- Email: [email protected]
Utilization and costs of glucose lowering therapies following health technology assessment for the new reimbursement scheme in Sweden
Health Policy 108 (2012) 207–215
Contents lists available at SciVerse ScienceDirect
Health Policy journal homepage: www.elsevier.com/locate/healthpol
Utilization and costs of glucose lowering therapies following health technology assessment for the new reimbursement scheme in Sweden Billie Pettersson a,∗ , Mikael Hoffmann b , David Andersson c , Per Wändell d , Lars-Åke Levin a a b c d
The Center for Medical Technology Assessment, CMT Linköping University, 581 83 Linköping, Sweden The NEPI Foundation – The Swedish Network for Pharmaco-epidemiology, 581 91 Linköping, Sweden The Department of Management and Engineering, Linköping University, 581 83 Linköping, Sweden Center for Family and Community Medicine, Karolinska Institute, 141 83 Huddinge, Sweden
a r t i c l e
i n f o
Article history: Received 21 November 2011 Received in revised form 4 October 2012 Accepted 18 October 2012 Keywords: Pricing and reimbursement policy Glucose lowering therapies Drug utilization Health technology assessment Health economics
a b s t r a c t Objectives: A new reimbursement scheme (RS) for glucose lowering therapies (GLT) was implemented in Sweden on March 1, 2010. Products on the market were retained, restricted, excluded or excluded for new courses in the new RS. The aim of this study was to compare utilization and costs of GLT for type 2 diabetes Mellitus (T2DM) before and after the implementation of the changed RS. Methods: This was a quasi-experimental study using data on dispensed GLT and costs from a database on dispensed individual based prescriptions in Sweden. Segmented regression analyses were used to assess utilization and costs. Results: Following the changed reimbursement status, there was an accelerated increasing trend in number of patients treated with restricted (P = 0.0007) or retained (P = 0.0021) insulins, as well as in costs for insulin based GLT (P = 0.0014). No impact was detected in the total number of patients treated with oral GLT, but a slightly negative trend in total costs for oral GLT was detected following the intervention (P = 0.0177). Conclusions: The new reimbursement scheme had a minor impact on utilization and costs of oral GLT. Despite restricted reimbursement for patients with T2DM, the utilization of insulin based GLT and related costs increased faster following the intervention. © 2012 Elsevier Ireland Ltd. All rights reserved.
1. Background Type 2 diabetes mellitus (T2DM) is a progressive disease associated with macrovascular and microvascular complications [1]. The risk of developing cardiovascular disease (CVD) is increased 2–4 times in subjects with T2DM independently of other concomitant risk factors [2]. Glycemic control in patients with T2DM reduces the risk of
∗ Corresponding author at: Pontongränd 13, 183 68 Täby, Sweden. Tel.: +46 768 85 00 10; fax: +46 8 626 14 22. E-mail addresses: Billie [email protected] (B. Pettersson), [email protected] (M. Hoffmann), [email protected] (D. Andersson), [email protected] (P. Wändell), [email protected] (L.-Å. Levin). 0168-8510/$ – see front matter © 2012 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.healthpol.2012.10.008
developing complications [3,4] and is the primary goal for T2DM treatments [5]. In international [6] and Swedish national guidelines [7,8], biguanides (metformin) is recommended as first line treatment, while sulfonylureas (SUs) and neutral protamine hagedorn (NPH) insulin are recommended for second line treatment in patients not adequately controlled on metformin. These are recommended to be used before newer treatments like thiazolidinediones (glitazones), dipeptidyl peptidase 4 (DPP-4) inhibitors and glucagon-like peptide-1 (GLP-1) analogues [7,8]. Utilization of glucose lowering therapies (GLT) has increased over time in many European countries, but at different rates and levels. In Sweden, the number of patients treated with GLT in 2010 was 372,000, an increase by 17%
208
B. Pettersson et al. / Health Policy 108 (2012) 207–215
since 2006. Number of patients treated with insulins was 183,000 and with oral GLT 265,000 in 2010, an increase by 15% and by 19% respectively since levels in 2006 [9]. In total 76,000 patients were treated with both insulins and oral GLT during 2010, either at the same time or due to a switch of therapy. In a comparison with ten European countries, the utilization of insulin in Sweden was highest up to the year 2000, while the utilization of oral GLT was on an average level [10]. The prevalence of T2DM has been estimated to around 3.5–4.5% in Sweden [11,12] and on average estimated to be around 3% among European countries, with variations between 1.7% in the Netherlands and 4.2% in Germany [13]. Earlier findings suggest, however, that this variation is overestimated and more due to variations in factors related to definitions, detection, registration among others [14]. The variation in prevalence does therefore not fully explain the variation in utilization of GLT. Another factor that has been suggested to explain the variation is differences in reimbursement schemes [10]. A new reimbursement scheme (RS) for GLT was introduced in Sweden 1 March 2010 [15] following a health technology assessment (HTA) of the glucose lowering therapies by the Swedish pricing and reimbursement agency, the Dental and Pharmaceutical Benefits Agency (Swedish acronym: TLV) [16]. The review was presented in a report of 2 December 2009 and it concluded that utilization of GLT in Sweden was cost-effective, with a few exceptions. Consequently, all glucose lowering therapies were assigned any of the following four different reimbursement statuses in the new RS; retained, restricted, no reimbursement or no reimbursement for new courses. TLV estimated that decisions made in the review could result in cost savings by at least 12 million Swedish Kronor [≈D1.3 million] (1 SEK ≈ D0.11, 5 November 2011 [17]) per annum [15]. An independent evaluation of the initial effects of the new RS for GLT, commissioned by the Department of Health, was published in a report in August 2010 and showed that there were fairly rapid treatment pattern changes following the new RS and that savings at constant volumes based on data for the first 6 months following the new RS were in line with those estimated by TLV [18]. The aim of this study is to compare utilization and costs of oral and insulin based GLT for type 2 diabetes mellitus in Sweden before and after the implementation of the new reimbursement scheme on March 1, 2010. 2. Methods 2.1. Setting The TLV, known before 2008 as “The Pharmaceutical Benefits Board” (Swedish acronym: LFN) was established in 2002 [16]. The new system and the task of the pricing and reimbursement agency has been described earlier [19]. In short, the task of the agency is to decide on which new prescription drugs should be included in the public pharmaceutical benefits scheme, and to review all drugs already included in the benefits. The primary purpose of the establishment of TLV was to replace the former system of “automatically” subsidizing drugs with market authorization towards a system that employed health technology
assessment (HTA) coupled with economic efficiency criteria (i.e. marginal cost-effectiveness) and evaluate this in relation to other ethical principles for prioritization in healthcare as the basis for reimbursement. The aim was to improve a rational and cost-effective use of medicines. From its foundation until 2010, after a priority plan for reviewing nearly 2000 drugs included in the national reimbursement scheme prior to the new system, the TLV was employing the same principles as the new reimbursement system [16,20,21]. The process and the sequence in the priority plan for the reviews is described elsewhere [21]. This plan was recently revised for a more flexible process and sequence for the reviews, when TLV announced a more dynamic approach with continuous reevaluation of the sequence of reviews based on the need of the healthcare system [16] in order to improve the efficiency of the reviews. 2.2. The new reimbursement scheme The TLV review of the GLT included all drugs in the pharmaceutical group A10 in the Anatomical Therapeutic Chemical (ATC 10) classification [22]. The reimbursement status of the products following the review is shown in Table 1. The new RS resulted in the assignment of four different reimbursement statuses to the products: retained, restricted, excluded, and excluded for new courses of treatment. 2.2.1. Retained 2.2.1.1. Insulins. Rapid-acting human insulin, rapid-acting insulin analogues, intermediate-acting human insulin (NPH), intermediate-acting insulin with rapid onset (2step). (generic metformin) (the 2.2.1.2. Oral. Biguanides patented metformin products had been excluded from the reimbursement system prior to the review), sulfonylureas (SU) (Mindiab, generic glimepiride, generic glibenclamide, Amaryl). 2.2.2. Restricted 2.2.2.1. Insulins. For type 2 diabetes patients the coverage of long-acting insulin analogues, insulin glargine and insulin detemir are restricted to patients for whom other insulin treatment is not sufficient to reach the treatment objectives due to recurring hypoglycemic episodes (the restrictions did not apply for patients with type 1 diabetes). Insulin detemir used to be restricted to patients with type 1 diabetes from 2004 until 19 June 2007, when the restriction was lifted [23]. rosiglitazone, pioglitazone, 2.2.2.2. Oral. Acarbose, sitagliptine, vildagliptine, repaglinide, exenatide and the combinations with metformin will only be reimbursed for patients who have first tried metformin, SUs or insulin, or if these treatments are not suitable. 2.2.3. Excluded 2.2.3.1. Oral. Nateglinide (Starlix) is no longer covered by the benefit scheme. Glibenclamide sold under the
B. Pettersson et al. / Health Policy 108 (2012) 207–215
209
Table 1 Status and categories in the new reimbursement scheme for glucose lowering therapies. Therapeutic group
Active substance
Product
The following products retain general reimbursement status Insulins All All (A10AB) Rapid-acting human insulin Rapid-acting insulin analogues All All (A10AB) All All (A10AC) Intermediate-acting human insulin (NPH) Intermediate-acting insulin with rapid onset (2-step) All All (A10AD) Oral Biguanides Metformin Generic metformin (A10BA02) Sulfonylureas (SU) Glipizide Mindiab (A10BB07) Sulfonylureas (SU) Glimepiride Generic glimepiride (A10BB12) Sulfonylureas (SU) Glimepiride Amaryl (A10BB12) Hypoglycemia drugs Glucagon Glucagon Novo Nordisk (H04AA01) The following products are restricted (for type 2 diabetes) to patients where other insulin treatment is not sufficient to reach the treatment objectives due to recurring hypoglycemic episodes. However, they shall be part of the pharmaceutical benefits system for type 1 diabetes Long-acting insulin analogues Insulin glargin Lantus (A10AE04) Long-acting insulin analogues Insulin detemir Levemir (A10AE05) The following products are restricted to patients who have first tried metformin, SUs or insulin, or where these options are not suitable Acarbose Glucobay (A10BF01) Alpha-glucosidase inhibitors Glitazones Rosiglitazone Avandia (A10BG02) Glitazones Pioglitazone Actos (A10BG03) DPP-4 inhibitors Sitagliptine Januvia (A10BH01) Vildagliptine Galvus (A10BH02) DPP-4 inhibitors Repaglinide NovoNorm (A10BX02) Meglitinides Exenatide Byetta (A10BX04) GLP-1 analogues Metformin and pioglitazone Avandamet (A10BD03) Combination drugs Combination drugs Metformin and pioglitazone Competact (A10BD05) Metformin and sitagliptine Janumet (A10BD07) Combination drugs Combination drugs Metformin and vildagliptine Eucreas (A10BD08) The following products lose their reimbursed status Glibenclamide Daonila (A10BB01) Sulfonylureas (SU) Meglitinides Nateglinide Starlix (A10BX02) Combination drugs Glimepiride and rosiglitazone Avaglim (A10BD04) The following products are not reimbursed for new courses of treatment Glibenclamide Glibenklamid Recip (A10BB01) Sulfonylureas (SU) a
Daonil was later given the status “not reimbursed for new courses of treatment”.
trade name of Daonil is excluded while generic products with glibenclamide are retained (but not for new courses of treatment). The combination of rosiglitazone and glimepiride is excluded from the reimbursement scheme, while the active substances as separate substances are retained. 2.2.4. Excluded for new courses of treatment 2.2.4.1. Oral. Glibenclamide (Glibenklamid Recip) is excluded for incident patients but retained for patients already treated with the drug. 2.3. Study design This is a quasi-experimental study [24] using segmented time series [25] design for analyzing the utilization and expenditures of GLT in Sweden. The study consisted of two time periods: before and after the new reimbursement scheme (intervention) was implemented. We studied 38 separate months in the period between February 2008 and March 2011 (25 months prior and 13 months after the intervention on March 1, 2010). The analyses were carried out on total effects from the intervention for oral and for insulin therapies respectively. Oral GLT were analyzed in total and for the following reimbursement status in the new RS: retained, restricted,
excluded, or excluded for new courses of treatment. Insulin therapies were analyzed in total and for reimbursement statuses ‘retained’ and ‘restricted’. 2.4. Data Drug dispensing data was collected from the Swedish Prescribed Drug Register held by the National Board of Health and Welfare [9] described elsewhere [9,26,27]. In short, the registry contains data for all dispensed prescriptions covering the whole population of Sweden (9.5 million inhabitants) but not drugs utilized in hospitals or purchased over the counter. It contains patient and drug specific data, reimbursement status, prescription and dispensation date and costs among others. It does not contain information on indications for treatment. 2.5. Statistical analysis and definitions Utilization of GLT is defined as volumes, expressed as numbers of patients/1000 inhabitants/month (TIM). Since the average period between dispensations for chronic medication in Sweden is 3 months the number of dispensations/TIM is not the same as the period prevalence of treated patients (which roughly can be estimated as three times as high as the number of patients dispensed
210
B. Pettersson et al. / Health Policy 108 (2012) 207–215
Insulin, patients/1000 inhabitants/month Level p=0.7100 Trend p=0.0021
Level p=0.5474 Trend p=0.0027
Time of intervenon
10
6 3 0 10
0
20
30
40
Month
B
6
Paents
Regress ion li ne
4 2 0 0
10
20
30
Month
A
Time of intervenon
9
8
Paents
Regress ion li ne
Level p=0.0422 Trend p=0.0007
Insulin restricted reimbursement
40 Paents/1000 inh./month
Paents/1000 inh./month
All reimbursed insulins
Paents/1000 Inh./month
Insulin retained reimbursement
Time of intervenon
3
2
1
0 0
C
10
20 Paents
30
40
Regress ion li ne
Fig. 1. (A–C) Segmented regression analyses of number of patients/1000 inh. (inhabitants)/month, all reimbursed insulins, insulins with retained reimbursement and insulins with restricted reimbursement before and after implementation of the intervention.
GLT/TIM). We used a cut off age level of ≥40 years at the time for dispensation as a proxy for patients with T2DM, since indications are not available in the registry and the onset of T2DM most often occurs in patients >40 years [1,12]. Costs were measured in Swedish Kronor (SEK) and all analyses were carried out for total costs (reimbursed expenditure and patient co-payment). We converted to Euros using current exchange rates; (1 SEK ≈ D0.11, 5th November 2011 [17]). Linear segmented regression analyses were used to analyze the changes in the levels and trends in utilization and costs before and after the intervention. Number of patients and costs were entered as dependent variables. A dichotomous indicator variable for the intervention was entered as independent variable in the regression models. The regression models allowed for a slope for the time period preceding the intervention and a slope and a level shift to account for the change after the intervention. Separate models were fitted for oral GLT and for insulin based GLT in total and for respective reimbursement category. Shifts in level (intercept) or slope related to the intervention with P < 0.05 was considered statistically significant. Total costs were computed one year before (March 2009 to February 2010) and one year after (March 2010 to February 2011) the new RS was implemented and results were used to analyze cost savings following the intervention. All analyses were performed using SPSS Statistical Package for Windows 16.0 and SAS institute Inc., North Carolina, USA, SAS version 9.1.3 and the graphs were produced in Excel-WindowsT.
3. Results 3.1. Utilization of glucose lowering therapies 3.1.1. Insulins Results from the regression analyses of number of patients/1000 inhabitants/month (TIM) treated with Insulin based GLT before and after the intervention are shown in Fig. 1A–C. There was an accelerated increasing trend in number of patients/TIM treated with insulins following the intervention (P = 0.0027) (Fig. 1A), for insulins with retained reimbursement as well as for restricted insulins (Fig. 1B and C). 3.1.2. Oral glucose lowering therapies Results from the regression analyses of number of patients/1000 inhabitants/month (TIM) treated with oral GLT before and after the intervention are shown in Fig. 2A–E. No significant differences in total number of patients treated with oral GLT were detected before versus after the intervention (Fig. 2A). There was a negative trend in number of patients/TIM treated with products with restricted reimbursement (P = 0.0035) (Fig. 2C) and a negative shift in products that were excluded from the reimbursement scheme following the intervention (P < 0.0001) (Fig. 2D). 3.2. Costs Total costs in the year following the intervention (March 2010 to February 2011) amounted to 1.1 billion SEK [≈D121 million], an increase by 58 million SEK [≈D6.4 million]
B. Pettersson et al. / Health Policy 108 (2012) 207–215
Paents/1000 Inh./month
All reimbursed oral GLT
Oral glucose lowering therapies, patients/1000 inhabitants/month
211
Level p=0.1610 Trend p=0.6084
Time of intervenon
15 10 5 0 10
0
20
30
40
Month
A
Paents
Level p=0.1255 Trend p=0.2625
Time of intervenon
8 4 0 0
10
20
30
40
Oral restricted reimbursement Paents/1000 Inh./month
Paents/1000 Inh./month
Oral retained reimbursement 12
Regress ion li ne
Time of intervenon
2 1,6 1,2 0,8 0,4 0 0
10
Oral no reimbursement Paents/1000 Inh./month
C
Regress ion li ne
Level p=<.0001 Trend p=0.1603
Time of intervenon
0,04 0,03 0,02 0,01 0 0
10
20
30
40
Paents
Paents
40
Regress ion li ne
Regress ion li ne
Level p=0.5867 Trend p=0.1880
Time of intervenon
2,5 2 1,5 1 0,5 0 0
10
Month
D
30
Oral no reimbursement for new courses Paents/1000 Inh./month
Paents
20 Month
Month
B
Level p=0.1348 Trend p=0.0035
20
30
40
Month
E
Paents
Regress ion li ne
Fig. 2. (A–E) Segmented regression analyses of number of patients/1000 inh. (inhabitants)/month treated with oral glucose lowering therapies: all reimbursed, those with retained-, those with restricted-, those with no- and those with no reimbursement for new courses before and after implementation of the intervention.
compared to a reference period 1 year before the intervention (March 2009 to February 2010). Total costs for insulins increased by 60 million SEK [≈D6.6 million] (7%), while costs for oral glucose lowering therapies decreased by 2 million SEK [≈D0.22 million]) (−0.07%) (data not shown). 3.2.1. Insulins Regression analyses of costs showed there was an accelerated increasing trend in total costs (Fig. 3A), both for insulins with retained reimbursement and for insulins with restricted reimbursement) following the intervention. There was an increase by 100 SEK [≈D 11]/TIM, which of 60 SEK[≈D6.6]/TIM for products with retained reimbursement and 40 SEK[≈D4.4]/TIM for restricted products (data not shown). 3.2.2. Oral glucose lowering therapies There were negative trends in costs for oral GLT in total (Fig. 3B) and for drugs which were retained or restricted in
the new RS and a negative level shift in products excluded from the RS (P < 0.0001)(data not shown). 4. Discussion Number of patients treated with insulin analogues increased at a higher rate following the decision to restrict reimbursement among patients with T2DM. There are several possible explanation to the increasing number of patients treated with restricted insulins following the new RS, but also the overall accelerated increasing trend of insulins. One explanation might be that the intervention with reimbursement restrictions only affected a subset of the insulin-treated patients–patients with T2DM who need insulin but fail to reach the treatment objectives with alternatives other than the long-acting insulin analogues. Another explanation might be due to factors influencing the use of the insulin analogues among patients with T1DM
212
B. Pettersson et al. / Health Policy 108 (2012) 207–215
Level p=0.3094 Trend p=0.0014
Cost/1000 Inh./month
Total costs insulins Time of intervenon
12000 8000 4000 0 0
10
20
30
40
Month
A
Cost
Regress ion li ne
Level p=0.2738 Trend p=0.0177
Cost/1000 Inh./month
Total costs oral GLT Time of intervenon
4000 3000 2000 1000 0 0
B
10
20
30
40
Month Cost
Regress ion li ne
Fig. 3. (A and B) Segmented regression analyses of total costs/1000 inh. (inhabitants)/month in Swedish Crown (SEK), for insulin and oral GLT before and after implementation of the intervention.
and that are probably more important than changes affecting the subset of patients with T2DM treated with the restricted insulins, even if this study used a cut off level of >40 years (see Section 2.5 for explanation) to limit the number of patients with type 1 diabetes mellitus (T1DM) in the analysis. Yet another explanation could be that the utilization prior to the intervention was restrained by alarm reports related to insulin analogues and risk of developing breast cancer [28], which was published during the period and that this effect waned in the later study period. Increased marketing activities, induced by the new RS, might have created an opportunity for the pharmaceutical companies to inform (and influence) the physicians and might be another explanation behind the increasing number of patients. Furthermore, the effect might be due to unclear reimbursement status for the existent insulin analogues, prior to the new RS. The reimbursement status differed between the two existing products and the reimbursement status had been changed earlier. However, the three last observations deviated considerably (Fig. 1C) and when they were excluded from the analysis the trend was similar to the trend before the intervention and therefore an alternative interpretation could be that the utilization of restricted insulins did not change. The new RS did not affect total number of patients treated with oral GLT. The effect on utilization of oral GLT that were restricted was minor, which could be explained by prescription adherence to existing restrictive guidelines (patients should have tried metformin, sulfonylureas or insulins first) has been reported to be about 90% already prior to the new RS [18]. In addition, concerns on adverse
effects related to rosiglitazone [29] were reported during the period before the implementation and was partly the motivation for TLV’s decision to restrict reimbursement of glitazones. The utilization of glibenclamide (marketed under the trade name of Daonil) and nateglinid (both excluded in the new RS initially) decreased steadily prior to the intervention, due to increased utilization of generic products, enhanced mainly by generic substitution and formulary guidelines both on the national and the regional level [7,8]. The intervention seems to have accelerated this trend slightly. 4.1. Utilization of glucose lowering treatment compared to other countries The utilization of insulin based treatment in Sweden widely exceeded the use in ten other countries, and as much as twice as high in Sweden compared to Denmark in subjects ≥45 years of age, but not in patients <45 years [10]. This might indicate a wider utilization of insulin-based treatment for T2DM in Sweden compared to Denmark and the difference might be explained by differences in reimbursement system. In Sweden insulins remained reimbursed for 100% while the reimbursement of oral GLT was reduced in 1997, when a new pharmaceutical benefits scheme was introduced. In Denmark they were also 100% reimbursed up to March 1, 2000 when they were included in a system where all reimbursable products have an equal status in terms of reimbursement rate [30]. 4.2. Costs and savings Cost differences between two time periods might depend on underlying trends in volume, price and product mix effects (assortment, strength, doses and pack sizes), besides effect of the intervention. The total effect on savings includes market dynamic effects related to prices due to increased competition of generics in combination with other market regulations for non patented drugs introduced by TLV, i.e. generic substitution, substitution mechanisms by pharmacies and differences in assortment and might not be directly associated with or be a consequence of the review. The generic substitution, has been shown to be an effective policy to reduce expenditures [31] and it was introduced in 2004, i.e. prior to the new RS. The dynamic effect from the intervention is that the price of glibenclamide marketed as Daonil was reduced to the level of generics available on the market, which contributed to the reduction of costs for oral GLT following the intervention. Total costs for insulins increased by 60 million SEK [≈D6.6 million] while costs for oral GLT decreased by 2 million SEK [≈D0.22 million] following the new RS. Price per day for glucose lowering treatment varies widely between insulins (7–21 SEK) [≈D0.8–2.3] [15], where the daily price depends on the weight of the patient and the dose required to achieve an effect. It also varies considerably for oral GLT (1–15 SEK) [≈D0.1–1.7], but less than and at a lower range than that for insulins. The utilization of oral GLT consists furthermore mainly of lower priced generic products and
B. Pettersson et al. / Health Policy 108 (2012) 207–215
therefore the costs for oral GLT are only a small part (about 20%) of the total costs for GLT and therefore changes in prices for oral GLT have limited effects on cost savings for GLT. In the review of GLT, the TLV concluded that the overall utilization of glucose lowering drugs was considered cost effective already prior to the review, and therefore the new RS was not designed to alter the overall utilization. The perception that the utilization of GLT is cost effective can however be subject of discussion, since the proportion of utilization of insulin based GLT for T2DM might be excessive in Sweden compared to other countries [10]. Furthermore, the cost-effectiveness of the long-acting insulin analogues compared to other insulins is mainly driven by the higher quality-of-life due to fewer number of episodes of hypoglycemia and/or the possibility to increase control of hyperglycemia without introducing more episodes of hypoglycemia in a subset of patients [32]. This is however mainly studied in T1DM [33] and can be questioned both on the basis on how to evaluate the loss of quality-oflife due to hypoglycemia, and an increasing use in T2DM where the doses of insulin (and thus the costs) tends to be higher in cases with higher weights and relative insulin-resistance.
213
was included at the end of the series but not at the beginning and thus the analysis could be influenced by seasonal fluctuations since February is a short month and therefore includes fewer dispensations than other months. Another option would have been to try to choose matching time periods surrendering the intervention but that would on the other hand have limited the total data set available for analysis. Linear regression models might not be appropriate because the linearity assumptions might be violated by phenomena of adaptation after abrupt changes in the system. However, we tested the normality assumptions on the residuals of all regressions with Shapiro–Wilk test. The test could not show significant non-normality in any of the residuals from the regressions (P-values > 0.05). The drug registry contains individual based level data on purchased drugs, which is ideal for evaluating impact of implementation of nationwide policies, since it includes all drugs dispensed and therefore provides possibilities to assess and estimate a wide range of measures. In our analyses we used only aggregated data, which could be a limitation, since data on new patients is a sensitive and immediate measure of adherence to interventions. 4.4. Summary of the RS and implication for the future
4.3. Strengths and limitations Time series design is a strong quasi-experimental alternative to randomization. The external validity of the design may be high as it occurs in a natural setting and segmented regression is a robust technique to test immediate and sustained changes associated with the intervention while controlling for trends [25,34]. Internal validity in time series analyses, as in other epidemiological techniques can be influenced by confounding factors. For instance, it could be threatened by historical events, i.e. some other event occurring at the same time as the intervention and so could explain the pattern of change over time. Time series design avoids however some of the limitations in before and after design, which is often employed to evaluate effects of interventions, since before and after design might produce results that are affected by either of the time points being atypical due to concurrent other events. Even though the time series design does not eliminate all problems neither of the influence of time periods studied, nor in interpreting changes in turnover rate, the extended time period strengthen the ability to attribute changes to the intervention [34]. Even though the prevalence of diabetes, and thus the aggregated need for treatment, is not susceptible to obvious seasonal changes, other factors might influence the actual amount of drugs dispensed over a given time period. The use of months as the time unit instead of longer time periods introduces methodological challenges such as seasonal variations including holidays, difference in number of days and especially working days with open pharmacies and/or access to prescribing physicians etc. The results are therefore sensitive to chosen months and years and have to be assessed cautiously. In this study, the starting point of the observations following the intervention was March 2010, and the endpoint was March 2011, and thus February
Enforcement has a stronger and more direct effect on utilization than soft demand policies such as education and guidelines, however restrictive demand side polices might trigger unexpected behaviour [35]. In this context the new RS for GLT was designed in line with other policies as generic substitution and harmonized with ongoing trends, along with national guidelines [8], which might have reduced the risks of the RS leading to unintended effects. In the review of GLT, the TLV concluded that the overall utilization of glucose lowering drugs was considered cost-effective already prior to the review, and therefore the new RS was not designed to alter the overall utilization. By contrast, the review of lipid modifying therapies (LMT) [36], was aiming at reducing the utilization of lower doses of patented statins towards equivalent doses of generic statins, which was considered to be more cost-effective. The LMT review with subsequent changes in reimbursement status of the LMT was implemented on June 1, 2009 [37], where the drugs were either retained, restricted or excluded from the RS. In products that were restricted in the new RS for LMT, an upward shift in level was detected following the new RS [36], while the accelerated trend increased for the restricted insulins in the new RS for GLT. For the restricted oral GLT the effect was that the trend was decreasing prior to the intervention, even if it was slightly accelerated following the new RS. The effects on utilization of restricted products were therefore non-conclusive, but they lead to possibly unintended effects for LMT and for insulins. The effect on overall utilization and for products which were retained or excluded was similar for oral GLT and for LMT, with expected rapid downward shifts. The effects on total utilization and on products with retained reimbursement were not affected following the RS for LMT and for oral GLT.
214
B. Pettersson et al. / Health Policy 108 (2012) 207–215
The effect on costs and expected saving differed for LMT, oral and insulin based GLT respectively, which could be dependent on timing of the review and the subsequent new RS. In the market for LMT, there was a large share of utilization of drugs that were still patent protected, while a major part of the utilization of oral GLT is in the generic market, therefore potential and realized savings were more substantial following the new RS for LMT, while costs even increased for insulins. Changes in reimbursement schemes to steer use of pharmaceuticals should be carefully evaluated with respect to timing and design, in order to avoid unintended and unwanted effects; this should be weighted against potential savings. Softer demand-side policies might be a better option to steer more precisely towards a cost-effective use of medicines. Independent follow-up is needed to assess the impact of new reimbursement decisions. At the same time, the TLV should have a follow-up plan to make their own assessments of the impact from their decisions, since our findings show that unintended effects might emerge. 5. Conclusions The new reimbursement scheme had a minor overall effect on utilization and costs of oral glucose lowering therapies. Despite more restrictive reimbursement for patients with T2DM, the utilization of insulin based glucose lowering therapies and related costs increased even faster after the intervention. References [1] International Diabetes Federation: IDF Diabetes Atlas. 4 edition Brussels: International Diabetes Federation, 2009. [2] Conroy RM, Pyorala K, Fitzgerald AP, Sans S, Menotti A, De Backer G, De Bacquer D, Ducimetiere P, Jousilahti P, Keil U, Njolstad I, Oganov RG, Thomsen T, Tunstall-Pedoe H, Tverdal A, Wedel H, Whincup P, Wilhelmsen L, Graham IM. Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE project. European Heart Journal 2003;24:987–1003. [3] Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group. Lancet 1998;352: 837–53. [4] Stratton IM, Cull CA, Adler AI, Matthews DR, Neil HA, Holman RR. Additive effects of glycaemia and blood pressure exposure on risk of complications in type 2 diabetes: a prospective observational study (UKPDS 75). Diabetologia 2006;49:1761–9. [5] IDF Clinical Guidelines Task Force: Global guideline for Type 2 diabetes. Brussels: International Diabetes Federation, 2005. [6] Nathan DM, Buse JB, Davidson MB, Ferrannini E, Holman RR, Sherwin R, Zinman B. Medical management of hyperglycemia in type 2 diabetes: a consensus algorithm for the initiation and adjustment of therapy: a consensus statement of the American Diabetes Association and the European Association for the Study of Diabetes. Diabetes Care 2009;32:193–203. [7] Medical Products Agency. Läkemedelsbehandling vid typ 2-diabetes–ny rekommendation;(English translation: Pharmaceutical treatment for type 2 diabetes- new recommendation). http://www.lakemedelsverket.se/upload/halso-och-sjukvard/ behandlingsrekommendationer/L%c3%a4kemedelsbehandling%20 vid%20typ%202-diabetes rek.pdf; 2010. [8] National board for health and welfare. Nationella riktlinjer för diabetesvården 2010–Stöd för styrning och ledning (English translation: National guidelines for diabetes care 2010). http://www. socialstyrelsen.se/publikationer2010/2010-2-2/Sidor/default.aspx; 2010.
[9] The National Board of Health and Welfare. The Swedish Drug Registry. http://www.socialstyrelsen.se/register/halsodataregister/ lakemedelsregistret; 2011. [10] Melander A, Folino-Gallo P, Walley T, Schwabe U, Groop PH, Klaukka T, Vallano A, Laporte JR, Gallego MR, Schiappa M, Roder M, Kampmann JP, de Swaef A, Aberg M, Mansson NO, Lindblad U. Utilisation of antihyperglycaemic drugs in ten European countries: different developments and different levels. Diabetologia 2006;49:2024–9. [11] Jansson SP, Andersson DK, Svardsudd K. Prevalence and incidence rate of diabetes mellitus in a Swedish community during 30 years of follow-up. Diabetologia 2007;50:703–10. [12] Ringborg A, Lindgren P, Martinell M, Yin DD, Schon S, Stalhammar J. Prevalence and incidence of Type 2 diabetes and its complications 1996–2003 – estimates from a Swedish population-based study. Diabetic Medicine 2008;25:1178–86. [13] Jonsson B. Revealing the cost of Type II diabetes in Europe. Diabetologia 2002;45:S5–12. [14] Fleming DM, Schellevis FG, Van Casteren V. The prevalence of known diabetes in eight European countries. European Journal of Public Health 2004;14:10–4. [15] The Dental and Pharmaceutical Benefits Agency (TLV). The review of medicines used for treating diabetes – A summary. http://www.tlv.se/Upload/Genomgangen/summary-diabetes.pdf; 2010. [16] The Dental and Pharmaceutical Benefits Agency (TLV). The Pharmaceutical reviews. http://www.tlv.se/in-english-old/in-english/; 2011. [17] Sveriges Riksbank Exchange rates. http://www.riksbank.com/ templates/Page.aspx?id=17182; 2011. [18] The NEPI Foundation – The Swedish Network for Pharmacoepidemiology. Diabetesmedel – effekter av Tandvårds – och läkemedelsförmånsverkets genomgång december 2009 (English translation: Diabetes treatments—effects of the TLV review December 2009). http://www.nepi.net/res/dokument/ NEPI-rapport-diabetesmedel-100831.pdf; 2010. [19] Wettermark B, Godman B, Andersson K, Gustafsson LL, Haycox A, Bertele V. Recent national and regional drug reforms in Sweden: implications for pharmaceutical companies in Europe. PharmacoEconomics 2008;26:537–50. [20] Sjögren E. Reasonable drugs – making decisions with ambiguous knowledge. The economic research institute. Stockholm: Stockholm School of Economics; 2006. [21] Wettermark B, Godman B, Neovius M, Hedberg N, Mellgren TO, Kahan T. Initial effects of a reimbursement restriction to improve the cost-effectiveness of antihypertensive treatment. Health Policy 2010;94:221–9. [22] WHO Collaborating Centre for Drug Statistics Methodology. Guidelines for ATC classification and DDD assignment 2010. http://www.whocc.no/atc ddd publications/guidelines/; 2009. [23] The Dental and Pharmaceutical Benefits Agency (TLV). Reimdecision, Levemir. http://www.tlv.se/Upload/ bursement Beslut 2007/BES 070618 levemir.pdf; 2007. [24] Grimshaw J, Campbell M, Eccles M, Steen N. Experimental and quasi-experimental designs for evaluating guideline implementation strategies. Family Practice 2000;17(Suppl. 1):S11–6. [25] Wagner AK, Soumerai SB, Zhang F, Ross-Degnan D. Segmented regression analysis of interrupted time series studies in medication use research. Journal of Clinical Pharmacy and Therapeutics 2002;27:299–309. [26] Wettermark B, Hammar N, Fored CM, Leimanis A, Otterblad Olausson P, Bergman U, Persson I, Sundstrom A, Westerholm B, Rosen M. The new Swedish Prescribed Drug Register – opportunities for pharmacoepidemiological research and experience from the first six months. Pharmacoepidemiology and Drug Safety 2007;16: 726–35. [27] Furu K, Wettermark B, Andersen M, Martikainen JE, Almarsdottir AB, Sorensen HT. The Nordic countries as a cohort for pharmacoepidemiological research. Basic and Clinical Pharmacology and Toxicology 2009;106:86–94. [28] Hemkens LG, Grouven U, Bender R, Gunster C, Gutschmidt S, Selke GW, Sawicki PT. Risk of malignancies in patients with diabetes treated with human insulin or insulin analogues: a cohort study. Diabetologia 2009;52:1732–44. [29] European Medicines Agency. European Medicines Agency update on ongoing benefit-risk review of Avandia, Avandamet and Avaglim. http://www.ema.europa.eu/ema/index.jsp?curl=pages/news and events/news/2010/07/news detail 001062.jsp&murl=menus/news and events/news and events.jsp&mid=WC0b01ac058004d5c1& jsenabled=true; 2010.
B. Pettersson et al. / Health Policy 108 (2012) 207–215 [30] The Danish association of the pharmaceutical industry. Pricing and reimbursement in Denmark. http://parno1.ipapercms. dk/LIF/Notater/PricingandreimbursementinDenmark/; 2007. [31] Andersson Swedish Pharmaceutical Benefit Reforms – analyses of implementation, pharmaceutical sales patterns and expenditures. Department of Public health and Community Medicine. Gothenburg: University of Gothenburg; 2006. [32] Cameron CG, Bennett HA. Cost-effectiveness of insulin analogues for diabetes mellitus. Canadian Medical Association Journal 2009;180:400–7. [33] Tran KBS, Li H, Cimon K, Daneman D, Simpson SH, Campbell K. Long-acting insulin analogues for diabetes mellitus: metaanalysis of clinical outcomes and assessment of cost-effectiveness technology Report number 92. Ottawa: Canadian Agency for Drugs and Technologies in Health; 2007.
215
[34] Polit D, Tatano Beck C, Nursing Research: Generating. Assessing evidence for nursing practice. Philadelphia: Lippincott Williams & Wilkins; 2008. [35] Wettermark B, Godman B, Jacobsson B, Haaijer-Ruskamp FM. Soft regulations in pharmaceutical policy making: an overview of current approaches and their consequences. Applied Health Economics and Health Policy 2009;7:137–47. [36] Pettersson B, Hoffmann M, Wändell P, Levin L-Å. Utilization and costs of lipid modifying therapies following health technology assessment for the new reimbursement scheme in Sweden. Health Policy 2011. [37] The Dental and Pharmaceutical Benefits Agency (TLV). The review of medicines for treating lipid disorders. http://www.tlv.se/Upload/Genomgangen/review-lipid-disorder.pdf; 2009.
Contents lists available at SciVerse ScienceDirect
Health Policy journal homepage: www.elsevier.com/locate/healthpol
Utilization and costs of glucose lowering therapies following health technology assessment for the new reimbursement scheme in Sweden Billie Pettersson a,∗ , Mikael Hoffmann b , David Andersson c , Per Wändell d , Lars-Åke Levin a a b c d
The Center for Medical Technology Assessment, CMT Linköping University, 581 83 Linköping, Sweden The NEPI Foundation – The Swedish Network for Pharmaco-epidemiology, 581 91 Linköping, Sweden The Department of Management and Engineering, Linköping University, 581 83 Linköping, Sweden Center for Family and Community Medicine, Karolinska Institute, 141 83 Huddinge, Sweden
a r t i c l e
i n f o
Article history: Received 21 November 2011 Received in revised form 4 October 2012 Accepted 18 October 2012 Keywords: Pricing and reimbursement policy Glucose lowering therapies Drug utilization Health technology assessment Health economics
a b s t r a c t Objectives: A new reimbursement scheme (RS) for glucose lowering therapies (GLT) was implemented in Sweden on March 1, 2010. Products on the market were retained, restricted, excluded or excluded for new courses in the new RS. The aim of this study was to compare utilization and costs of GLT for type 2 diabetes Mellitus (T2DM) before and after the implementation of the changed RS. Methods: This was a quasi-experimental study using data on dispensed GLT and costs from a database on dispensed individual based prescriptions in Sweden. Segmented regression analyses were used to assess utilization and costs. Results: Following the changed reimbursement status, there was an accelerated increasing trend in number of patients treated with restricted (P = 0.0007) or retained (P = 0.0021) insulins, as well as in costs for insulin based GLT (P = 0.0014). No impact was detected in the total number of patients treated with oral GLT, but a slightly negative trend in total costs for oral GLT was detected following the intervention (P = 0.0177). Conclusions: The new reimbursement scheme had a minor impact on utilization and costs of oral GLT. Despite restricted reimbursement for patients with T2DM, the utilization of insulin based GLT and related costs increased faster following the intervention. © 2012 Elsevier Ireland Ltd. All rights reserved.
1. Background Type 2 diabetes mellitus (T2DM) is a progressive disease associated with macrovascular and microvascular complications [1]. The risk of developing cardiovascular disease (CVD) is increased 2–4 times in subjects with T2DM independently of other concomitant risk factors [2]. Glycemic control in patients with T2DM reduces the risk of
∗ Corresponding author at: Pontongränd 13, 183 68 Täby, Sweden. Tel.: +46 768 85 00 10; fax: +46 8 626 14 22. E-mail addresses: Billie [email protected] (B. Pettersson), [email protected] (M. Hoffmann), [email protected] (D. Andersson), [email protected] (P. Wändell), [email protected] (L.-Å. Levin). 0168-8510/$ – see front matter © 2012 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.healthpol.2012.10.008
developing complications [3,4] and is the primary goal for T2DM treatments [5]. In international [6] and Swedish national guidelines [7,8], biguanides (metformin) is recommended as first line treatment, while sulfonylureas (SUs) and neutral protamine hagedorn (NPH) insulin are recommended for second line treatment in patients not adequately controlled on metformin. These are recommended to be used before newer treatments like thiazolidinediones (glitazones), dipeptidyl peptidase 4 (DPP-4) inhibitors and glucagon-like peptide-1 (GLP-1) analogues [7,8]. Utilization of glucose lowering therapies (GLT) has increased over time in many European countries, but at different rates and levels. In Sweden, the number of patients treated with GLT in 2010 was 372,000, an increase by 17%
208
B. Pettersson et al. / Health Policy 108 (2012) 207–215
since 2006. Number of patients treated with insulins was 183,000 and with oral GLT 265,000 in 2010, an increase by 15% and by 19% respectively since levels in 2006 [9]. In total 76,000 patients were treated with both insulins and oral GLT during 2010, either at the same time or due to a switch of therapy. In a comparison with ten European countries, the utilization of insulin in Sweden was highest up to the year 2000, while the utilization of oral GLT was on an average level [10]. The prevalence of T2DM has been estimated to around 3.5–4.5% in Sweden [11,12] and on average estimated to be around 3% among European countries, with variations between 1.7% in the Netherlands and 4.2% in Germany [13]. Earlier findings suggest, however, that this variation is overestimated and more due to variations in factors related to definitions, detection, registration among others [14]. The variation in prevalence does therefore not fully explain the variation in utilization of GLT. Another factor that has been suggested to explain the variation is differences in reimbursement schemes [10]. A new reimbursement scheme (RS) for GLT was introduced in Sweden 1 March 2010 [15] following a health technology assessment (HTA) of the glucose lowering therapies by the Swedish pricing and reimbursement agency, the Dental and Pharmaceutical Benefits Agency (Swedish acronym: TLV) [16]. The review was presented in a report of 2 December 2009 and it concluded that utilization of GLT in Sweden was cost-effective, with a few exceptions. Consequently, all glucose lowering therapies were assigned any of the following four different reimbursement statuses in the new RS; retained, restricted, no reimbursement or no reimbursement for new courses. TLV estimated that decisions made in the review could result in cost savings by at least 12 million Swedish Kronor [≈D1.3 million] (1 SEK ≈ D0.11, 5 November 2011 [17]) per annum [15]. An independent evaluation of the initial effects of the new RS for GLT, commissioned by the Department of Health, was published in a report in August 2010 and showed that there were fairly rapid treatment pattern changes following the new RS and that savings at constant volumes based on data for the first 6 months following the new RS were in line with those estimated by TLV [18]. The aim of this study is to compare utilization and costs of oral and insulin based GLT for type 2 diabetes mellitus in Sweden before and after the implementation of the new reimbursement scheme on March 1, 2010. 2. Methods 2.1. Setting The TLV, known before 2008 as “The Pharmaceutical Benefits Board” (Swedish acronym: LFN) was established in 2002 [16]. The new system and the task of the pricing and reimbursement agency has been described earlier [19]. In short, the task of the agency is to decide on which new prescription drugs should be included in the public pharmaceutical benefits scheme, and to review all drugs already included in the benefits. The primary purpose of the establishment of TLV was to replace the former system of “automatically” subsidizing drugs with market authorization towards a system that employed health technology
assessment (HTA) coupled with economic efficiency criteria (i.e. marginal cost-effectiveness) and evaluate this in relation to other ethical principles for prioritization in healthcare as the basis for reimbursement. The aim was to improve a rational and cost-effective use of medicines. From its foundation until 2010, after a priority plan for reviewing nearly 2000 drugs included in the national reimbursement scheme prior to the new system, the TLV was employing the same principles as the new reimbursement system [16,20,21]. The process and the sequence in the priority plan for the reviews is described elsewhere [21]. This plan was recently revised for a more flexible process and sequence for the reviews, when TLV announced a more dynamic approach with continuous reevaluation of the sequence of reviews based on the need of the healthcare system [16] in order to improve the efficiency of the reviews. 2.2. The new reimbursement scheme The TLV review of the GLT included all drugs in the pharmaceutical group A10 in the Anatomical Therapeutic Chemical (ATC 10) classification [22]. The reimbursement status of the products following the review is shown in Table 1. The new RS resulted in the assignment of four different reimbursement statuses to the products: retained, restricted, excluded, and excluded for new courses of treatment. 2.2.1. Retained 2.2.1.1. Insulins. Rapid-acting human insulin, rapid-acting insulin analogues, intermediate-acting human insulin (NPH), intermediate-acting insulin with rapid onset (2step). (generic metformin) (the 2.2.1.2. Oral. Biguanides patented metformin products had been excluded from the reimbursement system prior to the review), sulfonylureas (SU) (Mindiab, generic glimepiride, generic glibenclamide, Amaryl). 2.2.2. Restricted 2.2.2.1. Insulins. For type 2 diabetes patients the coverage of long-acting insulin analogues, insulin glargine and insulin detemir are restricted to patients for whom other insulin treatment is not sufficient to reach the treatment objectives due to recurring hypoglycemic episodes (the restrictions did not apply for patients with type 1 diabetes). Insulin detemir used to be restricted to patients with type 1 diabetes from 2004 until 19 June 2007, when the restriction was lifted [23]. rosiglitazone, pioglitazone, 2.2.2.2. Oral. Acarbose, sitagliptine, vildagliptine, repaglinide, exenatide and the combinations with metformin will only be reimbursed for patients who have first tried metformin, SUs or insulin, or if these treatments are not suitable. 2.2.3. Excluded 2.2.3.1. Oral. Nateglinide (Starlix) is no longer covered by the benefit scheme. Glibenclamide sold under the
B. Pettersson et al. / Health Policy 108 (2012) 207–215
209
Table 1 Status and categories in the new reimbursement scheme for glucose lowering therapies. Therapeutic group
Active substance
Product
The following products retain general reimbursement status Insulins All All (A10AB) Rapid-acting human insulin Rapid-acting insulin analogues All All (A10AB) All All (A10AC) Intermediate-acting human insulin (NPH) Intermediate-acting insulin with rapid onset (2-step) All All (A10AD) Oral Biguanides Metformin Generic metformin (A10BA02) Sulfonylureas (SU) Glipizide Mindiab (A10BB07) Sulfonylureas (SU) Glimepiride Generic glimepiride (A10BB12) Sulfonylureas (SU) Glimepiride Amaryl (A10BB12) Hypoglycemia drugs Glucagon Glucagon Novo Nordisk (H04AA01) The following products are restricted (for type 2 diabetes) to patients where other insulin treatment is not sufficient to reach the treatment objectives due to recurring hypoglycemic episodes. However, they shall be part of the pharmaceutical benefits system for type 1 diabetes Long-acting insulin analogues Insulin glargin Lantus (A10AE04) Long-acting insulin analogues Insulin detemir Levemir (A10AE05) The following products are restricted to patients who have first tried metformin, SUs or insulin, or where these options are not suitable Acarbose Glucobay (A10BF01) Alpha-glucosidase inhibitors Glitazones Rosiglitazone Avandia (A10BG02) Glitazones Pioglitazone Actos (A10BG03) DPP-4 inhibitors Sitagliptine Januvia (A10BH01) Vildagliptine Galvus (A10BH02) DPP-4 inhibitors Repaglinide NovoNorm (A10BX02) Meglitinides Exenatide Byetta (A10BX04) GLP-1 analogues Metformin and pioglitazone Avandamet (A10BD03) Combination drugs Combination drugs Metformin and pioglitazone Competact (A10BD05) Metformin and sitagliptine Janumet (A10BD07) Combination drugs Combination drugs Metformin and vildagliptine Eucreas (A10BD08) The following products lose their reimbursed status Glibenclamide Daonila (A10BB01) Sulfonylureas (SU) Meglitinides Nateglinide Starlix (A10BX02) Combination drugs Glimepiride and rosiglitazone Avaglim (A10BD04) The following products are not reimbursed for new courses of treatment Glibenclamide Glibenklamid Recip (A10BB01) Sulfonylureas (SU) a
Daonil was later given the status “not reimbursed for new courses of treatment”.
trade name of Daonil is excluded while generic products with glibenclamide are retained (but not for new courses of treatment). The combination of rosiglitazone and glimepiride is excluded from the reimbursement scheme, while the active substances as separate substances are retained. 2.2.4. Excluded for new courses of treatment 2.2.4.1. Oral. Glibenclamide (Glibenklamid Recip) is excluded for incident patients but retained for patients already treated with the drug. 2.3. Study design This is a quasi-experimental study [24] using segmented time series [25] design for analyzing the utilization and expenditures of GLT in Sweden. The study consisted of two time periods: before and after the new reimbursement scheme (intervention) was implemented. We studied 38 separate months in the period between February 2008 and March 2011 (25 months prior and 13 months after the intervention on March 1, 2010). The analyses were carried out on total effects from the intervention for oral and for insulin therapies respectively. Oral GLT were analyzed in total and for the following reimbursement status in the new RS: retained, restricted,
excluded, or excluded for new courses of treatment. Insulin therapies were analyzed in total and for reimbursement statuses ‘retained’ and ‘restricted’. 2.4. Data Drug dispensing data was collected from the Swedish Prescribed Drug Register held by the National Board of Health and Welfare [9] described elsewhere [9,26,27]. In short, the registry contains data for all dispensed prescriptions covering the whole population of Sweden (9.5 million inhabitants) but not drugs utilized in hospitals or purchased over the counter. It contains patient and drug specific data, reimbursement status, prescription and dispensation date and costs among others. It does not contain information on indications for treatment. 2.5. Statistical analysis and definitions Utilization of GLT is defined as volumes, expressed as numbers of patients/1000 inhabitants/month (TIM). Since the average period between dispensations for chronic medication in Sweden is 3 months the number of dispensations/TIM is not the same as the period prevalence of treated patients (which roughly can be estimated as three times as high as the number of patients dispensed
210
B. Pettersson et al. / Health Policy 108 (2012) 207–215
Insulin, patients/1000 inhabitants/month Level p=0.7100 Trend p=0.0021
Level p=0.5474 Trend p=0.0027
Time of intervenon
10
6 3 0 10
0
20
30
40
Month
B
6
Paents
Regress ion li ne
4 2 0 0
10
20
30
Month
A
Time of intervenon
9
8
Paents
Regress ion li ne
Level p=0.0422 Trend p=0.0007
Insulin restricted reimbursement
40 Paents/1000 inh./month
Paents/1000 inh./month
All reimbursed insulins
Paents/1000 Inh./month
Insulin retained reimbursement
Time of intervenon
3
2
1
0 0
C
10
20 Paents
30
40
Regress ion li ne
Fig. 1. (A–C) Segmented regression analyses of number of patients/1000 inh. (inhabitants)/month, all reimbursed insulins, insulins with retained reimbursement and insulins with restricted reimbursement before and after implementation of the intervention.
GLT/TIM). We used a cut off age level of ≥40 years at the time for dispensation as a proxy for patients with T2DM, since indications are not available in the registry and the onset of T2DM most often occurs in patients >40 years [1,12]. Costs were measured in Swedish Kronor (SEK) and all analyses were carried out for total costs (reimbursed expenditure and patient co-payment). We converted to Euros using current exchange rates; (1 SEK ≈ D0.11, 5th November 2011 [17]). Linear segmented regression analyses were used to analyze the changes in the levels and trends in utilization and costs before and after the intervention. Number of patients and costs were entered as dependent variables. A dichotomous indicator variable for the intervention was entered as independent variable in the regression models. The regression models allowed for a slope for the time period preceding the intervention and a slope and a level shift to account for the change after the intervention. Separate models were fitted for oral GLT and for insulin based GLT in total and for respective reimbursement category. Shifts in level (intercept) or slope related to the intervention with P < 0.05 was considered statistically significant. Total costs were computed one year before (March 2009 to February 2010) and one year after (March 2010 to February 2011) the new RS was implemented and results were used to analyze cost savings following the intervention. All analyses were performed using SPSS Statistical Package for Windows 16.0 and SAS institute Inc., North Carolina, USA, SAS version 9.1.3 and the graphs were produced in Excel-WindowsT.
3. Results 3.1. Utilization of glucose lowering therapies 3.1.1. Insulins Results from the regression analyses of number of patients/1000 inhabitants/month (TIM) treated with Insulin based GLT before and after the intervention are shown in Fig. 1A–C. There was an accelerated increasing trend in number of patients/TIM treated with insulins following the intervention (P = 0.0027) (Fig. 1A), for insulins with retained reimbursement as well as for restricted insulins (Fig. 1B and C). 3.1.2. Oral glucose lowering therapies Results from the regression analyses of number of patients/1000 inhabitants/month (TIM) treated with oral GLT before and after the intervention are shown in Fig. 2A–E. No significant differences in total number of patients treated with oral GLT were detected before versus after the intervention (Fig. 2A). There was a negative trend in number of patients/TIM treated with products with restricted reimbursement (P = 0.0035) (Fig. 2C) and a negative shift in products that were excluded from the reimbursement scheme following the intervention (P < 0.0001) (Fig. 2D). 3.2. Costs Total costs in the year following the intervention (March 2010 to February 2011) amounted to 1.1 billion SEK [≈D121 million], an increase by 58 million SEK [≈D6.4 million]
B. Pettersson et al. / Health Policy 108 (2012) 207–215
Paents/1000 Inh./month
All reimbursed oral GLT
Oral glucose lowering therapies, patients/1000 inhabitants/month
211
Level p=0.1610 Trend p=0.6084
Time of intervenon
15 10 5 0 10
0
20
30
40
Month
A
Paents
Level p=0.1255 Trend p=0.2625
Time of intervenon
8 4 0 0
10
20
30
40
Oral restricted reimbursement Paents/1000 Inh./month
Paents/1000 Inh./month
Oral retained reimbursement 12
Regress ion li ne
Time of intervenon
2 1,6 1,2 0,8 0,4 0 0
10
Oral no reimbursement Paents/1000 Inh./month
C
Regress ion li ne
Level p=<.0001 Trend p=0.1603
Time of intervenon
0,04 0,03 0,02 0,01 0 0
10
20
30
40
Paents
Paents
40
Regress ion li ne
Regress ion li ne
Level p=0.5867 Trend p=0.1880
Time of intervenon
2,5 2 1,5 1 0,5 0 0
10
Month
D
30
Oral no reimbursement for new courses Paents/1000 Inh./month
Paents
20 Month
Month
B
Level p=0.1348 Trend p=0.0035
20
30
40
Month
E
Paents
Regress ion li ne
Fig. 2. (A–E) Segmented regression analyses of number of patients/1000 inh. (inhabitants)/month treated with oral glucose lowering therapies: all reimbursed, those with retained-, those with restricted-, those with no- and those with no reimbursement for new courses before and after implementation of the intervention.
compared to a reference period 1 year before the intervention (March 2009 to February 2010). Total costs for insulins increased by 60 million SEK [≈D6.6 million] (7%), while costs for oral glucose lowering therapies decreased by 2 million SEK [≈D0.22 million]) (−0.07%) (data not shown). 3.2.1. Insulins Regression analyses of costs showed there was an accelerated increasing trend in total costs (Fig. 3A), both for insulins with retained reimbursement and for insulins with restricted reimbursement) following the intervention. There was an increase by 100 SEK [≈D 11]/TIM, which of 60 SEK[≈D6.6]/TIM for products with retained reimbursement and 40 SEK[≈D4.4]/TIM for restricted products (data not shown). 3.2.2. Oral glucose lowering therapies There were negative trends in costs for oral GLT in total (Fig. 3B) and for drugs which were retained or restricted in
the new RS and a negative level shift in products excluded from the RS (P < 0.0001)(data not shown). 4. Discussion Number of patients treated with insulin analogues increased at a higher rate following the decision to restrict reimbursement among patients with T2DM. There are several possible explanation to the increasing number of patients treated with restricted insulins following the new RS, but also the overall accelerated increasing trend of insulins. One explanation might be that the intervention with reimbursement restrictions only affected a subset of the insulin-treated patients–patients with T2DM who need insulin but fail to reach the treatment objectives with alternatives other than the long-acting insulin analogues. Another explanation might be due to factors influencing the use of the insulin analogues among patients with T1DM
212
B. Pettersson et al. / Health Policy 108 (2012) 207–215
Level p=0.3094 Trend p=0.0014
Cost/1000 Inh./month
Total costs insulins Time of intervenon
12000 8000 4000 0 0
10
20
30
40
Month
A
Cost
Regress ion li ne
Level p=0.2738 Trend p=0.0177
Cost/1000 Inh./month
Total costs oral GLT Time of intervenon
4000 3000 2000 1000 0 0
B
10
20
30
40
Month Cost
Regress ion li ne
Fig. 3. (A and B) Segmented regression analyses of total costs/1000 inh. (inhabitants)/month in Swedish Crown (SEK), for insulin and oral GLT before and after implementation of the intervention.
and that are probably more important than changes affecting the subset of patients with T2DM treated with the restricted insulins, even if this study used a cut off level of >40 years (see Section 2.5 for explanation) to limit the number of patients with type 1 diabetes mellitus (T1DM) in the analysis. Yet another explanation could be that the utilization prior to the intervention was restrained by alarm reports related to insulin analogues and risk of developing breast cancer [28], which was published during the period and that this effect waned in the later study period. Increased marketing activities, induced by the new RS, might have created an opportunity for the pharmaceutical companies to inform (and influence) the physicians and might be another explanation behind the increasing number of patients. Furthermore, the effect might be due to unclear reimbursement status for the existent insulin analogues, prior to the new RS. The reimbursement status differed between the two existing products and the reimbursement status had been changed earlier. However, the three last observations deviated considerably (Fig. 1C) and when they were excluded from the analysis the trend was similar to the trend before the intervention and therefore an alternative interpretation could be that the utilization of restricted insulins did not change. The new RS did not affect total number of patients treated with oral GLT. The effect on utilization of oral GLT that were restricted was minor, which could be explained by prescription adherence to existing restrictive guidelines (patients should have tried metformin, sulfonylureas or insulins first) has been reported to be about 90% already prior to the new RS [18]. In addition, concerns on adverse
effects related to rosiglitazone [29] were reported during the period before the implementation and was partly the motivation for TLV’s decision to restrict reimbursement of glitazones. The utilization of glibenclamide (marketed under the trade name of Daonil) and nateglinid (both excluded in the new RS initially) decreased steadily prior to the intervention, due to increased utilization of generic products, enhanced mainly by generic substitution and formulary guidelines both on the national and the regional level [7,8]. The intervention seems to have accelerated this trend slightly. 4.1. Utilization of glucose lowering treatment compared to other countries The utilization of insulin based treatment in Sweden widely exceeded the use in ten other countries, and as much as twice as high in Sweden compared to Denmark in subjects ≥45 years of age, but not in patients <45 years [10]. This might indicate a wider utilization of insulin-based treatment for T2DM in Sweden compared to Denmark and the difference might be explained by differences in reimbursement system. In Sweden insulins remained reimbursed for 100% while the reimbursement of oral GLT was reduced in 1997, when a new pharmaceutical benefits scheme was introduced. In Denmark they were also 100% reimbursed up to March 1, 2000 when they were included in a system where all reimbursable products have an equal status in terms of reimbursement rate [30]. 4.2. Costs and savings Cost differences between two time periods might depend on underlying trends in volume, price and product mix effects (assortment, strength, doses and pack sizes), besides effect of the intervention. The total effect on savings includes market dynamic effects related to prices due to increased competition of generics in combination with other market regulations for non patented drugs introduced by TLV, i.e. generic substitution, substitution mechanisms by pharmacies and differences in assortment and might not be directly associated with or be a consequence of the review. The generic substitution, has been shown to be an effective policy to reduce expenditures [31] and it was introduced in 2004, i.e. prior to the new RS. The dynamic effect from the intervention is that the price of glibenclamide marketed as Daonil was reduced to the level of generics available on the market, which contributed to the reduction of costs for oral GLT following the intervention. Total costs for insulins increased by 60 million SEK [≈D6.6 million] while costs for oral GLT decreased by 2 million SEK [≈D0.22 million] following the new RS. Price per day for glucose lowering treatment varies widely between insulins (7–21 SEK) [≈D0.8–2.3] [15], where the daily price depends on the weight of the patient and the dose required to achieve an effect. It also varies considerably for oral GLT (1–15 SEK) [≈D0.1–1.7], but less than and at a lower range than that for insulins. The utilization of oral GLT consists furthermore mainly of lower priced generic products and
B. Pettersson et al. / Health Policy 108 (2012) 207–215
therefore the costs for oral GLT are only a small part (about 20%) of the total costs for GLT and therefore changes in prices for oral GLT have limited effects on cost savings for GLT. In the review of GLT, the TLV concluded that the overall utilization of glucose lowering drugs was considered cost effective already prior to the review, and therefore the new RS was not designed to alter the overall utilization. The perception that the utilization of GLT is cost effective can however be subject of discussion, since the proportion of utilization of insulin based GLT for T2DM might be excessive in Sweden compared to other countries [10]. Furthermore, the cost-effectiveness of the long-acting insulin analogues compared to other insulins is mainly driven by the higher quality-of-life due to fewer number of episodes of hypoglycemia and/or the possibility to increase control of hyperglycemia without introducing more episodes of hypoglycemia in a subset of patients [32]. This is however mainly studied in T1DM [33] and can be questioned both on the basis on how to evaluate the loss of quality-oflife due to hypoglycemia, and an increasing use in T2DM where the doses of insulin (and thus the costs) tends to be higher in cases with higher weights and relative insulin-resistance.
213
was included at the end of the series but not at the beginning and thus the analysis could be influenced by seasonal fluctuations since February is a short month and therefore includes fewer dispensations than other months. Another option would have been to try to choose matching time periods surrendering the intervention but that would on the other hand have limited the total data set available for analysis. Linear regression models might not be appropriate because the linearity assumptions might be violated by phenomena of adaptation after abrupt changes in the system. However, we tested the normality assumptions on the residuals of all regressions with Shapiro–Wilk test. The test could not show significant non-normality in any of the residuals from the regressions (P-values > 0.05). The drug registry contains individual based level data on purchased drugs, which is ideal for evaluating impact of implementation of nationwide policies, since it includes all drugs dispensed and therefore provides possibilities to assess and estimate a wide range of measures. In our analyses we used only aggregated data, which could be a limitation, since data on new patients is a sensitive and immediate measure of adherence to interventions. 4.4. Summary of the RS and implication for the future
4.3. Strengths and limitations Time series design is a strong quasi-experimental alternative to randomization. The external validity of the design may be high as it occurs in a natural setting and segmented regression is a robust technique to test immediate and sustained changes associated with the intervention while controlling for trends [25,34]. Internal validity in time series analyses, as in other epidemiological techniques can be influenced by confounding factors. For instance, it could be threatened by historical events, i.e. some other event occurring at the same time as the intervention and so could explain the pattern of change over time. Time series design avoids however some of the limitations in before and after design, which is often employed to evaluate effects of interventions, since before and after design might produce results that are affected by either of the time points being atypical due to concurrent other events. Even though the time series design does not eliminate all problems neither of the influence of time periods studied, nor in interpreting changes in turnover rate, the extended time period strengthen the ability to attribute changes to the intervention [34]. Even though the prevalence of diabetes, and thus the aggregated need for treatment, is not susceptible to obvious seasonal changes, other factors might influence the actual amount of drugs dispensed over a given time period. The use of months as the time unit instead of longer time periods introduces methodological challenges such as seasonal variations including holidays, difference in number of days and especially working days with open pharmacies and/or access to prescribing physicians etc. The results are therefore sensitive to chosen months and years and have to be assessed cautiously. In this study, the starting point of the observations following the intervention was March 2010, and the endpoint was March 2011, and thus February
Enforcement has a stronger and more direct effect on utilization than soft demand policies such as education and guidelines, however restrictive demand side polices might trigger unexpected behaviour [35]. In this context the new RS for GLT was designed in line with other policies as generic substitution and harmonized with ongoing trends, along with national guidelines [8], which might have reduced the risks of the RS leading to unintended effects. In the review of GLT, the TLV concluded that the overall utilization of glucose lowering drugs was considered cost-effective already prior to the review, and therefore the new RS was not designed to alter the overall utilization. By contrast, the review of lipid modifying therapies (LMT) [36], was aiming at reducing the utilization of lower doses of patented statins towards equivalent doses of generic statins, which was considered to be more cost-effective. The LMT review with subsequent changes in reimbursement status of the LMT was implemented on June 1, 2009 [37], where the drugs were either retained, restricted or excluded from the RS. In products that were restricted in the new RS for LMT, an upward shift in level was detected following the new RS [36], while the accelerated trend increased for the restricted insulins in the new RS for GLT. For the restricted oral GLT the effect was that the trend was decreasing prior to the intervention, even if it was slightly accelerated following the new RS. The effects on utilization of restricted products were therefore non-conclusive, but they lead to possibly unintended effects for LMT and for insulins. The effect on overall utilization and for products which were retained or excluded was similar for oral GLT and for LMT, with expected rapid downward shifts. The effects on total utilization and on products with retained reimbursement were not affected following the RS for LMT and for oral GLT.
214
B. Pettersson et al. / Health Policy 108 (2012) 207–215
The effect on costs and expected saving differed for LMT, oral and insulin based GLT respectively, which could be dependent on timing of the review and the subsequent new RS. In the market for LMT, there was a large share of utilization of drugs that were still patent protected, while a major part of the utilization of oral GLT is in the generic market, therefore potential and realized savings were more substantial following the new RS for LMT, while costs even increased for insulins. Changes in reimbursement schemes to steer use of pharmaceuticals should be carefully evaluated with respect to timing and design, in order to avoid unintended and unwanted effects; this should be weighted against potential savings. Softer demand-side policies might be a better option to steer more precisely towards a cost-effective use of medicines. Independent follow-up is needed to assess the impact of new reimbursement decisions. At the same time, the TLV should have a follow-up plan to make their own assessments of the impact from their decisions, since our findings show that unintended effects might emerge. 5. Conclusions The new reimbursement scheme had a minor overall effect on utilization and costs of oral glucose lowering therapies. Despite more restrictive reimbursement for patients with T2DM, the utilization of insulin based glucose lowering therapies and related costs increased even faster after the intervention. References [1] International Diabetes Federation: IDF Diabetes Atlas. 4 edition Brussels: International Diabetes Federation, 2009. [2] Conroy RM, Pyorala K, Fitzgerald AP, Sans S, Menotti A, De Backer G, De Bacquer D, Ducimetiere P, Jousilahti P, Keil U, Njolstad I, Oganov RG, Thomsen T, Tunstall-Pedoe H, Tverdal A, Wedel H, Whincup P, Wilhelmsen L, Graham IM. Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE project. European Heart Journal 2003;24:987–1003. [3] Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group. Lancet 1998;352: 837–53. [4] Stratton IM, Cull CA, Adler AI, Matthews DR, Neil HA, Holman RR. Additive effects of glycaemia and blood pressure exposure on risk of complications in type 2 diabetes: a prospective observational study (UKPDS 75). Diabetologia 2006;49:1761–9. [5] IDF Clinical Guidelines Task Force: Global guideline for Type 2 diabetes. Brussels: International Diabetes Federation, 2005. [6] Nathan DM, Buse JB, Davidson MB, Ferrannini E, Holman RR, Sherwin R, Zinman B. Medical management of hyperglycemia in type 2 diabetes: a consensus algorithm for the initiation and adjustment of therapy: a consensus statement of the American Diabetes Association and the European Association for the Study of Diabetes. Diabetes Care 2009;32:193–203. [7] Medical Products Agency. Läkemedelsbehandling vid typ 2-diabetes–ny rekommendation;(English translation: Pharmaceutical treatment for type 2 diabetes- new recommendation). http://www.lakemedelsverket.se/upload/halso-och-sjukvard/ behandlingsrekommendationer/L%c3%a4kemedelsbehandling%20 vid%20typ%202-diabetes rek.pdf; 2010. [8] National board for health and welfare. Nationella riktlinjer för diabetesvården 2010–Stöd för styrning och ledning (English translation: National guidelines for diabetes care 2010). http://www. socialstyrelsen.se/publikationer2010/2010-2-2/Sidor/default.aspx; 2010.
[9] The National Board of Health and Welfare. The Swedish Drug Registry. http://www.socialstyrelsen.se/register/halsodataregister/ lakemedelsregistret; 2011. [10] Melander A, Folino-Gallo P, Walley T, Schwabe U, Groop PH, Klaukka T, Vallano A, Laporte JR, Gallego MR, Schiappa M, Roder M, Kampmann JP, de Swaef A, Aberg M, Mansson NO, Lindblad U. Utilisation of antihyperglycaemic drugs in ten European countries: different developments and different levels. Diabetologia 2006;49:2024–9. [11] Jansson SP, Andersson DK, Svardsudd K. Prevalence and incidence rate of diabetes mellitus in a Swedish community during 30 years of follow-up. Diabetologia 2007;50:703–10. [12] Ringborg A, Lindgren P, Martinell M, Yin DD, Schon S, Stalhammar J. Prevalence and incidence of Type 2 diabetes and its complications 1996–2003 – estimates from a Swedish population-based study. Diabetic Medicine 2008;25:1178–86. [13] Jonsson B. Revealing the cost of Type II diabetes in Europe. Diabetologia 2002;45:S5–12. [14] Fleming DM, Schellevis FG, Van Casteren V. The prevalence of known diabetes in eight European countries. European Journal of Public Health 2004;14:10–4. [15] The Dental and Pharmaceutical Benefits Agency (TLV). The review of medicines used for treating diabetes – A summary. http://www.tlv.se/Upload/Genomgangen/summary-diabetes.pdf; 2010. [16] The Dental and Pharmaceutical Benefits Agency (TLV). The Pharmaceutical reviews. http://www.tlv.se/in-english-old/in-english/; 2011. [17] Sveriges Riksbank Exchange rates. http://www.riksbank.com/ templates/Page.aspx?id=17182; 2011. [18] The NEPI Foundation – The Swedish Network for Pharmacoepidemiology. Diabetesmedel – effekter av Tandvårds – och läkemedelsförmånsverkets genomgång december 2009 (English translation: Diabetes treatments—effects of the TLV review December 2009). http://www.nepi.net/res/dokument/ NEPI-rapport-diabetesmedel-100831.pdf; 2010. [19] Wettermark B, Godman B, Andersson K, Gustafsson LL, Haycox A, Bertele V. Recent national and regional drug reforms in Sweden: implications for pharmaceutical companies in Europe. PharmacoEconomics 2008;26:537–50. [20] Sjögren E. Reasonable drugs – making decisions with ambiguous knowledge. The economic research institute. Stockholm: Stockholm School of Economics; 2006. [21] Wettermark B, Godman B, Neovius M, Hedberg N, Mellgren TO, Kahan T. Initial effects of a reimbursement restriction to improve the cost-effectiveness of antihypertensive treatment. Health Policy 2010;94:221–9. [22] WHO Collaborating Centre for Drug Statistics Methodology. Guidelines for ATC classification and DDD assignment 2010. http://www.whocc.no/atc ddd publications/guidelines/; 2009. [23] The Dental and Pharmaceutical Benefits Agency (TLV). Reimdecision, Levemir. http://www.tlv.se/Upload/ bursement Beslut 2007/BES 070618 levemir.pdf; 2007. [24] Grimshaw J, Campbell M, Eccles M, Steen N. Experimental and quasi-experimental designs for evaluating guideline implementation strategies. Family Practice 2000;17(Suppl. 1):S11–6. [25] Wagner AK, Soumerai SB, Zhang F, Ross-Degnan D. Segmented regression analysis of interrupted time series studies in medication use research. Journal of Clinical Pharmacy and Therapeutics 2002;27:299–309. [26] Wettermark B, Hammar N, Fored CM, Leimanis A, Otterblad Olausson P, Bergman U, Persson I, Sundstrom A, Westerholm B, Rosen M. The new Swedish Prescribed Drug Register – opportunities for pharmacoepidemiological research and experience from the first six months. Pharmacoepidemiology and Drug Safety 2007;16: 726–35. [27] Furu K, Wettermark B, Andersen M, Martikainen JE, Almarsdottir AB, Sorensen HT. The Nordic countries as a cohort for pharmacoepidemiological research. Basic and Clinical Pharmacology and Toxicology 2009;106:86–94. [28] Hemkens LG, Grouven U, Bender R, Gunster C, Gutschmidt S, Selke GW, Sawicki PT. Risk of malignancies in patients with diabetes treated with human insulin or insulin analogues: a cohort study. Diabetologia 2009;52:1732–44. [29] European Medicines Agency. European Medicines Agency update on ongoing benefit-risk review of Avandia, Avandamet and Avaglim. http://www.ema.europa.eu/ema/index.jsp?curl=pages/news and events/news/2010/07/news detail 001062.jsp&murl=menus/news and events/news and events.jsp&mid=WC0b01ac058004d5c1& jsenabled=true; 2010.
B. Pettersson et al. / Health Policy 108 (2012) 207–215 [30] The Danish association of the pharmaceutical industry. Pricing and reimbursement in Denmark. http://parno1.ipapercms. dk/LIF/Notater/PricingandreimbursementinDenmark/; 2007. [31] Andersson Swedish Pharmaceutical Benefit Reforms – analyses of implementation, pharmaceutical sales patterns and expenditures. Department of Public health and Community Medicine. Gothenburg: University of Gothenburg; 2006. [32] Cameron CG, Bennett HA. Cost-effectiveness of insulin analogues for diabetes mellitus. Canadian Medical Association Journal 2009;180:400–7. [33] Tran KBS, Li H, Cimon K, Daneman D, Simpson SH, Campbell K. Long-acting insulin analogues for diabetes mellitus: metaanalysis of clinical outcomes and assessment of cost-effectiveness technology Report number 92. Ottawa: Canadian Agency for Drugs and Technologies in Health; 2007.
215
[34] Polit D, Tatano Beck C, Nursing Research: Generating. Assessing evidence for nursing practice. Philadelphia: Lippincott Williams & Wilkins; 2008. [35] Wettermark B, Godman B, Jacobsson B, Haaijer-Ruskamp FM. Soft regulations in pharmaceutical policy making: an overview of current approaches and their consequences. Applied Health Economics and Health Policy 2009;7:137–47. [36] Pettersson B, Hoffmann M, Wändell P, Levin L-Å. Utilization and costs of lipid modifying therapies following health technology assessment for the new reimbursement scheme in Sweden. Health Policy 2011. [37] The Dental and Pharmaceutical Benefits Agency (TLV). The review of medicines for treating lipid disorders. http://www.tlv.se/Upload/Genomgangen/review-lipid-disorder.pdf; 2009.