RESUMO
There is no standard method for estimating annual malaria deaths in India. The degree of under-reporting in routine system is uncertain with a varied picture of 10, 15, 25 and 200 times as found in different studies/models. During October 2010 a community based cross-sectional survey was conducted by National Vector Borne Disease Control Programme (NVBDCP) in 32 villages of West Garo Hills district of Meghalaya. 'Gap filling', 'fever survey' and 'verbal autopsy' models, developed in NVBDCP were used on the data obtained from two surveys done simultaneously: 'fever survey' with 2 weeks recall period and 'death survey' with one year recall period. Inputs from demographic and epidemiological data of the district were also given. During 2010, the district has reported 32 malaria deaths. Degrees of under-reporting found in the study were 6.93, 7.53 and 12 times in three models respectively. The first two models yielded closer estimates. But, verbal autopsy model gave inflated value as the method was unreliable for diagnosing malaria deaths. First two models, if found to be replicable, can be incorporated in NVBDCP activities to evaluate malaria control measures at district level. Verbal autopsy model may be used after concordance study in those districts where Pf malaria cases occur but no death is reported.