Validation of Verbal Autopsy in Perinatal Deaths – A Prospective Study in Belgaum District, Karnataka, India

Background: The first goal of National Rural Health Mission (NRHM) and the fourth goal of Millennium Development are to reduce under-five mortality rate. In India, data is available for aprox. 3% of the perinatal deaths. The need for good quality data on perinatal deaths is mandatory for public health planning and resources allocation. This has lead to a renewed interest in Verbal autopsy (VA) in poor resource settings. Objective: To validate a simplified community-based VA instrument. Materials and Methods: A simplified community-based VA instrument was developed. The sample size of 50 was calculated at power – 80%, significance – 5% and sensitivity – 70%. ANMs were trained to assign COD according to the ICD – 10. Following which they administered Verbal Autopsies for every perinatal death occurring in their area. A panel of 3 Physicians was formed who reviewed the VA data and assigned cause of death according to the ICD–10. Kappa Statistics was used to measure the degree of agreement between the ANM and Physician panel assigned COD. Results: Out of 50 Perinatal Deaths, the COD assigned by ANM and the COD assigned by the Physician panel were same for 40 of the perinatal deaths, which showed the agreement of 80% [Kappa Statistics = 0.838; P = 0.000 (<0.001)] which indicates statistically significant agreement. There was agreement of 93% for 14 Stillbirths and 75% agreement for 36 Early Neonatal Deaths.