Estimating the intra-cluster correlation coefficient for evaluating aneducational intervention program to improve rabies awareness anddog bite prevention among children in Sikkim, India: A pilot study

Abstract

Background: Educational initiatives targeting at-risk populations have long been recognized as a mainstayof ongoing rabies control efforts. Cluster-based studies are often utilized to assess levels of knowledge,attitudes and practices of a population in response to education campaigns. The design of cluster-basedstudies requires estimates of intra-cluster correlation coefficients obtained from previous studies. Thisstudy estimates the school-level intra-cluster correlation coefficient (ICC) for rabies knowledge changefollowing an educational intervention program.Methods: A cross-sectional survey was conducted with 226 students from 7 schools in Sikkim, India,using cluster sampling. In order to assess knowledge uptake, rabies education sessions with pre- and post-session questionnaires were administered. Paired differences of proportions were estimated for questionsanswered correctly. A mixed effects logistic regression model was developed to estimate school-level andstudent-level ICCs and to test for associations between gender, age, school location and educational level.Results: The school- and student-level ICCs for rabies knowledge and awareness were 0.04 (95% CI: 0.01,0.19) and 0.05 (95% CI: 0.2, 0.09), respectively. These ICCs suggest design effect multipliers of 5.45 schoolsand 1.05 students per school, will be required when estimating sample sizes and designing future clusterrandomized trials. There was a good baseline level of rabies knowledge (mean pre-session score 71%),however, key knowledge gaps were identified in understanding appropriate behavior around scareddogs, potential sources of rabies and how to correctly order post rabies exposure precaution steps. Afteradjusting for the effect of gender, age, school location and education level, school and individual post-session test scores improved by 19%, with similar performance amongst boys and girls attending schoolsin urban and rural regions. The proportion of participants that were able to correctly order post-exposureprecautionary steps following educational intervention increased by 87%.Conclusion: The ICC estimates presented in this study will aid in designing cluster-based studies evaluatingeducational interventions as part of disease control programs. This study demonstrates the likely benefitsof educational intervention incorporating bite prevention and rabies education.