G Model
ARTICLE IN PRESS
HEAP-3210; No. of Pages 7
Health Policy xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Health Policy journal homepage: www.elsevier.com/locate/healthpol
Health care cost in Switzerland: Quantity- or price-driven?夽 Reto Schleiniger ∗ Zürich University of Applied Sciences, Center for Economic Policy, Bahnhofplatz 12, 8401 Winterthur, Switzerland
a r t i c l e
i n f o
Article history: Received 15 October 2013 Received in revised form 7 April 2014 Accepted 8 April 2014
Keywords: Costs and cost analysis Index calculation Regression analysis
a b s t r a c t In Switzerland, per capita health care costs vary substantially from canton to canton and rise considerably and steadily from year to year. Since costs are equal to the product of quantities and prices, the question arises whether regional cost variations and cost increase over time are quantity- or price-driven. Depending on the answer, the containment of health care costs must be approached differently. This article examines the cost of mandatory health insurance in Switzerland for the period from 2004 to 2010 and breaks it down into quantity and price effects. The main result of the cross-section analysis reveals that regional cost differences are mainly due to quantity differences. Similarly, the longitudinal analysis shows that the cost increase across all health care services is primarily caused by increasing per capita quantities. Any attempt to contain costs must therefore focus primarily on the extent of medical care utilization, and the key challenge to be met is how to identify medical care services which do not have a positive effect on patients’ health status. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Two observations serve as the starting point for the study: Firstly, there are large regional differences in per capita health care costs across Switzerland, reflected by the substantial variance in the premiums paid for mandatory health insurance in Swiss cantons. Secondly, there is a steady increase in per capita costs over time. Over the past fifteen years, costs for mandatory health care services outpaced overall GDP growth and amounted to 4.25% of GDP in 2010 [2]. In order to understand regional cost variation and be able to limit medical cost growth, we must first identify the sources of the regional disparity and of the cost growth.
夽 This study was commissioned by santésuisse, the Swiss association of health insurers. An earlier study covering only the period from 2004 to 2005 was published in 2008 and focuses on productivity and efficiency measures [1]. ∗ Tel.: +41 52 267 78 75. E-mail address: [email protected]
Since costs are the product of prices and quantities, the first step is to decompose costs into prices and quantities. Obviously, measures to contain costs must differ substantially depending on whether cost changes are driven by price or quantity. The present study examines mandatory health care services in Switzerland and covers the period from 2004 to 2010. Since mandatory health insurance covers a variety of services, prices and quantities across all services cannot be compared directly but must be calculated and expressed in terms of price and quantity indices. Interestingly, most studies on the decomposition of regional medical care cost variations are conducted in the USA [3]. Whereas in other countries regional variations in total health care expenditures is an issue, the cost differences are usually not split into prices and quantities [4,5]. In Switzerland, the decomposition of costs into prices and quantities is of special interest because, due to the federal state structure, the Swiss health care system is organized along cantonal borders. As a consequence, prices are not determined centrally and can vary from canton to
http://dx.doi.org/10.1016/j.healthpol.2014.04.004 0168-8510/© 2014 Elsevier Ireland Ltd. All rights reserved.
Please cite this article in press as: Schleiniger R. Health care cost in Switzerland: Quantity- or price-driven? Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.04.004
G Model HEAP-3210; No. of Pages 7
ARTICLE IN PRESS R. Schleiniger / Health Policy xxx (2014) xxx–xxx
2
canton. To what extent the decentralized price setting is responsible for regional cost variations is one of the questions that will be analyzed in the following.
2. Data and methodology 2.1. Data selection and preparation Most of the necessary data input has been extracted from the data pool of santésuisse, the Swiss health insurance association. The data pool comprises all mandatory health care transactions of the health insurance companies. To capture the full cost of mandatory health care services, the data supplied by santésuisse is complemented by data on public subsidies to hospitals as published by the Swiss Federal Statistical Office [6]. In addition, the price development of pharmaceuticals over time is described by the corresponding sub-index of the Swiss consumer price index, also published by the Swiss Federal Statistical Office. The study distinguishes various health care services such as out-patient services by physicians, physiotherapists, and hospitals, in-patient hospital services, pharmaceuticals, nursing homes, and laboratory services. The in-patient hospital services are further divided into ten sub-categories ranging from university hospitals to local hospitals and specific clinics such as psychiatric institutions. In total, the costs of the 16 categories amount to almost 20 billion Swiss francs. Together with the yearly subsidies to public hospitals of approximately eight billion Swiss francs, costs in the amount of 28 billion Swiss francs are included in the calculations. The only other service worth mentioning is home care with a cost share of a little less than 2%. Here no meaningful split into price and quantity was possible, because home care services are charged partly on a time tariff and partly in the form of a lump sum. Since the available data does not distinguish the two forms of payment, home care was excluded from the data. While 10 different hospital categories were distinguished, some of the smaller Swiss cantons do not have many of these categories. For the purpose of this study, the semi-cantons of Appenzell-Ausserrhoden and AppenzellInnerrhoden, as well as Obwalden and Nidwalden, have been merged with their larger neighbors, St. Gallen and Lucerne, respectively. As a consequence, the original number of 26 cantons and semi-cantons has been reduced to 22 regions. In order to calculate per capita results, special attention needs to be given to the determination of the cantonal population figures. When using these figures, it must be remembered that care providers such as hospitals and medical practitioners not only offer services to the resident population but also across cantonal borders. Dividing the cantonal cost of care providers by the number of the resident population, therefore, yields biased per capita results. To correct this bias, a matrix of inter-cantonal service flows is used to convert the resident population into the “medically cared for population”. Per capita measures, then, are calculated as cost and quantities per medicated population.
2.2. Index calculation The challenge in trying to break down costs into quantities and prices lies in the fact that mandatory health insurance covers a range of different services. Therefore, prices and quantities across all services must be reproduced as indices [7]. In order to derive such indices, the cost of each service must be expressed individually as the product of quantities and prices. To this end, a plausible quantity and price indicator must be selected for each service. The following briefly describes what price and quantity indicators were chosen for each health care category. 2.2.1. Ambulatory services (out-patient hospital services, practitioners, and physiotherapists) The information provided by TARMED, the Swiss medical tariff scheme for out-patient services, was used to break down the costs for all ambulatory services. The scheme assigns each medical service a specific number of tax points, based on the time required to render the service. The tax points serve as a quantity indicator for out-patient treatments. Medical costs are then determined by multiplying the tax points by so called tax point values which are to be understood as the price of a standardized service. The tax point values are the result of a negotiation process. Regionally, they vary by up to 20%, which is one reason why prices for medical services vary across Switzerland. 2.2.2. In-patient hospital services and nursing homes With in-patient hospital services, it was not possible to use a similar procedure to the one used for ambulatory services to split the costs because no country-wide DRG scheme was in place in the period from 2004 to 2010. Therefore, “hospital day per hospital category” was used as the smallest standardized quantity measure that could be inferred from the data. Such a procedure implies that a hospital day at a given hospital category represents the same output all over the country. It does not, however, suggest that a day in a university hospital is comparable to, say, a day in a local hospital. The implicit price for in-patient hospital services was then calculated as total hospital cost per hospital day and category. Again, the costs per hospital day and category vary considerably from canton to canton and constitute a second reason why prices are different across cantons. 2.2.3. Pharmaceuticals and laboratory services With pharmaceuticals and laboratory services, there are no regional price differences but only price changes over time. To account for these price changes on the national level, we again relied on tax point values for laboratory services and on the consumer price sub-index for pharmaceuticals as published by the Swiss Federal Statistical Office. This information on prices and quantities for each health care category was used to determine Paasche quantity and Laspeyres price indices. These were calculated in two ways, once without and once with public subsidies to hospitals. Since these subsidies vary substantially from canton to canton, the resulting price indices using net
Please cite this article in press as: Schleiniger R. Health care cost in Switzerland: Quantity- or price-driven? Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.04.004
G Model
ARTICLE IN PRESS
HEAP-3210; No. of Pages 7
R. Schleiniger / Health Policy xxx (2014) xxx–xxx
3
Fig. 1. Result of cross-section analysis: mean of quantity and price index from 2004 to 2010.
calculation without subsidies can differ considerably from gross indices which include the subsidies. 3. Results 3.1. Regional quantity and price indices To determine regional quantity and price indices, two types of reference regions are possible: either a specific region chosen at random or an artificially constructed average region. The former is known as the asymmetric star method, the latter the symmetric star method. This study uses the symmetric method. Therefore, the resulting indices must be interpreted relative to the Swiss average. For example, a per capita quantity index of 1.05 in a given canton signifies that the per capita health care demand is 5% larger than the average national demand.1 Fig. 1 exhibits the resulting indices of all mandatory health care services, calculated as the mean of the period from 2004 to 2010. The horizontal axis represents the regional quantity index per capita, the vertical axis the corresponding price index. As mentioned before, the price
1 This method belongs to the average basket methods and is also known as the ECLAC method (United Nations Economic Commission for Latin America and the Caribbean). See Ref. [8].
index was calculated as both a net and a gross concept, i.e. without and with public subsidies to hospitals. Also plotted are the isocost lines, which describe price–quantity combinations at a given cost level. Note that the product of Paasche quantity and Laspeyres price indices yields the value index, which stands, in the context of this study, for the cantonal per capita costs relative to the national average. It is now possible to classify the cantons according to their relative costs, with the high-cost cantons situated in the top right corner and the low-cost cantons in the bottom left corner. The figure shows that in Geneva per capita quantities are almost 30% and prices about 5% above average, resulting in per capita costs that are 35% higher than all over Switzerland. In Basel-Stadt, the other city canton besides Geneva, the quantities are also very high, but the prices are only average. Neuchâtel and Berne, on the other hand, exhibit high prices and average quantities, while Ticino, the Italianspeaking canton, is characterized by large quantities at prices slightly below average. The high gross price index of Neuchâtel is the result of the relatively high subsidies hospitals receive in this canton. By contrast, in Zurich the relative gross price is clearly below the relative net price, indicating that public subsidies to Zurich hospitals are relatively small.
Please cite this article in press as: Schleiniger R. Health care cost in Switzerland: Quantity- or price-driven? Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.04.004
G Model HEAP-3210; No. of Pages 7
ARTICLE IN PRESS R. Schleiniger / Health Policy xxx (2014) xxx–xxx
4
Table 1 Regression results to explain all mandatory health care services per capita: random effects method. All services including subsidies
2006–2010
Pharmaceuticals
2004–2010
Explanatory variable Share of women Age > 75 Population density Unemployment rate Density of specialists cor. R2
Coefficient 2.895*** 0.186** 0.007** 0.056** 0.170*** 0.703
Explanatory variable Density of GP Density of specialists Density of self-dispensing physicians Density of pharmacies Dummy variable indicating Latin language cor. R2
Coefficient 0.151*** 0.183*** 0.049*** 0.083* 0.150** 0.503
* ** ***
Level of significance: 90%. Level of significance: 95%. Level of significance: 99%.
On the other side of the cost spectrum, we find the cantons of Central Switzerland and the canton of St. Gallen. In all these regions, prices, and particularly consumed quantities, are below average, yielding per capita costs that are up to 20% lower than the Swiss average. Fig. 1 also shows that the quantity variance is larger than the price variance.2 Therefore, it can be concluded that the cost differences are mainly caused by quantity differences. However, there is a positive correlation between prices and quantities, which indicates that, without price differences, the cost differences would be smaller. 3.2. Panel data analysis Subsequently, an econometric panel analysis of regional quantity differences was performed. The regression model included spatial (22 regions) and time dimensions (7 years). The set of variables which are used as the determining factors can be divided into demand-side factors such as age and gender, supply-side factors such as physician density, and insurance-specific factors such as the size of the deductible [9,10]. Table 1 depicts the regression results with the per capita quantity index of all health care services as the explained variable and after eliminating all non-significant variables. It shows that the share of women, the share of the population aged above 75, the population density, and the unemployment rate, as well as the density of specialists, all yield the expected positive influence on consumed quantities. For the period from 2006 to 2010, the corrected determination coefficient amounts to 70% and is larger than the corresponding coefficient for the period from 2004 to 2010. Due to the introduction of a new accounting scheme for out-patient services in the year 2004, which led to a delayed booking of delivered services the data on outpatient services for the period from 2004 to 2005 might be biased, which reduces the explanatory power of the regression over the entire period of observation. Since the regression is based on indices, the resulting coefficients cannot be easily interpreted quantitatively. However, the strongest influence emanates from, in this order, the density of specialist practitioners, the share of
2 In logarithmic form, the cost variance can be decomposed into the price variance, the quantity variance, and twice the covariance of prices and quantities. In the gross calculation with subsidies, the quantity variance accounts for 77 and the price variance for only 6% of the total variance.
women, and the share of senior citizens. In the case of the specialists, this means that the quantity difference between a canton with a high density and a canton with a medium density of specialists is relatively large. In a next step, an additional regression is conducted with the quantity of pharmaceuticals as the explained variable. On the right hand side of Table 1, it can be seen that all the supply side factors exert a positive influence on quantities (with the density of pharmacies only significant at the 90% level). In addition, the dummy variable for Latin language is significant at the 95% level, which indicates that in the French- and Italian-speaking cantons more pharmaceuticals are consumed. The strongest influence on consumed pharmaceuticals is exerted by the density of self-dispensing physicians. However, it must be noted that this variable also varies strongly from canton to canton.
3.3. Price and quantity indices over time, 2004–2010 The results of the longitudinal section are shown for the entire country only since the regional dynamic patterns are in most cantons similar to the national development. All price and quantity indices were first calculated as yearly indices. Price and quantity changes over more than one year were then determined with a multiplicative chaining of the year-to-year values. Since the total costs are the product of prices, quantities per capita, and population, the cost index can be decomposed into the three indices for prices, per capita quantity, and population. Table 2 highlights the results of such decomposition for the total of health care services with 2004 and 2005 as the respective starting year. Thus it is shown that the total cost increase in 36% between 2004 and 2010 was caused by a 5% increase in population, a 22% increase in per capita consumption, and a 5.5% increase in prices. The price increase equals more or less the increase in the general consumer price level from 2004 to 2010 which amounts to 5.7%. In Table 2 it can also be seen that the quantity increase since 2005 is clearly smaller than it has been since 2004. Such a large difference is not very plausible and might be attributed to the above-mentioned introduction of a new accounting scheme for out-patient services in 2004. The delayed booking of services from 2004 in 2005 yields an upward bias from 2004 to 2005 and, consequently, a downward bias in the following period from 2005 to 2006.
Please cite this article in press as: Schleiniger R. Health care cost in Switzerland: Quantity- or price-driven? Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.04.004
G Model HEAP-3210; No. of Pages 7
ARTICLE IN PRESS R. Schleiniger / Health Policy xxx (2014) xxx–xxx
5
Table 2 Development of costs from 2004 (2005) to 2010: Price, quantity and population effects.
All services including subsidies
In-patient hospital services including subsidies Out-patient services by physicians Out-patient hospital services Pharmaceuticals
Prices
Quantity per capita
Population
Cost
1.0558 (1.0610)
1.2180 (1.1030)
1.0544 (1.0500)
1.3560 (1.2277)
Prices
Quantity per capita
Cost per capita
1.2476 (1.2231)
0.9875 (0.9776)
1.2320 (1.1956)
0.9824 (0.9994)
1.5448 (1.0850)
1.5176 (1.0844)
0.9747 (0.9854) 0.8300 (0.8374)
2.1578 (1.7105) 1.3311 (1.2802)
2.1031 (1.6855) 1.1049 (1.0720)
The lower part of Table 2 shows the national results of the longitudinal section for specific services with large cost shares. Looking at out-patient services alone, it can be seen that the quantity increase is very large, especially in case of out-patient hospital services. Here, per capita quantities more than doubled over six years. The strong increase is further analyzed in Section 3.4. For in-patient hospital services, on the other hand, the dynamic pattern is different. Here, prices, measured as costs per hospital day, rose by 25%, while quantities, measured as hospital days, remained more or less constant. One reason for higher implicit prices might be the observed reduction of the average length of stay in hospitals [11]. As a consequence, the intensity of medical care and, therefore, costs per day increased. Over time, the rise in the quantity of pharmaceuticals was partly offset by a decline in the real price. 3.4. Additional time series evidence The strong increase in out-patient hospital services over time as shown in Table 2 is striking and the question arises whether such an increase was the result of a substitution of out-patient hospital services for in-patient hospital services or for out-patient services by physicians. If such a substitution had taken place, a negative correlation between out- and in-patient hospital services or between out-patient hospitals and out-patient physician services would have been observed. Thus, one might expect that in years and cantons where out-patient services have grown strongly, in-patient hospitals services or out-patient physician services would have decreased or would have at least grown only marginally. However, Fig. 2, which depicts the quantity index (PQI) of out-patient hospital services against the quantity index of in-patient hospital services, shows no evidence of a negative correlation. In other words, the increase in out-patient hospital services is not accompanied by fewer in-patient services. The same result holds true in comparison to outpatient services by physicians. 4. Discussion As the cross-section analysis has shown, the regional cost differences in Switzerland are mainly quantity-driven. This result is in line with the literature on regional Medicare spending variation in the USA [12,13]. Gottlieb’s [12] conclusion can be adopted for Switzerland almost literally. He
writes: “The price-adjustment analysis resulted in less variation in what Medicare pays regionally, but not much. The findings suggest that utilization – not local price differences – drives Medicare regional payment variations”. However, for the privately insured medical care market in the USA, the evidence is different. Here, both price and utilization differences are important cost drivers [14]. Although some of the utilization variation in Switzerland can be explained by socio-economic factors, the remaining differences indicate that some regions use medical services more efficiently than others. And it is here where there is potential for cost containment. Econometric studies to explain regional health care variation typically apply explanatory variables that affect utilization rather than cost. Yet, since utilization on an aggregate level is often not available, the studies analyze cost variation instead. In this respect, the present regression analysis is an exception, at least for Switzerland, since it uses a per capita quantity index as the explained variable. However, compared to other studies for Switzerland which explain regional cost rather than regional quantity differences, the results of the panel data analysis do not come as a surprise. In Crivelli et al. [15] as well as in Reich et al. [16], density of physicians, population density, and age structure are shown to have the expected effect on cost. The present regression results on quantities therefore confirm existing findings with regard to costs. A somewhat unique factor in explaining utilization is the “Latin” variable. It stands for possible cultural differences between the German-speaking and the Frenchand Italian-speaking regions of Switzerland. As the results indicate, there is a difference in attitude toward public institutions, i.e. people in the “Latin” regions feel more entitled to claim benefits from the social security system than those in the German-speaking cantons. A major result of the longitudinal analysis is that the price increase in all medical services together equals almost precisely the increase in the general price level from 2004 to 2010, which means that the relative price of medical services remained constant during that period. Therefore, if the price increase in all health care services is adjusted with general inflation, the remaining cost increase is entirely due to larger quantities. By comparison, medical care prices in the privately insured health care market in the USA rose by 15.9% from 2003 to 2007 [17]. However, the general inflation rate for the same period amounted to 11.5%. Thus, real medical
Please cite this article in press as: Schleiniger R. Health care cost in Switzerland: Quantity- or price-driven? Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.04.004
G Model HEAP-3210; No. of Pages 7
6
ARTICLE IN PRESS R. Schleiniger / Health Policy xxx (2014) xxx–xxx
Fig. 2. Correlation between out-patient and in-patient hospital services, 2004–2010.
service prices rose only by 4% [18]. This example shows that, particularly with high inflation rates, a distinction between nominal and real price changes is important. In Switzerland, the prices for out-patient services as well as the prices for pharmaceuticals are determined politically by a negotiation process between care providers and the pharmaceutical industry, on the one hand, and the Swiss Insurance Association and the Federal Office of Public Health, on the other. Since both the high Swiss prices compared to other countries and the permanent rise of health care costs are perceived as serious problems by the Swiss population, the political pressure was obviously strong enough to prevent politically set prices from rising even higher. A final observation refers to the very strong increase in out-patient hospital services. The often cited explanation for this phenomenon is the fact that technical progress as well as innovation in prescription drugs is causing a shift of the mix of medical care services toward more out-patient treatment [19]. For example, glaucoma surgery with new laser technology or cancer therapy with new biochemical antibodies can nowadays be treated without the need for costly over-night stays. If this explanation for the increase in out-patient hospital utilization holds, we would expect to observe a substitution of out-patient for in-patient hospital services. However, the annual changes of out-patient and in-patient hospital services for the 22 analyzed regions show no evidence of such a substitution. In fact, the increase in out-patient hospital services appears to be an independent phenomenon that takes place independently from changes in other services. However, more research will be necessary to better understand the strong increase in out-patient hospital utilization.
5. Conclusion This study systematically breaks down the cost variation in the mandatory Swiss health care system into price and quantity effects. Since mandatory health insurance covers a multitude of services, the prices and quantities across all services must be reproduced as indices. The cross-section analysis shows that regional cost differences are mainly, although not exclusively, caused by quantity differences. Since prices and quantities are positively correlated, cost differences in case of regionally uniform prices would be smaller. A panel data analysis to explain the regional quantity differences suggests that the variables density of specialists, share of women, and age structure exert the strongest influence on the consumption of health care services. An additional regression analysis to explain differences in the consumption of pharmaceuticals shows a significant and positive effect of all supply-side factors, with the density of self-dispensing physicians as the strongest effect on the amount of pharmaceuticals consumed. The longitudinal analysis of all health care services together reveals that the prices for health care services in the period from 2004 to 2010 increased to the same extent as the consumer price index. Therefore, when the price increase in health care services is adjusted for general inflation, the remaining cost increase is entirely due to rising quantities. By far the largest increase in utilization can be observed in out-patient hospital services. In general, the analysis indicates that cost differences as well as cost growth are quantity-driven. Therefore, any attempt to contain costs must focus primarily on the level of medical care utilization. The key challenge will be to
Please cite this article in press as: Schleiniger R. Health care cost in Switzerland: Quantity- or price-driven? Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.04.004
G Model HEAP-3210; No. of Pages 7
ARTICLE IN PRESS R. Schleiniger / Health Policy xxx (2014) xxx–xxx
identify medical care services which do not have a positive effect on patients’ health status. Role of the funding source The present study was funded by santésuisse, the Swiss association of health insurance companies. Its involvement was limited and included the formulation of the research question (to break down costs into prices and quantities) and the contribution of data from their data pool. Otherwise, there was no further involvement by santésuisse, particularly not with regard to the method applied, the results, their interpretation, or their presentation. References [1] Schleiniger R. Regional quantity, productivity and efficiency measures of the Swiss health care system. Swiss Journal of Economics and Statistics 2008;144(3):459–76. [2] Bundesamt für Gesundheit, Zeitreihen Krankenversicherung, 2010, Bruttoleistungen pro Versicherten pro Jahr in Franken nach Kantonen, alle Versicherten. [3] Skinner J. Causes and consequences of regional variations in health care. In: Culyer AJ, Pauly MV, Newhouse JP, McGuire TG, Barros PP, editors. Handbook in economics, health economics, vol. 2. North Holland: Elsevier; 2012 [Chapter 2]. [4] WIC. Bibliography on international small-area health care variation studies. Wennberg International Collaborative; 2011. [5] Mangano A. An analysis of the regional differences in health care utilization in Italy. Health and Place 2009;16(2):301–8. [6] Swiss Federal Statistical Office. Nettofinanzbedarf der Kantone und Gemeinden im Gesundheitswesen; 2004–2010.
7
[7] Berndt ER, et al. Price indexes for medical care goods and services – an overview of measurement issues. In: Cutler DM, Berndt ER, editors. Medical care output and productivity, national bureau of economic research. University of Chicago Press; 2001. p. 141–200. [8] Hill RJ. A taxonomy of multilateral methods for making international comparisons of prices and quantities. Review of Income and Wealth 1997;43(1):49–69. [9] Gerdtham UG, Jönsson B. International comparisons of health expenditure: theory, data, and econometric analysis. In: Culyer AJ, Newhouse JP, editors. Handbook of Health Economics, vol. 1. Elsevier Science; 2000. p. 11–53. [10] Camenzind P. Erklärungsansätze regionaler Kostenunterschiede im Gesundheitswesen, Obsan Arbeitsdokument 30, Juni 2008. [11] Roth M, Roth S. Entwicklung der Ausgaben der obligatorischen Krankenversicherung von 1998 bis 2010. Neuenburg: OBSAN Bericht 53; 2012. [12] Gottlieb, Daniel J, et al. Prices don’t drive regional Medicare variations. Health Affairs 2010;29(3):537–43. [13] Medicare Payment Advisory Commission. Measuring regional variation in service use, report to the congress; 2009. [14] Dunn A, et al. Geographic variation in commercial medical-care expenditures: a framework for decomposing price and utilization. Journal of Health Economics 2013;32:1153–65. [15] Crivelli L, et al. Federalism and regional health care expenditures: an empirical analysis for the Swiss cantons. Health Economics 2006;15:535–41. [16] Reich O, et al. Exploring the disparities of regional health care expenditures in Switzerland: some empirical evidence. European Journal of Health Economics 2011;13:193–202. [17] Dunn A, et al. Decomposing medical-care expenditure growth, Federal Reserve Bank of San Francisco, Working Paper Series, 2012-26; 2012. [18] Shapiro AH. What’s driving medical care spending growth? FRBSF Economic Letter 2013;2013-7:1–5. [19] Aizcorbe A, Nestoriak N. Changing mix of medical care services: stylized facts and implications for price indexes. Journal of Health Economics 2011;30:568–74.
Please cite this article in press as: Schleiniger R. Health care cost in Switzerland: Quantity- or price-driven? Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.04.004
ARTICLE IN PRESS
HEAP-3210; No. of Pages 7
Health Policy xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Health Policy journal homepage: www.elsevier.com/locate/healthpol
Health care cost in Switzerland: Quantity- or price-driven?夽 Reto Schleiniger ∗ Zürich University of Applied Sciences, Center for Economic Policy, Bahnhofplatz 12, 8401 Winterthur, Switzerland
a r t i c l e
i n f o
Article history: Received 15 October 2013 Received in revised form 7 April 2014 Accepted 8 April 2014
Keywords: Costs and cost analysis Index calculation Regression analysis
a b s t r a c t In Switzerland, per capita health care costs vary substantially from canton to canton and rise considerably and steadily from year to year. Since costs are equal to the product of quantities and prices, the question arises whether regional cost variations and cost increase over time are quantity- or price-driven. Depending on the answer, the containment of health care costs must be approached differently. This article examines the cost of mandatory health insurance in Switzerland for the period from 2004 to 2010 and breaks it down into quantity and price effects. The main result of the cross-section analysis reveals that regional cost differences are mainly due to quantity differences. Similarly, the longitudinal analysis shows that the cost increase across all health care services is primarily caused by increasing per capita quantities. Any attempt to contain costs must therefore focus primarily on the extent of medical care utilization, and the key challenge to be met is how to identify medical care services which do not have a positive effect on patients’ health status. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Two observations serve as the starting point for the study: Firstly, there are large regional differences in per capita health care costs across Switzerland, reflected by the substantial variance in the premiums paid for mandatory health insurance in Swiss cantons. Secondly, there is a steady increase in per capita costs over time. Over the past fifteen years, costs for mandatory health care services outpaced overall GDP growth and amounted to 4.25% of GDP in 2010 [2]. In order to understand regional cost variation and be able to limit medical cost growth, we must first identify the sources of the regional disparity and of the cost growth.
夽 This study was commissioned by santésuisse, the Swiss association of health insurers. An earlier study covering only the period from 2004 to 2005 was published in 2008 and focuses on productivity and efficiency measures [1]. ∗ Tel.: +41 52 267 78 75. E-mail address: [email protected]
Since costs are the product of prices and quantities, the first step is to decompose costs into prices and quantities. Obviously, measures to contain costs must differ substantially depending on whether cost changes are driven by price or quantity. The present study examines mandatory health care services in Switzerland and covers the period from 2004 to 2010. Since mandatory health insurance covers a variety of services, prices and quantities across all services cannot be compared directly but must be calculated and expressed in terms of price and quantity indices. Interestingly, most studies on the decomposition of regional medical care cost variations are conducted in the USA [3]. Whereas in other countries regional variations in total health care expenditures is an issue, the cost differences are usually not split into prices and quantities [4,5]. In Switzerland, the decomposition of costs into prices and quantities is of special interest because, due to the federal state structure, the Swiss health care system is organized along cantonal borders. As a consequence, prices are not determined centrally and can vary from canton to
http://dx.doi.org/10.1016/j.healthpol.2014.04.004 0168-8510/© 2014 Elsevier Ireland Ltd. All rights reserved.
Please cite this article in press as: Schleiniger R. Health care cost in Switzerland: Quantity- or price-driven? Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.04.004
G Model HEAP-3210; No. of Pages 7
ARTICLE IN PRESS R. Schleiniger / Health Policy xxx (2014) xxx–xxx
2
canton. To what extent the decentralized price setting is responsible for regional cost variations is one of the questions that will be analyzed in the following.
2. Data and methodology 2.1. Data selection and preparation Most of the necessary data input has been extracted from the data pool of santésuisse, the Swiss health insurance association. The data pool comprises all mandatory health care transactions of the health insurance companies. To capture the full cost of mandatory health care services, the data supplied by santésuisse is complemented by data on public subsidies to hospitals as published by the Swiss Federal Statistical Office [6]. In addition, the price development of pharmaceuticals over time is described by the corresponding sub-index of the Swiss consumer price index, also published by the Swiss Federal Statistical Office. The study distinguishes various health care services such as out-patient services by physicians, physiotherapists, and hospitals, in-patient hospital services, pharmaceuticals, nursing homes, and laboratory services. The in-patient hospital services are further divided into ten sub-categories ranging from university hospitals to local hospitals and specific clinics such as psychiatric institutions. In total, the costs of the 16 categories amount to almost 20 billion Swiss francs. Together with the yearly subsidies to public hospitals of approximately eight billion Swiss francs, costs in the amount of 28 billion Swiss francs are included in the calculations. The only other service worth mentioning is home care with a cost share of a little less than 2%. Here no meaningful split into price and quantity was possible, because home care services are charged partly on a time tariff and partly in the form of a lump sum. Since the available data does not distinguish the two forms of payment, home care was excluded from the data. While 10 different hospital categories were distinguished, some of the smaller Swiss cantons do not have many of these categories. For the purpose of this study, the semi-cantons of Appenzell-Ausserrhoden and AppenzellInnerrhoden, as well as Obwalden and Nidwalden, have been merged with their larger neighbors, St. Gallen and Lucerne, respectively. As a consequence, the original number of 26 cantons and semi-cantons has been reduced to 22 regions. In order to calculate per capita results, special attention needs to be given to the determination of the cantonal population figures. When using these figures, it must be remembered that care providers such as hospitals and medical practitioners not only offer services to the resident population but also across cantonal borders. Dividing the cantonal cost of care providers by the number of the resident population, therefore, yields biased per capita results. To correct this bias, a matrix of inter-cantonal service flows is used to convert the resident population into the “medically cared for population”. Per capita measures, then, are calculated as cost and quantities per medicated population.
2.2. Index calculation The challenge in trying to break down costs into quantities and prices lies in the fact that mandatory health insurance covers a range of different services. Therefore, prices and quantities across all services must be reproduced as indices [7]. In order to derive such indices, the cost of each service must be expressed individually as the product of quantities and prices. To this end, a plausible quantity and price indicator must be selected for each service. The following briefly describes what price and quantity indicators were chosen for each health care category. 2.2.1. Ambulatory services (out-patient hospital services, practitioners, and physiotherapists) The information provided by TARMED, the Swiss medical tariff scheme for out-patient services, was used to break down the costs for all ambulatory services. The scheme assigns each medical service a specific number of tax points, based on the time required to render the service. The tax points serve as a quantity indicator for out-patient treatments. Medical costs are then determined by multiplying the tax points by so called tax point values which are to be understood as the price of a standardized service. The tax point values are the result of a negotiation process. Regionally, they vary by up to 20%, which is one reason why prices for medical services vary across Switzerland. 2.2.2. In-patient hospital services and nursing homes With in-patient hospital services, it was not possible to use a similar procedure to the one used for ambulatory services to split the costs because no country-wide DRG scheme was in place in the period from 2004 to 2010. Therefore, “hospital day per hospital category” was used as the smallest standardized quantity measure that could be inferred from the data. Such a procedure implies that a hospital day at a given hospital category represents the same output all over the country. It does not, however, suggest that a day in a university hospital is comparable to, say, a day in a local hospital. The implicit price for in-patient hospital services was then calculated as total hospital cost per hospital day and category. Again, the costs per hospital day and category vary considerably from canton to canton and constitute a second reason why prices are different across cantons. 2.2.3. Pharmaceuticals and laboratory services With pharmaceuticals and laboratory services, there are no regional price differences but only price changes over time. To account for these price changes on the national level, we again relied on tax point values for laboratory services and on the consumer price sub-index for pharmaceuticals as published by the Swiss Federal Statistical Office. This information on prices and quantities for each health care category was used to determine Paasche quantity and Laspeyres price indices. These were calculated in two ways, once without and once with public subsidies to hospitals. Since these subsidies vary substantially from canton to canton, the resulting price indices using net
Please cite this article in press as: Schleiniger R. Health care cost in Switzerland: Quantity- or price-driven? Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.04.004
G Model
ARTICLE IN PRESS
HEAP-3210; No. of Pages 7
R. Schleiniger / Health Policy xxx (2014) xxx–xxx
3
Fig. 1. Result of cross-section analysis: mean of quantity and price index from 2004 to 2010.
calculation without subsidies can differ considerably from gross indices which include the subsidies. 3. Results 3.1. Regional quantity and price indices To determine regional quantity and price indices, two types of reference regions are possible: either a specific region chosen at random or an artificially constructed average region. The former is known as the asymmetric star method, the latter the symmetric star method. This study uses the symmetric method. Therefore, the resulting indices must be interpreted relative to the Swiss average. For example, a per capita quantity index of 1.05 in a given canton signifies that the per capita health care demand is 5% larger than the average national demand.1 Fig. 1 exhibits the resulting indices of all mandatory health care services, calculated as the mean of the period from 2004 to 2010. The horizontal axis represents the regional quantity index per capita, the vertical axis the corresponding price index. As mentioned before, the price
1 This method belongs to the average basket methods and is also known as the ECLAC method (United Nations Economic Commission for Latin America and the Caribbean). See Ref. [8].
index was calculated as both a net and a gross concept, i.e. without and with public subsidies to hospitals. Also plotted are the isocost lines, which describe price–quantity combinations at a given cost level. Note that the product of Paasche quantity and Laspeyres price indices yields the value index, which stands, in the context of this study, for the cantonal per capita costs relative to the national average. It is now possible to classify the cantons according to their relative costs, with the high-cost cantons situated in the top right corner and the low-cost cantons in the bottom left corner. The figure shows that in Geneva per capita quantities are almost 30% and prices about 5% above average, resulting in per capita costs that are 35% higher than all over Switzerland. In Basel-Stadt, the other city canton besides Geneva, the quantities are also very high, but the prices are only average. Neuchâtel and Berne, on the other hand, exhibit high prices and average quantities, while Ticino, the Italianspeaking canton, is characterized by large quantities at prices slightly below average. The high gross price index of Neuchâtel is the result of the relatively high subsidies hospitals receive in this canton. By contrast, in Zurich the relative gross price is clearly below the relative net price, indicating that public subsidies to Zurich hospitals are relatively small.
Please cite this article in press as: Schleiniger R. Health care cost in Switzerland: Quantity- or price-driven? Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.04.004
G Model HEAP-3210; No. of Pages 7
ARTICLE IN PRESS R. Schleiniger / Health Policy xxx (2014) xxx–xxx
4
Table 1 Regression results to explain all mandatory health care services per capita: random effects method. All services including subsidies
2006–2010
Pharmaceuticals
2004–2010
Explanatory variable Share of women Age > 75 Population density Unemployment rate Density of specialists cor. R2
Coefficient 2.895*** 0.186** 0.007** 0.056** 0.170*** 0.703
Explanatory variable Density of GP Density of specialists Density of self-dispensing physicians Density of pharmacies Dummy variable indicating Latin language cor. R2
Coefficient 0.151*** 0.183*** 0.049*** 0.083* 0.150** 0.503
* ** ***
Level of significance: 90%. Level of significance: 95%. Level of significance: 99%.
On the other side of the cost spectrum, we find the cantons of Central Switzerland and the canton of St. Gallen. In all these regions, prices, and particularly consumed quantities, are below average, yielding per capita costs that are up to 20% lower than the Swiss average. Fig. 1 also shows that the quantity variance is larger than the price variance.2 Therefore, it can be concluded that the cost differences are mainly caused by quantity differences. However, there is a positive correlation between prices and quantities, which indicates that, without price differences, the cost differences would be smaller. 3.2. Panel data analysis Subsequently, an econometric panel analysis of regional quantity differences was performed. The regression model included spatial (22 regions) and time dimensions (7 years). The set of variables which are used as the determining factors can be divided into demand-side factors such as age and gender, supply-side factors such as physician density, and insurance-specific factors such as the size of the deductible [9,10]. Table 1 depicts the regression results with the per capita quantity index of all health care services as the explained variable and after eliminating all non-significant variables. It shows that the share of women, the share of the population aged above 75, the population density, and the unemployment rate, as well as the density of specialists, all yield the expected positive influence on consumed quantities. For the period from 2006 to 2010, the corrected determination coefficient amounts to 70% and is larger than the corresponding coefficient for the period from 2004 to 2010. Due to the introduction of a new accounting scheme for out-patient services in the year 2004, which led to a delayed booking of delivered services the data on outpatient services for the period from 2004 to 2005 might be biased, which reduces the explanatory power of the regression over the entire period of observation. Since the regression is based on indices, the resulting coefficients cannot be easily interpreted quantitatively. However, the strongest influence emanates from, in this order, the density of specialist practitioners, the share of
2 In logarithmic form, the cost variance can be decomposed into the price variance, the quantity variance, and twice the covariance of prices and quantities. In the gross calculation with subsidies, the quantity variance accounts for 77 and the price variance for only 6% of the total variance.
women, and the share of senior citizens. In the case of the specialists, this means that the quantity difference between a canton with a high density and a canton with a medium density of specialists is relatively large. In a next step, an additional regression is conducted with the quantity of pharmaceuticals as the explained variable. On the right hand side of Table 1, it can be seen that all the supply side factors exert a positive influence on quantities (with the density of pharmacies only significant at the 90% level). In addition, the dummy variable for Latin language is significant at the 95% level, which indicates that in the French- and Italian-speaking cantons more pharmaceuticals are consumed. The strongest influence on consumed pharmaceuticals is exerted by the density of self-dispensing physicians. However, it must be noted that this variable also varies strongly from canton to canton.
3.3. Price and quantity indices over time, 2004–2010 The results of the longitudinal section are shown for the entire country only since the regional dynamic patterns are in most cantons similar to the national development. All price and quantity indices were first calculated as yearly indices. Price and quantity changes over more than one year were then determined with a multiplicative chaining of the year-to-year values. Since the total costs are the product of prices, quantities per capita, and population, the cost index can be decomposed into the three indices for prices, per capita quantity, and population. Table 2 highlights the results of such decomposition for the total of health care services with 2004 and 2005 as the respective starting year. Thus it is shown that the total cost increase in 36% between 2004 and 2010 was caused by a 5% increase in population, a 22% increase in per capita consumption, and a 5.5% increase in prices. The price increase equals more or less the increase in the general consumer price level from 2004 to 2010 which amounts to 5.7%. In Table 2 it can also be seen that the quantity increase since 2005 is clearly smaller than it has been since 2004. Such a large difference is not very plausible and might be attributed to the above-mentioned introduction of a new accounting scheme for out-patient services in 2004. The delayed booking of services from 2004 in 2005 yields an upward bias from 2004 to 2005 and, consequently, a downward bias in the following period from 2005 to 2006.
Please cite this article in press as: Schleiniger R. Health care cost in Switzerland: Quantity- or price-driven? Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.04.004
G Model HEAP-3210; No. of Pages 7
ARTICLE IN PRESS R. Schleiniger / Health Policy xxx (2014) xxx–xxx
5
Table 2 Development of costs from 2004 (2005) to 2010: Price, quantity and population effects.
All services including subsidies
In-patient hospital services including subsidies Out-patient services by physicians Out-patient hospital services Pharmaceuticals
Prices
Quantity per capita
Population
Cost
1.0558 (1.0610)
1.2180 (1.1030)
1.0544 (1.0500)
1.3560 (1.2277)
Prices
Quantity per capita
Cost per capita
1.2476 (1.2231)
0.9875 (0.9776)
1.2320 (1.1956)
0.9824 (0.9994)
1.5448 (1.0850)
1.5176 (1.0844)
0.9747 (0.9854) 0.8300 (0.8374)
2.1578 (1.7105) 1.3311 (1.2802)
2.1031 (1.6855) 1.1049 (1.0720)
The lower part of Table 2 shows the national results of the longitudinal section for specific services with large cost shares. Looking at out-patient services alone, it can be seen that the quantity increase is very large, especially in case of out-patient hospital services. Here, per capita quantities more than doubled over six years. The strong increase is further analyzed in Section 3.4. For in-patient hospital services, on the other hand, the dynamic pattern is different. Here, prices, measured as costs per hospital day, rose by 25%, while quantities, measured as hospital days, remained more or less constant. One reason for higher implicit prices might be the observed reduction of the average length of stay in hospitals [11]. As a consequence, the intensity of medical care and, therefore, costs per day increased. Over time, the rise in the quantity of pharmaceuticals was partly offset by a decline in the real price. 3.4. Additional time series evidence The strong increase in out-patient hospital services over time as shown in Table 2 is striking and the question arises whether such an increase was the result of a substitution of out-patient hospital services for in-patient hospital services or for out-patient services by physicians. If such a substitution had taken place, a negative correlation between out- and in-patient hospital services or between out-patient hospitals and out-patient physician services would have been observed. Thus, one might expect that in years and cantons where out-patient services have grown strongly, in-patient hospitals services or out-patient physician services would have decreased or would have at least grown only marginally. However, Fig. 2, which depicts the quantity index (PQI) of out-patient hospital services against the quantity index of in-patient hospital services, shows no evidence of a negative correlation. In other words, the increase in out-patient hospital services is not accompanied by fewer in-patient services. The same result holds true in comparison to outpatient services by physicians. 4. Discussion As the cross-section analysis has shown, the regional cost differences in Switzerland are mainly quantity-driven. This result is in line with the literature on regional Medicare spending variation in the USA [12,13]. Gottlieb’s [12] conclusion can be adopted for Switzerland almost literally. He
writes: “The price-adjustment analysis resulted in less variation in what Medicare pays regionally, but not much. The findings suggest that utilization – not local price differences – drives Medicare regional payment variations”. However, for the privately insured medical care market in the USA, the evidence is different. Here, both price and utilization differences are important cost drivers [14]. Although some of the utilization variation in Switzerland can be explained by socio-economic factors, the remaining differences indicate that some regions use medical services more efficiently than others. And it is here where there is potential for cost containment. Econometric studies to explain regional health care variation typically apply explanatory variables that affect utilization rather than cost. Yet, since utilization on an aggregate level is often not available, the studies analyze cost variation instead. In this respect, the present regression analysis is an exception, at least for Switzerland, since it uses a per capita quantity index as the explained variable. However, compared to other studies for Switzerland which explain regional cost rather than regional quantity differences, the results of the panel data analysis do not come as a surprise. In Crivelli et al. [15] as well as in Reich et al. [16], density of physicians, population density, and age structure are shown to have the expected effect on cost. The present regression results on quantities therefore confirm existing findings with regard to costs. A somewhat unique factor in explaining utilization is the “Latin” variable. It stands for possible cultural differences between the German-speaking and the Frenchand Italian-speaking regions of Switzerland. As the results indicate, there is a difference in attitude toward public institutions, i.e. people in the “Latin” regions feel more entitled to claim benefits from the social security system than those in the German-speaking cantons. A major result of the longitudinal analysis is that the price increase in all medical services together equals almost precisely the increase in the general price level from 2004 to 2010, which means that the relative price of medical services remained constant during that period. Therefore, if the price increase in all health care services is adjusted with general inflation, the remaining cost increase is entirely due to larger quantities. By comparison, medical care prices in the privately insured health care market in the USA rose by 15.9% from 2003 to 2007 [17]. However, the general inflation rate for the same period amounted to 11.5%. Thus, real medical
Please cite this article in press as: Schleiniger R. Health care cost in Switzerland: Quantity- or price-driven? Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.04.004
G Model HEAP-3210; No. of Pages 7
6
ARTICLE IN PRESS R. Schleiniger / Health Policy xxx (2014) xxx–xxx
Fig. 2. Correlation between out-patient and in-patient hospital services, 2004–2010.
service prices rose only by 4% [18]. This example shows that, particularly with high inflation rates, a distinction between nominal and real price changes is important. In Switzerland, the prices for out-patient services as well as the prices for pharmaceuticals are determined politically by a negotiation process between care providers and the pharmaceutical industry, on the one hand, and the Swiss Insurance Association and the Federal Office of Public Health, on the other. Since both the high Swiss prices compared to other countries and the permanent rise of health care costs are perceived as serious problems by the Swiss population, the political pressure was obviously strong enough to prevent politically set prices from rising even higher. A final observation refers to the very strong increase in out-patient hospital services. The often cited explanation for this phenomenon is the fact that technical progress as well as innovation in prescription drugs is causing a shift of the mix of medical care services toward more out-patient treatment [19]. For example, glaucoma surgery with new laser technology or cancer therapy with new biochemical antibodies can nowadays be treated without the need for costly over-night stays. If this explanation for the increase in out-patient hospital utilization holds, we would expect to observe a substitution of out-patient for in-patient hospital services. However, the annual changes of out-patient and in-patient hospital services for the 22 analyzed regions show no evidence of such a substitution. In fact, the increase in out-patient hospital services appears to be an independent phenomenon that takes place independently from changes in other services. However, more research will be necessary to better understand the strong increase in out-patient hospital utilization.
5. Conclusion This study systematically breaks down the cost variation in the mandatory Swiss health care system into price and quantity effects. Since mandatory health insurance covers a multitude of services, the prices and quantities across all services must be reproduced as indices. The cross-section analysis shows that regional cost differences are mainly, although not exclusively, caused by quantity differences. Since prices and quantities are positively correlated, cost differences in case of regionally uniform prices would be smaller. A panel data analysis to explain the regional quantity differences suggests that the variables density of specialists, share of women, and age structure exert the strongest influence on the consumption of health care services. An additional regression analysis to explain differences in the consumption of pharmaceuticals shows a significant and positive effect of all supply-side factors, with the density of self-dispensing physicians as the strongest effect on the amount of pharmaceuticals consumed. The longitudinal analysis of all health care services together reveals that the prices for health care services in the period from 2004 to 2010 increased to the same extent as the consumer price index. Therefore, when the price increase in health care services is adjusted for general inflation, the remaining cost increase is entirely due to rising quantities. By far the largest increase in utilization can be observed in out-patient hospital services. In general, the analysis indicates that cost differences as well as cost growth are quantity-driven. Therefore, any attempt to contain costs must focus primarily on the level of medical care utilization. The key challenge will be to
Please cite this article in press as: Schleiniger R. Health care cost in Switzerland: Quantity- or price-driven? Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.04.004
G Model HEAP-3210; No. of Pages 7
ARTICLE IN PRESS R. Schleiniger / Health Policy xxx (2014) xxx–xxx
identify medical care services which do not have a positive effect on patients’ health status. Role of the funding source The present study was funded by santésuisse, the Swiss association of health insurance companies. Its involvement was limited and included the formulation of the research question (to break down costs into prices and quantities) and the contribution of data from their data pool. Otherwise, there was no further involvement by santésuisse, particularly not with regard to the method applied, the results, their interpretation, or their presentation. References [1] Schleiniger R. Regional quantity, productivity and efficiency measures of the Swiss health care system. Swiss Journal of Economics and Statistics 2008;144(3):459–76. [2] Bundesamt für Gesundheit, Zeitreihen Krankenversicherung, 2010, Bruttoleistungen pro Versicherten pro Jahr in Franken nach Kantonen, alle Versicherten. [3] Skinner J. Causes and consequences of regional variations in health care. In: Culyer AJ, Pauly MV, Newhouse JP, McGuire TG, Barros PP, editors. Handbook in economics, health economics, vol. 2. North Holland: Elsevier; 2012 [Chapter 2]. [4] WIC. Bibliography on international small-area health care variation studies. Wennberg International Collaborative; 2011. [5] Mangano A. An analysis of the regional differences in health care utilization in Italy. Health and Place 2009;16(2):301–8. [6] Swiss Federal Statistical Office. Nettofinanzbedarf der Kantone und Gemeinden im Gesundheitswesen; 2004–2010.
7
[7] Berndt ER, et al. Price indexes for medical care goods and services – an overview of measurement issues. In: Cutler DM, Berndt ER, editors. Medical care output and productivity, national bureau of economic research. University of Chicago Press; 2001. p. 141–200. [8] Hill RJ. A taxonomy of multilateral methods for making international comparisons of prices and quantities. Review of Income and Wealth 1997;43(1):49–69. [9] Gerdtham UG, Jönsson B. International comparisons of health expenditure: theory, data, and econometric analysis. In: Culyer AJ, Newhouse JP, editors. Handbook of Health Economics, vol. 1. Elsevier Science; 2000. p. 11–53. [10] Camenzind P. Erklärungsansätze regionaler Kostenunterschiede im Gesundheitswesen, Obsan Arbeitsdokument 30, Juni 2008. [11] Roth M, Roth S. Entwicklung der Ausgaben der obligatorischen Krankenversicherung von 1998 bis 2010. Neuenburg: OBSAN Bericht 53; 2012. [12] Gottlieb, Daniel J, et al. Prices don’t drive regional Medicare variations. Health Affairs 2010;29(3):537–43. [13] Medicare Payment Advisory Commission. Measuring regional variation in service use, report to the congress; 2009. [14] Dunn A, et al. Geographic variation in commercial medical-care expenditures: a framework for decomposing price and utilization. Journal of Health Economics 2013;32:1153–65. [15] Crivelli L, et al. Federalism and regional health care expenditures: an empirical analysis for the Swiss cantons. Health Economics 2006;15:535–41. [16] Reich O, et al. Exploring the disparities of regional health care expenditures in Switzerland: some empirical evidence. European Journal of Health Economics 2011;13:193–202. [17] Dunn A, et al. Decomposing medical-care expenditure growth, Federal Reserve Bank of San Francisco, Working Paper Series, 2012-26; 2012. [18] Shapiro AH. What’s driving medical care spending growth? FRBSF Economic Letter 2013;2013-7:1–5. [19] Aizcorbe A, Nestoriak N. Changing mix of medical care services: stylized facts and implications for price indexes. Journal of Health Economics 2011;30:568–74.
Please cite this article in press as: Schleiniger R. Health care cost in Switzerland: Quantity- or price-driven? Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.04.004
