The German and Japanese health care systems: an international comparison using an input–output model

p u b l i c h e a l t h 1 4 1 ( 2 0 1 6 ) 6 3 e7 3

Available online at www.sciencedirect.com

Public Health journal homepage: www.elsevier.com/puhe

Original Research

The German and Japanese health care systems: an international comparison using an inputeoutput model €ffski b A. Rump a,*, O. Scho a b

Institut fu¨r Radiobiologie der Bundeswehr, Germany € t Erlangen-Nu¨rnberg, Germany Lehrstuhl fu¨r Gesundheitsmanagement, Friedrich-Alexander-Universita

article info

abstract

Article history:

Objectives: The German and Japanese health care systems have common roots, but have

Received 24 February 2016

evolved differently. Whereas the German system is often considered as expensive and

Received in revised form

poorly efficient, people in Japan are viewed as healthy and health care as comparatively

6 June 2016

cheap. In this study, we compared the quality, the effectiveness and efficiency of the

Accepted 24 June 2016

German and Japanese health care systems.

Available online 28 September 2016

Study design: This study includes comparative health care data analysis. Method: The quality and effectiveness of the German and Japanese health care systems

Keywords:

were analyzed using an inputeoutput model including 12 countries based on health in-

Germany

dicators published by the OECD. Besides the invested resources, a risk-related input

Japan

dimension was used for risk adjustment. The efficiency of the systems was assessed by

Health system

relating the average output to the health expenses per capita.

Inputeoutput model

Results: Health risks seem qualitatively different in Germany and Japan, but at the aggre-

Quality

gate level, lifestyle does not seem to be an outstanding explanatory factor for health

Efficiency

outcome differences between both countries. For investments in health resources, Germany is in a top position, whereas in the international comparison, the outcome is rather poor. The resources invested in Japan are also high, but slightly less than in Germany, whereas on average, the outcome is better. However, in the international comparison, resources as well as results in Japan show a very high variability. Relating the average output to the health expenses per capita indicates that on the average, the health care system in Japan is more efficient than in Germany. Conclusion: Germany and Japan have a quality problem with their health care systems. In Germany there is a transmission failure from structural to outcome quality that might be related to coordination problems between the outpatient and inpatient sector. Japan shows an unbalanced system that may be suspected to have a quality problem as a whole. As the development of the remuneration system including quality requirements is under the

* Corresponding author. Institut fu¨r Radiobiologie der Bundeswehr, Neuherberg Str. 11, D e 80937, Mu¨nchen, Germany. FAX: þ49 89 992692 2255. E-mail address: [email protected] (A. Rump). http://dx.doi.org/10.1016/j.puhe.2016.06.023 0033-3506/© 2016 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved.

64

p u b l i c h e a l t h 1 4 1 ( 2 0 1 6 ) 6 3 e7 3

direct responsibility and guidance of the Ministry of Health in Japan, the issue might however be more easily solved in Japan than in Germany. Although on average, health care seems more efficient in Japan than in Germany, taking into account health as well as longterm care expenses and uncertainties related to exchange rate adjustments, the higher efficiency of the Japanese system becomes questionable. © 2016 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved.

Introduction The German and Japanese health care systems have common roots. Japan introduced German medicine at the Meiji period (1868e1912) as a part of the technological modernization programme of the country.1 The financing of both health care systems mainly relies on public health insurances (Bismarck system),2 although the contribution from tax funds is much higher in Japan (about 38%)3 than in Germany (V11.5/181.5 billion in the public health insurance ¼ 6.3% for 2013; billion is used for 109 throughout the text).4,5 Medical care is provided by self-employed independent physicians in private practice or hospitals, and freedom of access for everybody is a high value in Germany as well as in Japan. Nowadays, both countries share common challenges leading to constantly increasing health expenses: demography is characterized by high life expectancies, combined with low fertility rates. Health expenses constantly raise and at the same time the share of the working population contributing to the social systems constantly decreases. Moreover, technological improvements enhance the costs of medical care in both countries. Despite common roots and challenges, the health systems have nevertheless evolved quite differently over time and at present also show substantial differences. In Germany, a private insurance may be a substitute to social insurance for employees with incomes exceeding defined limits (V4125/month in 2015), self-employed professionals and in the case of public servants for costs not reimbursed by the state. About 10% of the population are covered by a substitutive private insurance.6 In Japan, all residents are affiliated to the public health insurance. Affiliation depends on occupation or the location of residence. A private health insurance may be held in addition, but is not a substitute to mandatory public insurance.7 In Germany, outpatient and inpatient care are regulated quite differently, and each sector of care even has its own budget.8 Apart from patients owning a private insurance, patients have to consult at first a registered practitioner, acting as a gate keeper for inpatient care. Patients may not attend hospitals without a referral, except for emergencies. In Japan, the flow of patients is very poorly regulated. Patients are free to consult a practitioner at his private practice or clinic or to directly attend the outpatient ward of a hospital. Thus, outpatient care is an important source of income for hospitals. Although there is no need to be referred by a registered practitioner, hospitals may charge an extra fee to outpatients consulting without a referral.3,9 Medical services for outpatients are billed differently depending on the insurance status (patients with a private

insurance or affiliated to the public health insurance) in Germany. Inpatient care is remunerated with a case-based lump sum according to the German DRG system. Health expenses are controlled by budgets negotiated between the public health insurances as payer organizations on one side and the practitioner associations and hospitals as the care providers on the other side.10 In Japan, there is a unique fee-for-service scheme that is independent of the insurance status of the patient.11,12 The billing of inpatient treatment at hospitals is very complex: depending on the diagnosis and procedure, a day rate that decreases over time in three steps is applied (DPC/PDPS system). However, this daily fee does not include expensive medical procedures (e.g. surgery, anaesthesia, etc.) that are billed in addition according to the general fee schedule. Moreover, the DPC/PDPS system is not mandatory, and hospitals are free to opt for a billing of the individual medical services provided.13e15 Health expenses are not controlled by setting strict global budgets, but by regularly revising the valuation of the individual medical items of the unique fee schedule according to overriding financial targets. Although care payer and provider organizations are also involved, the development of the fee schedule occurs under the lead management and responsibility of the Ministry of Health, Labour and Welfare. At the same time, the development of the valuation of individual service items of the fee schedule permits to set incentives to achieve public health goals.12 Regarding quality and efficiency, the German system is often described as being very expensive but as having a mediocre outcome (‘you pay for a Mercedes, but you get a Volkswagen’).6 The Japanese system is sometimes presented as exemplary, leading to an excellent outcome (life expectancy is the highest in the world!), and at the same time, health costs seem quite low.2 On the other side, several experts and insiders express harsh critics toward its quality.16e18 Thus, it seems worth while to examine the German and Japanese health care systems more closely to understand their special distinctive features and compare their relative advantages and drawbacks.

Methods To compare the German and Japanese health care systems, an international comparison using an inputeoutput model with a ranking was performed including the G7 states and the five European states with the highest health care quality according to the study of Penter and Schulze19 (Sweden, Norway, Iceland, Switzerland, The Netherlands). As indicators, data from

p u b l i c h e a l t h 1 4 1 ( 2 0 1 6 ) 6 3 e7 3

the OECD health data basis 2014 were used.20 The data largely refer to 2012 with some results for 2011 (e.g. 30-day mortality after acute myocardial infarction). In a few isolated cases, older data had to be used. The indicators selected were assigned to one of three dimensions: risk- or resource-related input or output. Because the data used refer to quality, the model can be described as a quality-oriented inputeoutput model. At first, all indicators of the OECD health database were excluded if data for Germany or Japan were not available, as these countries are in the focus of this study. In a second step, it was evaluated whether the indicators could be reasonably assigned to one of the three input or output dimensions in a plausible manner. Indicators whose assignment was perceived as ambiguous were excluded (e.g. the suicide rate can be considered as a culturally conditioned risk-related input factor or as a result of prevention and treatment measures, i.e. an outcome indicator reflecting output and thus it was excluded). Besides purely health-related lifestyle factors (e.g. alcohol consumption, smoking habits), social indicators which might have an impact on health (e.g. average annual working time) were deliberately included into the risk-related input. On the other side, prevention measures were assigned to the resource-related input as they require particular investments. In the case where several indicators contained

identical statements, only the indicators considered as most representative were used (e.g. the number of physicians and nurses were included, but not in addition the number of dentists, midwives or physiotherapists). The inputeoutput model with the indicators is shown in Fig. 1. For every indicator, a ranking was performed among the 12 examined countries. For example, the rank number 1 was assigned to the country with the lowest 30-day mortality after acute myocardial infarction and the rank number 12 to the country with the highest mortality. For the three dimensions (risk- and resource-related input, output), the mean and the standard deviations of the rankings were calculated and again ranked for each country. In the case where the mean values for two or more countries were identical (connection of the values), the arithmetic mean of the series of the respective ranking numbers was assigned to these countries before further calculations. The correlations between the ranks of the average risk- and resource-related input and output were calculated according to Spearman. Moreover, the correlations between the ranks in the respective dimensions and the variability of the individual indicators expressed by the ranks of the standard deviations were examined. In the next step, only Japan and Germany were considered. For every dimension, the distributions of the individual indicators were analyzed, in order to recognize patterns suited

Risk-related input Tobacco consumption Alcohol consumption Adiposity Employment rate Annual working hours Expenses on education Gini at disposable income Output

Process

Resource-related input Density of physicians Density of nurses Acute care hospital beds Psychiatric care beds Annual hospital discharges Annual doctor‘s consultations CT units MRT units Breast cancer screening rate Cervix cancer screening rate Immunization rate against DTP

65

Life expectancy Low birth weight rate Perinatal mortality Maternal mortality Breast carcinoma survival rate Colon carcinoma survival rate Myocardial infarction mortality Ischemic stroke mortality Perceived health status Decayed-missed-filled teeth in children

Fig. 1 e The inputeoutput model.

66

p u b l i c h e a l t h 1 4 1 ( 2 0 1 6 ) 6 3 e7 3

to give hints about the specific structural features of the respective health care system. In order to assess the efficiency of the health care systems, the reciprocal of the arithmetic means of the individual ranks of the output indicators was calculated and multiplied by the factor 10,000 to get an index for effectiveness.19 This index value was divided by the health expenses per capita as given in the OECD health data base to calculate an effectivenessecost ratio as an indicator of efficiency of the health care systems.

Results The health indices for Germany and Japan are shown in Table 1 and the rankings of the input and output of all the countries included in the study are shown in Table 2. There was no significant correlation at the 5% level between the risk- and resource-related input (r ¼ 0.193, P ¼ 0.527). Whereas the output was significantly correlated with the risk-related input (r ¼ 0.708, P ¼ 0.009), surprisingly there was no significant correlation with the resource-related input (r ¼ 0.228, P ¼ 0.456). Although a low variability is considered as a sign of

quality, there was no significant correlation between the rankings and the standard deviations in either input or output dimensions (risk-related input: r ¼ 0.539, P ¼ 0.066; resource-related input: r ¼ 0.193, P ¼ 0.527; output: r ¼ 0.09, P ¼ 0.766). Regarding the risk-related input, Germany as well as Japan lie on a middle and almost similar position (rank 8 and 7, respectively) and show a quite high variability of the individual indicators (among the 12 countries studied, rank 8 and 10, respectively). Nevertheless, the risk patterns of the two countries are very different. Whereas alcohol consumption and obesity are much more pronounced in Germany, the annual working hours and income disparities seem to be a particular issue in Japan. Germany is the country with the best ranking for the resource-related input (rank 1) and also shows a low variability of the indicators in this dimension (rank 2). This reflects the high standards set up for formal structure and process quality assurance. Japan also belongs to the top countries regarding the resource-related input on the average (rank 3). But at the difference of Germany, the variability is very high (rank 12). The distribution of the rankings of the individual indicators shows two peaks at both ends of the spectrum

Table 1 e Health indices included in the input-output analysis for Japan and Germany (source: OECD Health data 201420). In parenthesis, the rank among the 12 examined countries is given for each indicator, the arithmetic mean and the standard deviation of the individual indicator ranks. Indicator Risk-related input Daily smokers in % of the population over 15 years of age Alcohol consumption, litres per capita at the age over 15 Obese population, % of total population (measured for Japan, self-reported for Germany) Employment rate (all persons aged 15e64 years) (%) Average annual hours actually worked per worker Expenditures on education (all levels), % of GDP Gini at disposable income, post-taxes and transfers (for Japan 2009, for Germany 2011) Arithmetic mean of the ranks ± standard deviation of the ranks Resource-related input Physicians, density per 1000 population Nurses, density per 1000 population Curative (acute care) beds per 1000 population Psychiatric care beds per 1000 population Annual hospital discharges, all causes, per 100,000 population Annual doctor's consultations, number per capita CT units per 1 million population MRT units per 1 million population Breast cancer screening, in % of females aged 50e69 screened Cervix cancer screening, in % of females aged 20e69 screened Immunization rate against DTP, in % of children immunized Arithmetic mean of the ranks ± standard deviation of the ranks Output Life expectancy, total population at birth, years Low birth weight, in % of total live births Perinatal mortality, deaths per 1000 total births Maternal mortality, deaths per 1,00,000 live births Breast carcinoma 5-year survival rate, in % Colon carcinoma 5-year survival rate, in % Acute myocardial infarction mortality in house (30 days), rate per 100 hospital discharges Ischaemic stroke mortality in house (30 days), rate per 100 hospital discharges Perceived health status (good/very good), in % aged over 15 years Decayedemissingefilled teeth, average number at age 12 Arithmetic mean of the ranks ± SD of the ranks

Japan

Germany

20.7 (9) 7.2 (4) 3.6 (1) 70.8 (8) 1745 (10) 5.14 (10) 0.336 (10) 7.429 (7) ± 3.289 (10)

21.9 (10) 11 (11) 14.7 (8) 73.3 (6) 1393 (2) 5.12 (11) 0.293 (6) 7.714 (8) ± 3.057 (8)

2.3 (12) 10.5 (7) 7.94 (1) 2.7 (1) 11,055 (11) 13 (1) 101.3 (1) 46.9 (1) 36.4 (11) 37.7 (12) 98 (2.5) 5.500 (3) ± 4.838 (12)

4.0 (2) 11.3 (5) 5.38 (2) 1.3 (3) 25,093 (1) 9.7 (2) 18.6 (6) 11.3 (7) 68.4 (8) 78.7 (2) 96 (8) 4.182 (1) ± 2.552 (2)

83.2 (1) 9.6 (12) 2.7 (2) 4.8 (7.5) 87.3 (4) 68 (2) 12.2 (12) 3 (1) 28.6 (12) 1.1 (7) 6.05 (6) ± 4.441 (12)

81 (10.5) 6.9 (8) 5.3 (6.5) 4.6 (6) 85 (8) 64.3 (4) 8.9 (11) 6.7 (6) 63.8 (11) 0.7 (1.5) 7.25 (10) ± 2.952 (8)

67

5.15 ± 3.899 3 10 3.90 ± 1.319 1 1 4.55 ± 2.296 2 4 5.375 ± 2.342 7.05 ± 1.890 4 9 5 2 6.35 ± 3.399 7 9 6.40 ± 4.294 8 11 5.875 ± 2.804 8.90 ± 2.782 5 12 7 6 8.50 ± 2.062 11 3 7.25 ± 2.952 10 8

4.182 ± 2.552 5.773 ± 2.691 6.545 ± 2.872 7.545 ± 2.536 6.682 ± 3.910 8.591 ± 2.566 5.773 ± 2.733 5.909 ± 2.937 5.278 ± 3.172 5.571 ± 2.834 7.273 ± 2.957 1 5.5 8 11 9 12 5.5 7 2 4 10 2 4 7 1 11 3 5 8 10 6 9

7.714 ± 3.057 8.714 ± 2.814 8.571 ± 4.031 8.429 ± 1.990 8.429 ± 3.812 7.143 ± 2.232 5.857 ± 2.642 5.143 ± 2.167 2.429 ± 0.904 4.429 ± 1.678 3.714 ± 3.104 8 12 11 9.5 9.5 6 5 4 1 3 2 8 7 12 3 11 5 6 4 1 2 9

Risk-related input Mean ± SD 7.429 ± 3.289 Rank of the mean 7 Rank of the SD 10 Resource-related input Mean ± SD 5.500 ± 4.838 Rank of the mean 3 Rank of the SD 12 Output Mean ± SD 6.05 ± 4.441 Rank of the mean 6 Rank of the SD 12

ISL SWE NOR NL CH CND USA GB IT FRA

There is no universally accepted method to assess the performance or adequacy of a health care system. Thus, quite different criteria have been used for international comparisons, and rankings of health care systems are very controversial. In its worldwide comparison, the World Health Organization (WHO) has formulated five intrinsic goals, including the level and distribution of health and responsiveness as well as the fairness of financial contribution.21 A composite measure of overall goal achievement was constructed using weights for the five components. The overall performance was assessed by relating the achievement to health system expenditures. A very different approach was chosen by Beske et al.:22,23 indicators related to the structural quality (e.g. density of physicians) or of the level of benefits offered by the health insurances play an important role. Thus, the results strongly reflect the ‘generosity’ of the social health systems. Another interesting approach consists of applying a multiple indicator multiple cause model (MIMIC), including structural and outcome data, to extract an overall quality

GE

Discussion

JPN

(Fig. 2). Japan is particularly in the lead for the number of hospital beds and heavy technical equipment (CT, MRT), but poorly ranked for the number of hospitalizations (rank 11). The indicators on preventive measures did not give a conclusive picture. Despite its excellent positioning for the resource-oriented input, Germany ends only on one of the last ranks (10) for the output. The variability of the outcome indicators is also quite high (rank 8) and the distribution between the ranks 6 e 11 is quite broad, showing a skew on the right. These findings correspond to the mediocre outcomes reported in the literature for the German health care system.19 Japan is better positioned on the average (rank 6), but again displays a very high variability in outcomes (rank 12). As with the resourcerelated input, the output displays a ranking distribution of the individual indicators with two peaks at both ends of the spectrum. The health expenses per capita, the index for effectiveness and the efficiency of the respective health care systems expressed as effectivenessecost ratio are shown in Table 3. There is no significant correlation between the expenses per capita and the index for effectiveness (r ¼ 0.042, P ¼ 0.886). The efficiencies of the health care systems however significantly correlate with health expenses per capita (r ¼ 0.601, P ¼ 0.036) as well as with the indices for effectiveness (r ¼ 0.727, P ¼ 0.006). The efficiency ranking of the examined countries is shown in Fig. 3. With relative high expenses per capita (rank 5), Germany belongs to the countries with a quite poor efficiency (rank 9). On the other side, Japan shows relatively low expenses per capita (rank 9), but in the international comparison, its health care system has a rather good efficiency (rank 4). With relatively low expenses per capita (rank 8, just lower than in Japan), the health system of Sweden shows the highest efficiency (rank 1). Among the examined countries, the highest expenses per capita are found in the USA (rank 1), at the same time combined with the lowest efficiency (rank 12).

Table 2 e Rankings of the mean and the standard deviations (SD) of the risk- and resource-related input and the output of the 12 examined countries. Country code: JPN: Japan; GE: Germany; FRA: France; IT: Italy; GB: Great Britain; USA: United States of America; CND: Canada; CH: Switzerland; NL: The Netherlands; NOR: Norway; SWE: Sweden; ISL: Iceland.

p u b l i c h e a l t h 1 4 1 ( 2 0 1 6 ) 6 3 e7 3

68

p u b l i c h e a l t h 1 4 1 ( 2 0 1 6 ) 6 3 e7 3

Risk-related Input

Resource-related Input

Output

Japan

55

55

44

44

44

33

33

33

22

22

22

11

11

11

55

00

00 00

11

00

22 33 4 4 5 5 6 6 7 7 8 8 9 910 10 11 11 12 12 13 13

11

22 33 4 4 5 5 6 6 7 7 8 8 9 910 10 11 11 12 12 13 13

00 00

11

22 33 4 4 5 5 6 6 7 7 8 8 9 910 10 11 11 12 12 13 13

Rank Rang

Rank Rang

Rank Rang

Germany 55

55

5

44

44

4

33

33

3

22

22

2

11

1

00

0 00

11

22 33 4 4 5 5 6 6 7 7 8 8 9 910 10 11 11 12 12 13 13

1

1

0

0

0

1 1

2 2

3 3

Rank Rang

4 4

5 6 7 8 9 10 11 12 13 5 6 7 8 9 10 11 12 13 Rang Rank

0 0

1

2

3

4

5

6

7

8

9

10 11 12 13

Rank

Fig. 2 e Distribution of the ranks of the individual indicators of the risk- and resource-related input and output for Japan and Germany.

value considered as a latent variable characterizing a health care system.24 Different perspectives and methods lead to quite different rankings, as comparatively shown for Germany and Japan in Table 4. A construct to assess the performance of a health care system should ideally fulfil criteria as feasibility, acceptability, sensitivity, reliability and validity. The accepted key performance domains are population health, the health service outcomes, the patient's experience, financial protection, equity and efficiency.25 However, the explicit and/or implicit goals of health systems in different countries may be set differently for historical or cultural reasons, affecting results, but also limiting the universal validity of the performance measure.26 Thus, there is no single ‘gold standard’ to assess the validity of a construct measuring the performance of a

health system. Moreover, depending on the target of the study, the indicators used may also not necessarily be selected from all key performance domains. More importantly, the ultimate goal of the construct should be suited ‘to promote the achievements of health system objectives’.27 The classic input-output analysis developed by Wassily Leontief is based on a representation of values of goods and services that flow between industrial sectors during a defined period of time, and in many developed countries, it is incorporated into the national accounting system.28 However, because of differences between the functional and accounting classification systems, it is almost impossible to construct a proper flow analysis and input-output table within the health sector.29 That's why for our study, we used an input-processoutput model as established in health services research.30,31

Table 3 e Health expenses per capita in US-$ PPP as indicated by the OECD20 (ranking in parenthesis), index of effectiveness of the health systems based on the output of our analysis, as reported in Table 2, and efficiency of the health systems (quotient index of effectiveness/health expenses per capita) of the 12 examined countries (country code as given in the legend of Table 2).

Health expenses Output-mean Index of effectiveness Efficiency Rank of the efficiency

JPN

GE

FRA

IT

GB

USA

CND

CH

NL

NOR

SWE

ISL

3649 (9) 6.05 1653 0.4530 4

4811 (5) 7.25 1379 0.2866 9

4288 (7) 8.5 1177 0.2744 11

3209 (12) 5.875 1702 0.5305 3

3289 (11) 8.9 1124 0.3416 7

8745 (1) 6.4 1563 0.1787 12

4602 (6) 6.35 1575 0.3422 6

6080 (3) 5.375 1861 0.306 8

5099 (4) 7.05 1418 0.2782 10

6140 (2) 4.55 2198 0.3579 5

4106 (8) 3.9 2564 0.6245 1

3536 (10) 5.15 1942 0.5492 2

69

p u b l i c h e a l t h 1 4 1 ( 2 0 1 6 ) 6 3 e7 3

SWE ISL IT JPN NOR CND GB CH GE NL FRA USA

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Efficiency Fig. 3 e Comparison of the efficiency of the health systems of the 12 examined countries. The country code is given in the legend of Table 2.

Our analysis uses health risk-, social- and non-monetary resource indicators on the input side and indicators of health care outcome to reflect the output of the system.32 Thus, it is not surprising that our ranking shows a very strong similarity with the results found using an MIMIC approach (spearman correlation of the rankings of 10 countries r ¼ 0.84, P ¼ 0.0000002); however, Sweden and Norway were not included by Manouguian et al.24 There is a much looser correlation with the results of the WHO21 (r ¼ 0.29, P ¼ 0.34). At the difference of the WHO method, distribution and fairness aspects were not considered in our study. Our findings are also not significantly correlated with the results of Beske et al.22,23 (r ¼ 0.02, P ¼ 0.95), as benefit indicators are heavily weighted by these authors, but not included in our analysis. The particular feature of our study is that the distribution patterns of the rankings of the individual indicators were examined for the input and output dimensions for Germany and Japan. This approach seems to permit conclusions on the special features of a health care system that cannot be recognized if considering only the aggregate rank of a country. Lifestyle in Germany is generally not considered as being particularly healthy.33 Our results, reflected by the risk-related input and its median position in the international

comparison, correspond to the expectations. On the contrary, lifestyle in Japan has often been considered as healthy, and has been quoted as one explanation for the high life expectancy of the Japanese.2 This view cannot be confirmed by our findings. Apart from a low part of the population being obese, social indicators as annual working hours or disparities in income distribution show that there is a particular health stress on living in Japan. The inclusion of a risk-related input dimension in our input-output analysis can be considered as a risk-adjustment procedure using several indicators. Our findings seem to show that although health risks seem qualitatively different in Germany and Japan, at the aggregate level, lifestyle does not seem to be an outstanding explanatory factor for health outcome differences between both countries. Health resources in Germany are top positioned, regarding the mean as well as the low variability of the rankings of the indicators. This is not surprising and corresponds well to the high priority given to quality assurance systems. But in comparison to the resources invested, the output of the German health care system is really disappointing. This corresponds quite well to the fact that the system is often described as expensive, whereas the results do not correspond to the investments.19,34 Unfortunately, our findings do not permit to draw conclusions on the causes of this discrepancy. The transformation relating the input to the output is not represented in our model, but hidden in the ‘process’ (Fig. 1) being a black box, and it may not be assumed that the production process is optimal or even comparable for different countries. The Japanese system is also very well ranked regarding the resource-related input. The output is moderate, but better at the aggregate level than in the German system. However, the resource-related input as well as the output are characterized by an extreme high variability, from top to poor performances. This meets the critic formulated exemplarily for the field of cardiac surgery that in too many institutions too many doctors each perform only a small number of interventions (‘low volume, high risk’).17,18 As standardization and a low variability in outcome are usually considered as attributes of a high quality, this reflects that there is a serious quality problem with the Japanese health care system, as has been argued by several authors for years.16,35,36 On the bottom line, our findings show that Germany as well as Japan has a quality problem with their health care systems, although the issues at stake are quite different. In Germany, the transmission of a high structural quality into a corresponding outcome quality is the major issue. A better coordination between the outpatient and inpatient sector could be a

Table 4 e Ranking of the healthcare systems of Germany and Japan as reported in several international health systems comparisons. Source WHO

21

Countries compared 191 (worldwide comparison)

Beske et al.22,23

14 (European countries, USA, Canada, Australia, Japan)

Manouguian et al.24

28 (European countries, USA, Canada, Australia, New Zealand, Japan, Korea, Mexico, Turkey)

Criteria

Ranking Japan

Ranking Germany

Overall attainment Overall performance Benefits in kind Benefits in cash Overall benefits Quality index calculated with a MIMIC model

1 10 6 5 7 2

14 25 1 3 1 12

70

p u b l i c h e a l t h 1 4 1 ( 2 0 1 6 ) 6 3 e7 3

promising approach. In Japan, it seems that in a first step, the resources invested should be matched and streamlined to achieve a more balanced system. The inclusion of stricter requirements for the remuneration of medical services into the unique fee system may permit to set the right incentives. As the development of this fee system is under the direct responsibility of the Ministry of Health, Labour and Welfare, this might be possible without great disruptive reforms. Besides quality and effectiveness, the efficiency of a health system taking into account the health expenses is an important parameter. The efficiency of a unit, e.g. the health care system of a country, is often evaluated against an efficiency curve representing the maximal output that can be achieved with a defined input, e.g. health expenses per capita used for an optimal allocation in production factors. The efficiency curve can be constructed based on defined health outputs and health expenses of a sample of countries using a stochastic frontier analysis. Thus, maximal efficiency is not absolutely defined, but empirically derived and quite relative.24 We used a simpler method by computing an index of effectiveness from the mean output and related it to the health expenses per capita, to use this effectivenessecost ratio for a relative comparison. If comparing our efficiency ranking with the findings of Manouguian et al.24 using a stochastic frontier analysis based on a quality index as output, data correlate quite well (spearman correlation, r ¼ 0.80, P ¼ 0.0026). If comparing our data with the overall performance as determined by the WHO,21 correlation is quite poor (r ¼ 0.05, P ¼ 0.87). However, independently on the method used to quantify efficiency, two pitfalls related to the health expenses must be considered, and these may also affect the interpretation of our findings: differences in accounting standards and parity adjustment methods. The data reported by the OECD20 are defined as including the sum of private and public spending in different categories. Classification systems may differ between countries, but as far as expenses are only classified under different headings, the total expenses are not affected. At the difference of Germany, medical prevention is mainly not financed by public health insurances in Japan, but is the responsibility of local governments.37 Thus, expenses for preventive health care are often not included in official Japanese statistics.35 The total health expenses for 2012 as published by the Health Ministry amount to ¥39.2 trillion (1012),3 corresponding roughly to ¥308 ,661 per capita (¥39.2 trillion/127 million population), less than the figure given by the OECD (¥375,804 per capita).20 Expenditures on prevention and public health are explicitly included in the total health spending reported by the OECD (category HC.6),38 and thus it may be assumed that this difference in responsibilities and accounting does not bias our efficiency ranking. An important issue however is the classification of expenses on long-term care. According to the OECD,38 longterm care is ‘typically a mix of medical (including nursing care) and social services’, and only medically related activities should be included in the health expenses. The border lines are however ill-defined and differences in classification have for example been shown to explain in part the substantial differences in health spending between Norway and the other Scandinavian countries (c.f. Table 3 and Melberg39), although the health systems of these Nordic countries are quite similar.

As Japan as well as Germany has a high proportion of elderly in their respective populations, this point is of importance for a founded interpretation of differences in health expenses and efficiency between both countries. At the difference of Germany, in Japan some medical services are financed by the longterm care insurance (e.g. medical home services as well as in medical facilities and general hospitals).40 In 2013, Germany spent 11% of its GDP on health and 1% on long-term care (total 12%), whereas Japan spent only 10.2% of its GDP on health, but 2.1% on long-term care (total 12.3%).41 If comparing the total spending per capita on health and long-term care,41 Germany still spends more than Japan (4819 þ 438 ¼ 5257 > 3713 þ 764 ¼ 4477 US-$ PPP); however, the difference of the total spending (15% in Japan compared to Germany) is less than the difference of the health expenses alone (24%). These figures correspond quite well to the differences in long-term care spending reported for 2005 (Japan 1751 vs Germany 1185 US-$ PPP per capita).40 Whereas in Germany there are quite strict definitions of disabling conditions to qualify for benefits from the long-term care insurance, Japan is pushing toward an integration of all social resources, including medical as well as non-medical social care, to support the continuity of daily life of the elderly.42 The interrelations of non-medical social care and the health care system are complex and our data do not permit conclusions about the effects on health care utilization in both countries. Great caution is required when drawing such conclusions as it was shown that opposite to common belief, even an improvement of long-term care does not automatically reduce the number of in-patient days at hospitals.43 For a comparison of the performance of health systems across countries, health expenses must be converted to a common unit. Health spending can be compared using market exchange rates, purchasing power parities (PPP) based on general price levels or health-specific purchasing power parities. PPP should convert national expenditures into ‘real’ expenditures and reflect differences in the volume of goods and services.44 However, at a higher aggregate level, despite a same volume reflected by an identical valuation in a PPPadjusted currency unit, even the relative composition of the basket of general goods may vary between countries. Healthspecific power parities are much more problematic. That is why the OECD still relies on general PPP calculated on a broaddefined basket of goods, although it does not specifically reflect health-related services.45 Our efficiency ranking is based on standard PPP, and thus our findings that the health care system in Japan is more efficient than in Germany must be judged with caution. Applying market exchange rates,46 instead of standard US-$ PPP on the health expenses per capita, reduces the advantage of Japan over Germany (12% vs 24%) (Fig. 4). Using wages of physicians or nurses as a proxy for health-specific input prices to adjust expenses per capita shows that expenses in Japan are almost similar (based on physician wages) or even higher (based on nurse wages) than in Germany (Fig. 4). The comparison of wages is however associated with large uncertainties as these show a large variability and results depend on the data base used. In Germany, the real incomes of hospital-based specialists compared to physicians owning a private practice is discussed very controversially (annual net income V117,000 in private practice vs V135,000 for

71

p u b l i c h e a l t h 1 4 1 ( 2 0 1 6 ) 6 3 e7 3

% +30

+17.4 %

+20 +9.8 %

+10

0 -1.1%

-2.5 %

-10 -12.3 %

-20 -24.2 %

-30

Market exchange rate

Health PPP input-adjusted (physician wage)

Health PPP output-adjusted (hospital surgical treatment)

Standard PPP adjusted

Health PPP input-adjusted (nurse wage)

Health PPP output-adjusted (hospital medical treatment)

Fig. 4 e Health expenditures per capita in Japan compared to expenditures in Germany (benchmark) using different parity adjustment methods. Data in standard US-$ PPP are from the OECD health data base 2014.20 Historical market exchange rates were determined using the calculator on the homepage of the Oanda Corporation.46 Calculations of health-specific input-adjusted parities are based on information on wages in the Journal of the German Medical Association for physicians,47 the German public service tariff table for nurses,51 and the indications of the Japan Nursing Association for physicians and nurses in Japan.50 Health-specific output-adjusted parities are based on the findings of Okamoto et al.52 and have been calculated as a mean for nine surgical or six medical hospital treatments (2007 data for Germany and 2011 data for Japan where only hospitals using the DPC/PDPS system are considered).

consultants at hospitals).47 But according to the tariffs for public hospitals, a consultant (Oberarzt, Entgeltgruppe III, Stufe 3) earns only V95,000 annually,48 so that figures reporting higher incomes may result from additional payments for overtime work. In Japan, hospital-based physicians are reported to earn only half of physicians in private practice (average annual net income ¥21 million in private practice vs ¥11.4 million at hospitals in 2011, but ¥16.4 million in ‘upper tier hospitals’).49 The figures given by the Japan Nursing Associations for physicians amounts to ¥1.2 million per month without details on the activity compared to ¥380,000 per month for experienced nurses50 (vs V2771/ month for a nurse, level 7a, highest grade in a public hospital in Germany in 201051). Wage-based adjustments must therefore be judged very cautiously in view of the uncertainties related to the real income situations. If applying a health-specific output-adjusted PPP based on the remuneration of hospitals (in Japan, only hospitals using the DPC/PDPS system are included), it seems that compared to Germany, total health expenses per capita in Japan are quite similar if only surgical interventions are considered (2.5%), but even higher than in Germany if health-specific PPP are calculated on the base of medical treatments (þ17.4%)52 (Fig. 4). Thus, the choice of the adjustment method matters a great deal as described in other studies.39 On the bottom line, if considering the uncertainties of adjustment combined with the higher expenses for long-term care including medical

components, it is quite questionable whether health care in Japan is really cheaper and more efficient than in Germany. The health system is an important part of the social security system and the future of its financing depends on the economic system as a whole. On the average, health expenses between 2002 and 2012 grew annually by 2% in Germany, and expenses of the public health insurances as a percentage of GDP remained even constant for a decade (1998: 6.6%, 2008: 6.5%), indicating that the control of cost development seems quite effective.6 In comparison, the average annual growth rate of health expenses between 2002 and 2012 was 3.5% in Japan.53 This might also be due to the continuous ageing of the population combined with the fact that despite health expenses for younger people being less, expenses for the elderly have been reported to be higher in Japan than in Germany.54 On the other side, despite annual fluctuations, GDP growth in Germany is higher than in Japan, where even negative rates were recently observed (mean GDP growth rate from 2011 to 2014: 1.5% vs 0.7%).55 So, considering the still stagnant economy in Japan despite ‘Abenomics’,56 it may be difficult in the future to sustain the health system at its current level in Japan.

Conclusion Our findings show that the input-output model developed in this study can be used not just for an international ranking at

72

p u b l i c h e a l t h 1 4 1 ( 2 0 1 6 ) 6 3 e7 3

the aggregate level, but by analyzing the distribution of the individual indicators in the respective dimensions, particular features of a health system can be identified. This was used to compare the German and Japanese health care systems. Although Germany as well as Japan have established highlevel systems, both countries have to struggle with quality issues. Disappointing outcomes do not correspond to the expensive investments in Germany, whereas the Japanese system is very heterogeneous as a whole. Although health expenses per capita are lower in Japan in standard US-$ PPP, it is not certain that its system is really more efficient than in Germany, if considering accounting differences between both countries and the parity adjustment method.

Author statements Ethical approval Data used in this analysis are indicators published by the OECD. Patients or volunteers were not involved in this study. No ethical approval was required or sought for the analysis in this paper.

Funding sources There was no funding for this study.

Competing interests The authors declare that they have no conflicts of interest.

Author contributions This study is a part of the MHBA thesis of A. Rump accepted by the Faculty of Laws and Economics of the Friedrich-Alexander € ffski University of Erlangen-Nu¨rnberg in 2015. Professor Scho provided critical input into the redrafting of the manuscript. Both authors approved the final draft.

references

1. Nakahara T. The health system of Japan. In: Raffel MW, editor. Health care and reforms in industrialized countries. University Park Pennsylviana: Pennsylviana State University Press; 1997. p. 105e33. 2. Trambacz J. Japan e weltweites Vorbild mit zuku¨nftigen € hmen C, Herausforderungen. In: Trambacz J, Borner A, Do € blreiter S, Maaß P, Mauersberg C, Wehage A, Schmeier S, Ro editors. Internationale Gesundheitssysteme im Vergleich. Hamburg: Diplomica Verlag; 2015. p. 21e31. 3. Ministry of Health, Labour and welfare (MHLW). Annual Health, Labour and Welfare Report 2013e2014. MHLW; n.d. http://www.mhlw.go.jp/english/wp/wp-hw8/dl/02e.pdf (last accessed 06 January 2016). 4. Bundesministerium fu¨r Gesundheit (BMG). Finanzierungsgrundlagen der gesetzlichen Krankenversicherung. BMG (last accessed 06 January 2016), http://www.bmg.bund. de/themen/krankenversicherung/finanzierung/ finanzierungsgrundlagen-der-gesetzlichenkrankenversicherung.html; 2015.

5. Bundesministerium fu¨r Gesundheit (BMG). Daten des Gesundheitswesens 2015. BMG (last accessed 06 January 2016), http://www.bmg.bund.de/fileadmin/dateien/Publikationen/ Gesundheit/Broschueren/151104_BMG_DdGW_A4.pdf; 2015. € lkopf M, Pressel H. Das Gesundheitswesen im 6. Scho internationalen Vergleich. 2nd ed. Berlin: Medizinisch Wissenschaftliche Verlagsgesellschaft; 2014. 7. Kobayashi Y. Five decades of universal health insurance coverage in Japan: lessons and future challenges. JMAJ 2009;52:263e8. 8. Wille E. Die korporative Koordination als Allokationsmechanismus im Gesundheitswesen. In: Rebscher H, editor. Gesundheits€ okonomie und Gesundheitspolitik im Spannungsfeld zwischen Wissenschaft und Politikberatung. Festschrift fu¨r Gu¨nter Neubauer. Heidelberg: Economica Verlag Medizinrecht.de; 2006. p. 427e40. 9. Keio University Hospital. Information for patients. Keio University Hospital; n.d. http://www.hosp.keio.ac.jp/annai/ raiin/syoshin/(last accessed 01 July 2015). 10. Preuß O. Krankenhausmanagement. Aktionsfelder und Managementinstrumente. Mu¨nchen: Oldenbourg Wissenschaftsverlag; 2013. 11. Uetsuka Y. Characteristics of Japan's health care systems and the problems. JMAJ 2012;55:330e3. 12. Ikegami N. Controlling health expenditures by revisions to the fee schedule in Japan. In: Ikegami N, editor. Universal health coverage for inclusive and sustainable development. Lessons from Japan. A World Bank Study. Washington, DC: World Bank Group; 2014. p. 69e100. 13. Okamura S, Kobayashi R, Sakamaki T. Case-mix payment in Japanese medical care. Health Policy 2005;74:282e6. 14. Matsuda S, Ishikawa KB, Kuwabara K, Fujimori K, Fushimi K, Hashimoto H. Development and use of the Japanese case-mix system. Eurohealth:25e30 (last accessed 06 January 2016), http://www.euro.who.int/__data/assets/pdf_file/0016/80440/, 2008;14. 15. Ishii M. DRG/PPS and DPC/PDPS as prospective payment systems. JMAJ 2012;55:279e91. 16. Henke N, Kadonaga S, Kanzler L. Improving Japan's health care system. McKinsey Quaterly::54e63 (last accessed 01 July 2015), http://www.mckinsey.com/insights/health_ systems_and_services/improving_japans_health_care_ system; 2009. 17. Minami K. Nihon to doitsu no iryou systemu no chigai (The differences between the Japanese and German health systems). n.d. http://www.kenko-kenbi.or.jp/tkf/pdf/002_03/ 20070314/lecture_minamikazutomo.pdf (last accessed 06 January 2016). 18. Minami K. Konna iryoude ii desu ka e Nihon de okonawarete iru iryou doitsu de okonawarete iru (Is this health system a good one e The health system in Japan and Germany). Tokio: Haru Shobo; 2004. €t 19. Penter V, Schulze J. Das Deutsche Gesundheitssystem e Qualita und Effizienz. Berlin: KPMG AG Wirtschaftspru¨fungsgesellschaft (last accessed 01 July 2015), http://www.kpmg.com/DE/de/Documents/deutschegesundheitssystem-qualitaet-effizienz-2014-kpmgcompressed.pdf; 2014. 20. Organisation for Economic Cooperation and Development (OECD). OECD Health Statistics 2014. OECD; n.d. http:77stats. oecd.org/index.aspx?DataSetCode¼HEALTH_STAT (last accessed 18 July 2015). 21. World Health Organization (WHO). The world health report 2000. Health Systems Improving Performance. WHO; n.d. http:// www.who.int/whr/2000/en/whr00_en.pdf (last accessed 06 January 2016). 22. Beske F, Drabinski T, Golbach U. Leistungskatalog des Gesundheitswesens im internationalen Vergleich e Eine Analyse

p u b l i c h e a l t h 1 4 1 ( 2 0 1 6 ) 6 3 e7 3

23.

24.

25.

26.

27.

28.

29. 30.

31.

32. 33.

34.

35. 36.

37.

38.

€ndern. Band I: Struktur, Finanzierung und von 14 La Gesundheitsleistungen. Kiel: IGSF; 2005. Beske F, Drabinski T. Leistungskatalog des Gesundheitswesens im €ndern. Band II: internationalen Vergleich e Eine Analyse von 14 La Geldleistungen. Kiel: IGSF; 2005. € ver J, Verheyen F, Vo € pel H. Qualita €t und Manouguian MS, Sto Effizienz der Gesundheitsversorgung im internationalen Vergleich. HWWI policy paper 55. Hamburg: Hamburgisches Weltwirtschaftsinstitut; 2010. Papanicolas I, Cylus J. Comparison of healthcare systems performance. In: Kuhlmann E, Blank RH, Bourgealt IL, Wendt C, editors. The Palgrave international handbook of healthcare policy and governance. UK: Palgrave Macmillan Publisher; 2015. p. 116e34. Kleinman A. Concepts and a model for the comparison of medical systems as cultural systems. Soc Sci Med 1978;12:85e95. Smith PC, Mossialos E, Papanicolas I. Performance measurement for health system improvement: experiences, challenges and prospects. Copenhagen: World Health Organization and World Health Organization, on behalf of the European Observatory on Health Systems and Policies; 2008. Raa T. Input-output economics: theory and applications. Featuring Asian economics. Singapore: World Scientific Publishing Co. Ltd.; 2010. Jewczak M, Suchecka J. Application of input-output analysis in the health care. Comp Econ Res 2014;17(4):87e104. Pfaff H. Versorgungsforschung e Begriffsbestimmung, Gegenstand und Aufgaben. In: Pfaff H, Schrappe M, Lauterbach KW, Engelmann U, Halber M, editors. Gesundheitsversorgung und Disease Management. Grundlagen und Anwendungen der Versorgungsforschung. Bern: Verlag Hans Huber; 2003. p. 13e23. Busse R. Methoden der Versorgungsforschung. In: Hey M, Maschewsky-Schneider U, editors. Kursbuch Versorgungsforschung. Berlin: Medizinisch Wissenschaftliche Verlagsgesellschaft; 2006. p. 244e51. Ross TK. Health care quality management: tools and applications. San Francisco: John Wiley & Sons; 2014. € se I, Wallmann B. DKV-Report “Wie gesund lebt Frobro € ln: Zentrum fu¨r Gesundheit der Deutschland?” 2012. Ko € ln; 2012. Deutschen Sporthochschule Ko € ffski O. Die Qualita €t des deutschen Lauerer M, Emmert M, Scho Gesundheitswesens im internationalen Vergleich. Schriften zur Gesundheits€ okonomie 18. Burgdorf: Health Economics Research Zentrum; 2011. Ikegami N, Campbell JC. Medical care in Japan. N Engl J Med 1995;333:1295e9. Bittner F, Hehmann U. Das japanische Gesundheitssystem. Gesundheitssysteme im Vergleich. Ausarbeitung zum Referat. Fachhochschule Neubrandenburg, Fachbereich Soziale Arbeit und Gesundheit, Studiengang Gesundheitswissenschaften. Norderstedt: GRIN Verlag GmbH; 2004. Tsugawa Y, Ikegami N, Miake N, Reich MR. Improving population health through public health centers in Japan. In: Ikegami N, editor. Universal health coverage for inclusive and sustainable development. Lessons from Japan. Washington DC: A World Bank Study; 2014. p. 163e77. Organisation for Economic Cooperation and Development (OECD). OECD Health Statistics 2015. Definition, sources and methods. Health Expend financing. OECD; n.d. http://www. oecd.org/els/health-systems/Table-of-Content-MetadataOECD-Health-Statistics-2015.pdf (last accessed 06 January 2016).

73

39. Melberg HO. Some problems with international comparisons of health spending. Oslo: Health Economics Research Programme at the University of Oslo (HERO); 2011. 40. Campbell JC, Ikegami N, Gibson MJ. Lessons from public longterm care insurance in Germany and Japan. Health Aff 2010;29:87e95. 41. Organisation for Economic Cooperation and Development (OECD). Health at a glance 2015. OECD indicators. Paris: OECD publishing; 2015. http://dx.doi.org/10.1787/health_glance2015-en (last accessed 06 January 2016). 42. Morikawa M. Towards community-based integrated care: trends and issues in Japan's long-term care policy. Int J Integr Care 2014;14:1e10. 43. Deraas TS, Berntsen GR, Hasvold T, Førde OH. Does long-term care use within primary health care reduce hospital use among older people in Norway? A national five-year population-based observational study. BMC Health Serv Res 2011;11:287. 44. Organisation for Economic Cooperation and Development (OECD). Eurostat-OECD methodological manual on purchasing power parities. Eurostat methodologies and working papers. European Union/OECD; 2012. 45. Cylus J, Smith PC. Comparative measures of health system efficiency. In: Papanicolas I, Smith PC, editors. Health system performance comparison. An agenda for policy, information and research. Berkshire: Open University Press; 2013. p. 281e312. 46. Oanda Corporation. Average exchange rates. n.d. http://www. oanda.com/lang/de/currency/average (last accessed 06 January 2016). € 47. Deutsches Arzteblatt. Niedergelassene verdienen weniger als € Kliniker. Dtsch Arztebl 2015;112(46):A1912. € € 48. Marburger Bund (MB). Tarifvertrag fu¨r Arztinnen und Arzte an € €usern (TV-Arzte/VKA) kommunalen Krankenha vom 17.08.2006, Stand 23.04.2015. MB (last accessed 06 January 2016), www. marburger-bund.de/tarifpolitik/tarifvertraege; 2015. 49. Ikegami N. Factors determining the allocation of physicians in Japan. In: Ikegami N, editor. Universal health coverage for inclusive and sustainable development. Lessons from Japan. A World Bank Study. Washington, DC: World Bank Group; 2014. p. 119e32. 50. Japan Nursing Association (JNA). Nursing in Japan. Working Conditions in Japan. JNA; n.d. http://www.nurse.or.jp/jna/ english/nursing/employment.html (last accessed 06 January 2016). € ffentlichen Dienst (TVo € D). TV€oD-K51. Tarifvertrag fu¨r den o €user. n.d. http://oeffentlicher-dienst.info/tvoed/k/ Krankenha (last accessed 06 January 2016). 52. Okamoto E, Kumakawa T. Estimation of per-case hospitalization charges from Diagnosis-ProcedureCombination (DPC) data and an international comparison of hospital prices. J Nat Inst Public Health 2014;63:532e7. 53. Mossialos E, Wenzl M, Osborn R, Anderson C. International profiles of health care systems 2014. New York, Washington: The Commonwealth Fund; 2015. 54. Fukawa T. Macro evaluation of the Japanese healthcare system in comparison with Germany. Jpn J Soc Secur Policy 2007;6:31e42. 55. World Bank. Data. GDP growth (annual %). World Bank; n.d. http://data.worldbank.org/indicator/NY.GDP.MKTP.KD.ZG (last accessed 8 January 2016). 56. Hilpert HG. Japan nach zwei Jahren “Abenomics”. SWP (Stift Wiss und Polit)-Aktuell 2015;24:1e8.