Early Childhood Oral Health Impact Scale (ECOHIS) questionnaire: reliability generalization meta-analysis of...

To summarise the estimates of Cronbach's α for the Child Impact Section (CIS), Family Impact Section (FIS), and total Early Childhood Oral Health Impact Scale (ECOHIS) and to identify the significant predictors that may influence the pooled estimates.

A systematic search was conducted in PubMed, Scopus, EMBASE, Web of Sciences, Dentistry and Oral Science Source, and CINAHL from inception until 21st November 2024. Included were those reported in English with Cronbach's α values for the ECOHIS or its subscales. Abstracts, conference proceedings, letters to editors, pilot studies, brief communications, and commentary were omitted. Two review authors independently reviewed the title and abstracts and full-text. Data extraction included demographic characteristics, language, study design and setting, sampling, Cronbach's α, and questionnaire item count. COSMIN checklist was used for quality assessment and "Reliability Generalization Meta-analysis" (RGMA) was conducted to estimate the internal consistency of ECOHIS questionnaire.

Overall, 1,524 publications were retrieved, of which 454 and 304 were available for screening title and abstracts and full texts respectively. Among the included 66 publications, 74 estimates were obtained. The pooled Cronbach's α for the total scale was 0.86 with high heterogeneity (I² = 95.43). No significant impact of moderators was seen on the pooled estimate. Subgroup analysis showed little variation in the pooled estimates for continent, language, study setting, study design, and type of sampling. Only 46 estimates were included for the RGMA of CIS (0.85) and FIS (0.79). No significant effects of moderators were seen on the pooled Cronbach's α for both CIS and FIS. Similarly, subgroup analysis showed little variation in the pooled estimates for the CIS and FIS for various categorical predictors.

The pooled Cronbach's α of ECOHIS and its sub-scales were higher than the acceptable benchmark with no effect of moderators.