Discussion

Section:

ChooseTop of pageAbstractMethodsResultsDiscussion <<ReferencesCITING ARTICLES

Using a nationally representative population-based sample of Medicare beneficiaries, we found that the prevalence of PNTM, as estimated from diagnostic codes on medical claims, is increasing significantly across all regions of the United States and among both men and women. Additionally, we identified significant racial/ethnic differences that are independent of region. Although other studies have estimated the prevalence of PNTM within specific geographic areas in the United States (4–6), this is the first study to provide nationally representative estimates among older adults from all 50 states plus the District of Columbia and across various racial/ethnic groups. The ability to define the burden of PNTM across an 11-year period is important given the often chronic nature of this disease (12, 13).

Although clinic-based reports have suggested that the prevalence of PNTM has increased (14), few population-based studies have documented this (4, 15). du Moulin and colleagues (15) demonstrated a fivefold increase in the number of patients with M. avium complex isolates in Massachusetts from 1972 to 1983, suggesting that this trend of an increasing prevalence has been occurring before our study period, at least in some areas. More recently, an isolate-based study in four HMOs using American Thoracic Society (ATS) microbiologic criteria found an average annual prevalence of 30 and 23 cases/100,000 persons in women and men, respectively, which is similar to our average annual prevalence among women and men of 37 and 23 cases/100,000 persons, respectively (4). Although some individuals in our data set may have received a PNTM ICD-9 code without having met ATS criteria, our results most likely represent an underestimate of true disease prevalence, because using ICD-9 codes to identify PNTM cases meeting ATS criteria has been shown to miss approximately 25 to 75% of cases (4, 16). However, because NTM is a rare disease, those who do receive this diagnostic code are more likely to have clinical evidence of a pathogenic NTM species. Furthermore, using ICD-9 codes, which have been validated in other settings, allows for the estimation of prevalence in populations numbering millions of people, numbers that would not be feasible to study in depth in smaller clinical studies. Thus, despite the potential for some misclassification and underestimation of prevalence, ICD9 codes represent a critical tool for large-scale population studies of rare conditions. Outside the United States, increases in PNTM disease have also been noted in several Australian (17–19) and Asian studies (20–23). One Canadian study identified an annual increase in the isolation prevalence of PNTM of 8.4% per year from 1997 to 2003, which is consistent with our findings (24).

It is also important to note that the prevalence of PNTM in the United States exceeds that of tuberculosis (4), which has been decreasing in incidence in several major U.S. cities from 2000 to 2007 by 4 to 21% per year (25). It has been postulated that the decrease in tuberculosis may be contributing to the increase observed in NTM, as prior tuberculosis infection may possibly stimulate antimycobacterial immunity (26). Although 3.6% of the PNTM cases identified here also had an ICD-9 code for pulmonary tuberculosis, the use of diagnostic codes for this condition is considered to be highly unreliable, with a very low positive predictive value (27). This is because a patient suspected to have active tuberculosis may have been coded as such before the disease was confirmed; the patient may have been evaluated for a latent tuberculosis infection; the patient had a history of treated tuberculosis; or the patient was actually infected with another species of Mycobacterium (27), as might have often been the case here.

The increase in PNTM may represent a true increase in disease or a combination of factors, including improved diagnostics and heightened awareness for NTM. For instance, improved mycobacterial diagnostics and increased use of imaging technology, such as computed tomography scans (28), may have resulted in more accurate detection and increased diagnosis of NTM infections among patients. Still, increases in NTM isolation, sensitization, and disease prevalence have been noted worldwide and throughout various points in time (4, 17–24), suggesting that at least some of the observed increase is likely true and beyond the contribution of potential surveillance and diagnostic biases. Numerous hypotheses have been advanced for the increase in NTM prevalence, such as the Clean Water Act in 1970, which led to widespread disinfection of drinking water that may have increased NTM prevalence by selection due to increased resistance to chlorination relative to competing microorganisms (1). Additionally, therapies that inhibit tumor necrosis factor-α (TNF-α), a cytokine that mediates host immune defense against opportunistic pathogens, have become widely used within the last decade to treat autoimmune inflammatory conditions such as rheumatoid arthritis, increasing the risk of NTM infection in these patients and perhaps contributing to its increasing prevalence (16, 29). However, TNF-α inhibitor exposure is unlikely to account for the overall increases that we found.

Prior clinic-based studies in the United States have identified an increase in PNTM among white women with no identified risk factors (7). Although more than 90% of the PNTM cases in this study were white, Asians/Pacific Islanders were at significantly greater risk for disease, with a prevalence twofold that of whites. Additionally, whereas white women had a 50% higher prevalence than white men, among Asians/Pacific Islanders, men were more affected than women. In fact, Asian/Pacific Islander men were twice as likely as white men to have PNTM. The difference in risk between Asian/Pacific Islander men and women is notable, and may be attributed to behavioral or cultural factors associated with increased risk, such as frequent tobacco use, which among Asian/Pacific Islanders is higher in men than women (30, 31).

Assessing the role of race/ethnicity in disease can be complex, as racial/ethnic categorizations may be a surrogate measure for behavioral, cultural, and biological factors. Furthermore, the lack of clearly defined and consistently applied categorizations for racial/ethnic groups limits the ability to compare results across studies (32). In particular, the Asian/Pacific Islander category comprises a variety of subgroups that have significant health-related and demographic differences. Pacific Islanders are more likely than Asian Americans to be smokers, overweight, have lower income and education levels, and have higher chronic disease and mortality rates, despite no differences reported in access to care (33). It is also possible that some subgroups within the Asian/Pacific Islander population may share a common genetic ancestry and have increased susceptibility to PNTM disease. For instance, among patients in Japan with pulmonary M. avium complex infection, human leukocyte antigen–A33 and human leukocyte antigen–DR6 antigens, which have been linked to a role in host-defense mechanisms, were significantly more frequent than in the control population (34). PNTM disease dynamics are most likely driven by complex interactions among behavioral, cultural, genetic, and environmental factors.

In this study, the majority of Asian/Pacific Islander PNTM cases resided in Hawaii, which also had the highest PP at 396 cases/100,000 persons. However, even when Asians/Pacific Islanders were excluded, Hawaii still had the highest PP in the United States, suggesting that the regional exposure was dominant over the ethnic issues. Similarly, in 1979, a nationwide survey of state public health laboratories found Hawaii had the highest isolation rate of M. avium complex in the United States (35). Environmental conditions unique to Hawaii may contribute to greater levels of NTM exposure. For example, the soils of Hawaii are high in polycyclic aromatic hydrocarbons, environmental contaminants produced by fossil fuel combustion and human industry, but also by volcanic activity. Some mycobacteria have the unique property of being able to readily degrade polycyclic aromatic hydrocarbons, enabling them to thrive in this environment (36).

The higher prevalence of disease among older adults living in southeastern and western states is consistent with prior reports of higher human exposure and observed prevalence (5, 37, 38) and with studies isolating greater numbers of mycobacteria from southeastern states than from other locations surveyed (39). Although regional variation in medicare practices relevant to PNTM may exist, the strength of our findings indicates that these results are unlikely to be explained by differences in regional practices alone. PNTM is prevalent throughout the United States, and exposure may often occur through drinking water distribution systems (40). Households with average water heater temperatures 50°C or lower have higher rates of NTM recovery, but it is unclear whether their inhabitants are at higher risk (41). Therefore, regional factors, individual behaviors, and local exposures likely contribute to the true risk of infection.

PNTM cases identified here had more reported comorbidities than noncases. Almost all PNTM cases were diagnosed with at least one other lung-associated condition, including 44% with bronchiectasis, an otherwise uncommon condition. This frequency is much higher than what was previously reported. Among Oregon residents with PNTM, only 16% had bronchiectasis (42), and in a study of non-AIDS PNTM-associated hospitalizations, 15% also had bronchiectasis (5), although these studies both included patients less than 65 years old. Among patients in Canada more than 56 years old meeting ATS criteria for NTM infection, 62% had nodular bronchiectasis (43). Although bronchiectasis may increase susceptibility to NTM infection, infection with NTM likely also exacerbates bronchiectasis. Similarly, rheumatoid arthritis, which was twice as prevalent among PNTM cases as noncases, especially among women, may be associated with NTM infection, and this association may be increased by anti–TNF-α therapies (29). COPD is associated with PNTM, but this relationship could not be adequately assessed here due to identified challenges associated with its definition and subsequent coding (44). COPD is typically defined to include emphysema and chronic bronchitis, though other conditions, such as asthma and bronchiectasis, are also sometimes categorized as COPD by physicians (44) and may have been coded as such in some Medicare claims analyzed here, resulting in inflated estimates.

PNTM cases had higher mortality, especially among men and in those with greater numbers of comorbidities. Surprisingly, PNTM cases with bronchiectasis were half as likely to die as those without it. It is possible that those patients without bronchiectasis may be more likely to have fibrocavitary disease, which has been noted to have a worse prognosis (45). Black and Hispanic PNTM cases were more than twice as likely to die as whites, which may reflect racial/ethnic differences in the use of services among Medicare beneficiaries (46). Elderly white Medicare beneficiaries were more likely than blacks and Hispanics to receive care and seek out more services (46–48), regardless of urban or rural residence (46). Whites were also more likely to seek out care from private physicians, whereas blacks more frequently used community health centers (46), possibly resulting in barriers to access of specialized care, which may be more difficult to obtain in low-income areas. This might also account for the higher observed prevalence of bronchiectasis and PNTM in whites compared with blacks and Hispanics, as well as the lower death rate among those with a diagnosis of bronchiectasis, as whites might be more likely to obtain specialized testing, such as computerized tomography scans or X-rays, and to receive more intensive care and follow-up. In contrast, differences in use of Medicare services between whites and Asians/Pacific Islanders have not been consistently demonstrated (48–52).

Because PNTM is not a nationally notifiable disease, systematic monitoring of temporal trends for disease across all regions is difficult. However, with PNTM increasing among adults aged 65 years or older, a population that is expected to nearly double by 2030 (11), it is critical that this condition is recognized as an important public health issue with potentially significant consequences for affected patients.