Costs of treating cardiovascular events in Germany: a systematic literature review

Objective This study aims to systematically evaluate available evidence regarding direct medical costs of treating cardiovascular (CV) events in Germany after 2003 on an individual patient basis and from a payer perspective. The CV events of interest were myocardial infarction (MI), unstable angina, heart failure (HF), stroke, and peripheral artery disease (PAD). Method A systematic literature search was performed in the following databases according to Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines - Medline, Embase, Centre for Reviews and Dissemination, TIBORDER, and German dissertation database from January 2003 to October 2013. Both observational studies and randomized clinical trials were considered for the review. All values stated in € are inflation adjusted to 2014 € unless stated otherwise. Result This review included 13 articles. For newly occurred MI patients, the average hospitalization costs during the acute phase were reported to be between € 6790 and € 8918 per admission. In the first year after a MI event, direct medical costs were € 13,838–14,792 per patient. Direct medical costs of chronic HF patients were found to be between € 3417 and 5576 per patient per year. Treatment costs increase with disease progression. The average treatment costs for hospitalized PAD in the acute phase were reported to be € 4963 per admission, € 2535 per patient during month 1–6 after the initial hospitalization, € 1601 in month 7–12, and € 1390 in month 13–18. For stroke of all types, total direct medical costs in the 1st year after an event were reported to be € 13,273 per patient. Total direct medical costs during the 1st year after an ischemic stroke event were € 17,399–21,954 per patient, € 6260 in month 13–18, and € 6496 per year in the subsequent 4 years. Conclusion MI, unstable angina, HF, stroke and PAD have a high financial impact on the German health care system. Treatment costs of these diseases are mostly incurred during the acute phase of events and tend to decrease over time. Hospitalization and rehabilitation costs were two major cost drivers. Medication costs was one of the smallest cost component reported.

However, there is a gap in publications with respect to the individual treatment costs of CV events, especially after the implementation of German diagnostic-related groups (G-DRG) in hospitals in 2003 [7], which gave healthcare providers incentives to shorten the length of stay and be more cost-conscious [8].
This systematic literature review aims to provide an overview of the costs of treating individual CV event patients after 2003 from a third-party payer perspective. The CV events of interest were myocardial infarction (MI), unstable angina, heart failure (HF), stroke, and PAD.

Identification of studies
A systematic literature search was conducted to identify all available publications about individual-level direct medical costs of the CV events of interest in Germany after the G-DRG reform. We adopted the perspective of healthcare providers and German social third party payers, including statutory health insurance (SHI) and statutory nursing insurance (SNI) for cost studies. The scope of the costs was restricted to direct medical costs per affected patient, including hospitalization costs (based on the G-DRG system), outpatient costs (based on the points assigned to the particular treatment published in the German tariff list), medication costs (based on the reimbursement price of German SHI, rather than the price set by manufacturers), and rehabilitation and ambulance/paramedic costs (based on the costs reimbursed by the German SHI or SNI).
The review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [9].
Publications between January 1 st 2003 and October 1 st 2013 in English or German were searched in the following databases: Medline, Embase, Centre for Reviews and Dissemination (CRD) database, TIBORDER database, and German dissertation database.
Published papers, conference abstracts, and dissertations were all included in this review. Search terms are presented in Table 1.
Observational, epidemiological studies and randomized clinical trials (RCTs) with individual cost information were included. For RCT publications, the RCT was included in the review if there was original information about costs.
References of the RCT studies were reviewed by hand for sources of original cost data.
Exclusion criteria were: setting was not Germany; not SHI perspective or healthcare provider perspective adopted; not including the CV events of interest; costs data before 2003; costs on individual level not available or not calculable; costs reported originated from other studies (in this case the cited study was hand-searched and reviewed); and costs evaluation using health economic models or truncated approaches (no direct data).
One researcher conducted the search. If there was doubt regarding the inclusion/exclusion of an article, another independent researcher was consulted.
All values stated in € are inflation-adjusted to 2014 € unless stated otherwise.

Data extraction and analysis
Relevant data was extracted in a Microsoft Excel file. The extracted parameters included study identification, study characteristics (i.e. objective, setting, perspective), type of CV events, intervention (i.e. % of each cardiovascular treatments, such as drug treatment, PCI, CABG, etc.), participants characteristics (i.e. gender, age, comorbidities), direct medical costs (including cost in acute phase, costs within the first year after the event and follow-up costs from the second year onwards), conclusion and limitations of the study.

Results
The search resulted in 400 hits. The reviewer removed 50 duplicated papers, leaving 350 papers for title and abstract-based selection. There were 64 articles/abstracts (including 4 additional papers based on hand search from citations of relevant papers) included for full-text screening, among which 51 articles/abstracts were excluded based on the pre-defined exclusion criteria: 9 studies were excluded because the reported CV event (e.g. atherothrombosis) was not of interest for this study; 3 because German costs were not separately reported; 24 because they failed Table 1 Search terms for the systematic literature review a   Domain Search terms Subject "cost" OR "costs" OR "expenditure" OR "economic" OR "burden" OR "resource" Setting "German" OR "Germany" [full text] Cardiovascular events "myocardial infarction" OR "myocardial infarction" [MeSH] "unstable angina" OR "angina, unstable" [MeSH] "heart failure" OR "heart failure" [MeSH] "peripheral vascular disease" OR "peripheral vascular event" OR "peripheral artery disease" OR "peripheral vascular diseases" [MeSH] The terms are searched in title/abstract if there is no additional explanation to report individual costs, instead they reported incremental costs of interventions compared to base treatments; 8 due to lack of individual cost data; 4 studies reported costs based on data before 2003; 3 studies reported costs using a model or using truncated approach based on indirect data. Ultimately, there were 13 studies included in this review, with several studies addressing multiple CV events. The selection procedure was recorded following the PRISMA flowchart (Fig. 1).
In the literature, methods used to investigate costs of MI include a claims data analysis, a hospital database analysis, medical chart review, and calculations of costs based on healthcare resource utilization from expert panels (or literature review and expert interview) and SHI price. Although there was no clear definition in the included articles regarding acute and long-term treatment of MI, treatment costs can be roughly categorized into costs during acute phase (or hospitalization costs), costs within 1 year after event, and long-term costs after event (normally from the 2 nd year onwards).
During the first year after an AMI event, direct medical costs of treating AMI patients were reported to be from € 13,838 to 14,792/patient [10,17]. Total costs spent on male patients were about 21 % higher than those for females (p ≤0.0001) [13]. Treatment costs seemed to decrease over time during the 1 st year after an event. As estimated by Brüggenjürgen and Reinhold, 50-64 % of total costs in the first year were spent on treatment within the acute phase, and 80-85 % of total cost estimates were reached within 6 months after the event [13,17]. According to the study conducted by Brüggenjürgen et al. hospitalization costs covered the largest share (69 %) of total first-year costs, followed by rehabilitation costs (14 %) and a relatively small proportion on medication costs (8 %) [17].
Treatment costs were heavily influenced by treatment strategies and the medical devices used. For example, treatment costs of stent implementation depended on the type of stent used: the use of bare-metal stent (BMS) led to less resource use than drug-eluting stent (DES) (€ 13,629 vs. 14,792, p <0.05), with more than 50 % of costs occurring within 30 days after the implementation Fig. 1 Flow chart of included studies [10]. Among patients who underwent PCI, in-hospital treatment costs of contemporary antithrombotic strategies (between € 3545 and 4755, roughly 1/3 of the total treatment costs in the first year) were highly related to the anticoagulant drug(s) used [12]. Inpatient cardiac rehabilitation for MI in the first year was also a significant cost component of the total treatment costs, estimated to be € 2304/patient (around 14 % of the total 1 st year costs) [14].

Unstable angina
One study based on claims data was identified that examined the direct medical costs of treating unstable angina [18].
The published average costs for the inpatient treatment per unstable angina patient (acute phase) were € 4442, with inpatient cardiac rehabilitation being cheaper than inpatient treatment only [18]. For hospital-admitted patients, the leading cost drivers were hospitalization costs (€ 2414), followed by angiography costs (€ 829), diagnostic costs (€ 683), and drug costs (€ 391, equal to 9 % of the total costs) [18].

Heart failure
Two studies reported costs of treating HF from a SHI perspective in Germany [19,20]. Both were database analyses.
Direct medical costs of CHF were calculated to be between € 3417 and 5576/patient per year for patients of all New York Heart Association (NYHA) levels [19,20]. Biermann et al. found that the largest cost component was related to hospitalizations (€ 2525, 74 %); while costs of rehabilitation (€ 319, 9 %), medication (€ 315, 9 %) and outpatient contacts (€ 258, 8 %) were considerably less important cost components [19]. Compared with a cost of € 2683 per patient per year for NYHA level I patient, there was a cost increase in NYHA II, III and IV of 14, 48, and 71 %, respectively. About 76 % of the cost increase resulted from augmented hospital (inpatient) resource use [19].

Stroke
The review resulted in 5 papers regarding the direct medical costs of treating stroke in Germany [16,17,[21][22][23]. Different methods were applied in these studies, including claims data analyses, calculations based on resource utilization and SHI price, and retrospective cost-of-illness study.
For stroke of all types, total direct medical costs in the 1st year after an event were reported to be € 13,273 perpatient [22]. Male patients incurred approximately 10 % more costs than females. About half of the costs were incurred in the first 4 weeks after the event, 80 % of the costs were reached in the first 6 months [16,22]. Hospitalization costs accounted for the largest share (75 %) of the total costs of treating stroke [22].
For ischemic stroke, total direct medical costs in the 1 st year after the event were € 17,399-21,954 per patient [17,21,23]. 29 % of the total costs in the 1 st year incurred in the acute phase, 66 % were reached within 6 months after the event [17]. 1 st year treatment costs for ischemic stroke patients varied due to the treatment settings where they received rehabilitation treatments: hospitalized patients had around € 3352 more costs per patient per year compared to those treated in rehabilitation centers [21]. For patients admitted to a hospital for post-stroke rehabilitation, the leading components of 1 st year costs was hospitalization costs (€ 13,953), followed by medicines (€ 1915) and outpatient care (€ 1066). For patients who were admitted to a rehabilitation facility for post-stroke rehabilitation, the leading components were hospitalization costs (€ 8844), inpatient rehabilitation care (€ 4788), medicines (€ 1691), and outpatient care (€ 1317). For both groups, hospitalization (including rehabilitation) is the main cost driver for rehabilitation treatment of ischemic stroke patients [21]. Rehabilitation accounted for the largest share (37 %) of costs in the first year after an ischemic stroke event [23].
In month 13-18, costs of treating ischemic stroke were € 6260 per patient [17]. During year 2 to 5 after the event, direct medical costs decreased to € 6496 per patient per year [23].
No cost information regarding hemorrhagic stroke and transient ischemic attack was identified.

Peripheral artery disease
Only one paper was identified reporting direct medical costs of treating PAD [17].
Using a Delphi panel to obtain healthcare resource utilization and calculating costs based on reimbursed price, Brüggenjürgen et al. reported that the average treatment costs for hospitalized PAD in the acute phase (hospitalization) were € 4963 per admission, € 2535 per patient during month 1-6 after the initial hospitalization, € 1601 in month 7-12, and € 1390 in month 13-18 [17], meaning that 54.5 % of the 1 st year costs are consumed in the acute phase and 82.4 % within the first 6 months.
The key results of all included studies are presented in Table 2.

Conclusion
To summarize, MI, unstable angina, HF, stroke and PAD are leading to high direct medical costs in Germany. Treatment costs of MI, PAD and stroke mostly concentrate during the acute phase of the events and tend to decrease over time. Figure 2 shows the results of the studies that reported treatment costs in different time   frames. Hospitalization costs and rehabilitation costs are two major cost drivers for all CV events of interest. This suggests that it might be cost-saving to increase investment into the prevention of these CV events in order to avoid costly future hospitalization.

Discussion
This study is, to our knowledge, the first systematic literature review of direct medical costs of treating CV events on individual patient's level in Germany.
As what was found in this review, treatment costs of CV events are high in developed countries. However, the results are not comparable across studies from different countries because of potential various treatment patterns, reimbursement systems, different study years, etc. For example, it was reported that the treatment costs of AMI during 6 months post event were $ 2764-4953 (equal to 2014 € 2436-4365 [1]) in the US in 2005 [24]. In Sweden, the average costs of HF-related hospitalizations was SEK 72,613 (2014 € 8156) per patient per year; in contrast, annual prescription costs were low: on average 3 % of total cost (SEK 3503, 2014 € 393 per patient) [25]. Similar to that in Germany, in the US stroke seems to be the most expensive CV event among those included in this study. Based on 2.7,000 000 hospital admissions with a diagnosis of stroke in the US, the average total charge per admission were $ 46,518 (equal to 2014 € 35,202 1 ) in 2009 [26].
In Germany, although treatment costs of CV events are overall high, there was a substantial variance (around € 1000/patient per year) in the costs reported, especially in stroke treatment (up to € 3000/patient per year). Reported costs varied across studies due to the following reasons: 1) different study design, 2) different patient group, and 3) different treatment pattern.
Among the 13 studies identified in this review, research methods adopted included claims data analyses, which included relatively large sample sizes from multiple healthcare providers [10,11,13,21,22]; retrospective observational studies based on a few cardiovascular centers [12]; retrospective patient registry analyses, which composed of various RCTs and cohorts [19,23]; from expert interviews or medical chart review [16,17]; and RCTs [14,20].
As an example, Brüggenjürgen et al. [17] reported € 3780 and 4496 per patient more cost estimation for a stroke patient during the end of acute phrase to 6 months after event and 7-12 months after event, respectively, compared to Lindig et al. [22] (see Fig. 2). As the Lindig study is an observational one based on claims data, it represented the real costs in the German practice. The Brüggenjürgen study was based on experts' opinion regarding resource utilization and calculation using published price; it might represent the "recommended" treatment pattern and costs accordingly.
In general, results of RCTs are least representative for the treatment costs of general CV patients in Germany as they normally have strict recruitment procedures based on patients' baseline characteristics, health problems and medical history [27]. Results of the observational studies, including only a few high-volume cardiovascular centers, also need to be interpreted with caution as these hospitals are likely to treat more severe patients (referred to by secondary hospitals) leading to higher resource use and costs per case. Moreover, as all of these studies [12,14,20] evaluated the treatment costs of specific treatment strategies, e.g. use of anticoagulants, out-and inpatient rehabilitation, and a medical education program. With these differences across studies, it is difficult to ascertain representative data for direct medical costs in Germany. Due to different study objectives, researchers had variant inclusion and exclusion criteria (e.g. age, new disease vs. recurrent disease, disease of different levels, etc.) for cohorts. For example, Tiemann investigated the costs of primary treatments of AMI patients to compare variations in hospitalization costs over 9 European countries, including Germany [15]. In order to compare costs across countries; selection of patients was strict in this study. Only male patients between 50 and 60 years old without relevant co-morbidities (e.g. diabetes, hypertension, congestive heart failure, HIV infection) were eligible. Furthermore, patients were only included in the study if they came to an emergency department within 2 h of symptom(s) onset. Furthermore, in order to avoid the influence of very expensive procedures on the overall results, patients who needed bypass surgery and PCI with complications were also excluded from the analysis. Teaching hospitals have not been included because they might have greater resource intensity and higher costs per case. Due to the extensive restrictions in patient selection, the results of this study can hardly be generalized to all AMI patients in Germany.
Restrictions of patient groups were also implemented in several other studies. For example, Reinhold et al. and Kolominsky-Rabas et al. only included patients who experienced their first-ever event (AMI and stroke, respectively) and excluded recurrent events in their studies respectively [13,23]. How this restriction influences direct medical costs per case is not clear. In examining the costs of treating MI during acute phase, Fuchs et al. [16] made an assumption of a relatively old patient group for conducting the interviews, which makes their results less generalizable to a wider (younger) patient population.
Different mixes of CHF patients were included in the analyses with respect to NYHA level. The studies of Biermann et al. [19] included patients of all four NYHA categories, while the study of Peters-Klimms [20] included 98 % of NYHA II or III patients. As it was found by Biermann et al. [19] direct medical costs per case increased with severity of HF, the results based on patient groups of a certain NYHA level(s) might not be generalized to patient of other NYHA levels.
Different treatment patterns might influence the cost estimates reported in the studies. To compare the costs and effectiveness of different stents (DES and BMS), Bäumler et al. [10] only included patients who underwent PCI treatment due to their first incidence of AMI. As a result, the findings of this study do not represent costs of the general AMI patient population in Germany.
In order to compare resource utilization of ischemic stroke patients who underwent rehabilitation in a hospital-based unit or geriatric rehabilitation unit, Abbas et al. [21] only included patients who underwent rehabilitation treatment, whereas Lindig et al. [22] included patients who were hospitalized due to stroke. Therefore, costs of outpatient care were not included in these two studies.
Treatment costs are likely to be different across treatment settings (i.e. hospital, rehabilitation center, outpatient setting). For some diseases, such as AMI which is an acute and life-threatening event and most patients are admitted to a hospital, the results of studies including only hospitalized patients are likely to represent all patient groups. For example, the restriction of the study conducted by Stargardt et al. as inpatient only should not have significant influence on the representativeness of the result of this study [11]. However, for some diseases which can be treated in all settings, e.g. chronic heart failure, restrictions of treatment setting might have influence on treatment pattern and consequently treatment costs. For example, Biermann et al. included treatment in both primary and hospital settings [19]; while Peters-Klimm et al. only included patients treated in primary care setting [20]. The results of these two studies are therefore not comparable.
Despite the general awareness of the economic impact of CV events in Germany, the number of observational studies investigating the per-patient direct medical costs is relatively small. It should be noted that no study was identified for the costs of treating acute HF patients, haemorrhagic stroke, and transient ischemic attack from this literature review. There is a need for more research in this area. This literature review is limited to published studies through October 2013. Future literature reviews should be able to include more recent information on CV costs in Germany.