Prediction models and questionnaires developed to predict vitamin D status in adults: a systematic review

A systematic review of prediction models/questionnaires developed to identify people with deficient/insufficient vitamin D status shows the potential of self-reported information to estimate vitamin D status. The objective is to identify and compare existing screening tools, developed to identify vitamin D deficiency or insufficiency in adults. A systematic search of literature was conducted using MEDLINE, Scopus, Web of Science and CINAHL databases. Risk of bias and applicability concerns were assessed by quality assessment of diagnostic accuracy studies (QUADAS-2). Data were extracted on socio-demographic, anthropometric, risk factors, serum 25 hydroxyvitamin D [25(OH)D] levels, statistical methods and predictive ability. A total of 12 studies were considered for inclusion for this systematic review after screening of 4851 abstracts and 15 full-text articles. Ten of twelve studies developed prediction models and 2 studies developed questionnaires. The majority of studies had low risk of bias and applicability as assessed by QUADAS-2. All studies included only self-reported predictors of vitamin D status in their final models and development of scores. Sunlight exposure and related factors were important significant contributors to the predictive ability of the models and/or questionnaires. Sensitivity and specificity of the prediction models or questionnaires ranged from 55 to 91% and 35 to 84%, respectively. Six out of twelve studies converted final models to scores associated with vitamin D status. There was no evidence that any of these existing tools have been translated into clinical practice. The prediction models or questionnaires identified in this systematic review were moderately sensitive and specific for identifying people with vitamin D deficiency or insufficiency. The substantial contribution of sunlight exposure to the prediction of vitamin D status highlights the importance of including this information when developing vitamin D screening tools.