Economic impact of compliance to treatment with antidiabetes medication in type 2 diabetes mellitus: a review paper

Abstract

Objectives: Suboptimal compliance and failure to persist with antidiabetes therapies are of potential economic significance. The present research aims to describe the impact of poor compliance and persistence with antidiabetes medications on the cost of healthcare or its components for patients with type 2 diabetes mellitus (T2DM).

Methods: Literature search was conducted in PubMed for relevant articles published in the period between 1 January 2000 and 30 April 2009. Thus, it is possible that relevant articles not listed in PubMed, but available in other databases are not included in the current review. Studies describing economic consequence of compliance and/or persistence with pharmaceutical antidiabetes treatment were identified. The variability in the studies reviewed was high, making it extremely difficult to make a comparison between them.

Results: Of 449 articles corresponding to the primary search algorithm, 12 studies (all conducted in USA) fulfilled the inclusion criteria regarding the economic impact of compliance and/or persistence with treatment on the overall cost of T2DM care or its components. Compliance was assessed via medication possession ratio (MPR) in ten studies, where it ranged from 0.52 to 0.93 depending on regimen. Persistence was assessed in one study. Mean total annual costs per T2DM patient varied between the studies, ranging from $4570 to $17338. In seven studies, medication compliance was inversely associated with total healthcare costs, while in four other studies inverse associations between medication compliance and hospitalisation costs were reported. In one study increased adherence did not change overall healthcare costs.

Conclusions: Improved compliance may lead to reductions of the total healthcare costs in T2DM, Further research is needed in countries other than the US to assess impact of compliance and persistence to pharmacotherapy on T2DM costs in country-specific settings.

Key words::

• adherence
• compliance
• costs
• persistence
• type 2 diabetes mellitus

Introduction

Studies exploring associations among medication compliance and healthcare costs and utilisation in type 2 diabetes (T2DM) provide important information on the economic consequences of non-compliance to treatment and are essential in the planning of strategies reinforcing patients’ efforts to adhere in order to achieve therapeutic goals. It has been recognised that diabetes regimens are complex and, although consistent with practical guidelines and commonly prescribed practices, may pose a challenge to the patient, resulting in a patients’ inability to comply with them¹. Therapeutic potential of the pharmacological treatment in T2DM may be affected by the suboptimal compliance and/or failure to persist with drug therapy for the prescribed duration, ultimately accounting for the differences between efficacy and effectiveness in real-world settings².

According to the International Society for Pharmacoeconomics and Outcomes Research (ISPOR), medical compliance (syn. adherence) is ‘the extent to which a patient acts in accordance with the prescribed interval, dose and dosing regimen’, and medication persistence is ‘the length of time from initiation to discontinuation of therapy’³. Compliance is often measured by the medication possession ratio (MPR), and persistence as the number of days for which therapy was available³. According to Hughes et al, compliance has also to do with other issues such as whether medicines are taken with or after food, swallowed whole or not chewed, etc.⁴. Cramer et al defined ‘compliance’ and ‘persistence’ as two distinct constructs and supported the use of the term ‘compliance’ as the primary term and ‘adherence’ as a synonym of compliance³.

Non-compliance has been reported to lead to an increase in healthcare resource utilisation and overall expenditure due to T2DM and its complications⁵. In particular, non-compliance may lead to worsened glycaemic control, which in its turn may result in increased healthcare utilisation and costs (e.g. due to hospitalisations, temporary or permanent disability because of diabetic retinopathy or amputations caused by diabetic neuropathy), as reported in several studies^6,7.

Recently, a literature review⁸ (January 1997 to September 2007, ten studies) to identify the main methodological issues that explain the difference among reports in the relationship of non-compliance and costs in patients with diabetes was published. Salas et al ⁸ showed that research assessing that relationship is limited and of poor quality, and concluded that methodological differences among studies of costs of non-compliance in patients with diabetes make comparisons between the studies difficult. Another important review⁹ (1990–2004, seven studies) regarding the impact of non-compliance with diabetes treatment was conducted by Lee et al, indicating the negative impact of suboptimal compliance on costs and mentioning methodological differences among studies. The current review (January 2000 to April 2009, 12 studies) further explores the cost consequences of non-compliance to pharmaceutical treatment in T2DM, and provides important information for readers in addition to the above-mentioned review papers. The primary objective of the current review was to investigate the impact of non-compliance with pharmaceutical treatment on diabetes-related costs in patients with T2DM. Secondary objectives were as follows: (1) to determine the variability in the design, populations and drugs studied in studies that explore the association between non-compliance to diabetes treatment and costs; and (2) review the covariates associated with diabetes-related costs or resource utilisation in T2DM patients.

Materials and methods

Search strategy

A search was carried out in PubMed for studies describing economic consequences of adherence and/or compliance and/or persistence with pharmacological treatment of T2DM. The following search term in the PubMed included: ‘diabet* AND (complia* OR adhere* OR persiste*) AND (cost* OR econo*)’ with the limits ‘Publication Date from 2000/01/01 to 2009/04/30, Humans, English’. A very broad search criteria was used to retrieve as many relevant articles as possible. The references for retrieved publications were also reviewed to identify further relevant studies. References to textbooks in retrieved publications were not reviewed, because it was assumed that the likelihood of finding published studies on compliance and costs in T2DM in a textbook would be minimal.

Selection criteria

Studies were considered to be relevant if they involved patients with T2DM, examined adherence and/or compliance and/or persistence to pharmaceutical interventions, quantified the economic consequences of non-compliance, provided an economic evaluation or cost analysis, were in the English language, and were published between January 1 2000 and April 30 2009. The search period was based on American Diabetes Association recommendations for specific targets of glycaemic control introduced during the last decade, as well as market introduction of new antidiabetic medications in that period, which provided better opportunities for meeting these recommendations¹⁰. Earlier studies were not considered because: (1) results from those studies could not be compared with those from more recent studies, owing to changes in treatment patterns and healthcare resource prices, and (2) earlier studies were covered by a systematic literature search review paper by Lee et al ⁹. Studies were excluded from analysis if the economic consequences of adherence and/or compliance and/or persistence were not quantified.

Data extraction

The extracted data included the country where the study was performed, number of patients, patient characteristics at baseline, data source, study design, antidiabetic drug treatment, the duration of follow-up, definitions used for adherence, persistence and compliance, resource utilisation and costs. The data extraction was performed by one researcher. The data extraction process was subject to quality check, done by another researcher via comparison of extracted data with data from original publication.

Results

Search results

A total of 449 titles were identified and their abstracts were reviewed independently by two researchers based on pre-defined selection criteria. Based on the relevant information from abstracts, full-texts for 32 abstracts were retrieved and reviewed. Reasons that papers were found to be ineligible included the following: no costs provided, population without T2DM, neither adherence nor persistence nor compliance to pharmacotherapy described, and published protocol or design of the study without the results. Thirteen papers (referring to 12 studies) were identified that analysed the economic impact of adherence and/or compliance and/or persistence in the T2DM treatment. The extracted data from papers describing these 12 studies^11–22 is summarised in the Table 1.

Table 1. Studies describing the impact of compliance or persistence to treatment with antidiabetes medication on healthcare costs

Studies	Database	Study design	Measures of compliance or persistence	Patients	Baseline	Therapy	Compliance	Resource utilisation	Costs	Association between compliance and costs/ resource utilisation
Shenolikar et al, 2008^11USA3,137 patients	North Carolina Medicaid database	Retrospective, longitudinal cohort study 1-year follow-up	Medication possession ratio (MPR)*	T2DM patients initiating oral monotherapy and continued with the same medication	Mean age: 48.6 years	Metformin, sulfonylureas, and thiazolidinediones (TZD)	Mean MPR 0.56	Hospitalisations, emergency department (ED) visits, physician visits, medications	Mean total annual cost: $10,000	Increased medication compliance was associated with a decrease in resource utilisation
Pawaskar et al, 2007^12USA11,878 patients	North Carolina Medicaid database	Retrospective, longitudinal cohort study2-year follow-up	MPR*	T2DM patients who initiated insulin with, or converted to, a pen device versus a comparison syringe cohort	Mean age, weighted: 52.7 yearsMale: 26%	Insulin administered via pen or vial/syringe as add-on therapy to OAD	Mean overall MPR (syringe to pen) 0.88; (syringe only) 0.93; (OAD to syringe) 0.92; (OAD to pen) 0.93	Hospitalisations, ED visits, outpatient physician visits, antidiabetic medications, use of other medications for comorbid conditions	Mean total cost in year 1 excluding prescriptions (syringe to pen) $7,096; (syringe only) $7,013; (OAD to syringe) $17,338; (OAD to pen) $7,632	MPR during year 1 was significant predictor of total healthcare costs
Shenolikar et al, 2006^13USA1,073 patients	North Carolina Medicaid database	Retrospective, longitudinal cohort study3-year follow-up	MPR*	T2DM patients initiated pioglitazone	Mean age: 49.5 yearsMale: 26%African-American: 50%	TZD (pioglitazone)	Mean MPR 0.59	Hospitalisations, ED visits, outpatient physician visits, antidiabetic medication, other prescription drugs	Mean total healthcare costs in year 1 $7,906; year 2 $9,546	Increased compliance was associated with significant reductions in total healthcare costs
Cobden et al, 2007^14USA486 patients	PharMetrics database of medical and pharmaceutical claims	Retrospective, longitudinal cohort study2-year follow-up	MPR*	T2DM patients converted to the BIAsp 70/30 pen device for the 1st time	Mean age: 45.1 yearsMale: 56%	Insulin administered via BIAsp 70/30 device	Mean MPR prior to the conversion to pen device 0.59 and afterwards 0.68	Hospitalisations, ED visits, outpatient physician visit, medications	Mean annual total costs prior to the conversion to pen device $16,004 and $14,256 afterwards	In subjects with MPR of 0.80 or greater the frequency of hypoglycaemic events significantly decreased, resulting in reduced costs of hypoglycaemia- attributable ED resource utilisation
Karve et al, 2008^15USA4,943 patients	Arkansas Medicaid database of administrative claims	Retrospective, longitudinal cohort study1-year follow-up	MPR* and additional measures of compliance^15	T2DM patients receiving OADs	Mean age: 60.9 yearsMale: 24%	Monotherapy with OAD, ignoring medication switches within the oral hypoglycaemic class	Mean MPR 0.76	Hospitalisations, outpatient services, inpatient services, and nursing home	Mean diabetes- related hospitalisation and non-pharmacy costs per year $295 and $667, respectively	Compliance increases diabetes-related non-pharmacy costs and decreases the diabetes-related hospitalisation costs
Balkrishnan et al, 2006^16USA4,710 patients	North Carolina Medicaid database	Retrospective, longitudinal cohort studyBetween 2 and 4.5 years follow-up depending on study cohorts	MPR*Persistence†	T2DM patients receiving OADs	Mean weighted age: 48.6 yearsMale: 25%	Patients starting TZD, metformin or sulfonylureas	Mean MPR in year 2 was 0.59 in TZD group and 0.52 in other OADs group;Mean persistence rate in year 2 was 0.76 in TZD group and 0.71 in other OADs group	Hospitalisations, ED visits, outpatient physician visits, utilisation of antidiabetic medication, and medications for any comorbid conditions	Mean total healthcare costs in year 2: $9,323 in TZD group and $11,619 in other OADs group	Treatment compliance is associated with decreased hospitalisation risk and health-care cost reduction
Sokol et al, 2005^17USA3,260 patients	Administrative and prescription claims databases	Retrospective, longitudinal cohort study1-year follow-up	MPR*	T2DM patients receiving medication treatment to treat diabetes (further details on treatment not provided)	Mean age: 53.9 yearsMale: 55%	Patients receiving insulins and/or OAD prescriptions	In 55.2% of patients. MPR 0.80 and higher	Hospitalisations, ED visits, outpatient physician visits, medication prescriptions	Mean total healthcare costs per year $4,570 in patients with compliance rate of 0.80 and higher and $8,867 for those having compliance rate of 0.19 and lower	Increased compliance is associated with decrease in hospitalisation risk, decrease in total and diabetes-related healthcare costs, despite the increased drug costs
Hepke et al, 2004^18USA57,687 patients	Blue Cross Blue Shield of Michigan claims database	Retrospective, longitudinal cohort study1-year follow-up	MPR*	Patients with DM and/or having claims with a diabetes diagnosis and/or having a filled prescription for insulin or OAD	Mean age: 52.6 yearsMale: 55%	Patients receiving insulins and/or OAD prescriptions	In 45.6% of patients. MPR 0.80 and higher	Hospitalisations, emergency care, outpatient care, clinic visits, laboratory tests, professional services and pharmaceuticals	In patients with compliance rate of 0.80 and higher, mean total costs per year were about $7,300; in patients with compliance of 0.19 were about $7,100	Increased compliance is associated with decreased medical care costs, but not with decreased overall healthcare costs, because medication costs offset medical care cost savings
Balkrishnan et al, 2003^19USA775 patients	Administrative claims data from Medicare in North Carolina	Retrospective, longitudinal cohort study, up to 5 years of follow-up	MPR*	T2DM patients ≥65 yrs old	Mean age: 74.2 yearsMale: 42%	OADs (77% of subjects), insulin (13% of subjects), or a combination (10%)	Mean MPR ranged from 0.71 to 0.78 across the 5 years	Hospitalisations, emergency care, outpatient visits, other healthcare services, medication prescriptions	Mean total healthcare costs ranged from $5,043 to $8,306 per patient across the 5 years	Increased MPR was associated with decreased total annual healthcare costs
Cherry et al, 2002 ^20USA169 pts	Telemedicine diabetes disease management program, Texas	Prospective study with a retrospective component, 1 year of follow-up	Other measure of compliance ‡	Indigent or economically disadvantaged adult patients with diabetes mellitus	Mean age: 53 yMale: 23%	Antidiabetes medications not specified	Percentage of patients regularly taking their antidiabetic medication increased from 34% at baseline to 94% at study end	Hospitalisations, ED visits, postdischarge care visits, outpatient visits, charges for healthcare services	Mean total diabetes-related costs per year $7629, adjusted charge reductions per patient after the program intervention $747	Increase in percentage of patients taking their medicines regularly led to reduced mean total diabetes-related costs per year
Chiechanowski et al, 2000^21USA367 patients	Data from two primary care clinics of the Group Health Cooperative HMO, Washington	Cross-sectional study	Other measure of compliance^§	T1DM and T2DM patients receiving antidiabetic pharmacotherapy	Mean age: 61 yearsT2DM: 96%Male: 44%	Insulin (15.9%), OADs (51.5%), combination of both (19.5%), diet only (10.9%)	Percentage of days in interruption of antidiabetic pharmacotherapy was 14.9% in high depressive symptom severity tertile, and 7.1% in low severity tertile	Primary care, specialty care, ED visits, prescriptions, inpatient care, ambulatory care, mental health	Mean total 6-month healthcare costs $3,564 in high depressive symptom severity tertile, and $2,094 in low severity tertile	Higher depression severity in diabetes patients leads to the higher percentage of days in treatment interruption and to higher total healthcare costs
Lee et al, 2006^22USA1,156 patients	PharMetric database with medical and pharmacy claims of individuals from 57 managed care plans	Retrospective, longitudinal, switch-over study,2 years of follow-up	MPR*	T2DM patients whose treatment was converted from conventional human or analogue injection to a pre-filled insulin analogue pen	Mean age: 45.4 yearsMale: 53.8%	Insulin (100%), patients converted from conventional human or analogue insulin injection (vial/syringe) to a prefilled insulin analogue pen	Mean MPR increased from 0.62 to 0.69	Hospitalisations, ED visits, outpatient physician visits, pharmacy	Mean total annual treatment costs reduction from $16,359 to $14,769	Reduction in hypoglycaemic events occurred concurrently with improved compliance and with lower treatment costs

ED, emergency department; HMO, health maintenance organisation; MPR, medication possession ratio; OAD, oral antidiabetic drug; TZD, thiazolidinediones.

*Defined as the number of days of insulin prescription supply dispensed divided by the number of days between prescription refills.

†Estimated as number of days between start date of first prescription and end date of last prescription, divided by 365.

^‡Compliance reported as percentage of patients who report taking their medications regularly.

^§Compliance measured as percentage of days in interruption, by summing all days in interruption and dividing by the total number of days of intended treatment with antidiabetic medication.

Patient populations and antidiabetic drug treatment

All studies^11–22 were carried out in the US and involved analysis of inpatient, outpatient, and pharmacy claims data. Patient population size ranged widely in the studies, from 169²⁰ to 57,687¹⁸ patients. Mean age of patients across the studies ranged between 45.1¹⁴ and 74.2¹⁹ years; one of these studies included only patients aged 65 years or older¹⁹. Percentage of males ranged between 23%²⁰ and 56%¹⁴. One study included only indigent or economically disadvantaged patients²⁰. Three studies^12,14,22 provided the information on compliance and costs related to treatment with insulin, four studies^11,13,15,16 included patients receiving prescriptions for OADs, and four studies^17–19,21 included patients receiving prescriptions for insulin, OADs, or their combinations. One study did not specify types of antidiabetic medications administered to patients²⁰.

Compliance and persistence

Compliance was assessed via medication possession ratio (MPR) in ten studies^11–22. MPR was defined as number of days of antidiabetic prescription supply dispensed divided by the number of days between prescription refills as proposed by Steiner et al ²³. Other measurements of compliance included the percentage of patients who report taking their medications regularly²⁰, the percentage of days in interruption²¹, and proportion of days covered (PDC)¹⁵. MPR ranged from 0.52¹⁶ to 0.93¹² depending on regimen and drug administration (e.g. insulin pen vs. vial or syringe). Cobden et al ¹⁴ reported an increase in treatment compliance after conversion in premixed biphasic analogue and insulin glargine users who had received insulin analogue prescription in the preindex period. Regarding patients with human insulin prescription in preindex period, NPH users as well as previous premixed human insulin users showed an increase in compliance after the conversion to a premixed biphasic insulin analogue pen device¹⁴. Balkrishnan et al¹⁶ showed average compliance to new medications 0.59 in thiazolidinedione (TZD) cohort vs. 0.52 in patients with prescriptions for metformin or sulfonylureas. Average compliance to antidiabetic medications (OADs, insulin or their combination) prescribed to patients aged 65 years or older in another study ranged from 0.71 to 0.78¹⁹. Switch from conventional human or analogue insulin injection to a prefilled insulin analogue pen resulted in an improved compliance as assessed by MPR²². Persistence, measured as number of days between start date of the first prescription and end date of the last prescription and divided by 365, was assessed in one study and ranged between 0.71 per patient with TZD prescription and 0.76 per patient with prescriptions for metformin or sulfonylureas¹⁶.

Regarding associations between comorbidity severity and compliance there are some important controversies. Thus, according to Sokol et al ¹⁷, chronic disease score^24,25 showed significant positive correlations with compliance, but correlations between Charlson score²⁶ and compliance were not significant. On the contrary, Balkrishnan et al¹⁹ found that increased comorbidity severity according to Charlson score, self-reporting an emergency room visit in the pre-enrollment year, and use of injectable antidiabetic medications were associated with lower MPR. Chiechanowski et al²¹ found that depressive symptom severity was significantly associated with increased number of interruptions in refills of OADs.

Resource utilisation and costs

Mean total annual costs per patient having T2DM varied between the studies, ranging from $4,570¹⁷ to $17,338¹². The percentage of patients experiencing any-cause hospitalisation or emergency department visit per year ranged between 14%¹⁸ and 57%¹⁶. According to Balkrishnan et al ¹⁹, outpatient visits accounted for more than 50% of mean total reimbursed healthcare costs, hospitalisations and emergency room visits each accounted for about 20% of these costs, and the remaining 30% of these costs were reimbursements for prescriptions and costs for other healthcare services.

Comorbidity severity (e.g. Charlson score, chronic disease score) showed positive associations with diabetes-related healthcare cost¹⁷ and hospitalisation risk¹⁷. Sokal et al noted that age and gender had no effect on the diabetes-related healthcare cost and hospitalisation risk¹⁷. On the contrary, Hepke et al reported that younger age, female gender, higher illness severity (assessed by Diagnostic Cost Group relative risk score²⁷) and higher drug compliance level, were significant predictors of high diabetes-related costs¹⁸. Hepke et al also reported that younger patients, those having higher illness severity and being non-adherent to pharmacotherapy were more likely to have diabetes-related emergency department visit or an inpatient admission¹⁸. Higher levels of depressive symptom severity were significantly associated with increased total, ambulatory, and primary care 6-month costs compared to patients in the low-severity depression tertile²¹.

Impact of compliance on costs

In seven studies^{11–13,17,19–21} medication compliance was inversely associated with total healthcare costs per patient. According to Balkrishnan et al, an increase in antidiabetic MPR was associated with a decrease in total annual healthcare costs, after controlling for medication, comorbidity severity, and patient-reported emergency room visit in pre-enrolment year¹⁹. In particular, better compliance was associated with lower costs regardless whether patients received OADs or insulin therapy¹⁹. In four other studies^14–16,22 inverse associations between medication compliance and hospitalisation costs were reported. For example, Sokol et al ¹⁷ reported lowest hospitalisation rates (13%) for patients who had the highest level of medication compliance in diabetes (MPR ≥ 0.80). In patients with lowest compliance (MPR < 0.20) diabetes-related mean annual total costs and drug costs were $8867 and $55, respectively, while in patients with highest compliance (MPR ≥0.80), diabetes-related mean annual total costs and drug costs were $4,570 and $763, respectively¹⁷. Moreover, they observed linear decrease in hospitalisation rates from 30% to 13% as medication compliance increased¹⁷. According to Hepke et al ¹⁸, mean total, medical and drug costs per year and patient with MPR ≥ 0.80 were about $7,300, $4,800, and $2,500, respectively, while mentioned cost categories per year and patient having MPR < 0.20 were about $7,100, $6,000, and $1,100, respectively. As reported in the mentioned above study¹⁸, increased compliance did not significantly change overall healthcare costs, probably due to an increase in medication costs that offset medical care cost savings.

Discussion

The number of studies included in the review is very small, all studies were different in terms of patient population characteristics, time frames, and therapy regimen, and had complex study design which makes it extremely difficult to compare the studies. In other words, there was variability in the designs of the various studies evaluated and therefore there is likely to be inconsistency among the types of data available – resulting in potential difficulties in undertaking sound statistical analyses (e.g. meta-analyses) of these data. Thus, only very general statements can be made due to the high variability in the designs, populations, and drugs in the studies evaluated. The results of the current review suggest that treatment compliance may influence resource utilisation patterns and costs in T2DM. Definitions of compliance across the studies have not been uniform. Persistence was assessed in one study only, so any issues about persistence based on one article can not be discussed. Most studies described in this review show that increased antidiabetic drug compliance may reduce resource utilisation and overall healthcare costs in T2DM patients followed-up for 1 year or longer. These studies also suggest that higher medication costs associated with increased compliance regardless of the specific medications used may be offset by savings in direct medical costs due to reduced inpatient admissions. Literature suggests that the direct relationship between antidiabetic medication compliance, improved glycaemic control and reduced healthcare costs may exist^19,28. However, to the authors’ knowledge no study has specifically assessed that relationship between all three mentioned items and provided exact quantification of that relationship. Major limitation inherent in analyses of retrospective data is that causal inference cannot be made between compliance to treatment and resource utilisation or costs. It is important to bear in mind that in the studies included in this review, prescriptions filled were approximated to prescriptions taken. Thus, percentage of patients for whom prescriptions were filled out, but not actually consumed by the patients, is not known. Further limitation is that resource utilisation in studies is often underestimated, especially when patients self treat a hypoglycaemic event^29,30. And finally, only one database – PubMed was used in the current study. The authors believe that choice is justified by popularity of PubMed among researchers in the medical field, its practicality in use, optimal update frequency and the fact that it is free³¹. Due to budgetary constraints, widening the search to Embase or other databases requiring subscription was not considered. Therefore, it is possible that relevant articles not listed in PubMed, but available in other databases are not included in the current review.

It is important to mention, that variables such as age, gender, comorbidity severity, presence of diabetes complications, time since diagnosis of T2DM, presence of depressive symptoms and their severity, and treatment regimen may have considerable impact on compliance, resource utilisation or costs and, therefore, should be routinely collected and analysed in the studies investigating health economic aspects of compliance in T2DM.

Conclusions

Despite the high variability in the studies reviewed and the resulting difficulty in analysing these data, it may concluded that improved compliance can possibly lead to reductions of the total healthcare costs in T2DM. Studies assessing economic impact of persistence with pharmacotherapy in T2DM are limited, and studies investigating cost consequences of compliance are predominantly using MPR as a measure of compliance. Further research is needed in countries other than USA to assess impact of compliance and persistence to pharmacotherapy on T2DM costs in country-specific settings. To enable comparisons between results from different studies researchers should follow definitions of compliance and persistence proposed by the ISPOR Medication Compliance and Persistence Special Interest Group.

Acknowledgement

Declaration of interest: No funding for this study was provided. L.B., S.S., F-W.D. and O.S. have disclosed that they have no relevant financial relationships. Part of this work was presented as a poster at the ISPOR 12th Annual European Congress on 27th October 2009 in Paris.