Publication

Abstract : Objective To evaluate the pooled estimates of Cronbach’s alpha of the Oral Impacts on Daily Performance (OIDP) questionnaire and explore the moderators that could have influenced the overall estimate. Materials and methods A systematic search of common databases such as PubMed, Scopus, EMBASE, and CINAHL was performed from inception till 13th December 2024. Studies in English and those that reported Cronbach’s alpha values for the OIDP questionnaire were included. Studies reported as letters, conference proceedings, or abstracts; secondary analysis of the previous data; studies with alpha values reported for pilot studies; modified versions of the OIDP questionnaires; induced reliability estimates; retracted articles; short communications; and commentaries were excluded. Two review authors independently screened the publications. The information collected included year of publication, country, sample size, age, sex distribution, target population, language of administration, mode of administration, study setting, study design, patient selection, Cronbach’s alpha, and the number of items in the questionnaire. The risk of bias assessment was performed via the COSMIN checklist. Reliability Generalization Meta-analysis was performed via the random effects model (restricted maximum likelihood method) to obtain a pooled untransformed Cronbach’s alpha. Results A total of 1069 publications were available for screening, and 54 publications yielded 63 estimates with a sample size of 92,564 (sample size range: 47–12647). The overall pooled Cronbach’s alpha was 0.82 (95% CI = 0.8–0.84), with high heterogeneity (I^2 = 99.75%; Q = 26702.91). Meta-regression revealed no effects of moderators such as sex (coefficient: 0.02), age (coefficient: 0), language (coefficient: 0), population type (coefficient: 0), continent (coefficient: -0.02), or mode of administration (coefficient: -0.03) on the overall estimate. Conclusion The overall estimate of the Cronbach alpha for OIDP questionnaire was above the accepted benchmark. There was no effect of moderators such as sex, age, language, population type, continent, or mode of administration on the overall estimate.