Quality assessment of the 2014 to 2019 national survey on drug use and health (NSDUH) public use files

Abstract

National Survey on Drug Use and Health (NSDUH) public use files (PUFs) have been produced using a statistical disclosure control technique named MASSC, which stands for Micro Agglomeration, Substitution, Subsampling, and Calibration, to protect confidentiality and quality of data. To inform researchers that NSDUH PUFs maintain high data quality and comparability with NSDUH restricted-use files (RUFs), about 300 NSDUH published tables (based on RUF data) of substance use and mental health were selected and reproduced using PUF data. Key estimates and their respective standard errors (SEs) produced from the two sets of 2014 to 2019 data files were compared. Summary statistics of ratios of the estimates and ratios of their SEs from PUF and RUF data were produced. Out of 22,000 estimates compared, average ratios for estimated percentages across years were within the 0.99 to 1.01 range, and the average increase in SEs for the estimates produced from PUFs across years was about 7–11%, for both substance use and mental health measures. Multiyear trend comparisons between PUF and RUF estimates were also conducted graphically to demonstrate PUF estimates provide similar trend patterns to RUF estimates across years. This study will provide confidence to researchers and policymakers for making policies and public health decisions based on NSDUH PUFs.