original article

Wien Klin Wochenschr (2010) 122: 89–94 DOI 10.1007/s00508-009-1292-6 Printed in Austria © Springer-Verlag 2010

Wiener klinische Wochenschrift The Middle European Journal of Medicine

Impact of reimbursement changes on statin use among patients with diabetes in Austria Wolfgang C. Winkelmayer1, Markus Asslaber2, Anna Bucsics2, Thomas Burkhardt 2, Alexandra Schautzer3 , Peter Wieninger2, Michaela Pogantsch 3 , M. Alan Brookhart1; for the Pharmacoeconomics Advisory Council of the Austrian Sickness Funds* 1 Division

of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts, United States 2 Hauptverband der Österreichischen Sozialversicherungsträger, Vienna, Austria 3 Steiermärkische Gebietskrankenkasse, Graz, Austria

Received November 25, 2009, accepted after revision November 26, 2009

Hat die Freigabe der Verschreibbarkeit von Statinen für Diabetiker mit hohem kardiovaskulärem Risiko das ärztliche Verordnungsverhalten in Österreich beeinflusst? Zusammenfassung. Hintergrund: Lipidsenker der Kategorie 3-hydroxy-3-methyl-Co-Enzym A ReduktaseHemmer (Statine) bilden eine der Hauptsäulen der kardiovaskulären Sekundärprävention bei Personen mit Diabetes. Diese Medikamente sind effektiv und kosteneffektiv, können aber eine erhebliche Erhöhung der Medikamentenausgaben im niedergelassenen Bereich mit sich bringen. Der Hauptverband der österreichischen Sozialversicherungsträger steuert das Verschreibeverhalten ärztlicher Vertragspartner durch die Herausgabe­ des Erstattungskodex (vormals: Heilmittelverzeichnis) mit der Festlegung von Voraussetzungen für die Verschreibbarkeit von Medikamenten. Diese Voraussetzungen ermöglichen den Zugang zu evidenzbasierter Therapie unter Berücksichtigung ökonomischer Aspekte. Methoden: Das vorliegende Projekt evaluierte zwei Änderungen der Voraussetzungen für die freie Verschreibbarkeit von Statinen bei Patienten mit Diabetes im Zeitraum 4/2004 – 12/2005. Wir verwendeten Verrechnungsdaten von mehreren Versicherungsträgern um die Verschreiberate von Statinen bei Diabetikern zu evaluieren. Pseudoexperimentelle Zeit-Trend Analysen verglichen erwartete Verschreibungsmuster mit tatsächlich beobachteten. Ergebnisse: Während der Beobachtugsperiode stieg die monatliche Verschreibungsquote von Statinen von 20,6% auf 24,9%. Mit Oktober 2004 wurde die Verschreibung von Statinen für PatientInnen mit DiCorrespondence: Wolfgang C. Winkelmayer, MD, ScD, Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Women’s Hospital, 1620 Tremont Street, Suite 3030, Boston, MA 02120, USA, E-mail: [email protected]

wkw  3–4/2010

abetes mellitus mit hohem kardiovaskulärem Risiko freigegeben; diese Änderung der Verschreibbarkeit von Statinen führte zu keinen wesentlichen Änderungen im Verschreibeverhalten. Im Juli 2005 wurde eine verpflichtende Dokumentation über die Begründung einer Verschreibung von Atorvastatin 10 mg mit nachfolgender stichprobenartiger Überprüfung festgelegt. Die Verwendung von Atorvastatin nahm da­ raufhin­­ messbar zugunsten anderer bevorzugter Statine ab. Schlussfolgerung: Systematische Evaluationen von Änderungen in der Verschreibbarkeit von erstatteten Medikamenten können wertvolle Informationen für die Anwendung zukünftiger Steuerinstrumente geben. Summary. Background: Statins have evolved as cornerstones of cardiovascular prevention in patients with diabetes. They are effective and can be cost-effective therapies, but increased use imposes a sizeable shortterm burden on payors of health care. These have used various instruments to steer appropriate use of such treatment. It was the purpose of this study to examine the effect of two reimbursement policy changes of statin therapy in patients with diabetes in Austria. Methods: Retrospective cohort study; time-series analysis. From Austrian sickness funds claims, we identified a closed cohort of 68,953 patients receiving treatment for diabetes in the first quarter of 2004. From April * Participating Members of the Pharmacoeconomics Advisory Council of the Austrian Sickness Funds: Sonja Ikemeyer-Fuchs (Pro­v incial Sickness Fund Vienna); Jana Fischer, PhD; Michaela Stitz, MD (Provincial Sickness Fund Lower Austria); Berthold Reichardt, MSc (Provincial Sickness Fund Burgenland); Harald Klier, MD; Gerrit Uhl, MSc (Provincial Sickness Fund Styria); Margot Reiter, PhD; Ursula Riess, MD; Ulrike Wieser, MD (Provincial Sickness Fund Carinthia); Renato Kasseroller, MD, Michael Prossinger, PhD (Provincial Sickness Fund Salzburg); Monika Schumlits, MD; (Farmers’ Insurance Funds); Andreas Stummer, MD (Federal Employees’ Insurance Funds); Sarah Schnepf, MD; Anna Labek, MD (Provincial Sickness Fund Upper Austria).

Reimbursement change and statin use in Austria

89

original article

2004 – December 2005, we ascertained use of statins for each monthly interval. Patients were censored at death. We used pseudo-experimental time-series regression to evaluate the effect of two policy changes on statin use and cost overall, as well as on the use of preferred versus non-preferred statins. Results: Statin use among Austrian patients with diabetes increased from 20.6% to 24.9% during the time period. A policy change essentially expanding reimbursement for statins from secondary to primary prevention among patients with diabetes had no discernible effect on the observed trends in statin use. Another policy change that imposed random chart review for appropriateness of prescription of non-preferred statins including atorvastatin 10 mg yielded a marked drop in use of atorvastatin 10 mg and increase in the use of preferred statins, while leaving overall trends in statin use unaffected. Conclusions: Quantitative evaluation of new policies can provide important insights into the effectiveness und utility of such changes.

Methods

Key words: Drug policy, reimbursement, statins, utilization studies, diabetes.

For the purpose of this study, we used the complete claims data of individuals covered by five sickness funds: the provincial sickness funds (Gebietskrankenkassen) of Lower Austria, Upper Austria, Styria, and Salzburg, as well as the farmers’ insurance fund (Sozialversicherung der Bauern, SVB). Cumulatively, these sickness funds cover approximately four million members of the total Austrian population of 8.2 million. For the purpose of this study, we will refer to these five sickness funds in anonymized codes (i.e., Sickness Fund A, B, C, D, E). Each sickness fund separately identified from their insurance claims all patients who filled at least one claim for any medication used for treating diabetes: any sulfonylurea, biguanide, α-glucosidase inhibitor, thiazolidinedione (“glitazone”), repaglinid, or insulin. Each fund then provided us with a dataset containing the complete anonymized health care claims data and death dates covering the period from January 1, 2004 to December 31, 2005. We selected the study cohort by requiring at least 2 filled prescriptions of any diabetes drug during and survival beyond the first quarter of 2004 (January 1 to March 31). We described the cohort by providing the mean age in 2004, proportion of female patients studied, and by the proportion of study patients who used specific classes of anti-diabetes and cardiovascular drugs during the baseline period.

Introduction Patients with diabetes are at markedly higher cardiovascular risk compared to individuals without the disease [1]. Lipid abnormalities are an important part of the metabolic abnormality that accompanies the spectrum of impaired glucose tolerance including diabetes [2]. These disturbances in lipid metabolism contribute to the excess cardiovascular risk of these patients [3]. Lipid-lowering agents are an important tool to ameliorate the cardiovascular burden among patients with diabetes. Especially 3-hydroxy-3-methyl-glutaryl-coenzyme-A inhibitors, better known as statins have been demonstrated to be efficacious in secondary [4], but also primary cardiovascular prevention in diabetes [5]. Until the introduction of multi-source products, chronic treatment with statins was comparatively costly. So, payors of health care have sought to direct use of these medications towards those patients who were expected to benefit most from such treatment, while also negotiating on price with the competing manufacturers of these drugs. Austria, a provider of universal health care coverage to its residents, first made statins available for unrestricted use to its residents with established coronary heart disease in 1995. In 2004 and 2005, however, changes in medical evidence and renegotiated prices of individual statins led to the implementation of two significant changes in statin reimbursement policy among patients with diabetes. It is the purpose of this study to use data from several Austrian sickness funds to formally evaluate the impact of these two policy changes on the temporal trends in statin utilization among patients with diabetes. As a secondary analysis, we evaluated temporal trends in statin treatment cost among patients with diabetes.

90

Reimbursement change and statin use in Austria

Data sources Austria is a social welfare state that mandates universal health care to its residents [6]. Patients are assigned membership in one of several sickness funds depending on their type and location of current or former employment. At the time covered by this study, there were 19 sickness funds in Austria, with the 9 provincial sickness funds covering the majority of all residents. Most inpatient and outpatient medical services are covered and, basically, each encounter generates a claim to the health care system. All medications that are deemed efficacious by a national panel are reimbursed, so that access to prescription drugs is equal across sickness funds. The copayment for medicines per package was € 4.35 in 2004 and € 4.45 in 2005. Sickness fund premiums, other coinsurance and copayments vary slightly, and indigent patients can apply for waiver of the copayment for prescription drugs. Few medications are subject to prior authorization by the sickness funds. Prescription claims contain a unique identifier for the specific drug, the strength, and the quantity dispensed.

Study population

Outcomes Monthly statin use was ascertained from all health care claims and recorded from a filled prescription during each monthly interval from April 2004 (Month 4) to December 2005 (Month 24). Since up to two 28-day or 30-day supplies could be dispensed at a time, we carried forward one of these packages to the subsequent month if the claim indicated that two packages were indeed dispensed. Patients were removed from the monthly risk sets in and after the month of their death. We further distinguished between preferred (simvastatin 20 and 40 mg, lovastatin, pravastatin, fluvastatin, and atorvastatin 10mg) and nonpreferred statins (atorvastatin > 10 mg, all strengths of rosuva­ statin). As a secondary outcome, we calculated the mean costs of statin use per patient per month for each interval.

Statistical analysis We assessed policy change effects on statin use through segmented linear regression. The unit of analysis was a patientmonth indicator variable for whether the patient filled a pre Springer-Verlag  3–4/2010  wkw

original article

scription for a statin in the given month. The regression model allowed for a slope for the time period preceding both policy changes (applied to months 4–9), a slope to account the months after the first policy change but before the second policy change (applied to months 10–18), and finally a slope from the months after the second policy change studied (applied to months 19– 24). We also allowed for level changes at each policy change. Separate models were fit for preferred statins, non-preferred statins, and all statins. Parameters were estimated using a generalized estimating equations approach with an identity link, binomial error structure, and within-patient working correlation matrix with an independence structure. Consistent standard errors were estimated using the empirical variance-covariance matrix that accounted for the within-patient correlation of outcomes. We assessed policy change effects on total statin expenditures among statin users also using a segmented linear regression model. This model was similar to the previous models except that parameters were estimated by assuming a Gaussian error structure. Standard errors were estimated robustly to account for the within-patient correlation of outcomes. Due to the four categories of utilization parameters (overall statin, preferred statin, non-preferred statin, and atorvastatin 10mg) and four sets of statistical tests conducted, we adjusted the significance threshold using Bonferroni’s rule of 0.05/N, where N is the number of statistical tests conducted; since N = 4, statistical significance was set at alpha < 0.0125.

Results

insulin, 52.1% used sulfonylureas, and 55.9% filled at least one prescription for a biguanide (Table 1). Other diabetes drugs such as α-glucosidase inhibitors (5.4%), repaglinide (7.9%), or glitazones (1.5%) were used infrequently. One quarter of patients (25.6%) filled at least one prescription for a statin during the baseline period, and 3.2% for a fibrate. Other lipid-lowering medications were used in < 0.5% of patients. Table 1 displays baseline use of these and of several cardiovascular medication classes overall, as well as by individual sickness fund.

Trends in statin use Monthly rates of filling a prescription for a statin in this closed cohort are displayed in Fig. 1. In April 2004, 14,174 (20.6%) diabetic patients filled a prescription for a statin; 13.9% for a preferred statin, 6.7% for a non-preferred statin or atorvastatin 10 mg. Over the 21 months of observation, monthly statin filling rates rose to 24.9%, with preferred statins (21.2%) responsible for all of the increase. Non-preferred statins including atorvastatin 10 mg decreased to 3.7% during that time. Overall increase in statin use was estimated at 0.18% per month (P < 0.0001).

Impact of first policy change (10/1/2004)

Study population From the claims received from the five sickness funds, we ascertained a cohort of 68,953 patients who fulfilled the inclusion criteria. During study follow-up until December 2005, 3,525 patients died (5.1%). The average age was 66.2 years (standard deviation, SD: 13 years) and 54% were women. In the first quarter of 2004, the cohort identification period, 27.7% of these patients received

From Fig. 1, it is evident that overall trends in increasing statin use did not change as a result of the first policy change. From the parameterized time-series model, the intercept and slope of the statin increase did not differ significantly prior to versus after the first policy change (difference in intercept: –0.008, P = 0.035; difference in slope: 0.0007, P = 0.113). Regarding preferred statins, there was no change in intercept (P = 0.02), but the slope

Table 1. Characteristics of Study Population (N = 68,953) Sickness fund All patients

A

Number of patients 68,953 22,821 Age (mean) 66.2 65.7 Female (%) 54.0 54.3 Prior use of diabetes medications (%, in first quarter of 2004) Insulin 27.7 25.7 Sulfonylureas 52.1 54.0 Biguanides 55.9 57.8 α-Glucosidase inhibitors 5.4 5.4 Repaglinid 7.9 9.6 Thiazolidinedione 1.5 1.1 Prior use of lipid-lowering medications (%, in first quarter of 2004) Statins 25.6 29.7 Fibrates 3.2 3.8 Prior use of cardiovascular medications (%, in first quarter of 2004) α-Receptor blocker 6.8 8.6 ACE-inhibitors 46.8 46.5 ARBs 9.9 12.0 b-Receptor blockers 31.1 33.1 Calcium channel blockers 19.9 21.7 Diuretics 18.1 16.4 Vasodilators 11.0 11.4 wkw  3–4/2010   Springer-Verlag

B

C

D

E

17,787 65.1 49.9

5,416 64.9 51.8

15,170 66.2 55.4

7,759 71.5 60.0

26.8 53.2 57.1 4.9 7.5 2.6

23.5 53.4 60.6 6.1 6.9 1.2

35.7 44.4 50.6 5.9 6.4 1.1

22.6 58.6 54.6 5.4 7.5 1.1

22.8 2.2

25.4 3.2

24.5 3.6

21.9 2.5

5.8 44.6 7.9 29.7 17.5 16.4 7.2

5.5 46.1 9.3 26.7 19.1 16.6 11.7

5.4 47.1 10.0 32.7 19.7 20.1 13.0

7.3 52.7 8.8 28.8 20.8 23.7 13.6

Reimbursement change and statin use in Austria

91

original article

0.50 0.45 0.40 0.35

27

26 Total Cost (Euro)

Proportion of Cohort on a Statin

28

All Stains (Observed) Predicted Preferred Statins (Observed) Predicted Atorvastatin 10 mg (Observed‘) Predicted Other Non-preferred (Observed) Predicted

0.30 0.25 0.20

25

24

0.15 0.10

23

0.05 0.00

22 0

10

20

30

Month

10

Month

20

30

Fig. 1.  Monthly rates of statin use in a cohort of patients with diabetes (N = 68,953)

Fig. 2.  Mean monthly per-member statin costs in a cohort of patients with diabetes (N = 68,953)

increased marginally by 0.07% per month after the policy change (P = 0.003). Correspondingly, the slope of non-preferred statins including atorvastatin 10 mg declined by 0.13% per month after the policy change (P < 0.0001). Thus, expanding the indication for preferred statin use from secondary prevention to primary prevention in patients at high cardiovascular risk on October 1, 2004 did not meaningfully change the overall trends of increasing statin use over time, but a trend towards greater use of preferred statins was detectable.

statins from secondary prevention to patients with diabetes with a high cardiovascular risk but without manifest CHD. It generated a statistically detectable trend, but the size of this trend was overwhelmed by preexisting trends in statin use. The second policy change, which established random chart reviews for appropriateness of use of non-preferred statins with sanctions for violations yielded an acute reallocation of statin treatment from atorvastatin 10 mg to preferred statins. This policy appeared to have led to a small drop from long-term utilization trends, but by the end of the study period (December 2005), utilization was yet again in congruence with the rate that could be expected from extrapolation of October 2004 – June 2005 data. Thus, these policy changes appear to not have influenced access to these medications overall, an important quality measure of policy interventions. While not a direct aim of these policy interventions, our data demonstrate that mean monthly per patient statin costs have declined substantially during the observation period, due to price negotiations and competitive behavior by manufacturers of these drugs, including manufacturers of generic statins. Austria provides near universal health care coverage including generous reimbursement of prescription medications to its residents via a Bismarck system [7]. Currently, payment of outpatient care including pharmaceuticals is administered through 19 sickness funds. Reimbursement of medications may be unrestricted or subject to prior approval by a medical officer of the individual sickness fund. For certain drugs, prior approval is waived if patients fulfill certain criteria and this is documented by the prescriber. Whether the specified criteria are met is spot-checked by the individual sickness funds. Of note, whether a certain drug is reimbursed, and among those, whether prior authorization is required, is regulated centrally and is therefore homogenous across sickness funds. Reimbursement decisions are reviewed regularly and are revised upon arrival of new medical evidence that may change the indi-

Impact of second policy change (7/1/2005) From Fig. 1, it appears that compared to the time period between the two policy changes, there was a slight drop in statin utilization after July 1, 2005, which was accompanied by a greater monthly increase in statin use after that date. Again, using the parameterized model and compared with the time period from October 2004 to June 2005, the intercept of statin utilization declined by 2.36% (P = 0.009), but the slope increased by 0.1% per month, which was not significant (P = 0.025). A strong redistribution from atorvastatin 10 mg to preferred statins was observed as a consequence of the new reimbursement policy affecting atorvastatin 10mg (all slope and intercept estimates P < 0.0001). Average statin costs for each treated diabetic patient declined during the study period from € 27.85 in April 2004 to € 22.75 in December 2005 (P < 0.001; Fig. 2). While the results are shown for the aggregate cohort from all 5 sickness funds, the findings were very similar for each individual fund (not shown).

Discussion To our knowledge, this study provides the first evaluation of reimbursement policy changes in patients in Austria which is based on a formal analysis of patientlevel data. The first policy change expanded the indication for free prescription without prior authorization of

92

0

Reimbursement change and statin use in Austria

 Springer-Verlag  3–4/2010  wkw

original article

cation for a given drug, or following renegotiations of drug prices. During the years 2004 and 2005, reimbursement rules for statins in patients with diabetes were changed twice. In early 2004, most statins (“preferred”) were readily reimbursed for patients with diagnostically verified coronary heart disease and hypercholesterolemia, i.e. for secondary prevention only. Some statins (“nonpreferred”), notably those that were more expensive, were subject to prior approval for reimbursement upon demonstration that preferred statins were insufficient to control a patient’s lipid abnormalities. Atorvastatin 10 mg was originally in the group of “preferred” statins, and freely reimbursed for patients with coronary artery disease. Only preferred statins were included in the “Heilmittelverzeichnis” (list of medicinal products), which was considered by the European Court of Justice a positive list. On October 1, 2004, reimbursement criteria for some statins, mostly the previously preferred ones, were changed to include “secondary prevention in the presence of clinically manifest atherosclerosis and/ or diabetes mellitus with high cardiovascular risk”. Since essentially all patients with diabetes can be considered at high cardiovascular risk, this was in effect a decision to provide statins for most patients with diabetes even without overt coronary artery disease. For all other patients, who did not fulfill these criteria, and for non-preferred statins, prior approval by a medical officer of the responsible sickness fund was needed for reimbursement. Such an approval was granted in selected cases. The change of the reimbursement criteria was carried out on the recommendation of the Advisory Board (“Fachbeirat für Arzneimittelwesen”, consisting representatives of the sickness funds, academic pharmacologists, and representatives of stakeholders. It was accompanied by a reduction of the price of products whose reimbursement criteria were expanded. The recommendation for change was made on the basis of current evidence at the time [8]. Notably, the criteria for atorvastatin 10 mg were not expanded, but remained unchanged at this point. The Hauptverband der österreichischen Sozialversicherungsträger (umbrella organization of all Austrian sickness funds) decided to evaluate the effect of this policy change on the use of statins among patients with diabetes in Austria. We wished to see if the less restrictive criteria for waiving prior approval led to an increase in the use of statins, specifically for patients with diabetes. Atorvastatin 10 mg should have served as a “natural control” since reimbursement criteria were not expanded for this product. As can be seen from the present study, the effect of the policy measure was overwhelmed by the existing temporal trend towards greater statin use in patients with diabetes. This finding is surprising, since opening up reimbursement for primary-type rather than secondary-type prevention should have led to a marked increase in statin use. The pattern of antidiabetic drug use reported in this Austrian cohort (Table 1) is similar to that reported in Germany for the year 2004 [9]. The prevalence of statin use by diabetics is somewhat lower in this study wkw  3–4/2010   Springer-Verlag

compared to data from Belgium, where 39,9% of selected patients with diabetes were reported to have used statins in 2005 [10]. However, the participants of the Belgian study were selected as having a BMI > 25, age between 45 and 75 years, being regular visitors of the pharmacy and being willing to participate in the study. The numbers reported here resemble those from Germany more closely, where the prevalence of statin use in diabetics in 2004 is shown as 25,2% although the German study by Rathmann et al. excluded diabetics under the age of 30. Our finding is in contrast to the findings of Trifiro et al. who reported an increase in the use of statins particularly in patients with diabetes mellitus after a health policy intervention which provided patients with hypercholesterolemia and diabetes mellitus who required treatment with statins free of charge [11]. We cannot discern why the first policy change had so little impact. Physicians may have anticipated this policy change and practiced based on already existing evidence supporting primary prevention, thus preempting the action by health policy makers. On the other hand, data from Belgium (see above) and Hong Kong, where more than 30% of patients with diabetes mellitus received one or more lipid lowering agents for the primary prevention of cardiovascular events, indicate that there may be still be underutilization of statins use among diabetics which would mean that clinical evidence is incompletely incorporated into practice [12]. Coincidentally, during the course of the study period, there was a major reform of the drug reimbursement system. A system of so-called “boxes” was introduced. This consisted of the “red box” (a list in which drugs under current evaluation could be temporarily included), the “green box” (a list of drugs which could be prescribed without prior approval, either unconditionally or if certain conditions were met – essentially similar to the former “Heilmittelverzeichnis” and of the “yellow box” – a list of drugs which were conditionally reimbursed. This “yellow box” is subdivided into two parts: a so-called “dark yellow box”, which is a list of drugs subject to prior approval, and a so-called “light yellow box” which is a list of drugs whose prior approval is waived if the doctor adequately documents that certain conditions are met. Conditions for both the “light yellow” and the “dark yellow” box are published on the public announcement website of the Austrian Social Insurance (www.avsv.at). Most of the statins in the so-called “Heilmittelverzeichnis” were transferred to the “Green Box”. Some of the statins which were previously not included in.the “Heilmittelverzeichnis”, (corresponding to the later “Green Box”) were subsequently (as of January 1, 2005) transferred to the “light yellow box”. The status of Atorvastatin 10 mg was changed from readily reimbursed for patients with diagnostically verified coronary heart disease and hypercholesterolemia to “light yellow”. Thus, physicians wishing to prescribe atorvastatin 10 mg had to document that the criteria for reimbursement were met and had to expect random chart reviews by sickness fund officials. Certain professional sancReimbursement change and statin use in Austria

93

original article

tions were established for violations of the regulations (the “Heilmittel-Kontroll- und Bewilligungsverordnung”, see www.avsv.at). As can be seen in Fig. 1, implementation of the new reimbursement rules on July 1, 2005 led to a marked reduction in the use of atorvastatin 10 mg from 6% in the last pre-policy month to 3% in September to December 2005. By contrast, use of preferred statins increased beyond projection during that time from 17% (June 2005) to 21% (November and December 2005) and overall statin use by patients with diabetes, after a small but significant dip immediately after the policy change, was back to pre-policy projection (25%) in December 2005. Thus, this policy achieved its goal of directing users of atorvastatin 10mg to less costly preferred statins which are considered equally effective. This seems to indicate that mandatory documentation, with subsequent random review of medical records can be a useful tool in ensuring cost-effective and policy-adherent prescribing practice. Marketing certainly plays an important role in this field, and the public relations measures accompanying the first policy change may have been insufficient for a meaningful impact. When an Austrian doctor looks up a drug to be prescribed in the appropriate database or brochure, he or she will see a restriction, if there is one. It may be that the fact that there is a restriction at all and the sanctions accompanying this restriction have much more impact than the actual text of the restriction. Recently, all disease-based restrictions for prescribing many brands of simvastatin, pravastatin and lovastatin have been removed. It will be interesting to compare the effect of this most recent policy change (which was again accompanied by price reductions) to the less radical one implemented in 2004, especially since the changes in 2008 were rather heavily promoted by the marketing authorization holders. When conducting time-series analyses, it is certainly important to consider alternative explanations for any detected changes in prescription trends. It is the inherent assumption of time-series analyses, also often called pseudo-experimental design, that no other possible causes of a prescription behavior change occurred. For the purpose of this study, one would need to consider any influential studies or newly released guidelines during the observation period. We could not discover any changes in guidelines or publications that would have steered providers away from using atorvastatin and towards other statins during the evaluated time-period. Thus, we remain rather certain that the observed trends in intra-class allocation of statin use were in fact causally related to the implemented policy changes.

Conclusion The present study demonstrates that a national change in statin prescribing policy from secondary to primary prevention did not affect already occurring trends towards greater statin use. By contrast, a policy targeted at directing prescribing behavior from non-preferred to preferred statins that incorporated random post hoc review of medical records and potential penalties for vio-

94

Reimbursement change and statin use in Austria

lations appeared to have had the desired effect, whilst leaving overall use of this class unaffected. Conflict of Interest None of the authors has a competing interest to declare. The opinions presented here are exclusively those of the authors. They are not necessarily identical with those of the Department, the Main Association of Austrian Social Security Institutions, its Advisory Committees or its management. Financial support was provided by the Hauptverband der Österreichischen­Sozialversicherungen. Authors contributions WCW, AB, PW and MP designed the study; MA, TB, AS, and MP coordinated the acquisition, de-identification and formatting of the data; WCW and MAB conducted the data analysis; WCW, AB, PW, MP and MAB drafted the manuscript; MA, TB, and MS provided critical review and edits to the manuscript.

References 1. Laakso M (1999) Hyperglycemia and cardiovascular disease in type 2 diabetes. Diabetes 48 (5): 937–942 2. Haffner SM, Valdez RA, Hazuda HP, Mitchell BD, Morales PA, Stern MP (1992) Prospective analysis of the insulin-resistance syndrome (syndrome X). Diabetes 41 (6): 715–722 3. Game FL, Jones AF (2001) Coronary heart disease risk assessment in diabetes mellitus – a comparison of PROCAM and Framingham risk assessment functions. Diabet Med 18 (5): 355–359 4. Haffner SM, Alexander CM, Cook TJ, Boccuzzi SJ, Musliner TA, Pedersen TR, Kjekshus J, Pyorala K (1999) Reduced coronary events in simvastatin-treated patients with coronary heart disease and diabetes or impaired fasting glucose levels: subgroup analyses in the Scandinavian Simvastatin Survival Study. Arch Intern Med 159 (22): 2661– 2667 5. Collins R, Armitage J, Parish S, Sleigh P, Peto R (2003) MRC/BHF Heart Protection Study of cholesterol-lowering with simvastatin in 5963 people with diabetes: a randomised placebo-controlled trial. Lancet 361 (9374): 2005–2016 6. Bennett CL, Schwarz B, Marberger M (1993) Health care in Austria. Universal access, national health insurance, and private health care. Jama 269 (21): 2789–2794 7. Euro Health Consumer Index (2007) Health Consumer Powerhouse 8. Snow V, Aronson MD, Hornbake ER, Mottur-Pilson C, Weiss KB (2004) Lipid control in the management of type 2 diabetes mellitus: a clinical practice guideline from the American College of Physicians. Annals of internal medicine 140 (8): 644–649 9. Rathmann W, Haastert B, Icks A, Giani G (2007) Trends in outpatient prescription drug costs in diabetic patients in Germany, 1994–2004. Diabetes care 30 (4): 848–853 10. Mehuys E, De Bolle L, Van Bortel L, Annemans L, Van Tongelen I, Remon JP, Giri M (2008) Medication use and disease management of type 2 diabetes in Belgium. Pharm World Sci 30 (1): 51–56 11. Trifiro G, Alacqua M, Corrao S, Moretti S, Tari DU, Galdo M, Caputi AP, Arcoraci V (2008) Lipid-lowering drug use in Italian primary care: effects of reimbursement criteria revision. Eur J Clin Pharmacol 64 (6): 619–625 12. Lee VW, Ho IC, Chan WS, Tam KY, Lee KK (2008) Statin utilization patterns for the primary prevention of cardiovascular events: a retrospective study in patients with diabetes mellitus in Hong Kong. Am J Cardiovasc Drugs 8 (3): 199–205  Springer-Verlag  3–4/2010  wkw