The Lancet, August 2011, vol.378, iss.9793, pp.795-803. Available From:Link to Source
This study evaluates the effectiveness of an intervention in which mobile phone text-message reminders are sent to Kenyan health workers with the aim of enhancing their adherence to malaria treatment guidelines. The authors designed and implemented a cluster-randomised controlled trial involving all 107 rural government health facilities of 11 districts presenting high malaria prevalence. The sample of health facilities was randomly divided into an intervention and a control group. During a 6-month period, health workers from the intervention area were sent a series of 10 different text messages (two messages for each weekday) reminding them of key recommendations from the Kenyan national malaria guidelines and from training manuals. The same 10 messages were sent over the study period. The program was implemented from May 2009 to October 2009 and involved 119 health workers and approximately 2,100 children from Greater Kwale, Greater Kisii, and Gucha.
The data originate from three health facility surveys conducted before the intervention started and again 6 and 12 months after baseline. On the day of the survey, study nurses identified all sick children younger than 5 years and gathered data on children's age, weight, temperature, history of fever, absence of routine negative malaria tests, and signs of severe disease. A structured interview with children's caregivers was also conducted for children eligible for artemether-lumefantrine (AL) treatment. The information collected includes the previous use of anti-malarials, routine diagnostic results, and drugs prescribed. The research team also gathered data on health workers (for example, demographic information) and health facilities (for example, availability of drugs, retrospective malaria morbidity). The analysis is performed using a difference-in-differences model, which includes a treatment dummy variable, two time dummy variables, and two interaction terms between treatment status and each time dummy variable.
The intervention appears to have improved the quality of AL management. For the first follow-up survey (immediately after the intervention ended), significant results were found on the following outcomes: prescription of AL treatment (with a 16.2-percent increase), providing the first dose at the health facility (31.6-percent increase), counselling to take the second dose after 8 hours (32-percent increase), and counselling on what to do in the case of vomiting (20.1-percent increase). Fairly similar results were found at the second follow-up survey (6 months after the intervention ended): providing the first dose at the health facility (24.1-percent increase), counselling to take the second dose after 8 hours (27.5-percent increase), counselling to take AL after meal (24.5-percent increase), and counselling on what to do in the case of vomiting (24.5-percent increase).
The authors compare these results to those obtained in the literature assessing alternative strategies aimed at improving health workers' adherence to treatment guidelines. They notably point out a Kenyan study in which in-service training and passive distribution of job aids had failed to improve case-management practices (Wasunna et al., 2010) . However, the authors stress that their findings do not imply that text-messaging should be favoured over traditional case-management interventions such as in-service training, supervision, dissemination of guidelines, and job aids. Indeed, they argue that text-messaging should be used as a complement to existing strategies rather than as an alternative. The authors conclude by mentioning that a cost-effectiveness analysis was being undertaken at the time of publication and should provide additional insights to evaluate the pertinence and feasibility of scaling up the intervention.