Empirical Evidence on Impact by Accounting for Cluster Effect Within Patients in Cohort Studies Evaluating Comparative Effectiveness of Dental Treatments: A Meta-Epidemiological Study
Author(s)
Areum Han, BS1, Hee-Kyung Park, DDS, PhD2, SEOKYUNG HAHN, PhD3.
1Interdisciplinary Program of Medical Informatics, Seoul National University College of Medicine, Seoul, Korea, Republic of, 2Department of Oral Medicine and Oral Diagnosis, School of Dentistry and Dental Research Institute, Seoul National University, Seoul, Korea, Republic of, 3Department of Human Systems Medicine, Seoul National University College of Medicine, Seoul, Korea, Republic of.
1Interdisciplinary Program of Medical Informatics, Seoul National University College of Medicine, Seoul, Korea, Republic of, 2Department of Oral Medicine and Oral Diagnosis, School of Dentistry and Dental Research Institute, Seoul National University, Seoul, Korea, Republic of, 3Department of Human Systems Medicine, Seoul National University College of Medicine, Seoul, Korea, Republic of.
OBJECTIVES: In dental research, clustering naturally occurs when multiple teeth are treated within the same patient. However, this is rarely considered in the design or analysis of studies.
METHODS: Comparative effectiveness studies of dental treatments using a cohort design from major dental journals were included. We collected study characteristics relating to design, cohort information, and analytic approaches. Studies that accounted for patient clusters in design or analysis were categorized as 'cluster-considered' (CC); otherwise, 'non-cluster-considered' (NCC). We identified the proportion of studies that accounted for patient clusters. We then explored how the CC group differed from the NCC group in terms of their characteristics. We also examined the difference in outcome significance between the two groups and its potential interaction with other factors. Multiple logistic regression was applied to assess the relationship between consideration of patient clusters and outcome significance, adjusting for correlated factors.
RESULTS: Of 95 cohort studies included, 52 studies (55%) were categorized in the NCC group. In the CC group, 24 (56%) considered patient clusters when designing cohort. The two groups differed in terms of data source, cohort size and publication period. The proportion of studies considering clusters in the last decade was statistically significantly higher than in the previous period. A simple comparison showed that the proportion of studies with positive results was similar in both groups. However, studies in the NCC group involving endodontic treatments, retrospective designs, large clusters or large cohorts tended to show positive results more than those in the CC group, after adjusting for correlated factors. Large-sized clusters were more prevalent in retrospective or large cohort studies, a greater proportion of which focused on endodontic treatments.
CONCLUSIONS: Our meta-epidemiological study demonstrated that statistically significant results were more frequently observed in studies that did not consider clusters when the cluster size was not small.
METHODS: Comparative effectiveness studies of dental treatments using a cohort design from major dental journals were included. We collected study characteristics relating to design, cohort information, and analytic approaches. Studies that accounted for patient clusters in design or analysis were categorized as 'cluster-considered' (CC); otherwise, 'non-cluster-considered' (NCC). We identified the proportion of studies that accounted for patient clusters. We then explored how the CC group differed from the NCC group in terms of their characteristics. We also examined the difference in outcome significance between the two groups and its potential interaction with other factors. Multiple logistic regression was applied to assess the relationship between consideration of patient clusters and outcome significance, adjusting for correlated factors.
RESULTS: Of 95 cohort studies included, 52 studies (55%) were categorized in the NCC group. In the CC group, 24 (56%) considered patient clusters when designing cohort. The two groups differed in terms of data source, cohort size and publication period. The proportion of studies considering clusters in the last decade was statistically significantly higher than in the previous period. A simple comparison showed that the proportion of studies with positive results was similar in both groups. However, studies in the NCC group involving endodontic treatments, retrospective designs, large clusters or large cohorts tended to show positive results more than those in the CC group, after adjusting for correlated factors. Large-sized clusters were more prevalent in retrospective or large cohort studies, a greater proportion of which focused on endodontic treatments.
CONCLUSIONS: Our meta-epidemiological study demonstrated that statistically significant results were more frequently observed in studies that did not consider clusters when the cluster size was not small.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
MSR85
Topic
Methodological & Statistical Research
Topic Subcategory
Confounding, Selection Bias Correction, Causal Inference
Disease
No Additional Disease & Conditions/Specialized Treatment Areas