OBJECTIVES: MSUD was recommended by the ACMG as a Primary Target condition for NBS, based on its total score (1483) on an 18 question survey which determined a (high) entry point into an algorithm (EPA), the threshold for which was a score > 1200. Different EPAs resulted in different follow up algorithm questions that led to recommendations. ACMG did not acknowledge any uncertainty around scores. The objective of this research is to examine survey responses of the only two survey questions Incidence and Simplicity of Therapy (Simplicity) for which MSUD data were reported and assess the potential influences of missing data and sampling variation on scoring, EPA and recommendations.

METHODS: The missing responses were determined based on total respondents and the observed responses to the two MSUD questions. The influence of missing data was estimated through boundary estimates (Manski, 1989); bootstrapping allowed estimation of sampling variation. The combined influences of both uncertainties was also estimated by bootstrapping around the lower bound (LB). As only a lower score could decrease the score in the direction of a new EPA, the LB was most relevant.

RESULTS: Missing responses were 2 and 1 for Incidence and Simplicity. For Incidence the LB was 14.58 (0.42 points below the original mean. For Simplicity, the LB was 26.46 (0.54 points below the original mean). Neither separately nor together was the score change sufficient to alter the EPA. Also, bootstrapping the original mean or around the LB did not result in a challenge to the original EPA.

CONCLUSIONS: Confidence in the EPA for MSUD was not altered by considering these two types of uncertainties due to the high initial score for MSUD. It is possible that a full consideration of the other 16 questions (with an additional 31 missing responses) could raise more significant challenges.