Abstract
RATIONALE: State-specific case numbers and costs are critical for quantifying the burden of pulmonary nontuberculous mycobacterial disease in the United States. OBJECTIVES: To estimate and project national and state annual cases of nontuberculous mycobacterial disease and associated direct medical costs. METHODS: Available direct cost estimates of nontuberculous mycobacterial disease medical encounters were applied to nontuberculous mycobacterial disease prevalence estimates derived from Medicare beneficiary data (2003-2007). Prevalence was adjusted for International Classification of Diseases, 9th Revision, undercoding and the inclusion of persons younger than 65 years of age. U.S. Census Bureau data identified 2010 and 2014 population counts and 2012 primary insurance-type distribution. Medical costs were reported in constant 2014 dollars. Projected 2014 estimates were adjusted for population growth and assumed a previously published 8% annual growth rate of nontuberculous mycobacterial disease prevalence. MEASUREMENTS AND MAIN RESULTS: In 2010, we estimated 86,244 national cases, totaling to $815 million, of which 87% were inpatient related ($709 million) and 13% were outpatient related ($106 million). Annual state estimates varied from 48 to 12,544 cases ($503,000-$111 million), with a median of 1,208 cases ($11.5 million). Oceanic coastline states and Gulf States comprised 70% of nontuberculous mycobacterial disease cases but 60% of the U.S. population. Medical encounters among individuals aged 65 years and older ($562 million) were twofold higher than those younger than 65 years of age ($253 million). Of all costs incurred, medications comprised 76% of nontuberculous mycobacterial disease expenditures. Projected 2014 estimates resulted in 181,037 national annual cases ($1.7 billion). CONCLUSIONS: For a relatively rare disease, the financial cost of nontuberculous mycobacterial disease is substantial, particularly among older adults. Better data on disease dynamics and more recent prevalence estimates will generate more robust estimates.