Nontuberculous mycobacterial pulmonary disease: a nationwide cost analysis of healthcare costs, foregone earnings and public benefits in a universal healthcare setting

Abstract

Background

Nontuberculous mycobacterial pulmonary disease (NTM-PD) is associated with a significant economic burden due to complex diagnostics, prolonged multidrug treatments and hospitalisations. However, cost data are limited, particularly from countries with universal healthcare. We evaluated the economic burden of NTM-PD in Denmark.

Methods

We conducted a nationwide cost analysis of all adults diagnosed with NTM-PD between 2005 and 2017, identified through International Classification of Diseases, 10th Revision codes. Each case was matched 1:4 to comparators by age, sex, marital/cohabitation status and municipality of residence. We assessed direct healthcare costs (primary and secondary care, and medications) and indirect costs (foregone earnings and public benefits for those aged 18–64 years) 3 years before and after the diagnosis. Annual costs were estimated using a generalised linear model with a gamma distribution and log link.

Results

We included 545 NTM-PD patients and 2150 comparators. Patients had lower education levels and employment rates. Direct healthcare costs peaked in the diagnosis year at EUR 24 454 (95% CI 22 378–26 723) for patients versus EUR 4233 (95% CI 4057–4417) for comparators. Employment income declined before diagnosis and remained significantly lower (EUR 9547 versus 18 720), while public benefit payments were higher (EUR 18 005 versus 8978). Net costs peaked at EUR 29 394 (95% CI 27 523–31 408) in the diagnosis year, with comorbidity-adjusted direct healthcare costs being 4.8 times higher for patients (p<0.001).

Conclusion

NTM-PD is associated with a substantial economic burden in Denmark, with nearly 5-fold higher direct healthcare costs, reduced income, increased public benefit payments and with net costs exceeding EUR 29 000 in the diagnosis year.

Shareable abstract

Despite universal healthcare, NTM-PD is associated with a substantial economic burden in Denmark https://bit.ly/47AMz2u

Introduction

Nontuberculous mycobacterial pulmonary disease (NTM-PD) is an increasingly recognised clinical challenge, mainly among middle-aged to elderly patients with pre-existing lung diseases [1, 2]. NTM-PD is associated with high morbidity and mortality [3], and places a considerable economic burden on healthcare systems due to the excessive costs associated with complex diagnostics, prolonged multidrug treatments and hospitalisations [4–6]. For example, a German study estimated that direct healthcare expenditures for NTM-PD patients were nearly 4 times higher than for matched controls, with additional indirect costs arising from reduced work capacity and premature mortality [5]. Moreover, increased healthcare utilisation has been observed not only after diagnosis but also in the years leading up to it, reflecting the cumulative disease burden [7].

In Denmark, NTM-PD incidence and prevalence are also increasing, coinciding with an ageing population [8, 9]. However, the financial implications remain undescribed. The Danish healthcare system is tax funded, aiming to provide free and universal access to medical care through a highly organised and decentralised structure, with a general practitioner-led gatekeeper model [10]. Denmark's welfare model aims to provide financial support to all citizens while addressing the social determinants of health. Those unable to financially support themselves are most often entitled to certain public benefits [11].

Given differences in healthcare financing, delivery and social support across countries, we aimed to evaluate the economic burden of NTM-PD in Denmark, a country with a unique healthcare system, using nationwide registers.

Methods

Study design, patient eligibility, follow-up and case–comparator matching

We performed a nationwide cost analysis, retrospectively including all adults (≥18 years old) in Denmark with a first-time diagnosis code of NTM-PD from 2005 to 2017. NTM-PD was defined as a code of A31 (Pulmonary mycobacterial infection) and subcodes A31.8 (Other mycobacterial infections) or A31.9 (Mycobacterial infection, unspecified) in combinations with procedural codes and comorbidity codes in line with earlier methods (supplementary table S1) [8]. Diagnosis codes were based on the International Classification of Diseases, 10th Revision (ICD-10) system [12]. The date of the first diagnosis code was considered the index date. Patients were followed for 3 years before and after the index date (supplementary figure S1).

Patients were compared to a matched group of comparators, randomly selected from the Danish general population, excluding patients with an A31 (or subcode) diagnosis code. Patients were matched in a 1:4 ratio to comparators by age, sex, marital/cohabitation status and municipality of residence at the time of diagnosis. Comparators entered the study on the same calendar date as their matched patient's index date.

Economic evaluations

Direct healthcare costs included expenses from: 1) primary care, 2) inpatient and outpatient secondary care, and 3) prescription medication. As a measure of indirect costs, foregone earnings and public benefits were estimated. Foregone earnings were calculated as the difference in employment income between comparators and patients. Net costs, defined as the sum of direct costs and foregone earnings, were computed to assess the overall economic burden of disease. For those in the workforce (aged 18–64 years), public benefits considered included temporary unemployment benefits (unemployment benefits, social security and sick pay), pensions (disability and early retirement pension) and other public benefits (housing benefits, child benefits, “green cheques” and student grants). All costs were presented as 2020 prices in EUR, adjusted for inflation using the general price index [13].

Data sources

All NTM-PD patients in Denmark are managed exclusively in hospitals, either as inpatients or outpatients, and were identified using the Danish National Patient Registry [14]. The general population used for matching was identified through the Danish Civil Registration System, which includes all individuals legally residing in Denmark [15]. Similar to previously used methods, primary care costs were obtained from the Danish Health Insurance Registry, which documents health service consumption and costs within the Danish public insurance system [16, 17]. Secondary care costs, including somatic inpatient admissions and outpatient visits, were obtained from the Danish National Patient Registry [14]. Diagnosis-related group rate tariffs were applied to calculate the cost of these services [18]. Prescription medication costs were obtained from the Danish National Prescription Registry, which has recorded prescription sales in Denmark since 1995 [19]. Income and public benefits were extracted from the Danish Income Registry [20].

Statistical analyses

Sociodemographic characteristics were summarised using frequency and percentage for categorical variables or median with interquartile range (IQR) for continuous variables. Differences between patients and comparators were assessed using Chi-squared tests for categorical variables or Mann–Whitney U-tests for continuous variables.

To select the comparator group, the SAS SURVEYSELECT procedure was used to randomly sample individuals from the general Danish population, excluding those with any A31 diagnosis.

Annual direct costs, income and public benefits were estimated using a two-step generalised linear model with a gamma distribution and log link function. Unadjusted costs were predicted separately for patients and comparators. Least squares means were computed using the SAS LSMEANS statement to estimate predicted annual average direct costs, income and public benefits for patients and comparators, along with 95% confidence intervals and p-values. Costs were predicted for 3 years before and after the index date (year −3, −2, −1, 0, 1, 2) (supplementary figure S1). Costs incurred during the diagnosis year (year 0) were included in the succeeding year's calculations. To account for patient mortality during follow-up, we incorporated weights for patient-years, where each individual's contribution to annual estimates was weighted by the proportion of the year they were alive.

In an adjusted model, cost ratios (values >1 indicating higher costs for patients than comparators) were calculated, adjusting for comorbidity burden using the Charlson Comorbidity Index (CCI), as described by Quan et al. [21], and lung disease. To ensure accurate representation, lung disease (defined as ICD-10 codes J40–J47 (chronic lower respiratory diseases, including chronic bronchitis, emphysema, COPD, asthma and bronchiectasis) and J60–J67 (lung diseases caused by external agents, including pneumoconiosis due to dusts, asbestosis, silicosis and hypersensitivity pneumonitis)) was excluded from the CCI score and modelled separately as a binary variable (yes/no). Comorbidity burden was categorised as a binary variable, indicating whether the patient had a CCI score ≥1 (yes/no).

For public benefit estimations, only individuals aged 18–64 years (workforce population age) were included.

Statistical significance was defined as a two-sided p-value <0.05. Analyses were performed with Stata 16.1 (StataCorp, College Station, TX, USA) and SAS 9.4 TS Level 1M5 (SAS Institute, Cary, NC, USA). Data were visualised in R version 4.3.2 (www.r-project.org) using the package ggplot2 version 3.5.0.5.

Results

In total, we included 545 patients and 2150 comparators. Sociodemographics of patients and comparators are presented in table 1. The median age was 66 years in both groups, and approximately half were female. Patients had a lower level of education, with 41.3% having only primary education compared to 35.4% of comparators (p<0.001). Employment status also differed significantly, with fewer patients being employed (16.5% versus 32.5%) and a higher proportion receiving disability pensions (17.2% versus 6.1%) or age pensions (52.7% versus 50.7%) (p<0.001). Comorbidities were more common among patients, with a higher proportion having a CCI score ≥1 (24.2% versus 12.4%) and lung disease (45.1% versus 3.7%) (p<0.001).

TABLE 1.

Sociodemographics of patients and comparators in the year of diagnosis

	Patients (n=545)	Comparators (n=2150)	p-value#
Sex			Matched
Female	281 (51.6)	1105 (51.4)
Male	264 (48.4)	1045 (48.6)
Median (IQR) age, years	66 (18)	66 (17)	Matched
Age group			Matched
18–29 years	21 (3.9)	81 (3.8)
30–39 years	21 (3.9)	82 (3.8)
40–49 years	35 (6.4)	139 (6.5)
50–59 years	88 (16.1)	350 (16.3)
60–69 years	156 (28.6)	620 (28.8)
≥70 years	224 (41.1)	878 (40.8)
Marital/cohabitation status			Matched
Single	222 (40.7)	872 (40.6)
Married or cohabiting	323 (59.3)	1278 (59.4)
Region			Matched
North	69 (12.7)	274 (12.7)
Central	137 (25.1)	544 (25.3)
Southern	123 (22.6)	486 (22.6)
Capital	129 (23.7)	504 (23.4)
Zealand	87 (16.0)	342 (15.9)
Education			<0.001
Primary	225 (41.3)	761 (35.4)
Secondary	22 (4.0)	61 (2.8)
Vocational	147 (27.0)	735 (34.2)
Short college	14 (2.6)	78 (3.6)
Medium college	52 (9.5)	318 (14.8)
Master/PhD	22 (4.0)	123 (5.7)
Unknown	63 (11.6)	74 (3.4)
Employment status			<0.001
Employed	90 (16.5)	698 (32.5)
Unemployed	33 (6.1)	72 (3.3)
Disability pension	94 (17.2)	131 (6.1)
Early retirement	26 (4.8)	86 (4.0)
Age pension	287 (52.7)	1091 (50.7)
Education	5 (0.9)	34 (1.6)
Other	10 (1.8)	38 (1.8)
Comorbidity
CCI score (excluding lung disease) ≥1¶	132 (24.2)	267 (12.4)	<0.001
Lung disease+	246 (45.1)	79 (3.7)	<0.001

Data are presented as n (%), unless otherwise stated. IQR: interquartile range; CCI: Charlson Comorbidity Index. ^#: Chi-squared test p-values for patients versus comparators are for comparisons across the whole dataset for each variable. “Matched” indicates variables on which patients and comparators were matched, for which statistical testing is not applicable. ^¶: CCI as defined by Quan et al. [21]. ⁺: International Classification of Diseases, 10th Revision: chronic lower respiratory diseases (J40–J47) and lung diseases due to external agents (excluding respiratory conditions due to inhalation of chemicals, gases, fumes and vapours, pneumonitis due to solids and liquids, and respiratory conditions due to other external agents) (J60–J67). In patients versus comparators, COPD (J43–J44) was registered in 206 (37.8%) versus 56 (2.6%) (p<0.001), asthma (J45–J46) in 22 (4.0%) versus 23 (1.1%) (p<0.001), bronchiectasis (J47) in 40 (7.3%) versus three (0.1%) (p<0.001) and other lung diseases in 77 (14.1%) versus 14 (0.7%) (p<0.001). Cystic fibrosis was registered in 12 (2.2%) of the patients compared to less than three (0.1%) of the comparators (p<0.001).

Direct healthcare costs were significantly higher for patients than for comparators, peaking around the year of diagnosis (year 0) and then declining (figure 1a). The average direct healthcare costs in the diagnosis year were EUR 24 454 (95% CI 22 378–26 723) compared to EUR 4233 (95% CI 4057–4417) for the comparators. Patients had a decline in employment income in the years leading up to the diagnosis, remaining consistently lower, with an average employment income of EUR 9547 (95% CI 8736–10 433) in the diagnosis year compared to EUR 18 720 (95% CI 17 939–19 535) observed for comparators. Public benefit costs were higher for patients, particularly in the diagnosis year and the following year, with an average of EUR 18 005 (95% CI 15 782–20 542) in the diagnosis year compared to EUR 8978 (95% CI 8419–9574) for comparators. Foregone earnings dropped from EUR 10 121 (95% CI 10 059–10 142) at year −3 to EUR 9173 (95% CI 9102–9203) at the diagnosis year and EUR 8755 (95% CI 8577–8755) at year 2 (figure 1b). Conversely, net costs increased sharply, peaking at EUR 29 394 (95% CI 27 523–31 408) in the diagnosis year, and then declined in the subsequent years.

FIGURE 1.

Annual average a) total direct healthcare costs, employment income and total public benefits for patients and comparators, and b) foregone earnings and net costs for patients, 3 years before and after the nontuberculous mycobacterial pulmonary disease diagnosis (year 0). The black dashed vertical line represents the year of diagnosis.

Inpatient secondary care was the largest contributor to healthcare costs, followed by outpatient care (figure 2a). Disability pension accounted for the largest share of public benefits, followed by social security and early retirement pensions (figure 2b).

FIGURE 2.

Annual average a) direct healthcare costs and b) public benefits (i.e. temporary unemployment benefits, pensions and other public benefits (housing benefits, child benefits, “green cheques” and student grants) 3 years before and after the nontuberculous mycobacterial pulmonary disease diagnosis (year 0).

After adjusting for CCI and lung disease, total direct healthcare costs were 4.8 (95% CI 4.3–5.4) times higher for patients than for comparators in the diagnosis year (figure 3). Secondary inpatient care costs were 6.1 (95% CI 5.5–6.8) times higher in the diagnosis year. Primary care (1.2 (95% CI 1.1–1.4)) and secondary outpatient care (4.7 (95% CI 4.2–5.2)) peaked the year before diagnosis, while prescription medication costs rose gradually before diagnosis and peaked around year 1 (5.2 (95% CI 4.7–5.8)). Total public benefits were consistently higher for patients throughout the period, reaching 1.6 (95% CI 1.4–1.9) times higher by year 1. Unemployment benefits and sick pay increased after diagnosis, while social security and disability pension costs decreased slightly before diagnosis but remained considerably higher than those of comparators. Early retirement pension costs decreased for patients before diagnosis and were lower than for comparators, but increased post-diagnosis (0.9 (95% CI 0.8–1.1) by year 1). Employment income was lowest for patients 2 years before diagnosis (0.6 (95% CI 0.6–0.7)) but increased slightly over time (0.7 (95% CI 0.6–0.8) by year 2), although it remained consistently lower than in comparators. Cost ratios remained comparable when the model was further adjusted for education (supplementary figure S2). Estimates adjusted for CCI alone and for lung disease excluding CCI are shown in supplementary figures S3 and S4.

FIGURE 3.

Cost ratios for patients versus comparators, adjusted for Charlson Comorbidity Index (CCI) (excluding lung disease) and lung disease, 3 years before and after the nontuberculous mycobacterial pulmonary disease diagnosis (year 0). CCI calculated as described by Quan et al. [21], and lung diseases defined as International Classification of Diseases, 10th Revision codes for chronic lower respiratory diseases (J40–J47) and lung disease due to external agents (excluding respiratory conditions due to inhalation of chemicals, gases, fumes and vapours, pneumonitis due to solids and liquids, and respiratory conditions due to other external agents) (J60–J67). Public benefits were only calculated for those in the workforce (aged 18–64 years). The black dashed horizontal line represents no cost difference between groups, while values >1 indicate higher costs for patients.

Discussion

In this nationwide Danish study, we observed that the net costs associated with NTM-PD were EUR 29 394 in the diagnosis year, with direct healthcare costs nearly 5 times higher than for matched comparators. NTM-PD was also associated with lower employment income and higher public benefit payments, reflecting indirect patient costs.

Although our study provides overall estimates of the economic burden, the specific cost components of NTM-PD in Denmark remain unknown. In practice, treatment often involves the use of multiple concurrent antimicrobials over prolonged periods, together with repeated microbiological testing, monitoring for toxicity and management of adverse events. These components contribute to the overall expenditures captured in our data, but they were not disentangled as separate cost categories. The introduction of newer, high-cost agents such as novel β-lactamase inhibitors, tedizolid, omadacycline and liposomal amikacin, among others, is also likely to increase future expenditures. The cost and availability of these drugs may vary considerably across settings, potentially influencing treatment decisions and access. This highlights the need for detailed cost studies in Denmark to inform future healthcare planning.

Evidence from other countries shows that NTM-PD is associated with substantial and, in some settings, increasing healthcare costs. In South Korea, annual direct expenditures for NTM diseases increased by more than 5.8-fold from 2010 to 2021 [22], suggesting increased healthcare utilisation due to rising NTM-PD prevalence or a growing number of vulnerable patients with prolonged disease courses. Similar upward trends in hospital costs have been reported in the USA and Germany [23, 24]. A study from the USA estimated that NTM, along with Pseudomonas and Legionella, accounted for the majority of hospitalisations and deaths from waterborne diseases, with annual direct healthcare costs exceeding USD 2 billion [25]. A similar pattern could occur in Denmark if the prevalence of NTM-PD continues to increase [26].

Consistent with our findings, previous studies have shown that costs are not only elevated at diagnosis and during treatment but also in the years preceding and following diagnosis [4, 7, 27]. The observed reductions in employment income and increases in public benefits indicate the broader socioeconomic impact of NTM-PD, including reduced work capacity and increased reliance on social support, in line with earlier reports [5]. The early decline in employment income and increased indirect costs even before diagnosis suggest that the disease may impair work ability well before its formal recognition. Evidence on longer term economic impacts beyond a few years is lacking, making it speculative whether the financial burden persists for many years after diagnosis. This aligns with our observation that total direct healthcare costs were already twice as high among those with pre-existing lung disease 3 years before, as shown in the supplementary material. Managing pre-existing lung disease and conducting diagnostic workup may contribute substantially to pre-diagnostic costs.

Compared to our findings, a German study reported mean direct expenditures of EUR 39 560 per NTM-PD patient, nearly 4 times higher than those of matched controls, with hospitalisations accounting for 63% of total costs [5]. Attributable annual direct and indirect costs per patient were estimated at EUR 9093 and 1221, respectively. A study from South Korea estimated a median cumulative total medical cost of USD 5044 over 49.7 months of follow-up, with diagnostic testing and medication accounting for 59.6% of the total costs [6]. Notably, up to 50% of these costs were patient co-payments due to limited insurance coverage, highlighting the financial burden on individuals in healthcare systems with less comprehensive coverage than in Denmark. Another study in three European countries and Canada estimated the average direct medical costs per person-year at approximately EUR 17 900 (France), EUR 11 600 (Germany), GBP 9700 (UK) and CAD 16 200 (Canada) [28], based on a survey of physicians treating Mycobacterium avium complex (MAC)-PD. Costs undoubtedly also vary across healthcare system logistics and treatment settings, such as inpatient versus outpatient care and reimbursement structures. A recent Danish cohort study found that 42% of treated NTM-PD patients were never admitted, and among those admitted, the median (IQR) duration of hospitalisation was only 9 (23) days [29]. In Germany, outpatient drug costs accounted for >70% of total outpatient expenditures for patients with MAC-PD [30], while in other settings, hospitalisations represented the primary cost driver [27, 28]. Among older adults, the economic burden is particularly high, with hospitalisation and medication accounting for most costs [31].

Beyond differences between healthcare systems, patient characteristics may also influence the observed variation in costs. Differences in sociodemographic and clinical characteristics between patients and controls, such as the observed lower educational level, fewer employed individuals and a higher comorbidity burden, may heighten vulnerability and contribute to increased costs in certain subgroups. A large South Korean cohort study found that patients infected with NTM had 1.5 times higher total medical costs and 4.5 times higher respiratory disease-related costs compared to controls, with the highest healthcare expenditures occurring in the 6 months preceding diagnosis [7]. Certain comorbidities, such as gastro-oesophageal reflux disease, have also been associated with significantly higher healthcare use and costs among NTM-PD patients [32]. Costs were also influenced by pathogen species, treatment complexity and adherence to guidelines. Higher costs have been linked to M. abscessus and M. xenopi infection, intravenous therapy, and extensive NTM disease [6, 33, 34]. A US study reported that patients with M. abscessus had a 9.5-fold increased likelihood of incurring high treatment costs [34]. Treatment is often prolonged and complex, with a median of five antibiotics and a treatment burden of up to 7689 drug-days [34]. Additionally, adverse effects are frequent, affecting 50–100% of patients, depending on the type of antibiotic [34]. Interestingly, guideline-based treatment has been associated with lower healthcare expenditures compared to non-guideline-based treatment [35], demonstrating the potential public health benefits of adhering to standard-of-care recommendations, as well as improved outcomes [36]. However, deviations from guideline-based treatment may reflect more severe or vulnerable patients who cannot tolerate standard regimens, requiring treatment modifications or prolongation.

While all these factors help explain part of the variation in healthcare use and costs, certain methodological constraints should also be acknowledged. This study has several strengths, including its large, nationwide, population-based design and the use of matched comparators, enhancing internal validity. We comprehensively assessed direct and indirect costs over multiple years, adjusting for comorbidity, including lung disease. Misclassification due to coding errors, including the potential conflation of tuberculosis (which has its own specific codes) and NTM-PD, may have occurred. However, in Denmark these codes are typically used by physicians working with both tuberculosis and NTM. Limitations also include the potential for residual confounding (e.g. smoking, nutritional status, severity of underlying respiratory disease and functional status) despite matching and adjustment as these factors are not fully captured by CCI or register data. Other possible unmeasured factors include variations in healthcare-seeking behaviour, frequency of exacerbations or respiratory co-infections and physician practice patterns, all of which may contribute to differences in healthcare utilisation and costs. We also lacked detailed data on specific diagnostic and therapeutic components (e.g. microbiological sampling, species, high-resolution computed tomography scans and antimicrobial regimens) and on subgroups such as disease phenotype (fibrocavitary versus nodular bronchiectatic disease) and immunosuppression. Informal care costs were not included, which may underestimate the broader societal impact. Finally, while the findings may be generalisable to similar welfare-based healthcare systems, the magnitude of costs and healthcare utilisation patterns may differ in settings with different healthcare structures, financial models or labour market conditions.

Our findings underline the complex and ongoing resource needs of NTM-PD patients, driven by inpatient and outpatient care, diagnostic testing, medication, adverse events, and poor outcomes, as well as the importance of early diagnosis, guideline-based management, and interventions aimed at reducing both healthcare costs and broader socioeconomic burdens. Future research should investigate long-term economic outcomes and evaluate strategies to mitigate costs while enhancing care for this vulnerable patient population, including assessing the impact of disease severity, NTM species and different treatment regimens (e.g. two- versus three-drug regimens for MAC-PD).

In conclusion, this nationwide Danish study demonstrates that NTM-PD imposes a substantial and sustained economic burden, with healthcare costs nearly 5 times higher than those of matched comparators in the year of diagnosis, and elevated costs observed both before and after diagnosis. In addition to direct healthcare costs, NTM-PD was associated with reduced employment income and increased public benefit payments, indicating significant socioeconomic impacts.

Data availability

The data cannot be shared under Danish law.

Supplementary material

Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.

Supplementary material

00870-2025.SUPPLEMENT