Options
Jeanrenaud, Claude
Nom
Jeanrenaud, Claude
Affiliation principale
Fonction
Professeur.e émérite
Email
Claude.Jeanrenaud@unine.ch
Identifiants
Résultat de la recherche
Voici les éléments 1 - 1 sur 1
- PublicationAccès libreThree essays on cantonal variations in the utilization and costs of healthcare services in Switzerland(2013)
Part I of the document continues with an overview on Switzerland’s health data situation and reveals that its most critical weaknesses exist in the areas of outpatient healthcare provision and of epidemiology. Part I terminates with a presentation of the literature overviewing international and national differences in regional healthcare costs. The review concludes that it is challenging for health economics to provide consistent answers to many of the important research questions pertaining to the field of regional healthcare cost and utilization differences. The most frequently cited causes of this difficulty are the complexity of the healthcare systems and the crucial dearth of broadly recognized theoretical models and available data. A recently presented model (Chandra and Skinner 2012; Skinner 2012) was considered to be a good starting point for a more systematic analysis of geographic variations in healthcare costs and utilization.
One agreement in the literature about the methodological findings is obvious: when explaining regional health differences, it is advantageous to account for levels of and trends in healthcare costs (or utilization) simultaneously. However, one has to be aware that different sets of variables can influence each dimension. Thus, by being able to combine cross-section and time-series analyses, a panel econometric approach seems to be the most promising statistical-technical instrument for tackling these types of research questions.
Moreover, because the prices paid in the health sector (e.g., the cantonal taxpoint values in Switzerland), the volumes of care (e.g., the number of per capita GP consultations), and the medical practices applied (e.g., the average number of taxpoints used per consultation) can—again—be influenced by different sets of factors, separate analyses of these three main components of healthcare costs is preferable. Moreover, the literature review identifies individual data (i.e., the individual patient, the individual insurance policyholder, or the individual healthcare provider) and geo-coded information 1 as the statistical and geographical level that offers the most possibilities for such research. Unfortunately, the Swiss health data normally do not allow in-depth analysis on such individual levels.
In Part II of the document, three essays containing concrete analyses of regional differences in costs and in actual and future utilization of healthcare services in Switzerland are presented; none of these empirical investigations goes beyond the cantonal level. The first essay investigates the factors associated with cantonal differences in the utilization of mandatory health insurance (MHI) services between 2000 and 2007 by applying panel econometric (fixed effects) models. For variations in utilization for each of the six largest MHI service provider domains—viz., general practitioners, medical specialists, hospital inpatient care, hospital outpatient care, medication and nursing homes—significant factors that are correlated with utilization frequency over time and across cantons can be identified.
In particular, a greater density of service providers tends to be significantly associated with the per-capita consumption of healthcare services. On the demand side, older, more urban and wealthier populations with more deprivation problems summarize the principal positively correlated factors. Financing characteristics in the form of high deductibles or managed care instruments can also play a role in the utilization level of healthcare services, although some large difficulties 2 were faced in confirming their effects empirically. Finally, the general time trends describing the accelerating utilization of outpatient drugs, nursing homes and outpatient hospitals are presented in contrast to the declining trends observed for inpatient hospitals, GPs, and specialist services in private practices.
The main contribution of the first essay of the thesis is its being the first such work to analyze spatial and temporal differences in quantities instead of costs of healthcare service domains in Switzerland. Moreover, the testing of a constant set of 12 explanatory variables across the six healthcare service domains allows a bi-directional interpretation of the results. In addition to understanding how each of the six target variables is interrelated with the whole set of regressors, one can learn more about how each purportedly influential factor is individually associated with all six healthcare service domains.
The second essay in Part II begins by presenting the large differences in average annual per-bed costs between individual nursing homes and between nursing homes grouped by cantons. The paper tries to identify empirically some explanations for these sizable per-bed cost variations. At the same time, the assumed existence of two-levels of explanatory factors (viz., individual and cantonal levels) is taken into consideration by modeling them with regression estimations in multilevel form. Moreover, besides the variation of total costs per bed and per year, the variations of the annual per-bed costs of accommodation and assistance and the annual per-bed costs covered by the MHI are calculated separately.
Because the data from 2006 alone were available for the research presented in the second essay at the time of its conception, it was decided to approach this study with only a single year being analyzed in a cross-level setting. This approach clearly has its limitations, but it did not preclude employing more sophisticated panel data approaches at a later date. Such a limited model explains variations in the annual per-bed costs between cantons fairly well, but quite a share of variation within cantons remains unexplained. Because no ideal indicator was available for the data on the hotel service standards of Swiss nursing homes, this result was not surprising—especially regarding annual per-bed costs of accommodation and assistance. However, the operationalized variables—such as the number of days invoiced per bed and year (i.e., occupancy rate), the intensity of nursing time spent per patient and day, the qualification level of the personnel, the relative number of non-medical employees, and the cantonal wage index—were significantly correlated with all three cost indicators.
The essay admits to the difficulty faced in deriving recommendations for policy-making authorities from these results. Cantons should at least monitor their nursing home costs and financing continuously—in particular, their costs for assistance and accommodation. Should increasingly large numbers of individuals and their families out of the growing number of people with chronic illnesses be unable to pay these costs out of their own pockets in the coming years, the cantons might be forced to intervene.
The third essay of Part II analyzes regional differences in the utilization of somatic acute care beds in Swiss hospitals. A description of cantonal population age structures and trends and hospital utilization patterns in 2010 is followed by calculations of ranges of cantonal acute care hospital volumes through 2030. Originally developed by researchers at the Swiss Health Observatory and the Statistical Office of Canton Vaud (VD) for the statistical support of hospital planning processes in individual cantons, a projection model is applied for the first time in this study for all 26 cantons simultaneously and allows a direct comparison of the results across cantons and with the country as a whole as well as calculations of national-level results for medical branches. The projection model realizes a systematic link between Swiss medical statistics for hospitals and official cantonal population scenarios. Various hypotheses on future length-of-stay (LOS) trends in Swiss acute care hospitals are simulated with the model.
The most important results of the study are the following: the national number of hospital days required through 2030 should increase no more (or only slightly more) than the population increases. While an increase of hospital days between 5 percent and 13 percent is expected in the two “middle” scenarios of the model, the population will grow 11 percent between 2010 and 2030 in the official “average” demographic scenario. This rather positive outcome on the national level is the result of major differences between cantons. Some cantons will have to deal with increases of hospital days of approximately 30 percent, whereas in other cantons hospital days will rise less than 5 percent. Moreover, treatments typical for older patients, such as cardiology and vascular medicine treatments, will clearly be more necessary in 2030 than medical branches with very young patients, such as neonatology or obstetrics.
Part III provides some specific conclusions for the Swiss healthcare system with its characteristics of regulated competition and strong federalist structure. As a strategy of analyzing and comparing healthcare cost components defined as being low aggregated (e.g., individual health service domains or providers and individual cost components, such as quantities and prices) is favored and targeted in the three empirical essays of Part II, it seems important that such detailed analyses should afterwards be complemented by an attempt to draw an overall picture of the results. Accordingly, an applied synthesis of the results for two exemplary cantons—one canton with low (Obwalden OW) and one canton with high per capita healthcare costs (Geneva GE)—is presented in the Excursus of Part III.
Without yet having executed the necessary empirical work, proposals are made in the Excursus about how cantons might be distinguished by some of their characteristics on the demand and supply sides of the healthcare system. On the demand side, cantons may have a more “integrated” or a more “globalized” population. Concrete characteristics that assign a cantonal population to one or the other type could be derived from their different economic conditions (e.g., income, assets, and their distributions), the importance of social-economic exclusion (e.g., unemployment, receipt of benefits), the average physical and mental health status, and the actual and future age structure (including future requirements of health care).
On the supply side, two different types of cantons are proposed as well. First, there might be the “peripheral-type scheme” of a cantonal health provision system. Such a health system is focused on primary care and nursing homes, and it is characterized by a modest health provision infrastructure with only a few active specialists, with many health services being purchased in other cantons. Second, the “center-type scheme” of a cantonal health provision system is proposed. Such a system is characterized by a large (university or principal) hospital that is surrounded by many independent specialists and pharmacies. This system, in contrast, attracts patients from other cantons.
The document concludes by offering a few suggestions for future research. Rather concrete propositions were made in the three empirical articles of the thesis. First, they were mainly about more sophisticated methodological approaches: the use of instrumental variables and two-stages least squares estimations in the first article, the use of panel data models with additional regressors in the second article, and the integration of additional variables such as sex, morbidity trends, and future changes of regional patient flows in the third article. Second, extensions concerning the data were required: the necessity of always working with statistical data on an aggregated level might be the most significant problem this thesis faces throughout. Moreover, more detailed domains of health service providers could be analyzed in the first article, a search for better variables concerning hospitality standards and cantonal regulations in nursing homes is proposed in the second article, and a more precise analysis of hospital cases with very long LOSs would be useful in the third article. As a more general recommendation, more research about possible trends in the future health status of aging populations in Western societies was proposed. Special models to simulate such quantitative calculations exist and could be translated to the case presented by Switzerland. .
1 Of course, in reality, the exact research question actually determines whether individual-level data are needed; however, most of the time, data aggregation (if needed) is possible, but desaggregation is not.
2 The main methodological challenges were endogeneity problems (omitted variables and variable selection biases) and ecological fallacies (see Chapter 4).