We conducted a case-control study using JMDC data and investigated the coexistence of NTM disease and comorbid conditions in Japanese adults. JMDC has collected reimbursement data since 2005 and the size of the database makes it valuable resource for estimating the prevalence of diseases. This is the first study evaluating comorbidities associated with NTM disease in Japanese adults based on a case-control analysis using Japanese health insurance claims database. We have newly added comorbidities such as CKD, DPB, osteoporosis, and Sjögren syndrome to previously reported comorbidities in the US claims-data analysis . Human immunodeficiency virus (HIV) and cystic fibrosis are well-known risk factors of NTM disease . However, our data did not contain individuals with these diseases because of their low prevalence in Japan. The calculated prevalence and the proportion of females with NTM disease were similar to those reported in a previous Japanese study  of NTM lung disease (NTM-LD). It is likely that the majority of cases in our study had NTM-LD despite of definition it in two codes, A31.0 and A31.9, which are used essentially interchangeably in Japan. The database in JMDC was not contain the result of culture or the residence of individuals; therefore, the analyses including those information was unavailable in this study. A strength of this study is that it included 10 sex-and-age-matched non-NTM controls per case. In addition, we performed Bonferroni-adjustment to resolve the problem of multiple hypothesis testing.
Respiratory diseases and a variety of systemic conditions are comorbid with NTM disease. In terms of respiratory diseases, Marras et al. [9, 16] found an association between NTM disease and aspergillosis, asthma, COPD, lung cancer, and tuberculosis in the US claims-data analysis. Our data showed that interstitial lung disease and asthma were comorbid with NTM disease in addition to the classical risk factors such as COPD. A Korean study with 810 NTM-LD patients revealed that 42 patients (5.2%) had interstitial lung disease, which is a similar proportion to that found in our study . Asthma and COPD, especially if treated with inhaled corticosteroid therapy, and the use of systemic immunosuppressive agents have previously been identified as strongly associated with NTM-LD [8, 9, 18, 19], which might also contribute to the development of NTM-LD in individuals with interstitial lung disease.
Tsuji et al.  conducted a retrospective study of 33 patients with diffuse panbronchiolitis (DPB) and found that 7 (21.2%) of the patients also had pulmonary NTM disease. In our analysis, 1.9% of patients with NTM disease also had DPB. This fact is more characteristic in Japanese or Asian patients than it is among individuals in the US or Europe because DPB mainly affects East Asian people . We speculated that DPB is a risk factor for NTM disease, but does not occur due to NTM disease. Mycobacterium spp. also easily colonize the respiratory tract of DPB patients because DPB is accompanied mucociliary dysfunction and various bacteria colonize the respiratory epithelium . NTM is a well-recognized pathogen in cystic fibrosis patients [21, 22]. However, the prevalence of cystic fibrosis in Japan is very limited, and there were no patients with cystic fibrosis in our data.
Kusumoto et al.  pointed out the high incidence of lung cancer in patients with NTM-LD, with an estimated incidence rate of 124.6 per 100,000 patient-years. This incidence is much higher than that reported in Japan. These association between NTM diseases and lung cancer could be a detection bias, as patients with NTM-LD have chest CT scans more frequently than healthy individuals and are therefore more likely to have other lung nodules discovered incidentally. Kusumoto et al. also speculated that chronic lung inflammation might cause lung cancer, however, the pathogenesis of lung cancer in relation to NTM disease is unknown .
In terms of non-pulmonary comorbidities, patients with NTM had a significantly higher prevalence of chronic heart failure, GERD, cancer other than breast, lung, ovary, or prostate cancer, and rheumatoid arthritis than controls in both males and females. Females with NTM disease also had a significantly higher prevalence of CKD, osteoporosis, and Sjögren syndrome than controls. The coexistence between NTM disease and chronic heart failure, GERD, other cancer, and rheumatoid arthritis has been reported previously [8, 9, 16]. Chao et al.  reported an increased risk of NTM disease in patients with Sjögren syndrome based on an analysis of Taiwan’s National Health Insurance Research Database analysis. Pulmonary abnormalities have been described as an extra-glandular involvement of Sjögren syndrome . NTM-LD may be an underlying condition in patients with Sjögren syndrome because some patients with Sjögren syndrome have been reported to develop bronchiectasis [25, 26], and pulmonary involvement in patients with Sjögren syndrome can be difficult to distinguish from NTM-LD using CT.
Menopause is a well-known risk factor for NTM disease , and osteoporosis also frequently occur in those population. Low serum estradiol had been related to development of NTM-LD , and it might be a confounding factor between osteoporosis and NTM diseases. In addition, Jeon et al.  reported the association between severe vitamin D deficiency and development of NTM disease; thus, vitamin D deficiency may also be a confounding factor of these diseases. Measurement of serum vitamin D level was not supported by the health insurance system in Japan in 2014, so we were unable to assess whether there was an association between vitamin D deficiency and NTM disease. COPD, a common chronic pulmonary disease, has been reported to be an independent risk factor of osteoporosis  and so there is a need to investigate the mechanism of development of osteoporosis in patients with NTM disease.
This study should be interpreted in light of several limitations. First, the analysis was based on data from a claims database. Since we did not have access to the clinical or microbiological information of the participants, the prevalence of some disease complexes might differ from that based on a true diagnosis and it needs to be interpreted with caution and would be preferred to be validated by clinical study. Previous researchers have defined their targeted diseases by ≥2 claims in several claims-data analyses [9, 16, 30-33]; however, we defined by ≥3 claims to increase the specificity. Our more stringent definition of comorbidities may have led to a bias toward the null, making it less likely to detect true associations. However, this feature of our study likely ensured that the detected associations were accurate and robust. This issue is further strengthened by the use of Bonferroni correction for multiple comparisons. The calculated prevalence and proportion of women were similar to that reported in the previous study ; thus, the accuracy of our diagnosis of NTM disease is likely to be comparable to that of previous studies. Second, we were unable to differentiate the Mycobacterium species due to unavailability of microbiological information. Third, JMDC database did not include patients aged ≥75 years belonging to National Health Insurance. Thus, the result should be interpreted as people <75 years of age and the association of older patients could differ. It would be ideal to have included patients aged ≥75 years as comorbidities increases with advanced age. In addition, there is another selection bias of JMDC database including only employee and their dependent family members and not including business owners or welfare recipients. Fourth, it should be interpreted to be careful to the differences with clinical significance in such a big data analysis instead of performance of Bonferroni-adjusted analysis. In addition, the prevalence of aspergillosis or DPB are 0 in non-NTM group; thus, it should be validated with larger datasets.