Burden of non-tuberculous mycobacterial pulmonary disease in Germany

Discussion

The objective of this study was to estimate the burden of NTM-PD in Germany in incident patients. This was accomplished by a retrospective observational design comparing medical and productivity-related spending for 125 patients with PNTM to a 1:10 matched comparison group of patients without NTM-PD over a 39-month period. The economic burden attributed to NTM-PD is expected to be great, because it is a chronic disease that can require frequent medical visits, long-term treatment with multidrug regimens and hospitalisations to minimise the risk of further progression.

To date, only a few North American studies have been published presenting detailed cost data for treatment of NTM-PD in a prevalence-based approach: in the US, the total cost for inpatients varied between USD 19 876 [20] (median cost) and USD 37 579 (mean cost) [21] for inpatients, whilst Leber et al. [22] reported a median of only CAD 4916 per inpatient in Toronto, Canada.

Mirsaiedi et al. [23] calculated an aggregated cost of USD 903 767 292 for 20 048 hospital discharges of NTM-PD patients from 2001 to 2012 in the USA, whereas Strollo et al. [24] estimated a mean annual cost of USD 815 million for 86 244 national NTM-PD cases in 2010 alone.

To our knowledge, our study represents the first investigation of healthcare costs brought about by incident NTM-PD and of respective antibiotic prescription patterns in Europe.

Whilst previous studies covered only prevalent costs, our work covers the accumulation of costs that follow diagnosis. These were determined as incremental costs for NTM-PD in excess of those incurred for control patients. Furthermore, it includes, for the first time, both mortality and costs due to lost patient productivity.

In line with the previous studies cited above, we found approximately fourfold higher direct costs in German NTM-PD patients compared with a control group without NTM-PD, and 63% of incremental costs were driven by inpatient stays alone. Despite in-depth analysis, it remains unclear whether the continuously decreasing number of hospital days over the 39 months, from a total 1918 days in the index quarter to 835 days in the third year of observation, is primarily the desired result of successful treatment or, rather, the result of decreasing claims for healthcare services by fewer NTM-PD patients who were still alive.

In fact, mortality in our NTM-PD patients was nearly four times higher (3.73%, 95% CI 2.52–5.53) than in the matched patient group without the disease, impressively demonstrating the severity of the disease, and was even higher for NTM-PD patients with associated COPD (6.92%, 95% CI 4.82–9.95). Because mortality was highest at 3.2% within the first 3 months after diagnosis, immediate and appropriate medical treatment should be offered.

However, 25.6% of NTM-PD patients (32 out of 125) did not receive any treatment. In those who were treated, outpatient treatment started with a median delay of 32 days between the date of coding discharge diagnosis to the receipt of an outpatient prescription for drug therapy. A high scatter represented by an interquartile range of 250 days in which 50% of all prescriptions were filled shows that many patients began or continued medical treatment even later.

International guidelines (e.g. those described in references [6] and [7]) recommend long-term multidrug treatment until patients achieve 12 consecutive months of negative sputum cultures,. The most recommended treatments mostly comprise a macrolide, rifampicin, and ethambutol, modified according to the phenotypic resistance patterns of the causative strains, if necessary.

Surprisingly, 29 different combinations of drugs were prescribed at least once in the 39 months, and the short number of subsequent intervals in which 23 of the 29 different combinations were prescribed seems to reveal some uncertainty about which basic regimens should be chosen. The most frequently used combination of clarithromycin plus ethambutol alone (used in 26% of the patients) is not sufficient for treatment of NTM-PD, irrespective of the underlying species, and consequently is not recommended in any guideline. Furthermore, a considerable number of patients received an inadequate type of macrolide regimen that could have increased the probability of macrolide resistance during the course of treatment [4] (i.e. clarithromycin monotherapy, in 11 out of 93 patients (11.8%), or macrolide plus rifampicin only and/or a macrolide plus a fluoroquinolone) (table 6). No injectable drugs were prescribed in addition to any orally administered drug regimen in outpatients, although this may be indicated in refractory courses of M. avium Complex-PD and difficult-to-treat species such as M. abscessus. Finally, the mortality rate did not differ between those 32 patients who were not prescribed an antibiotic (7 out of 32, 21.9%) and those 93 patients prescribed at least one antibiotic (21 out of 93, 22.6%). Although the sicker patients may have been in the group that started treatment, this observation suggests that medication in the NTM-PD group may have partly been ineffective and thus may have contributed to the poor prognosis within the follow-up period.

Reduced productivity may also have a substantial economic impact on NTM-PD patients: although indirect costs cannot be captured directly from administrative claims, we calculated cost due to absenteeism from work productivity loss by Monte Carlo simulation, which amounted to €1221.05 per NTM-PD patient per year. As this cost is equal to 13.4% of the direct total costs, indirect costs cannot be considered negligible. Because short-term disability with sick leave of <3 days is not captured by the SHI, these costs are probably underestimated in our calculations. Together, annual direct costs of care (€9093.20) plus those indirect costs total €10 314.25 per year from a societal perspective. This figure of societal cost for a NTM-PD patient clearly exceeds the annual per patient cost for many other chronic respiratory diseases, such as COPD (€1013) or asthma (€1950) [25]. When extrapolating the incidence rate of 2.6 per 100 000 (95% CI 2.2–3.1) that we found in our study, on average a total of 2137 (range 1808–2548) newly diagnosed NTM-PD patients per year can be expected in the 2015 German population of 82.2 million [26], and the direct and indirect expenditures attributable to the disease amount to more than €22.04 million (range €18.65–€26.28 million).

Our analysis was limited by several factors. First, we only followed NTM-PD patients for 3 years after the index quarter in which diagnosis was assured and thus did not capture the long-term burden of the disease. Accordingly, future studies of the long-term costs of NTM-PD are warranted. Second, other costly comorbid conditions that were not accounted for via the CCI in the matching processes might have influenced the cost burden estimates. Whilst the CCI also addresses COPD and bronchiectasis comorbid conditions, cystic fibrosis is not included. However, because only one patient with both NTM-PD and cystic fibrosis was found in our database in 2010/2011, an overestimation of cost burden due to cystic fibrosis alone appears to be unlikely. Third, although the pool of insured persons from which our sample of NTM-PD patients and their controls were taken was large at more than 7 million, our patients were not formally selected as part of a representative sample. Thus, it is not certain whether our results are generalisable to the entire German population. Nevertheless, the database used has been shown to be representative of the German population in terms of measures of morbidity, mortality and drug usage [27]. Fourth, classification of NTM-PD patients is based on correct diagnostic coding and may thus lead to potential selection bias with unknown NTM-PD cases in the control group that could have affected the cost estimates in favour of the control patients. Accordingly, our approach of billing data may have resulted in an underestimate of the true costs associated with NTM-PD. Therefore, to address concerns of possible underestimation in our outcomes, a validation cohort study assessing the sensitivity and specificity of the ICD-10 code “A31.0” by investigating a random selection of a large number of inpatient medical records should be undertaken in the near future.

Nevertheless, our results not only demonstrate that per patient spending on NTM-PD is high but also suggest that treatment is not based on guidelines in a considerable proportion of cases; thus, appropriate and early guideline-based treatment of NTM-PD may both lower mortality and reduce costs. Improving and expanding the information available about guideline-based treatment options may lead to more effective treatment choices and help to avoid rapid progression of the disease.