Background
Non-tuberculous mycobacteria (NTM) are ubiquitous environmental and biologically diverse micro-organisms, some of which are associated with chronic and progressive pulmonary infections in susceptible individuals [
1,
2]. In general,
Mycobacterium avium complex (MAC) is considered the most common causative agent of pulmonary NTM (PNTM) infection worldwide [
3‐
6].
Recent epidemiologic data suggest that both the incidence and the prevalence of NTM infections are increasing worldwide [
5,
7‐
12]. Furthermore, there is growing evidence that the epidemiology of PNTM infections is changing, with immunocompetent subjects suffering from pre-existing structural lung disease and adults aged ≥50 years with a substantial proportion of never-smoking females without previous lung disease being increasingly reported from outside Europe [
5,
7,
13,
14]. Most of these studies were either solely laboratory-based or used the frequency of NTM isolates from clinical specimens in order to determine prevalence and to reason the significance of NTM infection in conjunction with clinical and radiological data according to the American Thoracic Society (ATS) and Infectious Diseases Society of America (IDSA) statement on the diagnosis of NTM disease [
3,
7,
10,
11,
13‐
15].
However, while representative population-based data for Europe are limited [
11,
15], the epidemiology of PNTM infections in Germany is largely unknown. In Germany, population-based data regarding hospitalisations are available at a federal level. Although primarily used for the billing of health services, some important epidemiological evidence has originated from the analysis of
International Classification of Diseases hospital discharge diagnosis codes and health insurance claims [
9,
16].
Epidemiological and health services research on PNTM infections is required in order to optimise patient care and to inform health policy. The aim of the present study was to provide an overview of the current burden and show the trends of PNTM infection-associated hospitalisations in Germany.
Discussion
Our study provides a detailed picture of PNTM infection-associated hospitalisations in Germany and emphasises the considerable interaction with chronic respiratory diseases, particularly COPD as the most frequent associated condition. Between 2005 and 2011, elderly men had the highest age-specific hospitalisation rate. However, while the prevalence of PNTM infection-associated hospitalisations increased significantly and regardless of the increasing overall number of hospitalisations for both sexes, it showed the most pronounced increases among females, especially those of young and middle age, and hospitalisations associated with cystic fibrosis (CF). These findings may be relevant to patient care and the allocation of resources in healthcare.
Population-based data on the epidemiology of PNTM infections in Europe are limited [
11,
15]. Two recent studies from France and Denmark, which analysed periods from 2001 to 2003 and from 1997 to 2008 and linked NTM isolates from respiratory specimens to population-based clinical data, found similar average annual rates of 0.7 and 1.1 PNTM infection per 100,000 population [
11,
15]. However, the incidence remained unchanged during the respective study periods. In this concern, our results are consistent with the upward trends of PNTM infections from other laboratory- and population-based studies around the world [
8,
13,
16,
19‐
21]. Our overall average annual age-adjusted hospitalisation rate of 0.9 PNTM infections per 100,000 population as well as the average age-adjusted increase of 6%/year is slightly lower, but comparable to the findings of a recent study from the United States (US) [
9]. However, in line with other European studies that have shown consistently lower rates of PNTM than in North American studies [
3,
11,
14,
15], age and sex specific hospitalisation rates among elderly men and women were considerably lower in our study than in the aforementioned US study by Billinger and colleagues [
9]. Furthermore, in agreement with other US studies we found that the prevalence of PNTM infection was most significantly increasing among females [
3,
9,
13,
16]. In fact, males carried the major burden of PNTM-associated hospitalisations in our study, whereas elderly women with nodular bronchiectatic disease have been considered as the predominant patient population outside Europe [
3,
14,
16,
22]. This finding supports the existence of a unique epidemiology of PNTM infections in accordance with the distinct distribution of NTM species isolated from respiratory samples within Europe and recent reports from other European countries [
6,
11,
15,
19,
20,
23], where PNTM infections appear to be . However, it remains to be studied whether changing host factors or a changing NTM species distribution over time account for these observations [
6].
Notably, we observed a significant decline of immunodeficiencies, tuberculosis (TB), alcohol-related disorders and influenza and pneumonia among PNTM infection-associated hospitalisations. In line with previous studies observing an increase in PNTM infections, the incidence of TB and with it the number of TB-associated hospitalisations declined steadily in Germany during the study period [
8,
12,
17,
21,
24,
25]. However, in contrast to reports from the US and Canada [
3,
13,
14,
21], the incidence of TB may still outrange that of PNTM infections in most European countries [
25].
The present study is a nationally representative population-based analysis of the burden and trends among almost 6,000 PNTM infection-associated hospitalisations, including 80% of hospitals across the country and >125 million hospitalisations over a 7-year period. However, our study has some inherent limitations. Firstly, ICD-10 diagnosis codes are primarily used for reimbursement purposes and only secondarily for epidemiological research. They may be subject to potential sources of errors, lack validation for PNTM infections and, in general, are considered to have high specificity, but only moderate sensitivity, thus being prone to an underestimation of disease prevalence [
9,
20,
26]. Secondly, the diagnosis of PNTM infection did not require compliance with the ATS/IDSA diagnostic criteria for NTM disease [
1]. Details on isolated mycobacterial species and disease presentations were unavailable. However, the fact that PNTM infection itself was the primary diagnosis in >50% of hospitalisations and severe complications like respiratory failure were frequent and significantly increasing over time suggests clinically relevant disease in the majority of hospitalisations. Thirdly, we were unable to account for readmissions, which may have had an impact on hospitalisation rates, though a substantial overestimation appears unlikely. Fourthly, several previous studies have suggested that the epidemiology of PNTM infections may be influenced by environmental, geographical and sociodemographic patterns [
22,
24,
27,
28]. However, we did not analyse the data stratified for individual federal states or urban and rural areas, mainly due to the potential bias related to regional differences of health care utilisation of medical services, and the unavailability of information on NTM species. Finally, our results apply to hospitalised populations only. PNTM infections are chronic in nature and usually require long-term follow-up care in the outpatient setting, where disease prevalence is likely different. A recent report on patient-centred care in outpatient respiratory medicine, which was based upon the billing data of 30 representative German private respiratory practices in 2010, showed that ICD-10 category A31 (infection due to other mycobacteria) accounted for on average 0.058% of outpatient diagnoses and correspondingly roughly 1,850 consultations [
29]. Moreover, the fact that none of the few deaths occurred during hospitalisation, further emphasises that the major burden of disease is managed out-of-hospital. Therefore, our data are likely to underestimate the overall burden of PNTM infections. However, as data are limited, our results are the best currently available surrogate for the epidemiological trends of PNTM infections across Germany.
Altogether, the reasons for increases in PNTM cases are unknown. Increasing hospitalisation rates do not necessarily indicate increasing true incidence or prevalence of PNTM infections. They may rather be attributable to advances in laboratory culture and molecular speciation methods, increasing awareness, increasing environmental exposure, decreasing immunity to mycobacteria and/or aging and increasingly prevalent susceptible populations [
4,
10,
12,
24]. In this respect, our findings affirm the close link between PNTM infection and COPD, which has also been observed by others [
11,
13,
14]. A recent Danish population-based case–control study demonstrated that chronic respiratory disease, particularly COPD, is a strong risk factor for PNTM disease [
23]. On the other hand, a Canadian study found that the increase in pulmonary MAC disease was likely multifactorial and could not be explained exclusively by population aging, COPD and other risk factors [
22]. Our finding that hospitalisation rates were not continuously increasing with age is contrary to recent studies from the US and Canada, which found the highest rates among the very advanced age group [
9,
13,
22], and argues against more severe disease and frequent readmissions in this age group. It may rather indicate age-related differences regarding the access to health services and suggest that PNTM infections are still underdiagnosed during hospitalisations of the very advanced age group in Germany. The increasing number of unrelated comorbidities, along with the upward trend regarding the average number of secondary diagnoses per primary diagnosis of PNTM infection, may reflect the increasingly aged and comorbid population during the study period. This is supported by the fact that both the average age of hospitalised subjects and the average number of secondary diagnoses per case were steadily increasing in general for hospitalisations during the study period (from 52.5 to 54.6 years and from 3.9 to 4.8 secondary diagnoses between 2005 and 2011, respectively). Moreover, the observation that cachexia, defined as a body mass index <18.5 kg/m
2 according to common DRG coding guidelines, was a frequent secondary condition in 11% of hospitalisations with PNTM infection as the primary diagnosis supports the existence of a distinct morphotype in a subset of patients, as recently suggested by Kim and colleagues [
30]. In this study, respective subjects were taller and leaner than controls and had high rates of CF transmembrane conductance regulator (CFTR) mutations in 36%.
Our findings have several important implications. Although we were unable to estimate the costs for hospitalisations associated with PNTM infection, our data on the mean LOS confirm that PNTM infections represent an underappreciated economic healthcare burden, with considerable associated treatment costs comparable to that of other chronic infectious diseases [
31]. Furthermore, the liaison between PNTM infection and chronic respiratory diseases like COPD, bronchiectasis and CF, which were all significantly increasing as associated conditions during the study period, calls for caution regarding the increasingly broad use of macrolide antibiotics as an adjunct immunomodulatory treatment [
32‐
34]. It has been shown that macrolides block autophagy and inhibit the intracellular killing of mycobacteria [
35]. As a consequence, their chronic use may hypothetically predispose to PNTM infection and, moreover, promote the emergence of macrolide resistance, which may be associated with worse outcomes [
1,
35]. A novel finding from our study is that young and middle-aged females showed the most pronounced increases of average age-specific hospitalisation rates of up to 17%/year. This observation may highlight the importance of genetic susceptibility, particularly CFTR mutations, and warrant extensive work-up and continued surveillance of these patients. Lastly, it is important to note that our results may not fully apply to other European countries and settings.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
FCR conceived and designed the study, performed the statistical analysis, interpreted the data and drafted the manuscript. RMA participated in the study design and revised the manuscript critically for important intellectual content. FCB revised the manuscript critically for important intellectual content. ADR participated in the study design and revised the manuscript critically for important intellectual content. MWP interpreted the data and contributed to the drafting of the manuscript. JR participated in the study design and revised the manuscript critically for important intellectual content. HS revised the manuscript critically for important intellectual content. DW interpreted the data and contributed to the drafting of the manuscript. TW contributed to the study design, supervised the study and revised the manuscript critically for important intellectual content. All authors read and approved the final manuscript.