Pharmacoeconomics 2012; 30 (9): 763-777 1170-7690/12/0009-0763/$49.95/0

ORIGINAL RESEARCH ARTICLE

Adis ª 2012 Springer International Publishing AG. All rights reserved.

Cost of Illness of Cystic Fibrosis in Germany Results from a Large Cystic Fibrosis Centre Mareike Heimeshoff,1 Helge Hollmeyer,2 Jonas Schreyo¨gg,1,3 Oliver Tiemann1,3 and Doris Staab2 1 Institute for Health Care Management and Health Economics, Faculty of Economics and Business Administration, University of Hamburg, Hamburg, Germany 2 Department of Paediatric Pneumology and Immunology, Charite´ University Medicine Berlin, Berlin, Germany 3 Institute of Health Economics and Health Care Management, Helmholtz Zentrum Mu¨nchen, German Research Center for Environmental Health, Munich, Germany

Abstract

Background: Cystic fibrosis (CF) is the most common life-shortening genetic disorder among Whites worldwide. Because many of these patients experience chronic endobronchial colonization and have to take antibiotics and be treated as inpatients, societal costs of CF may be high. As the disease severity varies considerably among patients, costs may differ between patients. Objectives: Our objectives were to calculate the average total costs of CF per patient and per year from a societal perspective; to include all direct medical and non-medical costs as well as indirect costs; to identify the main cost drivers; to investigate whether patients with CF can be grouped into homogenous cost groups; and to determine the influence of specific factors on different cost categories. Methods: Resource utilization data were collected for 87 patients admitted to an inpatient unit at a CF treatment centre during the first 6 months of 2004 and 125 patients who visited the centre’s CF outpatient unit during the entire year. Fifty-four patients were admitted to the hospital and also visited the outpatient unit. Since all patients were exclusively treated at the centre, data could be aggregated. Costs that varied greatly between patients were measured per patient. The remaining costs were summarized as overhead costs and allocated on the basis of days of treatment or contacts per patient. Costs of the outpatient and inpatient units and costs for drugs patients received at the outpatient pharmacy were summarized as direct medical costs. Direct non-medical costs (i.e. travel expenses), as well as indirect costs (i. e. absence from work, productivity losses), were also included in the analysis. Main cost drivers were detected by the analysis of different cost categories. Patients were classified according to a diagnosis-related severity model, and median comparison tests (Wilcoxon-Mann-Whitney tests) were performed to investigate differences between the severity groups. Generalized least squares (GLS)

Heimeshoff et al.

764

regressions were used to identify variables influencing different cost categories. A sensitivity analysis using Monte Carlo simulation was performed. Results: The mean total cost per patient per year was h41 468 (year 2004 values). Direct medical costs accounted for more than 90% of total costs and averaged h38 869 (h3876 to h88 096), whereas direct non-medical costs were minimal. Indirect costs amounted to h2491 (6% of total costs). Costs for drugs patients received at the outpatient pharmacy were the main cost driver. Costs rose with the degree of severity. Patients with moderate and severe disease had significantly higher direct costs than the relatively milder group. Regression analysis revealed that direct costs were mainly affected by the diagnosis-related severity level and the expiratory volume; the coefficient indicating the relationship between costs for mild CF patients and other patients rose with the degree of severity. A similar result was obtained for drug costs per patient as the dependent variable. Monte Carlo simulation suggests that there is a 90% probability that annual costs will be lower than h37 300. Conclusions: The share of indirect costs as a percentage of total costs for CF was rather low in this study. However, the relevance of indirect costs is likely to increase in the future as the life expectancy of CF patients increases, which is likely to lead to a rising work disability rate and thus increase indirect costs. Moreover we found that infection with Pseudomonas aeruginosa increases costs substantially. Thus, a decrease of the prevalence of P. aeruginosa would lead to substantial savings for society.

Key points for decision makers  The cost of cystic fibrosis rises with the degree of severity  Costs for drugs patients receive at outpatient pharmacies are the main cost driver  A decrease of the prevalence of infection rates with Pseudomonas aeruginosa would lead to substantial savings for society

Introduction Cystic fibrosis (CF) is the most common lifeshortening genetic disorder among Whites worldwide. The number of newborns with CF among all births varies between 1 in 2500 and 1 in 3500 depending on the ancestry of the population and on the intensity of screening performed.[1-4] In recent decades, life expectancy of patients with CF has improved considerably.[5] Due to the introduction of a number of therapeutic measures, including inhaled antibiotics and inhaled enzyme Adis ª 2012 Springer International Publishing AG. All rights reserved.

therapy, CF is no longer simply a paediatric disease.[6] Today, more than half of the patients are aged 18 years and older[5] and the mean life expectancy has increased from approximately 20 years in 1980 to nearly 40 years today.[7] CF is an autosomal recessive disease in which the modified protein product of a defective gene causes abnormal chloride ion transport, resulting in increased viscosity of mucus. The unusually thick, sticky mucus clogs the lungs and leads to lung infections, disturbances of the digestive tract, and high sodium concentration in sweat as well as

Pharmacoeconomics 2012; 30 (9)

Cost of Illness of Cystic Fibrosis

obstructions of the pancreas and inhibition of natural enzymes that would help the body overcome pancreatic insufficiency.[7] Therapy for CF is still restricted to treating symptoms, and the patient’s health worsens over time. Pulmonary disease is the primary cause of morbidity and mortality among patients with CF, and more than 70% of all patients aged 18 years and older test positive for Pseudomonas aeruginosa.[8] Many of these patients experience chronic endobronchial colonization and have to take antibiotics and be treated as inpatients. Therefore, societal costs of CF, including direct medical costs as well as direct non-medical and indirect costs (resulting from work disability of CF patients), may be high. As the disease severity varies considerably among patients, costs may be quite unequal between patients. Furthermore, previous studies have identified demographic and especially clinical variables, such as lung function and level of disease severity, as influencing total costs per patient.[9-12] Depending on the perspective, an economic evaluation should also include indirect costs. There are a number of arguments from an economic point of view to follow the societal perspective, which Jo¨nsson[13] summarized. For instance, it is often argued that, particularly in healthcare systems with highly fragmented healthcare financing, e.g. Germany and other social health insurance countries, the societal perspective is very important, because in these countries the statutory health insurance only covers one part of the total healthcare costs while other payers and private households cover the remaining part. Moreover, in a review on pharmacoeconomic guidelines required by official evaluation agencies in different countries, Zentner et al.[14] found that the majority of these (e.g. Canada, the Netherlands, Norway) prefer and require the societal perspective. There are only a few exceptions, e.g. Britain, that require the payer perspective. In this article, we took the societal perspective including indirect costs following the German recommendations on health economic evaluation.[15] The societal perspective requires the inclusion of indirect costs. So far, there is only one cost-ofillness study on CF that includes some indirect Adis ª 2012 Springer International Publishing AG. All rights reserved.

765

costs. DeWitt et al.[16] calculate cost of illness of patients with CF with mild impairment in lung function and include costs attributable to absence from school or work. However, costs resulting from work disability of patients with CF were not included in their analysis. Moreover, most studies calculating the costs of CF are based on a gross-costing approach. Only Schreyo¨gg et al.,[10] who calculated hospital costs of CF, Eidt-Koch et al.,[17,18] who focused on costs for outpatient treatment, and DeWitt et al.[16] used a micro-costing approach. There are pros and cons for using a micro-costing versus a gross-costing approach. Micro-costing has the advantage of considering certain cost components in greater detail. For instance, costs are often underestimated by gross costing for conditions requiring expensive drugs, e.g. CF. When comparing gross costing with micro-costing, Clement et al.[19] found that micro-costing leads to higher cost estimates than the gross-costing approach, which has to be considered for interpretation of results. The same authors also found that microcosting leads to greater variance than gross costing. An important advantage of gross costing is that it is less expensive than micro-costing, but it is less sensitive. Thus, micro-costing is particularly recommended in contexts requiring a high sensitivity of cost estimates.[20] We believe that in the context of a cost-of-illness study on CF a high sensitivity of cost estimates is needed to be able to allocate drug costs with high precision to patients with different severity levels. The objectives of this study were to calculate the average total costs of CF per patient and per year from a societal perspective; to include all direct medical and non-medical costs as well as indirect costs; to identify the main cost drivers; to investigate whether patients with CF can be grouped into homogenous cost groups according to defined severity levels; and to determine the influence of certain factors on different cost categories. Methods The study was conducted at the Department of Paediatric Pneumology and Immunology at the Charite´ Medical School, Berlin, Germany. Pharmacoeconomics 2012; 30 (9)

Heimeshoff et al.

766

The department maintains one of the largest centres for treatment of CF in Europe. Data for inpatient treatment were collected for 87 patients. All patients admitted to the unit over a period of 6 months from January to July 2004 were included. A patient was defined as any person with CF who was admitted to the unit during the 6-month period. Since all patients were discharged from hospital within the study period, the full costs for each patient were captured. Costs for the second 6-month period (July–December 2004) were assumed to be the same as for the first 6 months. Thus, costs for the first 6-month period were doubled. At the same centre, data for outpatient treatment of 125 patients with CF were collected. All patients who visited the outpatient unit during the year 2004 were included. For both groups of patients (i.e. inpatient and outpatient), the collected data included information on utilization and prices of drugs that were prescribed by the centre and that patients received from pharmacies outside the centre. Fifty-four patients visited the outpatient unit and were also admitted to the hospital during the period observed. Since the outpatient and inpatient units belong to the same hospital, costs for all patients could be summarized. Furthermore, it was verified that all CF patients were exclusively treated by this specialized centre. Thus, resource consumption for inpatient and outpatient treatment of 158 patients was included in the analysis. Resource Unit Costs

For both inpatient and outpatient treatments, costs were measured on the basis of the individual patient care data. In order to calculate total costs of CF per patient per year, we performed a bottomup collection of resource use for particular cost components. Hence, our method can be classified as a micro-costing bottom-up approach. Indirect and direct non-medical costs were also included in the analysis. Our calculations reflect the societal perspective. Costs that were identified in previous studies[11,21,22] as cost categories that vary greatly in terms of individual resource utilization were measured separately for each patient, while the reAdis ª 2012 Springer International Publishing AG. All rights reserved.

maining costs were summarized as overhead costs. Those cost categories with high impact on cost variation were staff costs for patient care, drug costs and laboratory costs. For inpatient and outpatient treatment, staff costs for patient care were calculated by measuring the time devoted to each patient for all professions employed in the unit (clinicians, nursing staff, physiotherapists and dieticians). For this reason, each room was equipped with a stopwatch, which staff members pressed when entering and leaving the room. Staff members were then required to document the exact time spent in contact with each patient. Costs were calculated by multiplying the measured time by the hourly wages of the respective professions. Moreover, all kinds of further costs from the employer perspective (i.e. mandatory contributions, social security contributions and pension contributions as well as sick and annual leave) were captured in our staff costs. We also took all other staff costs for other activities (e.g. staff meetings), which do not greatly vary in individual resource utilization, into account. The latter were estimated and added to the overhead costs for each patient. In order to exactly measure the drug costs for inpatients, all drugs consumed were documented in medical record files and accuracy was checked with selected members of the medical staff. Purchasing prices supplied by the hospital pharmacy were used as the drug prices. These were often lower than the official listed prices in Germany because every hospital is able to negotiate bulk drug discounts.[10] When drugs were prescribed in the outpatient unit, individual drug consumption, including name and quantity of the drug prescribed, was documented. Hospitals in Germany are not allowed to dispense drugs for outpatient treatment and patients must purchase their drugs from outpatient pharmacies instead. Purchasing prices were taken from the list with official selling prices that applied to all pharmacies in Germany.[23] To calculate laboratory costs for inpatient treatment, purchasing costs for substances, devices and other materials, and hourly wages for each profession at the study hospital were obtained. If services from external laboratories were Pharmacoeconomics 2012; 30 (9)

Cost of Illness of Cystic Fibrosis

utilized, the prices paid were recorded as costs. Laboratory costs for outpatient treatment were calculated identically because services were provided by the same laboratory. All other cost categories were classified as costs with low variation in individual resource utilization: staff costs for other activities and certain technical equipment and imputed costs (e.g. rent for the building, which was determined at market rate) were summarized as overhead costs. Furthermore, overhead costs for the hospital not exclusively attributable to the CF unit (e.g. maintenance, sterilization) were proportionally included. All of these costs were obtained from the accounting department. As we expected no major variation in individual resource utilization for the cost categories mentioned above, they were allocated on hospital days per patient for the inpatient treatment. According to the accounting department, overhead costs for the outpatient treatment were one-third of those for inpatient treatment. These costs were allocated based on outpatient contacts per patient. Moreover, costs for radiological services provided by other units were also distributed evenly between patients since there were only minor variations. Capital costs were included in the calculations. All costs incurred in the inpatient and outpatient units, as well as costs for drugs patients received at outpatient pharmacies, were summarized as direct medical costs. Direct non-medical costs (i.e. travel expenses) and indirect costs (i.e. productivity losses due to CF) were also taken into account. Travel expenses were calculated for every patient based on the measured distance between the patient’s home and the hospital multiplied by 0.3 h/km and the times the patient was admitted to the hospital or visited the outpatient unit multiplied by two. We used the human capital approach to calculate costs associated with productivity losses. These costs could not be calculated at the individual level due to missing data. Therefore, the average rate of patients obtaining employment disability pensions (i.e. the number of patients unable to work due to CF) was taken from the CF patient registry for Germany provided by the Mukoviszidose eingetragener Verein (e.V.).[8] The German Adis ª 2012 Springer International Publishing AG. All rights reserved.

767

registry[8] covers 82% of all CF patients in Germany.[24] CF registries in the UK (70%) and the US (85.6%) have a similar coverage in their respective countries.[1,25] The disability rate was then multiplied with the average yearly earned income per person for Germany in 2004 (h32 569).[26] The average adult patient sick days due to CF were obtained from the Mukoviszidose e.V. (unpublished data, 2007) and multiplied with the average income per day (h89). The method for calculating the average yearly income and the average income per day corresponds to the German recommendations on health economic evaluation of the Hanover Consensus Group.[15] Furthermore, the productivity loss of parents taking care of their sick children was considered. The average number of parents concerned and the average days of absence from work were also obtained from the Mukoviszidose e.V. (unpublished data, 2007) and multiplied by the average income per day. Data Analysis

In order to investigate whether patients with different levels of CF severity can be classified into homogenous cost groups, severity levels were defined according to a diagnosis-related severity model and Wilcoxon-Mann-Whitney tests were performed for different cost categories. Patients without colonization of the lungs with P. aeruginosa were grouped as mild, while patients with chronic colonization of the lungs, but without pulmonary hypertension and respiratory insufficiency, were classified as moderate. Patients with P. aeruginosa infection, pulmonary hypertension and global respiratory insufficiency were classified as severe. Generalized least squares (GLS) regression analysis was performed to determine the impact of diagnosis-related severity, forced expiratory volume in 1 second (FEV1) and the variables age and sex on costs per patient for total direct and drug costs as well as the remaining costs (all direct costs except drug costs). A Gamma regression with log-link function fitted best according to residual plots and the Akaike Information Criterion (AIC) and was therefore chosen. For Pharmacoeconomics 2012; 30 (9)

768

the regression analysis, the statistical software STATA[27] was used. To increase generalizability of the results found at our study hospital, we performed a sensitivity analysis using Monte Carlo simulation. The relative number of patients with mild, moderate and severe disease and all direct medical costs except for outpatient drug costs were identified as variables that could be different for other CF centres in Germany and influence the costs calculated. In order to assume a representative distribution of patients with different severity levels for Germany, we obtained the information from the CF patient registry of the Mukoviszidose e.V.,[8] which contains data of 82% of all CF patients in Germany.[24] The relevant information was the relative number of patients infected with P. aeruginosa. Forty-nine percent of all CF patients in Germany were not infected with P. aeruginosa.[8] Hence, the relative number of patients in the mild group was 49%, whereas the relative number of moderate and severe patients (patients of both groups were infected with P. aeruginosa) was 51%. We are confident that the share of patients colonized with P. aeruginosa reported by the German registry (51%) is reasonable, as the US CF registry, which has a similar coverage, reports a share of 52.5%.[1] The UK CF registry, which, however, has a lower coverage, reports a share of 39.4%.[25] As the percentage of patients being infected can vary over time, we defined a confidence interval of 20% around the value of 49% of patients not infected with P. aeruginosa. For every iteration of the Monte Carlo simulation, one value was randomly drawn of the Uniform distribution between the values 0.39 (0.49 - 20%) and 0.59 (0.49 + 20%). As the sum of the relative numbers of patients in the different severity groups has to be 1, the respective relative number of patients infected with P. aeruginosa was 1 minus the randomly drawn number of patients not infected with P. aeruginosa. The data on hospital costs from the Federal Statistical Office of Germany suggest that size and ownership status are two important factors that cause variation of hospital costs in addition to disease-related factors.[28] According to the Adis ª 2012 Springer International Publishing AG. All rights reserved.

Heimeshoff et al.

Mukoviszidose e.V., 60% of all CF patients in Germany were treated at university hospitals in 2005.[8] University hospitals are usually large (more than 500 beds) and publicly owned, which was true for our study hospital as well. Hence, the costs of these hospitals might be comparable with the costs of our study hospital. To increase generalizability beyond our study hospital, we applied different sizes and ownership statuses derived from the Federal Statistical Office of Germany on staff and material costs (including laboratory costs, overhead costs and drug costs for the inpatient treatment)[28] found in our study. Thus, if the staff costs for patient care of private hospitals with less than 100 beds were 83% of those of public hospitals with more than 500 beds (according to the data of the Federal Statistical Office of Germany),[28] we multiplied the staff costs calculated for our study hospital by 0.83. The same procedure was performed for laboratory costs, overhead costs and drug costs for the inpatient treatment, which were summarized as material costs. This was done separately for costs of the mild CF patients (not infected with P. aeruginosa) and for the average costs of the moderate and severe CF patients (infected with P. aeruginosa). Hence, the costs calculated for our study hospital served as the basis for the calculation of costs for hospitals of different size and ownership status. As drug costs for the outpatient treatment are the same for every pharmacy in Germany, they were not subject to Monte Carlo simulation. As 60% of the CF patients in Germany were treated at university hospitals in 2005,[8] we assumed that the remaining patients were distributed among the other types of hospitals according to the relative number of these hospital types in Germany. The estimated staff and material costs of different types of hospitals, as well as the estimated relative number of CF patients who visited the different hospitals, served as basis for the creation of the distributions of costs. The distributions that fitted best according to chi-squared statistics were chosen. For staff and material costs of patients not infected and infected with P. aeruginosa, exponential distributions with different parameters fitted best. Hence, the fixed values of staff and material costs of Pharmacoeconomics 2012; 30 (9)

Cost of Illness of Cystic Fibrosis

patients not infected and of patients infected with P. aeruginosa calculated for our study hospital were replaced by estimated distributions of costs for different types of hospitals. For each iteration of the Monte Carlo simulation, the randomly drawn value of the distribution of the relative number of patients not infected with P. aeruginosa was multiplied by the sum of the randomly drawn values of the distributions of staff and material costs and the outpatient drug costs of patients not infected with P. aeruginosa, and the respective relative number of patients infected with P. aeruginosa was multiplied by the sum of the randomly drawn values of the distributions of staff and material costs and the outpatient drug costs of patients infected with P. aeruginosa. Finally, both terms were summarized and supplemented by indirect costs and travel expenses (travel expenses were not subject to Monte Carlo simulation as they were rather low). For the generation of the distribution of the estimated total costs per patient in Germany, this procedure was repeated 10 000 times. For the Monte Carlo simulation, the statistical software @RISK[29] was used. For further details of the Monte Carlo simulation, please see the technical appendix in the Supplemental Digital Content, http://links.adisonline.com/PCZ/A146. Results Data Description

Data from 158 patients were collected and analysed. The mean age of all patients was 20.09 years, with 27 patients (17%) aged 6 years and younger, 37 patients (23%) aged 7–18 years, and 94 patients (60%) aged 19 years and older. Fifty-two percent of all patients were female and the mean FEV1 value was 58.32%. According to the diagnosisrelated severity index, 46 patients (29%) were classified as mild, while most of the patients (n = 86; 54.5%) were defined as moderate. Twenty-six patients (16.5%) formed the severe group (table I). Seventy-one patients received outpatient treatment only during the observed time period, while 33 patients received inpatient treatment only and 54 patients received both types of treatment. Adis ª 2012 Springer International Publishing AG. All rights reserved.

769

Table I. Demographics and clinical variables (n = 158) Variable

n (%)

Age 0–6 years

27 (17)

7–18 years

37 (23)

>18 years

94 (60)

Sex Female

82 (52)

Male

76 (48)

Diagnosis-related severity level 1 = mild (no Pseudomonas aeruginosa)

46 (29)

2 = moderate (P. aeruginosa)

86 (54.5)

3 = severe (pulmonary hypertension and global respiratory insufficiency)

26 (16.5)

Treatment Outpatient

71 (45)

Inpatient

33 (21)

Inpatient and outpatient

54 (34)

Patients being treated as inpatients only received drugs from outpatient pharmacies as well, which were prescribed by the hospital when patients were discharged from hospital. Patients classified as mild according to the diagnosis-related severity index were mainly treated in the outpatient unit only. Almost half of the patients classified as moderate were admitted to the inpatient unit and also visited the outpatient unit, whereas almost half of those patients classified as severe were treated as inpatients only (table II). The mean total cost per patient per year was h41 468 (year 2004 values). Direct medical costs (inpatient and outpatient care including outpatient drugs) averaged h38 869 per patient per year and accounted for 94% of total costs. Direct medical costs ranged from h3876 among mild patients to h88 096 among patients with severe disease. The largest single cost factor were drug costs (76% of total costs), which is due to the high costs for drugs patients received at the outpatient pharmacy (72% of total costs). Overhead costs accounted for 14% of total costs, while staff costs for patient care and laboratory costs were rather low (2% of total costs). Direct non-medical costs (i.e. travel expenses) were marginal (<1% of total costs). Indirect costs accounted for 6% of total costs and amounted to h2491 (table III). Pharmacoeconomics 2012; 30 (9)

Heimeshoff et al.

770

Table II. Number of patients who received outpatient treatment only, inpatient and outpatient treatment, or inpatient treatment only in the different severity groups [n (%)]a Treatment

Mild [n = 46]

Moderate [n = 86]

Outpatient only [n = 71]

36 (78)

27 (31.5)

8 (31) 7 (27)

Inpatient and outpatient [n = 54]

8 (17.5)

39 (45.5)

Inpatient only [n = 33]

2 (4.5)

20 (23)

a

Severe [n = 26]

11 (42)

The table shows the treatment of patients in different severity groups: mild patients were mostly treated as outpatients only (78% of all mild patients); almost half of the moderate patients were treated as inpatients and outpatients (45.5% of all moderate patients); and almost half of the severe patients were treated as inpatients only (42% of all severe patients).

The most expensive drugs (according to the product of consumption and price) for the outpatient treatment are shown in table IV. Antibiotics such as tobramycin, dornase alfa and colistin were identified as the drugs with the highest resource utilization. As the consumption of these drugs rises with the degree of severity, drug costs for outpatient care increase with the progression of the disease. Analyses of Variance

The classification of patients according to the diagnosis-related severity model revealed that costs rose with the degree of severity. The relative cost increase compared with the relatively milder group was the highest between patients infected

with P. aeruginosa (moderate patients) compared with patients without infection (mild patients) and all cost categories rose with the degree of severity (table V). The Wilcoxon-Mann-Whitney tests revealed that all cost categories for mild patients differed significantly (p £ 0.001) from those of moderate patients. Patients with moderate disease had significantly lower total direct and drug costs than patients with severe disease, but laboratory costs and staff costs were not statistically different. Generalized Least Squares Regression

The results of the regression analysis are presented in table VI. In a multiple regression with

Table III. Costs and cost variation for different cost categories (h, year 2004 values) [n = 158]a Cost categoriesb

Mean

Median

Minimum

Maximum

SDc

Drug costsd

31 667

37 782

854

72 291

15 120

Laboratory costs

731

355

26

5231

852

Staff costs for patient care

696

404

30

4113

739

5775

4263

1103

26 205

4352

42 262

3876

88 096

18 943

97

15

455

80

42 287

3940

88 175

18 939

2491

NA

NA

NA

NA

41 468

NA

NA

NA

NA

Overhead costs Direct medical costs Direct non-medical costs Total direct costs Indirect costse Total costs

38 869 108 38 977

a

Detailed costing data from our hospital were used.

b

Drug costs, laboratory costs, staff costs for patient care and overhead costs were summarized as direct medical costs. Total costs were calculated as the sum of direct medical costs, direct non-medical costs (i.e. travel expenses) and indirect costs (costs of absence from work and productivity losses due to CF).

c

SD indicates the variability of costs.

d

For the calculation of drug costs, purchasing prices were taken from the list with official selling prices that applied to all pharmacies in Germany.[23]

e

Only the average is available for indirect costs. For the calculation of indirect costs, data from the CF patient registry provided by the Mukoviszidose e.V.,[8] data from the Federal Statistical Office of Germany[26] and data from the Mukoviszidose e.V. (unpublished data, 2007) were used.

CF = cystic fibrosis; e.V. = eingetragener Verein; NA = not available; SD = standard deviation.

Adis ª 2012 Springer International Publishing AG. All rights reserved.

Pharmacoeconomics 2012; 30 (9)

Cost of Illness of Cystic Fibrosis

771

Table IV. Drug costsa according to severity level (h, year 2004 values) [mean – standard deviation] Drug (trade name)b

Diagnosis-related severity level (n = 72) Mild (n = 38)

Moderate (n = 21)

Severe (n = 13)

Tobramycin (Tobi)

1307 – 5252

9146 – 12 864

13 754 – 17 090

Dornase alfa (Pulmozyme)

4470 – 5513

6528 – 5789

9887 – 6943

2340 – 4227

5631 – 8330

Colistin (Colistin CF) Itraconazole (Sempera)

230 – 941

333 – 743

2123 – 2968

Pancreatin (Kreon)

396 – 724

1160 – 1005

2318 – 1798

33 – 148

684 – 621

710 – 612

255 – 514

476 – 525

413 – 572

Ciprofloxacin (Ciprobay)

54 – 209

442 – 694

304 – 375

Pantoprazole (Pantozol)

12 – 75

208 – 339

568 – 2047

511 – 1728

526 – 1627

Azithromycin (Zithromax) Salmeterol, fluticasone (Viani)

Tobramycin (Gernebcin)

164 – 1012

a

For the calculation of drug costs, purchasing prices were taken from the list with official selling prices that applied to all pharmacies in Germany.[23]

b

The use of trade names is for product identification purposes only and does not imply endorsement. The manufacturers are as follows: Tobi, Chiron; Gernebcin, Infectopharm; Pulmozyme, Genentech; Colistin CF, Gru¨nenthal; Sempera, Janssen-Cilag; Zithromax, Pfizer; Viani, GlaxoSmithKline; Ciprobay, Bayer; Pantozol, Altana Pharma Deutschland; and Kreon, Solvay Arzneimittel GmbH.

CF = cystic fibrosis.

age, sex, diagnosis-related severity levels (with mild patients as the reference) and FEV1 as independent variables and total direct costs per patient as the dependent variable (regression 1), the diagnosis-related severity coefficients for moderate and severe patients (patients with P. aeruginosa and patients with pulmonary hypertension and global respiratory insufficiency), as well as FEV1, were highly significant (p £ 0.01). Furthermore, the coefficients rose with the degree of severity, while the increase compared with the relatively milder group was higher for the coefficient for moderate patients. Age and sex were not significant. The result of the regression with FEV1 excluded from the analysis (regression 2) was almost the same, but the coefficients indicating the relative increase in costs for moderate and severe

patients compared with the costs for mild patients rose even more and the coefficient for age was positive and significant. For the regression with FEV1, age and sex as dependent variables (regression 3), FEV1 was negative and highly significant as well. In order to show the effects of drug costs, which accounted on average for 76% of total costs, two more models were tested, one with drug costs per patient (regressions 4–6) and the other with remaining costs per patient (regression 7–9) as the dependent variables. For drug costs (regressions 4–6), the results were comparable with those of the regressions with total direct costs as the dependent variable (regressions 1–3): sex had almost no influence and age was highly significant for the regression with FEV1 excluded (regression 5), whereas FEV1 was negative and highly significant in

Table V. Results of variance analyses for diagnosis-related severity levels (h, year 2004 values) [mean – standard deviation]a Cost categories

Diagnosis-related severity level (n = 158) Mild (n = 23)

Moderate (n = 83)

Severe (n = 25)

Total direct costs

16 721 – 7169

46 087 – 13 076

54 804 – 15 109

Drug costs

13 032 – 6590

37 257 – 9042

46 146 – 10 725

Laboratory costs

157 – 143

879 – 858

1256 – 1026

Staff costs for patient care

238 – 156

849 – 773

1118 – 956

a

All cost categories of mild patients differ significantly (p < 0.001) from those of moderate patients. Total direct costs and drug costs of moderate patients differ significantly (p < 0.001) from those of severe patients, while laboratory costs and staff costs of moderate patients were not statistically different from those of severe patients.

Adis ª 2012 Springer International Publishing AG. All rights reserved.

Pharmacoeconomics 2012; 30 (9)

Heimeshoff et al.

772

Table VI. Results of generalized least squares regressions.a Values in parentheses are z-values Dependent variable Total direct costs/patient

Drug costs/patient

Remaining direct costs/patient

Reg. 1

Reg. 2

Reg. 3

Reg. 4

Reg. 5

Reg. 6

Reg. 7

Reg. 8

Reg. 9

N

134

157

134

134

157

134

134

157

134

Constantb

10.43**

9.72**

11.22**

10.10**

9.46**

11.00**

9.33**

8.24**

9.71**

(73.64)

(98.71)

(71.39)

(76.94)

(93.66)

(71.26)

(28.54)

(42.02)

(34.30)

Age

-0.00

0.00

0.00

-0.00

0.01

0.00

-0.01

0.00

-0.00

(-1.57)

(2.24)*

(0.19)

(-1.34)

(2.80)**

(0.26)

(-0.94)

(0.06)

(-0.01)

Sex

0.00

-0.05

0.07

-0.01

-0.06

0.06

0.06

-0.04

0.11

(0.92)

(-1.02)

(-1.06)

(-0.04)

(1.02)

FEV1 (%)

(1.07)

(-0.18)

(0.96)

(0.54)

-0.01**

-0.01**

-0.01**

-0.01**

-0.02**

-0.02**

(-5.08)

(-8.55)

(-3.92)

(-7.93)

(-5.19)

(-6.83)

Diagnosis-related severity level Mildc

0.00

0.00

0.00

0.00

0.00

0.00

Moderate

0.79**

0.94**

0.81**

0.95**

0.66**

0.89**

(11.81)

(13.28)

(13.08)

(13.35)

(4.14)

(5.96)

Severe

0.83**

1.11**

0.93**

1.16**

0.28

0.87**

(8.60)

(12.26)

(10.10)

(12.73)

(1.20)

(4.58)

a

The coefficients are the results of a GLS model using Gamma regression with log-link function. Regressions 1–3, as well as 4–6 and 7–9, differ regarding the independent variables included.

b

Where the regression line intercepts the y-axis, representing the amount the dependent variable ‘y’ will be when all independent variables are 0.

c The reference category is mild. FEV1 = forced expiratory volume in 1 second; GLS = generalized least squares; Reg. = regression; * p < 0.05, ** p < 0.01.

all regressions. The coefficients indicating the relative cost increase rose with the degree of severity as well and were higher than those of the regression with total direct costs as a dependent variable. For the model without drug costs (regressions 7–9), age and sex had a minor influence as well and FEV1 was highly significant in all cases, but the results for the coefficients indicating the relative increase were different (regressions 7 and 8): the coefficient for the moderate patients was positive and higher than the one for severe patients in both regressions and the diagnosis-related coefficient for severe patients was not significant if FEV1 was included (regression 7). With FEV1 excluded, both coefficients, for moderate and severe patients, were highly significant (regression 8). Sensitivity Analysis

The results of Monte Carlo simulation are presented in figure 1. The figure shows the cumulaAdis ª 2012 Springer International Publishing AG. All rights reserved.

tive distribution of the estimated yearly costs per patient with CF for 10 000 iterations. The main result was that total yearly costs per patient were estimated to be lower than h37 300 with a probability of 90%. The minimum total yearly cost per patient was estimated to be h30 104 and the maximum h46 085. The mean distribution of total costs per patient per year was h34 474 and thus about h7000 lower than the calculated total costs of our study population. Discussion In this study, the total averaged costs per patient with CF, including all direct medical and non-medical costs as well as indirect costs, were calculated on the basis of measured individual resource utilization in Germany from a societal perspective and the main cost drivers were identified. Furthermore, it was tested if patients could be classified into different severity groups Pharmacoeconomics 2012; 30 (9)

Cost of Illness of Cystic Fibrosis

773

according to a diagnosis-related severity model, and the influence of certain factors on different cost categories was investigated. Moreover, a sensitivity analysis was performed. The main results of the study were the following: the mean total cost per patient per year was h41 468, with direct costs accounting for 94% of total costs. Indirect costs amounted to h2491 per patient per year and thus accounted for a rather low share of total costs. If we had used the frictioncost method, which is preferred by some health economists, indirect costs would have most likely been even lower.[30] Drug costs for prescribed drugs from the outpatient pharmacy were identified as the main cost driver, accounting for 72% of total costs and leading to higher outpatient costs than inpatient costs. The relative cost increase was highest for patients infected with P. aeruginosa compared with the relatively milder group (patients without P. aeruginosa), which was

30 100

⎧ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎨ ⎪ ⎪ ⎪ ⎪ ⎪ ⎩

9.6%

⎧ ⎪ ⎪ ⎪ ⎨ ⎪ ⎪ ⎪ ⎩

90.4% 1

37 300

Share of patients

0.8 0.6 0.4 0.2 0 30

32

34 36 38 40 42 44 46 Yearly costs per patient ( × 1000)

48

Fig. 1. Cumulative distribution of yearly costs per patient. The assumptions of the Monte Carlo simulation were as follows: the relative number of patients in a severity group can vary about 20%; drug costs for inpatient treatment, laboratory costs, overhead costs (summarized as material costs) and staff costs for patient care vary due to differences in the ownership status and size of the hospital; and drug costs for outpatient treatment and indirect costs are fixed. For every iteration of the Monte Carlo simulation, we randomly drew the number of patients not infected and infected with Pseudomonas aeruginosa, as well as the staff costs of patient care and the material costs, and we calculated the sum of the relative number of patients not infected with P. aeruginosa multiplied by the costs of patients not infected and the relative number of patients infected with P. aeruginosa multiplied by the costs of patients infected and supplemented these with travel expenses and indirect costs. The yearly costs per patient with cystic fibrosis in Germany were estimated 10 000 times. The figure shows the share of patients with lower costs than the respective value of the x-axis. Ninety percent of all estimates showed costs lower than h37 300.

Adis ª 2012 Springer International Publishing AG. All rights reserved.

confirmed by the results of the regression analysis. Furthermore, the impact of the CF severity level on drug costs was higher than on total direct costs. Monte Carlo simulation showed that total yearly costs were lower than h37 300 with a probability of 90%. Therefore, costs were overestimated in this study, which was mainly due to the higher prevalence of P. aeruginosa in our study population. Seventy-one percent of our study population was infected with P. aeruginosa compared with only 51% of all CF patients in Germany.[8] As mentioned earlier, the costs calculated in this study are difficult to compare with those calculated in other studies because either the method or the costs included are different. However, there are a few studies measuring individual resource use and focusing on costs of CF in Germany.[9,10,17] Eidt-Koch et al.[17] analysed outpatient and drug costs from an economic perspective and took the burden (time and money) for patients into account. Their results for average yearly drug costs per patient (h21 604) were different from the drug costs for outpatients in our study (h29 664). These differences might be due to different study settings and study designs. Eidt-Koch et al.[17] calculated costs based on data that were collected over a 4-week period in different CF treatment units and micro-costing was not performed in such detail as in our study. Schreyo¨gg et al.[10] measured hospitalization costs of CF per case from a healthcare provider’s perspective. The present study and that of Schreyo¨gg et al.[10] were based on the same data and should thus have led to the same results for hospitalization costs; however, Schreyo¨gg et al.[10] calculated costs per case and not per patient, thus results are difficult to compare. Furthermore, Baumann et al.[9] calculated direct costs for inpatient and outpatient treatment and for drugs patients received from the pharmacy from a payer perspective. They found that costs per patient per year amounted to h23 989. The lower costs compared with our study, even after sensitivity analyses were performed, may be due to costs being calculated using reimbursement rates. There is evidence that inpatients, particularly those treated with costly drugs, are underpaid in the current German diagnosis-related groups (G-DRG) system due to the compression Pharmacoeconomics 2012; 30 (9)

774

effect that typically occurs in early stages of newly introduced DRG systems.[31] This effect is due to the possibly inaccurate assignment of costs to different DRGs by the participating calculation hospitals. Cost calculations might be based on length of stay, as more accurate measures are often missing at the time the new reimbursement system is introduced. This leads to an underestimation of costs of cases with high drug costs or expensive technologies but a relatively short length of stay. The result of high outpatient drug costs was confirmed by Baumann et al.,[9] who found home drug treatment to be the most important single cost factor. This may be explained by different regulations for inpatient and outpatient drugs in Germany. Prices for inpatient drugs can be negotiated freely between hospitals and manufacturers, while there are fixed margins for outpatient pharmacies and wholesalers as well as uniform exmanufacturer prices throughout Germany for drugs dispensed in outpatient care. The share of drug costs in our study (76% of total costs) is rather high compared with other studies.[3,32,33] This difference may be explained by a combination of different factors. First, other studies from Germany also found higher shares of drug costs than in non-German studies, pointing to structural differences in healthcare systems. Hence, Baumann et al.[9] found drug costs for outpatient treatment to be 47% of all total costs. As costs for outpatient treatment accounted for 59% of total costs in their study, drug costs for outpatient treatment accounted for about 80% of all outpatient treatment costs.[9] Moreover, EidtKoch et al.[17] came to a similar result as they found drug costs to be 91% of all outpatient treatment costs. In general, several studies have shown that Germany has higher drug prices than most other countries.[34,35] Second, most other studies relied on a gross-costing approach. As mentioned in the Introduction section, micro-costing usually leads to higher cost estimates than gross-costing approaches.[19] Third, as most other studies used the payer perspective, drug costs are likely to be underestimated due to the compression effect occurring in reimbursement systems.[31] Fourth, Krauth et al.[6] found earlier studies underestimated Adis ª 2012 Springer International Publishing AG. All rights reserved.

Heimeshoff et al.

actual drug costs, as cost-intensive therapies have increased drug costs substantially in recent years. This is in line with the findings of DeWitt et al.,[16] who found costs of care for CF patients with mild impairment in lung function to exceed $US43 000 and drug costs to account for more than 85% of total costs. The authors argued that one reason for their finding of relatively high costs may reflect the evolution of practice patterns over time or variations in clinical practice between different CF treatment centres.[16] The finding that costs rise with progression of the disease is also confirmed by previous studies. Robson et al.[21] divided patients into four severity levels according to frequency of hospitalization and medical utilization. Their ratio of mild to severe cases was much higher than in our study, which may be due to different definition of the four groups. Lieu et al.[32] classified patients into three groups according to FEV1 values. Because very mild and mild cases were grouped together in their first group, their results are difficult to compare with ours. Moreover, Eidt-Koch et al.[36] divided patients into subgroups according to their co-morbidities, infection status (bacterial colonization of the lung vs no colonization) and FEV1 values and found significant differences in outpatient drug costs between the patient groups for all criteria. Finally, the result that costs are higher for patients suffering from a chronic infection with P. aeruginosa is confirmed by Braccini et al.[37] The authors compared the costs of treatment of initial infection and chronic infection with P. aeruginosa and found the latter to be considerably higher. A regression approach to explore the impact of patients’ demographic and clinical variables was conducted in five other studies. Johnson et al.[11] ran an ordinary least squares (OLS) regression model including the variables age, sex, body mass index (BMI), FEV1, the presence or absence of P. aeruginosa, Burkholderia cepacia or other organisms, and therapy of recombinant human deoxyribonuclease (DNase) as independent variables and ‘Ln (annual costs)’ as the dependent variable. The model explained 82% of the variance of annual costs. Moreover, Baumann et al.[9] constructed a stepwise regression including colonization Pharmacoeconomics 2012; 30 (9)

Cost of Illness of Cystic Fibrosis

status of the lung with P. aeruginosa and FEV1 as dependent variables. The coefficient of determination was 0.64. Schreyo¨gg et al.[10] investigated the influence of the variables age, sex, BMI, FEV1 and diagnosis-related severity and explained 31% of the variance of their dependent variable ‘Ln (hospitalization costs per case)’. Ouyang et al.[12] performed an OLS regression for patients with CF who were stratified by age into three groups and included gender and complication dummy variables as independent variables. They found various complications (e.g. malnutrition, transplantation, diabetes and infection with P. aeruginosa) to influence total expenditures. DeWitt et al.[16] used a generalized linear model with a Gamma error distribution and log link to detect factors influencing total costs. Almost all studies support our finding that the influence of demographic variables (age, sex, BMI) is rather small, while clinical variables explain a larger part of the cost variation. Our study has a number of strengths compared with previous approaches. To our knowledge, this is the first study to measure total individual costs of illness per patient with CF from a societal perspective. Moreover, our approach is more comprehensive than previous ones. Besides the consideration of direct medical costs for 158 patients, direct non-medical costs (i.e. travel expenses) and indirect costs (i.e. productivity losses due to CF and costs of absence from work) were taken into account. A further strength is that we measured individual resource consumption for relevant cost categories and thus followed a microcosting approach. Although previous studies were also based on individual patient data, resource use was usually based on expert estimates rather than on actual resource consumption. In contrast, our study relied on data based on actual resource consumption. Name and proportion of drugs taken were registered for every patient and multiplied by the relevant prices. This was very important for the accuracy of the calculation because drug costs accounted on average for 76% of total costs and there were large differences between the patients. Staff costs for patient care and laboratory costs were also based on measured resource consumption. In order to increase the Adis ª 2012 Springer International Publishing AG. All rights reserved.

775

robustness of our findings and to be able to generalize the results to the societal perspective, a sensitivity analysis using Monte Carlo simulation was performed. One limitation of the study is that data were collected in only one hospital and that the Monte Carlo simulation was the only method to generalize our findings with regard to Germany. Similar to other cost studies utilizing primary data, only a comparatively small number of patients could be included in the analysis. Despite our demonstration of results comparable with other studies, the general validity of our results would increase if data were also collected from other study sites. Data for conducting cost-of-illness studies should ideally be obtained from a large number of patients at several hospitals. Moreover, although we performed a Monte Carlo simulation, uncertainty regarding drug costs for inpatient care remains. We only have information on drug prices for our study hospital and do not have information on drug prices for other hospitals with specialized centres for CF. Finally, changes in treatment patterns of CF patients since the study year, especially a shift from inpatient therapy to outpatient and home-based therapy, should be considered when interpreting our results. It can be expected that the trend towards outpatient and home-based therapy may have reduced costs, particularly for patients with low disease severity. Finally, we could have included pancreas insufficiency as a potential cost driver, which is used as an additional indicator of severity in clinical studies.[12] However, according to our review of cost studies on CF, none of the studies have identified a significant impact from pancreas insufficiency on costs in a multiple regression, which was confirmed in a recent study by Ouyang et al.[12] Conclusion The share of indirect costs as a percentage of total costs for CF was rather low at the time our study was conducted. However, the relevance of indirect costs is likely to increase in the future, as life expectancy increases, which is likely to lead to a rising work disability rate and thus increase Pharmacoeconomics 2012; 30 (9)

Heimeshoff et al.

776

indirect costs. Moreover, calculating individual costs per patient is important because there are great differences in individual resource utilization (especially for drug costs), a result that was confirmed by previous studies. Hence, direct costs rise with the degree of severity. Compared with the relatively milder group, the largest increase in costs was found for patients infected with P. aeruginosa. Chronic infection with the bacterium leads to an increase in drug costs, which were identified as the main cost driver. Thus, substantial savings could be achieved by decreasing the prevalence of patients chronically infected with P. aeruginosa. Acknowledgements This work was planned and written by Mareike Heimeshoff, Helge Hollmeyer, Jonas Schreyo¨gg, Oliver Tiemann and Doris Staab, and the final version was approved by all the authors. There was no external funding for this study, and none of the authors have any potential conflicts of interests that are directly relevant to the content of this paper. The conceptual work was mainly done by Jonas Schreyo¨gg and Oliver Tiemann. Doris Staab and Helge Hollmeyer were responsible for the data collection process and provided all medical information. Helge Hollmeyer calculated part of the costs. Mareike Heimeshoff was responsible for the data analysis and for the preparation of a draft of the paper, which was supported by Jonas Schreyo¨gg and Oliver Tiemann. Results were interpreted and written by all authors. The guarantor for the overall content of this paper is Mareike Heimeshoff. The authors would like to thank the Munich Center of Health Sciences, based at Ludwig-Maximilians University of Munich, for their support.

References 1. Cystic Fibrosis Foundation. Patient registry report [online]. Available from URL: http://www.cff.org/LivingWithCF/ CareCenterNetwork/PatientRegistry/ [Accessed 2012 April 19] 2. Assael BM, Castellani C, Ocampo MB, et al. Epidemiology and survival analysis of cystic fibrosis in an area of intense neonatal screening over 30 years. Am J Epidemiol 2002 Sep; 156 (5): 397-401 3. Horvais V, Touzet S, Bourdy S, et al. Cost of home and hospital care for patients with cystic fibrosis followed up in two reference medical centers in France. Int J Technol Assess Health Care 2006; 22 (4): 525-31 4. Dodge JA, Morison S, Lewis PA, et al. Incidence, population and survival of cystic fibrosis in the UK, 1968-95. Arch Dis Child 1997 Dec; 77 (6): 493-6 5. Mukoviszidose e.V. Annual report 2008/09 [in German; online]. Available from URL: http://muko.info/fileadmin/ Dokumentenablage/Wer-wir-sind/Publikationen/Jahresbe richt_08-09.pdf [Accessed 2010 Oct 05]

Adis ª 2012 Springer International Publishing AG. All rights reserved.

6. Krauth C, Jalivand N, Welte T, et al. Cystic fibrosis: cost of illness and considerations for the economic evaluation of potential therapies. Pharmacoeconomics 2003; 21 (14): 1001-24 7. Cystic Fibrosis Foundation. About cystic fibrosis: what you need to know [online]. Available from URL: http://www. cff.org/AboutCF/ [Accessed 2010 Mar 18] 8. Stern M, Sens B, Wiedemann B, et al., editors. Qualita¨tssicherung mukoviszidose: u¨berblick u¨ber den gesundheitszustand der patienten in Deutschland 2005. Hanover: Zentrum fu¨r Qualita¨t und Management im Gesundheitswesen and Mukoviszidose e.V., 2005 9. Baumann U, Stocklossa C, Greiner W, et al. Cost of care and clinical condition in paediatric cystic fibrosis patients. J Cyst Fibros 2003 Jun; 2 (2): 84-90 10. Schreyo¨gg J, Hollmeyer H, Bluemel M, et al. Hospitalisation costs of cystic fibrosis. Pharmacoeconomics 2006; 24 (10): 999-1009 11. Johnson JA, Connolly MA, Jacobs P, et al. Cost of care for individuals with cystic fibrosis: a regression approach to determining the impact of recombinant human DNase. Pharmacotherapy 1999 Oct; 19 (10): 1159-66 12. Ouyang L, Grosse SD, Amendah DD, et al. Healthcare expenditures for privately insured people with cystic fibrosis. Pediatr Pulmonol 2009 Oct; 44 (10): 989-96 13. Jo¨nsson B. Ten arguments for a societal perspective in the economic evaluation of medical innovations. Eur J Health Econ 2009 Jul; 10 (4): 357-9 14. Zentner A, Velasco-Garrido M, Busse R. Methods of comparative evaluation for pharmaceuticals [in German]. 13th rev. ed. Berlin: German Agency for Health Technology Assessment at the German Institute for Medical Documentation and Information, 2005 15. Graf von der Schulenburg JM, Greiner W, Klusen N, et al. German recommendations on health economic evaluation: third and updated version of the Hanover Consensus. Value Health 2008 Jul-Aug; 11 (4): 539-44 16. DeWitt EM, Grussemeyer CA, Friedman JY, et al. Resource use, costs, and utility estimates for patients with cystic fibrosis with mild impairment in lung function: analysis of data collected alongside a 48-week multicenter clinical trial. Value Health 2012; 15 (2): 277-83 17. Eidt-Koch D, Mittendorf T, Wagner T, et al. Evaluation von kosten der ambulanten behandlung bei mukoviszidose in Deutschland: u¨bersicht u¨ber die ergebnisse einer prospektiven studie. Med Klin 2009 Jul; 104 (7): 529-35 18. Eidt-Koch D, Wagner TOF, Mittendorf T, et al. Resource usage in outpatient care and reimbursement for cystic fibrosis in Germany. Pediatr Pulm 2011; 46: 278-85 19. Clement FM, Ghali W, Donaldson C, et al. The impact of using different costing methods on the results of an economic evaluation of cardiac care: micro-costing vs grosscosting approaches. J Health Econ 2009 Apr; 18 (4): 377-88 20. Raftery J. Costing in economic evaluations. BMJ 2000 Jun; 320: 1597 21. Robson M, Abbot J, Webb K, et al. A cost description of an adult cystic fibrosis unit and cost analyses of different categories of patients. Thorax 1992 Sep; 47 (9): 684-9 22. Wildhagen MF, Verheij JB, Verzikl JG, et al. Cost of care of patients with cystic fibrosis in the Netherlands in 1990-1

Pharmacoeconomics 2012; 30 (9)

Cost of Illness of Cystic Fibrosis

23. 24.

25.

26.

27. 28.

29. 30.

[published erratum appears in Thorax 1997; 52 (2): 204]. Thorax 1996 Mar; 51 (3): 298-301 Rote Liste Service GmbH. Rote liste 2004. Frankfurt: Rote Liste, 2004 Stern M, Wiedemann B, Wenzlaff P, et al. From registry to quality management: the German cystic fibrosis quality assessment project 1995-2006. Eur Respir J 2008 Jan; 31 (1): 29-35 Cystic Fibrosis Trust. UK CF registry: annual data report 2008 [online]. Available from URL: http://www.cftrust. org.uk/aboutcf/publications/cfregistryreports/UK_CF_Reg istry-Annual_Data_Report_2008.pdf [Accessed 2010 Sep 20] Federal Statistical Office of Germany. Earnings and labour costs: results for Germany [in German; online]. Available from URL https://www.destatis.de/DE/Publikationen/ Thematisch/VerdiensteArbeitskosten/Arbeitskosten/Arbeits kostenBund2163201089004.pdf?__blob=publicationFile [Accessed 2012 April 19] STATA [computer program]. Version 10. College Station (TX): Statacorp, 2007 Federal Statistical Office of Germany. Report browser for hospital data [in German; online]. Available from URL https://www.destatis.de/DE/Publikationen/Thematisch/Ges undheit/AlteAusgaben/KostennachweisKrankenhaeuserAlt. html?nn=72108 [Accessed 2012 April 19] @RISK [computer program]. Version 5.5. Ithaca (NY): Palisade Corporation, 2009 Brouwer WB, Koopmanschap MA. The friction-cost method: replacement for nothing and leisure for free? Pharmacoeconomics 2005; 23 (2): 105-11

Adis ª 2012 Springer International Publishing AG. All rights reserved.

777

31. Schreyo¨gg J, Tiemann O, Busse R. Cost accounting to determine prices: how well do prices reflect costs in the German DRG-system? Health Care Manag Sci 2006 Aug; 9 (3): 269-80 32. Lieu TA, Ray GT, Farmer G, et al. The cost of medical care for patients with cystic fibrosis in a health maintenance organization. Pediatrics 1999 Jun; 103: e72 33. Ireys HT, Anderson GF, Shaffer TJ, et al. Expenditures for care of children with chronic illnesses enrolled in the Washington State Medicaid Program, fiscal year 1993. Pediatrics 1997 Jun; 100 (2): 197-204 34. Simoens S. International comparison of generic medicine prices. Curr Med Res Opin 2007 Nov; 23 (11): 2647-54 35. Garattini L, Motterlinia N, Cornagoa D. Prices and distribution margins of in-patent drugs in pharmacy: a comparison in seven European countries. Health Policy 2008 Mar; 85 (3): 305-13 36. Eidt-Koch D, Wagner T, Mittendorf T, et al. Outpatient medication costs of patients with cystic fibrosis in Germany. Appl Health Econ Health Policy 2010 Mar; 8 (2): 111-8 37. Braccini G, Festini F, Boni V, et al. The costs of treatment of early and chronic Pseudomonas aeruginosa infection in cystic fibrosis patients. J Chemother. 2009 Apr; 21 (2): 188-92

Correspondence: Mrs Mareike Heimeshoff, Institute for Health Care Management and Health Economics, Faculty of Economics and Business Administration, University of Hamburg, Esplanade 36, 20354 Hamburg, Germany. E-mail: [email protected]

Pharmacoeconomics 2012; 30 (9)