DDD vs. PDD: Calculation of daily therapy costs using liraglutide as an example

The Defined Daily Dose (DDD) is a theoretical quantity that was developed as an international standard comparison quantity for drug consumption research. It indicates the assumed mean maintenance dose per day in adults for an active ingredient in its main indication. In Germany, the official German version of the ATC classification (Anatomical-Therapeutic-Chemical Classification) including the DDD information is published by the DIMDI (German Institute for Medical Documentation and Information) (DIMDI 2010). This is based on the version of the WHO, which is responsible for creating the internationally standardized ATC/DDD system (Fricke et al. 2011). Both the WHO and the AOK Scientific Institute (WIdO), which in turn publishes the methodology of the ATC classification and DDD determination for Germany, point out that the DDD is a calculated quantity that is not related to the therapeutic or prescribed dose may be confused. For longitudinal drug consumption research, the DDDs must be kept stable and are therefore unsuitable as a basis for decision-making for questions regarding drug reimbursement, pricing and therapy (Fricke et al. 2011).

>> The DDD information published by the WHO is not identical to the defined daily doses in the official German version of the ATC classification. This is due, among other things, to adjustments made to reflect the specifics of the healthcare situation in Germany. In particular, the DDD information is aligned with the dosage information in the official product information (DIMDI 2010). Contrary to the statement by the WHO and the WIdO, the DDD information from DIMDI should nevertheless be used as a uniform reference for calculating daily therapy costs in accordance with Section 73 Paragraph 8 of the Social Code (SGB V). Section 84 Paragraph 7a of the Social Code (SGB V), which was repealed as part of the German Drug Reorganization Act (AMNOG) on January 1, 2011, provided for the use of other reference values in cases where a price comparison based on the DDD is not appropriate. However, this approach was practically never used in the past. Despite conflicting statements regarding the applicability of DDDs in cost analyses, this remains common practice: For example, DDDs are used, among other things, for cost comparisons in the Drug Prescription Report (AVR) and thus for statements regarding the cost-effectiveness of a drug. It was recently clarified again that an uncritical adoption of the ATC classification cannot achieve the intended control effects on costs because the ATC classification system is not linked to it (Zuck 2010).
As a theoretical calculation, the DDD does not reflect actual consumption under real-world care conditions; however, the reality of care can be reflected by the Prescribed Daily Dose (PDD). The IMS® LRx database contains nationwide prescription data from patients with statutory health insurance and can therefore evaluate the actual daily dosage.
Economic aspects are playing an increasingly important role in healthcare. Chronic diseases, such as diabetes mellitus, in particular, cause high costs for the German healthcare system due to the high prevalence and incidence of the disease, its chronic course, and the associated serious and costly long-term consequences. It is estimated that approximately 8 million people in Germany currently suffer from diabetes mellitus (DDU 2008). Diabetes-associated direct costs rose from EUR 2,400 per patient in 2000 to EUR 2,605 in 2007 (Köster et al. 2011). With the rising incidence of type 2 diabetes, the cost-effectiveness of various treatment options, often measured using DDD-based cost comparisons, is becoming increasingly important. For example, the AVR compares the DDD costs of the GLP-1 analogues liraglutide and exenatide at EUR 4.00 and EUR 3.26, respectively, with the DDD costs of insulin (EUR 1.26) (Schwabe/Paffrath 2010). The DDDs of liraglutide and exenatide are given in the official German version of the ATC classification as 1.2 mg and 15 μg (DIMDI 2010). In the corresponding product information, however, the maintenance doses of liraglutide and exenatide are given as 1.2 mg once daily and 5 μg and 10 μg twice daily, respectively (Product Information Byetta 2010; Product Information Victoza 2010).
Therefore, the present study aims to use the GLP-1 analogue liraglutide as an example to determine whether the DDD can reflect the actual daily consumption – and thus the real treatment costs – of liraglutide under everyday care conditions. The aim of this study was therefore to use real prescription data to illustrate the PDD of liraglutide and the resulting calculated treatment costs based on the pharmacy selling price (AVP).
methodology
This retrospective analysis was conducted using the IMS® LRx database, which includes 95% of all billed statutory health insurance prescriptions. The database allows for the analysis of longitudinal prescription data by age and gender.
From the date of liraglutide's market launch (July 2009) to November 2010 (database closure), patients with type 2 diabetes mellitus who had initiated initial therapy with liraglutide and who had at least two prescriptions of liraglutide were identified. The actual daily dose (PDD) of liraglutide was calculated for each prescription by dividing the prescribed amount (mg) by the time in days (d) until the next prescription. The active ingredient amount in mg was calculated taking into account the different available package sizes (N1, N2, N3) and the active ingredient strength (6 mg per ml) (Fig. 1).
The PDD of the total population was descriptively analyzed using mean, standard deviation, and median. The mean was calculated using two methods: first, taking all values into account and then as a trimmed mean. When calculating the trimmed mean, the extreme values (3 %) of prescriptions with particularly short and long prescription intervals were not taken into account. This statistical method is used to prevent distortion of the mean by extreme values (Huber 1981; Hampel et al. 1986). The extreme values of the prescription intervals lie in a range that is hardly possible in reality (e.g., 10 days, 190 days) and indicate the influence of external circumstances such as vacation or hospitalization, which should not be taken into account when calculating the PDD.
To describe the study population, the variables age, gender, health insurance, region and prescription of other antidiabetic drugs were also recorded.
Results
A total of 22,906 patients (22,182 patients for the calculation of the trimmed mean) with a total of 67,171 (63,140) liraglutide prescriptions were included. The mean age of the study population was 56.6 ± 11.6 years; the sexes were equally distributed (Table 1). Most patients received liraglutide in combination with metformin (n=9,352). In addition, liraglutide was used in combination with one or more other antidiabetic agents. The mean duration of therapy was 132 days. The median PDD of liraglutide was 1.20 mg/day (Table 2). The median PDD of liraglutide did not differ between age groups; the median daily dose was slightly higher in women than in men (Table 3).
To administer a PDD of 1.2 mg liraglutide, 0.2 ml of the injection solution must be administered per day at a strength of 6 mg/ml.
Accordingly, the PDD costs of liraglutide are EUR 3.87 at an AVP of EUR 580.15 (N3 pack) (Table 4).
discussion
Contrary to the statement of the WIdO, the DDD data from the DIMDI are used as a uniform reference for calculating daily treatment costs (Section 73 (8) SGB V). This approach is repeatedly controversial. It has already been shown for some substance classes that the DDD is not a suitable metric for reflecting real treatment costs (Wasem/Bramlage 2008, 2009). This is particularly true for therapies with variable treatment durations, such as treatment with antibiotics, colony-stimulating factors, and anthelmintics: Here, the application of the DDD can lead to an overestimation of real costs (Wasem/Bramlage 2009).
Based on the present analysis, it was shown that the DDD is identical to the PDD. The resulting PDD/DDD cost of liraglutide is EUR 3.87. In the case of liraglutide, the correct calculation of daily treatment costs based on the DDD would therefore be possible. However, this does not mean that a cost comparison with another GLP-1 analogue for which such a study has not been conducted based on the DDD is generally appropriate.
In principle, comparing DDD costs can only be meaningful if both substances are equally effective, the DDD corresponds to actual consumption, and the treatment duration, side effect profile, and any concomitant therapies are identical. Another important criterion regarding the application of DDD for comparing real treatment costs is the question of whether the DDD could be applied at all. The example of the GLP-1 analogue exenatide demonstrates that this question is also relevant and, if not considered, can lead to an underestimation of treatment costs.
The DDD of exenatide is given as 15 μg in the official German version of the ATC classification (DIMDI 2010). According to the product information, exenatide must be administered twice daily; the daily dose is given as 10–20 μg (Schwabe/Paffrath 2010). To administer the DDD of 15 μg, theoretically, a 5 μg and a 10 μg dose would have to be administered, thus requiring a 5 μg and a 10 μg pack to be prescribed, as no 7.5 μg pack is available on the market that would allow twice-daily administration of 7.5 μg. The 5 μg pack is only available as N1; prescribing the DDD of exenatide can only be done at additional cost. The DDD costs calculated in this way amount to EUR 4.12 with an AVP of EUR 247.40 (prescription of two N1 packs each of the two different strengths).
This specific example shows that the use of the DDD is unsuitable for comparing costs of the GLP-1 analogue substance class. In the case of liraglutide, the DDD costs do not differ from the PDD costs, as the DDD and PDD are identical. However, in the case of exenatide, the DDD does not correspond to the PDD: It cannot reflect the actual dosage or the dosage interval and is therefore not suitable for reflecting real costs. In the case of exenatide, the use of the DDD leads to an underestimation of costs. Applying the PDD or the dosage information in the product information would be a better tool for cost comparison in this case.
In general, calculations of therapy costs should be based on the reality of care. The PDD would be a suitable tool for this, even if determining the PDD has potential limitations. One limitation of any secondary data analysis is the impossibility of randomizing patients. It is therefore entirely conceivable and likely that the patient groups for which PDDs for different substances are calculated differ in their basic characteristics (e.g., age and gender). However, this should not be viewed as a potential bias of the results, but rather as a reflection of the current reality of care and, in this sense, a desired effect. Furthermore, the validation of diagnoses and prescriptions is not possible for data protection reasons.
The presentation of actual annual treatment costs is becoming increasingly important, particularly in light of the early benefit assessment introduced on January 1, 2011, with the Pharmaceutical Market Reorganization Act (AMNOG). According to the procedural rules for assessing the benefit of medicinal products pursuant to Section 35a of the Social Code Book V (Federal Joint Committee 2011), the dossier must include information on the costs of the therapy or appropriate comparator therapy incurred by statutory health insurance providers through its use. Costs must be presented based on the annual average consumption per patient, which in turn is measured based on the DDDs consumed. The results of this study show that this type of calculation distorts the cost representation and does not reflect the reality of care. To avoid this, a uniform metric that is equally suitable for all substances and can reflect the costs in the reality of care is essential as a basis for an objective cost comparison. Despite the limitations mentioned, the PDD appears to be the preferred instrument for cost comparisons compared to the DDD.