The cost of medical management of pulmonary nontuberculous mycobacterial disease in Ontario, Canada

RESULTS

A total of 172 patients, seen at the UHN respiratory clinic for pulmonary NTM infection in 2003–2008, were reviewed. 91 patients were followed for ≥1 month and treated with prolonged antimicrobial therapy and, therefore, included in subsequent analysis. Baseline characteristics of the 91 patients are presented in table 1. The majority of patients were thin, elderly females, consistent with previously reported results 1. Physician-defined pre-existing chronic obstructive pulmonary disease (COPD) was present in ∼40% of cases; however, obstructive airway disease, as defined by pulmonary function testing, was present in a much higher proportion. This discrepancy could be due to the direct effect of NTM infection on the airways, the development of bronchiectasis or a pre-existing but undetected abnormality of the airways that predisposes patients to pulmonary NTM infection.

All patients met ATS criteria for pulmonary NTM disease. The patients’ clinical, radiological and microbiological features are illustrated in table 1. Radiologic nodular bronchiectasis was two- to three-fold more common than cavitation. The majority of patients were infected with NTM of the Mycobacterium avium complex (MAC) (70%), followed by Mycobacterium xenopi (17%) and others (11%). The median duration of treatment was 14 months (interquartile range (IQR) 9–23 months). Median treatment duration was not significantly different between MAC and M. xenopi (15.5 versus 12 months; p = 0.06). The small number of patients infected with other species of NTM limited further comparisons. Cavitation on CT was not significantly associated with the duration of treatment in bivariate analysis (14 months with cavitation versus 12 months without cavitation; p = 0.66).

The majority of patients were treated with one out of six different oral antibiotic regimens, the choice of which was largely determined by patient tolerability and medication toxicities. i.v. therapy usually comprised amikacin, which was used in 23 cases. The individual costs of antibiotics and nonmedication treatment component costs are presented in the online supplement. Daily drug costs, for average doses of the commonly used medications, were as follows. Macrolides: CAD 2.77–4.94 (varies by agent); ethambutol: CAD 0.62; rifampin: CAD 2.48; fluoroquinolones: CAD 2.82–6.21 (varies by agent). Two patients with Mycobacterium abscessus received 3 months of carbepenem therapy (approximate wholesale cost of CAD 50 per dose) in conjunction with i.v. amikacin. This cost was not included in the analysis. No other i.v. antibiotics were used. The frequency of commonly prescribed antibiotic regimens is shown in figure 1a. The median monthly cost of common antibiotic regimens is shown in figure 1b. Clofazimine was used in 18 patients; however, it was not included in cost calculations, as it is made available free of charge and cost information was not made available to us. There was no significant difference in cost between the different oral antibiotic regimens. A significant increase in cost occurred with i.v. amikacin.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1–

a) Frequency of commonly prescribed antibiotic regimens. 10 cases were treated with other antibiotic regimens. b) Monthly cost of common antibiotic regimens (drug acquisition and administration costs only). M: macrolide; FQ: fluoroquinone; R: rifampin; E: ethambutol; C: clofazimine.

The median monthly and total medication and nonmedication costs are shown in table 2. Greater than two-thirds of the cost was incurred from medications. Table 3 highlights cost differences between patients treated with exclusively oral therapy and patients treated with parenteral plus oral therapy. The difference in cost of nearly CAD 900 per month was driven almost exclusively by the increase in drug acquisition and administration costs. In multivariable analysis, total monthly cost was significantly associated only with infection with M. abscessus, use of i.v. antibiotics and infection with M. xenopi. Multivariable model results and interpretation are summarised in table 4.

DISCUSSION

In our study of the treatment cost of pulmonary NTM infections, we observed an average monthly cost of approximately CAD 500. Drug costs were responsible for ∼70% of the total treatment cost. As expected, treatment costs rose dramatically with the use of i.v. antibiotics and in the presence of M. abscessus, two variables that were often, but not always, associated (i.v. therapy was invariably used for M. abscessus but also for numerous patients with other NTM species). In multivariable modelling, parenteral therapy added ∼CAD 700 to the monthly treatment cost, independent of other variables. Additional results from multivariable modelling included finding that M. xenopi was associated with greater treatment costs than MAC, but there was no clear cost association with cavitation on CT scan, age or sex.

A recently published retrospective cohort study 4 examining the medication costs associated with the treatment of 25 cases of pulmonary NTM infection in the USA found monthly and annual medication costs 44% and 15% higher, respectively, compared with our study. There are several factors that probably contributed to the large cost difference. The prior study had a higher prevalence of M. abscessus (22%, compared with 7% in the present study), an organism that is associated with a high treatment cost. In addition, the median number of antimicrobials employed in the prior study was five, compared with only three in our study, and we did not include the treatment cost of clofazimine in 18 patients. Finally, the drug costs were much higher in that study. The monthly cost of a standard first-line regimen comprising a macrolide, rifampin and ethambutol was ∼USD 470 in that study, compared with CAD 245 in our study. The present study also differed from the prior study in that we included all readily quantifiable treatment costs, rather than focusing exclusively on drug acquisition costs. We think that our study provides a reasonable estimate of the total costs of treating pulmonary NTM infection in Ontario.

Particularly relevant to the overall financial burden of treating pulmonary NTM disease in a population, is choosing the appropriate patients to treat. In our NTM database, only 91 of the 172 patients were actively treated with antibiotic therapy. It is well recognised that making the distinction between pulmonary “colonisation” and “disease” is not always easy, and has been addressed through the creation of explicit diagnostic guidelines. Despite these guidelines, the decision to treat patients with pulmonary NTM disease remains difficult, even for experts in the field 1. Therefore, we feel that obtaining the expertise of specialised physicians during the course of diagnosis and/or treatment would better identify those who required treatment, and could have a substantial impact on the overall disease cost.

In considering whether the financial cost of a medical treatment is acceptable, it can be helpful to compare it with other accepted therapies. Pulmonary NTM infections are probably less expensive to treat than other treatment-resistant infections, such as multidrug-resistant tuberculosis (MDR-TB), or chronic infections, such as HIV. The median outpatient costs of treating MDR-TB in seven HIV-seronegative patients in San Francisco, CA, USA was USD 21,929, over a mean duration of therapy of 98 weeks 6. US AIDS surveillance data in 2002–2003 estimated the annual average cost of antiretroviral therapy was USD 12,665 per patient 7.

We feel the cost of treating pulmonary NTM disease is closer in cost to the outpatient treatment of more common chronic diseases, such as diabetes mellitus and congestive heart failure. In a large cross-sectional study published by the American Diabetes Association in 2007, the annual per capita expenditure for outpatient care (not including emergency room visits) for individuals with diabetes who were >65 yrs of age was USD 3,319 8. In 2004, the average annual cost of outpatient treatment of >1,500 patients with congestive heart failure was USD 3,837 9. We estimated the drug cost of treating COPD, type 2 diabetes mellitus and symptomatic coronary artery disease using common drug combinations and doses. In Ontario, the approximate monthly drug costs for treating COPD (using fluticasone/salmeterol, tiotropium and salbutamol) is CAD 323, symptomatic coronary artery disease (using a β-blocker, acetylsalicylic acid, statin and an angiotensin-converting enzyme (ACE) inhibitor) is CAD 196, and type 2 diabetes mellitus (using a biguanide, sulfonylurea, thiazolidinedione, statin, ACE inhibitor and acetylsalicylic acid) is CAD 319. In a similar cost range, we observed that a standard first-line regimen for treating pulmonary NTM infection (using a macrolide, rifampin and ethambutol) was CAD 245 per month. Although data regarding objective benefits in treating pulmonary NTM disease are not very well established, the cost of treatment does not appear to be inconsistent with accepted costs of ongoing treatment of common chronic diseases.

We speculate that the use of “guideline therapy”, namely a macrolide, rifampin and ethambutol, as first-line treatment is the most cost-effective approach to the treatment of NTM disease in Canada. Assuming an 18-month course of treatment, the cost savings (based on our observed costs) of guideline therapy per treatment course versus other regimens, such as floroquinolone/rifampin/ethambutol, macrolide/fluroquinolone/ethambutol or fluroquinolone/rifampin/macrolide, is CAD 1,026, CAD 198 and CAD 1,566, respectively. Obviously the effectiveness of the regimen must be considered when considering cost saving approaches. However, at present, there is a paucity of studies that compare the effectiveness of different regimens. Jenkins et al. 10 compared the efficacy of a 2-yr treatment course of clarithromycin, rifampin and ethambutol versus ciprofloxacin, rifampin and ethambutol. There was no statistically significant difference between the two regimens in their primary end-points of death due to NTM and treatment failure (defined as positive sputum cultures on two separate occasions during the previous 3 months of treatment). As a secondary end-point, Jenkins et al. 10 compared the tolerability of these regimens and again found no statistically significant difference in the two regimens. Accepting the results from this complex study at face value, we would conclude that the combination of a macrolide, rifampin and ethambutol is the most cost-effective regimen due to its lower price, effectiveness and tolerability profile. It is even more difficult to address the cost-effectiveness of amikacin (or another aminoglycoside), the major factor driving large increases in cost of therapy. Guidelines recommend considering the use of injectable aminoglycosides in the initial therapy of cavitary MAC disease and advise their use in severe, advanced or recurrent disease. However, controlled data regarding the effectiveness of aminoglycosides in pulmonary NTM are limited. Kobashi et al. 11 randomised patients with pulmonary MAC to clarithromycin/rifampin/ethambutol plus either streptomycin or placebo injections. Patients in the streptomycin group did slightly better, although the differences were not statistically significant. Streptomycin appeared to be particularly beneficial (for clinical and microbiological outcomes) among patients with radiographically extensive disease. It appears that an injected aminoglycoside is beneficial in extensive pulmonary MAC, but there are inadequate data to assess its cost-effectiveness.

It is interesting to note that a minority of patients were receiving the first-line recommended therapy for MAC (macrolide/rifampin/ethambutol). This observation was present despite our general practice to introduce a three-drug first-line regimen when possible. Even when including regimens that additionally contained amikacin or a fluoroquinolone, the total number of patients with these regimens comprised only 44%. There are probably several reasons for this observation. Although we did not formally study regimen tolerability, we think that a siginificant proportion of patients had been intolerant of rifampin, explaining the substantial proportion of patients using nonrifampin-containing regimens. Furthermore, a very small proportion could not tolerate ethambutol, usually due to ocular toxicity. Also, patients with M. abscessus were not generally prescribed rifampin, due to the resistance of the isolates, and often received carbapenem therapy. Finally, many of our patients had recurrent or difficult-to-treat disease that led to the addition of more drugs, including fluoroquinolones, amikacin and clofazimine. Our regimen choice is, therefore, reflective of the spectrum of NTM species treated in our clinic and possibly drug intolerance that is common in treating pulmonary NTM 2.

The frequency of pulmonary NTM has been increasing substantially in Ontario 3, suggesting that the cost of treatment will have a mounting impact on the health resources, underscoring the relevance of studies like the present investigation. We think that our results are generally applicable in Canada and, with some modification for differences in healthcare costs, in other jurisdictions as well. Our study is the largest and most comprehensive investigation of the cost of treating pulmonary NTM disease to date, including drug acquisition and administration costs, as well as physician, facility and testing fees (audiology, biochemistry, haematology, microbiology and radiology). Our patients are probably representative of other populations of pulmonary NTM patients, as we ensured that all cases met the ATS diagnostic criteria for NTM disease. The drug regimens were also generally in close accordance with ATS/IDSA guidelines, although fluoroquinolones, drugs not recommended in first-line regimens, were used extensively. The fraction of patients in our clinic who were treated (53%) is probably in accordance with the observation that pulmonary NTM disease, even in the presence of significant symptoms, radiographic abnormalities and microbiologic evidence of disease, may be relatively indolent and progress only very slowly. Patients were offered antimycobacterial drug treatment if the clinical opinion was such that the benefits of therapy were likely to outweigh the toxicities. Generally, patients were not treated if they had relatively mild and nonprogressive symptoms, and nonprogressive lesions on serial chest imaging.

Although our study offers the first comprehensive analysis of the cost of treating pulmonary NTM, there are several limitations. We did not have the cost of clofazimine available to us, so this cost was omitted from our analysis. The impact of these missing data is probably small, however, since <20% of our patients were treated with this drug. We also did not include the cost of treatment periods with inhaled amikacin, since this therapy is less commonly used and we had few patients who received it. The latter would likely make our costs estimates thereof unreliable. Because only three patients received inhaled amikacin during the study period, it is unlikely that omitting the periods of inhaled amikacin greatly affected our results. Although we did not determine the costs surrounding the use of inhaled amikacin, we think its use would be less than for i.v. (assuming similar doses), despite the costs of an air compressor, saline and nebulisers, since inhaled therapy eliminates nursing costs and, in our practice, reduces biochemical and audiographic monitoring for toxicity. Also, if amikacin is used exclusively by inhalation, costs of central line insertion are avoided. We did not include the cost of the adverse effects of treatment. The multidrug regimen and prolonged course of therapy of pulmonary NTM is well known for high rates of adverse reactions 2. Ballarino et al. 4 recorded the frequency of adverse drug reactions ranged from 18% with azithromycin to 75% with levofloxacin. Adverse drug reactions may lead to more clinic visits, additional diagnostic tests and the use of more expensive, second-line agents, but this cost would have been captured using our methodology. In our cohort, most toxicities were managed by telephone, the costs of which are not included, since this is not a reimbursable service in our health system. One of our patients developed rifampin-related systemic illness, diffuse petichiae and ecchymoses with severe thrombocytopenia, for which hospitalisation was required. The in-patient costs of this toxicity were not available to us and not included in our analysis Although the in-patient cost of pulmonary NTM disease was also not explored, hospitalisation for initiation or modification of therapy did not occur in patients, outside of the unusual situation of pulmonary resection for localised disease, an intervention that is used in the minority of patients. The majority of patients probably do not require hospitalisation; however, hospitalisation in even a small proportion of patients could raise the average costs substantially. Notably, in our population, eight patients underwent surgery (surgical biopsy, lobectomy or pneumonectomy) as part of their assessment or treatment. In our total treatment cost measurements and our cost estimates for a single, uncomplicated 18-month treatment course (table 5), we did not account for the high rates of treatment failure and disease recurrence that are known to occur 1 and undoubtedly contribute heavily to the overall economic burden of pulmonary NTM. However, we did find that monthly costs of oral-only therapy are not inconsistent with the commonly accepted costs of treating chronic diseases, where therapy is continued indefinitely. We suggest that it may be just as important to focus on monthly costs, because treatment is often recurrent and sometimes chronic. Finally, the modelling of treatment cost may have been limited by the absence of information regarding the extent of radiographic disease (we had only data regarding cavitation) and macrolide resistance (most of our patients’ isolates were not tested for drug susceptibility).

The generalisability of our work to other healthcare systems may be questioned. We have made a limited comparison to the USA, with regard to drug costs only. The applicability of comparisons between the USA and Canada is not clear, given the differences in medical cost-payers (society as a whole versus individuals). Regardless, we have expanded the comparison between the costs of treatment in Ontario, with those of treatment in the USA in table 5, wherein the cost of a projected uncomplicated 18-month treatment course for pulmonary MAC is presented. The modelled cost in the USA was 1.7–2.3-fold greater than the cost in Ontario, a difference driven largely by medication costs, which comprise the bulk of treatment costs and were expected to be 1.9–2.4-fold greater in the USA. The difference between the USA and Ontario may be even greater than we modelled, since we used Medicare reimbursement rates for nondrug costs, which are lower than reimbursed rates from private insurers: a study in 1993 estimated that Medicare reimbursement was 76% that of private insurance carriers 13. Even though our model is probably a gross simplification, it appears to be clear that the cost of treating pulmonary NTM disease in the USA would be at least double the cost in Ontario.

The applicability of our work to systems where medical costs are generally borne by society is much greater. Our healthcare system in Ontario provides most prescription drugs free of charge to patients who are ≥65 yrs of age or who are receiving social assistance. Furthermore, the provincial drug plan includes automatic substitution of generic drugs and our calculations utilised costs of generic drugs when available. 38 of our patients were ≥65 yrs of age and several additional patients were receiving provincial drug benefits as part of their social income assistance. Finally, the costs of physicians’ visits, hospital facility fees, home nursing visits, ambulatory parenteral drug therapy and diagnostic tests requested by a physician are borne completely by the Ontario Ministry of Health and Long Term Care. As a result, nearly half of our patients’ treatment comprised exclusively societal costs, rather than costs borne by an individual or a private health insurance programme. In this context, we think that our costs reflect societal costs of the ambulatory medical management of pulmonary NTM disease very well. Based on the differing recommendations between the ATS 1 and the British Thoracic Society (BTS) 12 NTM guidelines regarding choice of specific drugs, we have presented in table 5 projected costs of an 18-month course of therapy. It is evident that the first-line regimen from the BTS guidelines (rifampin and ethambutol) is by far the least expensive. However, because the choice of regimen usually depends more upon tolerability and goals of therapy (cure versus suppression) and because we cannot adequately assign relative toxicity or efficacy of these regimens (greater toxicity or less effective regimens may lead to additional costs), the values in table 5 should not be used to make clinical decisions.

The rising prevalence of pulmonary NTM infections will have an increasingly large impact on population health and heath expenditures. We observed that the cost of treating this disease is substantial, but not inconsistent with the costs of well-established therapies for other chronic diseases. Furthermore, the cost varied greatly with the use of parenteral therapy and with the presence of M. abscessus infection. In light of our findings, we do not think that the cost of treating pulmonary NTM disease should discourage therapy. The identification of new, less expensive alternative therapies may be most helpful for M. abscessus infection and when intravenous antibiotics are deemed necessary.