Evaluating the Impact of Programmatic Mass Drug Administration for Malaria in Zambia Using Routine Incidence Data

Abstract

Background: In 2016, the Zambian National Malaria Elimination Centre started programmatic mass drug administration (pMDA) campaigns with dihydroartemisinin-piperaquine as a malaria elimination tool in Southern Province. Two rounds were administered, 2 months apart (coverage 70% and 57%, respectively). We evaluated the impact of 1 year of pMDA on malaria incidence using routine data.

Methods: We conducted an interrupted time series with comparison group analysis on monthly incidence data collected at the health facility catchment area (HFCA) level, with a negative binomial model using generalized estimating equations. Programmatic mass drug administration was conducted in HFCAs with greater than 50 cases/1000 people per year. Ten HFCAs with incidence rates marginally above this threshold (pMDA group) were compared with 20 HFCAs marginally below (comparison group).

Results: The pMDA HFCAs saw a 46% greater decrease in incidence at the time of intervention than the comparison areas (incidence rate ratio = 0.536; confidence interval = 0.337-0.852); however, incidence increased toward the end of the season. No HFCAs saw a transmission interruption.

Conclusions: Programmatic mass drug administration, implemented during 1 year with imperfect coverage in low transmission areas with suboptimal vector control coverage, significantly reduced incidence. However, elimination will require additional tools. Routine data are important resources for programmatic impact evaluations and should be considered for future analyses.

Keywords: impact evaluation; malaria; mass drug administration; routine data.

Figure 1.

Health facility catchment areas included in the analysis, Southern Province.

Figure 2.

Aggregated incidence across intervention (programmatic mass drug administration [pMDA]) and comparison groups with model fits: points are observed values, and lines are fitted values from the generalized estimating equations model. The vertical line indicates the timing of the first round of pMDA. The model fit varied by health facility, with the fit notably poor in some health facilities in 2016.

Figure 3.

Unadjusted percentage change between baseline and postintervention period by health facility for the programmatic mass drug administration and comparison groups.

Figure 4.

Cumulative monthly cases postintervention as percentage of average number of cases from the previous 3 seasons, by programmatic mass drug administration and comparison locations. For example, by November, only 67% of cases we would have expected in the entire season (based on the average from previous seasons) have occurred. The gray box indicates the duration of the malaria season.