The scale-up of HMM maintained the high adherence of HCPs to case management guidelines and the absence of malaria-attributed deaths seen in the pilot phase. HCPs tested 93% of suspected cases in both 2009 and 2010, and treated more than 95% of confirmed cases in both years, with no deaths attributed to malaria, showing that the rapid scale-up did not compromise performance. A high proportion of patients with negative RDTs were reported to have been referred for further evaluation.
The increase in overall consultations from 2008 to 2009, with the simultaneous decreases in suspected and confirmed cases as well as malaria-attributed hospitalizations in both areas, is likely due to increased access to health care in the setting of simultaneous scale-up of malaria prevention, primarily long-lasting insecticide treated nets (LLINs), which took place during the same period. A nationwide LLIN distribution campaign targeting children 6–59 months was conducted in June 2009
The intervention regions experienced a statistically significant decrease in all deaths and deaths attributed to malaria; no such decrease was seen in comparison regions. No other known differences in malaria management were present between comparison and intervention regions (diagnostics, treatment, supportive interventions, and costs). This pattern is consistent with an intervention such as improved access to case management that, while not having an impact on number of cases in the short term, prevents progression to severe disease and death. While there was not a relative decrease in hospitalizations, the HCPs also facilitated timely referral of severe cases, which may have enabled those patients to receive inpatient care while in relatively good condition. While not proof of a causal relationship, it is plausible that a large proportion of deaths are due to severe disease in those who have difficult access to health care and thus have long delays in seeking care, and that home-based diagnosis and treatment for these remote populations prevents the progression of uncomplicated malaria to severe disease and death.
While costs were not formally measured, at a cost of less than $1 per person at risk, programme cost was reasonable. A formal cost-effectiveness analysis of HMM has found it a cost-effective strategy, even when not taking into account the decreased expenses for the users
. The cost-effectiveness of HMM would be increased by expanding the services provided by HCPs to include management of diarrhoea and pneumonia, further decreasing the numbers of patients requiring health facility referral. Other roles for home-based care providers have been explored, such as combining home-based management with intermittent preventive treatment in children
This analysis faces a number of limitations. In a research setting, one would pre-define and possibly randomize intervention and control areas, and record a great deal more information. In a large-scale implementation in the majority of the regions of the country, this is not feasible, thus analysis of a limited set of programmatic data is necessary. Proving causality is not possible, but sources are triangulated to arrive at an argument for the plausibility of the impact. Data reported by the HCPs are subject to bias as they may inflate their performance, although all data reported by the HCPs were required to be verified by the health post nurse. Data reported by health facilities does not capture those not treated by the public health sector. Only deaths that occurred in public health facilities are measured; deaths occurring at private health facilities, at home or prior to arrival cannot be measured by this method. While a shift of deaths away from public health facilities in the intervention regions is possible, the increase in outpatient consultations in both regions, as well as the identical decrease in hospitalizations in the intervention and comparison regions argues against this. Analysis of impact is limited to the malaria transmission season of 2009, when the intervention was being scaled-up, which likely decreases impact measured. In addition, the implementation was limited to a subset of the most remote communities, and did not cover the entire population for which district level data was reported, possibly further diluting the impact. Given the simultaneous scale-up of other malaria control interventions, the amount of impact is impossible to measure due to the inability to firmly attribute decreases to the HMM programme. Nonetheless, it appears that the implementation of HMM is likely to have been at least partially responsible for the decrease in deaths attributed to malaria in the regions in which it was implemented.