Authors' response.

Sir,

We thank Dr. Gozdas for his valuable comments on our article1. The article describes the results of a follow up screening of a subsample of a cohort that participated in an earlier study of immunological response to an indigenous hepatitis B vaccine2. The original cohort was from the Nicobarese community of Car Nicobar Island and consisted of individuals who were negative for both HBsAg and anti-HBsAg irrespective of their anti-HBc status.

We agree that ideally the subjects should have been screened for anti-HBc antibodies and those negative only should have been selected into the study. But, the purpose of the original study2 was to assess the effect of a vaccination programme in terms of the trend in seroprotection among vaccinated subjects. In view of resource constraints, it might not be practical in a programme to screen all the subjects for HBsAg, anti-HBsAg antibody and anti-HBc to determine eligibility for vaccination. Murhekar et al23 described the immunological response of the cohort till the end of three years and these articles had presented seroprotection rates by pre-vaccination anti-HBc status. Anti-HBc positive individuals had anamnestic response only after the first dose of vaccine. The proportion of seropositives and geometric mean titre (GMT) after the second and third doses of vaccine and at the end of second and third years after vaccination were comparable in both the groups.

The trend in seroprotection was assessed during the first three years by Murhekar et al23 and by us now by conducting cross-sectional surveys of subsamples of the vaccinated cohort. There could be a probability that various characteristics including the proportions of isolated anti-HBc positives among the subsamples were different due to sampling error. This is a limitation of non-stratified sampling. Non-stratified sampling is done with the assumption that when the sample size is fairly large, the probability of the sample being drastically different from the population in terms of various characteristics is small. Initial anti-HBc status cannot be termed as a confounder because a confounder needs to influence both the exposure and the outcome. Anti-HBc status may influence anti-HBsAg response, but not selection into the vaccination cohort.

The screening of the subjects during the initial phase was one month after each dose of vaccination2. The labelling of the third point of the X-axis of the Figure should have been ‘7 months’ instead of ‘6 months1’.