI) or control (pIMCI) groups. Computers were deployed in eIMCI clinics, and one IMCI trained nurse was randomly selected to participate from each clinic. eIMCI participants received a one-day computer training, and all participants received a similar three-day IMCI update and two mentoring visits. A quantitative survey was conducted among mothers and sick children attending participating clinics to assess the quality of care provided by IMCI practitioners. Sick child assessments by participants in eIMCI and pIMCI groups were compared to assessment by an IMCI expert. Results Self-reported computer skills were poor among all nurse participants. IMCI knowledge was similar in both groups. Among 291 enrolled children: 152 were in the eIMCI group; 139 in the pIMCI group. The mean number of enrolled children was 9.7 per clinic (range 7–12). eIMCI consultations took longer than pIMCI consultations (median duration 28 minutes vs 25 minutes; p = 0.02). eIMCI participants were less likely than pIMCI participants to correctly classify children for presenting symptoms, but were more likely to correctly classify for screening conditions (TB, HIV and nutrition). However, this did not increase identification of children who screened positive. eIMCI participants were less likely to provide all required medications (124/152; 81.6% vs 126/139; 91.6%, p = 0.026), and more likely to prescribe unnecessary medication (48/152; 31.6% vs 20/139; 14.4%, p = 0.004) compared to pIMCI participants. Conclusions Implementation of eIMCI failed to improve management of sick children, with poor IMCI implementation in both groups. Further research is needed to understand barriers to comprehensive implementation of both pIMCI and eIMCI. (350) Clinical Trials Registration: BFC157/19