Effect of intervention on drug use and healthcare use for various disease syndromes during the intervention period from household morbidity surveillance (HMS), western Kenya, February 2008–March 2009.

*p-value calculation includes a random effect variable for village, and adjusts for pre-intervention rates of seeking care outside the home by village - F-test in SAS PROC GLIMMIX.†% given in table is the % of biweekly household visits with illness that resulted in medication use or care-seeking outside the home. Denominator for sought care was slightly lower than for medication use due to some missing data for that variable.¶For primary outcome, intracluster correlation (ICC) = 0.030.