Pulmonary Nontuberculous Mycobacterial Disease, Ontario, Canada, 1998–2010

We measured the prevalence and temporal trends of pulmonary nontuberculous mycobacterial disease among residents of Ontario, Canada, during 1998–2010. Five-year prevalence increased from 29.3 cases/100,000 persons in 1998–2002 to 41.3/100,000 in 2006–2010 (p<0.0001). Improved laboratory methods did not explain this increase, suggesting a surge in disease prevalence.

The traditional method of identifying cases for NTM disease epidemiology studies by using mycobacterial laboratory databases and measuring annual prevalence is not ideal. Such studies assume that, in patients with pNTM disease, the organism is isolated during every year of disease, an invalid assumption (6). Recent investigators have focused on prevalence within a defined period (period prevalence) as an improved estimate of pNTM disease, including a 2-year study in Oregon (5), 3-year sampling of 4 US health care delivery systems (6), and <11-year USwide sample of Medicare beneficiaries (7). Important limitations of these studies included the patient populations and geographic regions selected and the limited data about temporal prevalence changes. Expanding on methods of previous studies to overcome some prior limitations, we performed a population-based study of pNTM disease in Ontario, Canada, using 5-year periods for prevalence calculations and compared prevalence from 1998-2002 to 2006-2010.

The Study
We performed a retrospective cohort study of all Ontario residents who had pulmonary nontuberculous Mycobacterium spp. isolated during 1998-2010, identified from the records of the Public Health Ontario Laboratory, capturing ≈95% of NTM disease in Ontario. Culture was performed by using Bactec 460 TB system until 2000 and thereafter with BACTEC MGIT 960 (Becton Dickinson, Baltimore, MD, USA). Before 2008, speciation was performed by using a combination of DNA probes (AccuProbe, Gen-Probe, San Diego, CA, USA) for Mycobacterium avium complex (MAC) and M. gordonae and high-performance liquid chromatography for other species and thereafter solely by DNA probes (AccuProbe, Gen-Probe) or line-probe assays (GenoType, Hain Lifescience, Germany). Because MAC was not identified to individual species for most of our study, we present data only for MAC.
Full criteria for pNTM disease include the presence of all clinical (symptoms and radiology) and microbiological components (1). We defined surrogate criteria as microbiological criteria only (1), (>2 positive sputum cultures or 1 bronchoscopic or lung biopsy culture), which has a positive predictive value of 70%-100% (5,6,9,10). Period prevalence of disease was calculated as the number of persons who fulfilled the disease criteria during a 5-year period (1998-2002 or 2006-2010), divided by the Ontario population at the period midpoint. We left a 3-year gap (2003)(2004)(2005) between periods to minimize patient overlap. We excluded M. gordonae from period prevalence because it is rarely pathogenic (1). We selected a conservative 5-year period on the assumption that the median survival with pNTM disease is 5-10 years (10,11), using the low end of the survival range based on assumptions that a small proportion of the cohort would not have true disease (misclassified by surrogate definition) and disease of an additional small proportion would be cured.
Annual isolation prevalence (number of persons in a calendar year with >1 pulmonary Mycobacterium spp. isolate divided by the contemporary population) and annual disease prevalence (number of persons in a calendar year whose illnesses fulfilled criteria for disease divided by the contemporary population) are presented for illustrative purposes. A generalized linear model with negative binomial distribution was used to assess annual rate changes, and a simple model binomial approach was used to compare 5-year period prevalence rates by using SAS 9.2 (SAS Institute, Cary, NC, USA). This study was approved by the University of Toronto Research Ethics Board with the requirement for informed consent waived.

Conclusions
The 5-year prevalence of pNTM disease was substantial and increased significantly during our populationbased assessment in Ontario, Canada. Our measurements of period prevalence (29.3 and 41.3 cases/100,000 persons) were substantially higher than observed in Oregon (8.6/100.000), probably partially because of the shorter period (2 years) and more stringent definition for disease (medical records review) used in the Oregon study (5). Other studies did not present period prevalence for the entire study populations, only by age strata, and used durations of 3 years (6,7) or <11 years (7). We selected a 5-year period assuming it would provide the most accurate estimate of disease prevalence based on the chronic nature of pNTM disease. Prior studies provided age-stratified data, with high period prevalence in older patients (20.4/100,000 to >200/100,000, depending on period length and specific age range) (5-7), as expected, because pNTM disease is a disease of the elderly (1,4,6,12). Although age data were unavailable for our study, annual disease prevalence of pulmonary MAC in Ontario has a strong age association, with an average increase of 14/100,000 per decade increase during 50-80 years (4).
Changes in microbiological methods and the number of samples submitted annually did not account for the increases in pulmonary nontuberculous Mycobacterium isolation (13). The attenuation in the rate of increase in isolation prevalence around the middle of the study corresponded with a previously reported plateau in the annual number of specimens submitted (13 isolation prevalence continued to rise, and the annual disease prevalence rose steadily throughout the study period. We suspect a multifactorial explanation for the increase in pNTM disease: an increase in susceptible hosts (aging, chronic lung disease) contributes (4); decades-old increases in water aerosol exposure could cause recent increases in pNTM disease, given the potential latency of pNTM disease; more computed tomographic scanning probably leads to sampling patients with previously unidentified abnormalities; and reduced tuberculosis, with an associated reduction in cross-immunity, may play a role. The latter is supported by observations of increased extrapulmonary NTM infection in children not vaccinated with M. bovis BCG (14,15). pNTM disease in Ontario is substantial and increased greatly from early (1998)(1999)(2000)(2001)(2002) to recent (2006-2010) periods. Dr Marras is an attending staff physician in the Division of Respirology, Department of Medicine, University Health Network and Mount Sinai Hospital, Toronto, and assistant professor of medicine, University of Toronto. His primary research interest is nontuberculous mycobacterial disease. Search past issues of EID at wwwnc.cdc.gov/eid