Measles elimination - Is it achievable? Lessons from an immunisation coverage survey

Objectives. To determine routine measles coverage at district level and to explore reasons for immunisation failure in Mpumalanga Province, South Afri ca. Design. An adaptation of the World Health Organisation (WHO) Expanded Progr amme on Immunisation (EPI) duster sampling method was used to make a random selection of 30 dusters in each of 21 health districts, 630 clusters in to tal. Seven individuals from the age group 12 - 23 months were randomly sele cted from each cluster and immunisation status and source were transcribed from their child health records. Where no immunisations were administered, reasons provided by parents or guardians were recorded. Setting. Mpumalanga Province, South Africa. Results. The weighted valid population coverage with measles vaccine for ch ildren aged 12 - 23 months in Mpumalanga Province was 71.1% (95% confidence interval 64.9 - 78.5%); this was the lowest of all EPI antigens. There was marked heterogeneity in measles coverage across the province, with a coeff icient of variation of 22.2%. Districts with the lowest coverage shared bor ders with neighbouring provinces. District measles coverage was highly posi tively correlated with diphtheria, pertussis and tetanus (DPT3) coverage (r = 0.960, P = 0.000). There was a strong negative correlation between ranke d measles campaign coverage and routine measles immunisation coverage. Obst acles to immunisation accounted for nearly half (49%) of all reasons for im munisation failure, white lack of information and lack of motivation accoun ted for 30% and 22%, respectively. Conclusions. Survey results highlight the need for supplementary immunisati on including non-selective campaigns, if Mpumalanga is to achieve the South African goal of measles elimination by 2002. The value of determining dist rict resolution coverage in order to identify areas with low measles covera ge requiring supplementary intervention was also demonstrated. A strong neg ative correlation between routine and campaign coverage deserves further st udy in other settings.