Intracluster Correlation Coefficient in A College-Based Cluster Randomised Controlled Trial

Muhammad Adil Zainal Abidin, Hayati Kadir, Rosliza Abdul Manaf: Intracluster Correlation Coefficient in A College-Based Cluster Randomised Controlled Trial. 2019, (Type: POSTER PRESENTATION; Organisation: Department of Community Medicine, Kulliyyah of Medicine, IIUM, Department of Community Health, Faculty of Medicine and Health Science, UPM).

Abstract

INTRODUCTION: Most of the public health research is conducted at the population level. More study has utilized cluster design in which a group unit is randomized instead of individuals to avoid contamination effect and more practical. However, not many studies publish their Intracluster correlation coefficient (ICC) which is the measure of the relatedness of clustered data. It is important to report the ICC that can be used to calculate effective sample size in future cluster study. Here, we aim to report the ICC from a smoking intervention among young adult. METHODS: Based on a college-based quit smoking intervention, a single level model – student nested within colleges was used. There are 10 cluster which is the community college and 16 smokers within each college. The Intracluster correlation coefficients were measured for three outcomes namely motivation to quit smoking, number of cigarette smoke per day and number of quit attempts. The ICC was measured in R studio using ICCbin package. For the binary outcome ICC was measured using variance components from the ANOVA method and Smith confidence limit equation. RESULTS: For the proportion of motivation to quit smoking, the ? (rho) was 0.026 (95%CI: 0.00, 0.11), mean number of quit smoking, ?=0.013 (95%CI: 0.00, 0.08) and mean number of quit attempts, ?=0.01 (95%CI: 0.00, 0.07). DISCUSSION: The ? in this study was small and almost similar to other reported study. The researcher must calculate and report the ICC to enable others to use in future research.

    BibTeX (Download)

    @proceedings{APCPH-2019-196,
    title = {Intracluster Correlation Coefficient in A College-Based Cluster Randomised Controlled Trial},
    author = {Muhammad Adil Zainal Abidin and Hayati Kadir and Rosliza Abdul Manaf},
    year  = {2019},
    date = {2019-07-22},
    urldate = {2019-07-22},
    journal = {6th Asia-Pacific Conference on Public Health 2019 Proceedings},
    issue = {6},
    abstract = {INTRODUCTION: Most of the public health research is conducted at the population level. More study has utilized cluster design in which a group unit is randomized instead of individuals to avoid contamination effect and more practical. However, not many studies publish their Intracluster correlation coefficient (ICC) which is the measure of the relatedness of clustered data. It is important to report the ICC that can be used to calculate effective sample size in future cluster study. Here, we aim to report the ICC from a smoking intervention among young adult. METHODS: Based on a college-based quit smoking intervention, a single level model – student nested within colleges was used. There are 10 cluster which is the community college and 16 smokers within each college. The Intracluster correlation coefficients were measured for three outcomes namely motivation to quit smoking, number of cigarette smoke per day and number of quit attempts. The ICC was measured in R studio using ICCbin package. For the binary outcome ICC was measured using variance components from the ANOVA method and Smith confidence limit equation. RESULTS: For the proportion of motivation to quit smoking, the ? (rho) was 0.026 (95%CI: 0.00, 0.11), mean number of quit smoking, ?=0.013 (95%CI: 0.00, 0.08) and mean number of quit attempts, ?=0.01 (95%CI: 0.00, 0.07). DISCUSSION: The ? in this study was small and almost similar to other reported study. The researcher must calculate and report the ICC to enable others to use in future research.},
    note = {Type: POSTER PRESENTATION; Organisation: Department of Community Medicine, Kulliyyah of Medicine, IIUM, Department of Community Health, Faculty of Medicine and Health Science, UPM},
    keywords = {cluster analysis, research design, Statistics},
    pubstate = {published},
    tppubtype = {proceedings}
    }