Synthesis framework estimating prevalence of MCADD and sensitivity of newborn screening programme in the absence of direct evidence

• Published in: Journal of clinical epidemiology 2014-10, Vol.67 (10), p.1131-1138
• Main Authors: Leal, Jose, Wordsworth, Sarah, Oerton, Juliet, Khalid, Javaria M, Dezateux, Carol
• Format: Article
• Language: English
• Subjects: Acyl-CoA Dehydrogenase - deficiency; Acyl-CoA Dehydrogenase - genetics; Adolescent; Bayes Theorem; Child; Child, Preschool; Confidence intervals; Dehydrogenases; Disease; England - epidemiology; Epidemiology; Homozygote; Humans; Inborn errors of metabolism; Infant; Infant, Newborn; Internal Medicine; Lipid Metabolism, Inborn Errors - epidemiology; Lipid Metabolism, Inborn Errors - genetics; MCADD; Medical screening; Medium chain acyl-CoA dehydrogenase deficiency; Metabolism; Mutation; Neonatal Screening; Population; Prevalence; Program Evaluation; Screening; Sensitivity; Sensitivity and Specificity; Studies; Tandem mass spectrometry
• ISSN: 0895-4356; 1878-5921
• DOI: 10.1016/j.jclinepi.2014.05.011

Abstract Background and Objectives Several countries have included medium-chain acyl-CoA dehydrogenase deficiency (MCADD) in their newborn screening programs. However, the sensitivity of the programs cannot be estimated directly as only individuals with a positive result undergo a definitive diagnostic test. We propose a framework to overcome this limitation and estimate the prevalence of disease, sensitivity of screening, and its yield relative to no screening. Study Design and Setting A Bayesian model simultaneously combined available prevalence data on the most common mutation of MCADD (c.985A>G) in screened and nonscreened populations using the relationship between true and apparent prevalence of disease. Data originated from screening pilots in England, disease surveillance studies, and published literature. Model validity and consistency were formally checked. Results True prevalence of c.985A>G homozygotes in England was 6.2 per 100,000 individuals, and the sensitivity of the screening program was 94% (95% confidence interval [CI]: 74, 100%) compared with a detection rate in nonscreened areas of 48% (95% CI: 30, 68%) by age of 5 years. Hence, the screening program detected 47% (95% CI: 30, 60%) additional cases compared with no screening. Conclusion The sensitivity of the screening program in England was high and our estimation approach could be adapted to inform other jurisdictions, rare diseases, and newborn screening programs.