Health Policy 115 (2014) 180–188
Contents lists available at ScienceDirect
Health Policy journal homepage: www.elsevier.com/locate/healthpol
Length of hospital stay in Japan 1971–2008: Hospital ownership and cost-containment policies Naoko Kato a,b , Masahide Kondo c,∗ , Ichiro Okubo c , Toshihiko Hasegawa d a b c d
International University of Health and Welfare, Department of Social Services and Healthcare Management, Otawara, Tochigi, Japan University of Tsukuba, Graduate School of Comprehensive Human Sciences, Tsukuba, Ibaraki, Japan University of Tsukuba, Faculty of Medicine, Tsukuba, Ibaraki, Japan Nippon Medical School, Department of Health Service Administration, Bunkyo, Tokyo, Japan
a r t i c l e
i n f o
Article history: Received 28 February 2013 Received in revised form 19 December 2013 Accepted 2 January 2014
Keywords: Cost containment Hospital behaviour Length of stay Private hospital Public hospital
a b s t r a c t The average length of stay (LOS) is considered one of the most significant indicators of hospital management. The steep decline in the average LOS among Japanese hospitals since the 1980s is considered to be due to cost-containment policies directed at reducing LOS. Japan’s hospital sector is characterised by a diversity of ownership types. We took advantage of this context to examine different hospital behaviours associated with ownerships types. Analysing government data published from 1971 to 2008 for the effect of a series of costcontainment policies aimed at reducing LOS revealed distinctly different paths behind the declines in LOS between privately owned and publicly owned hospitals. In the earlier years, private hospitals focused on providing long-term care to the elderly, while in the later years, they made a choice between providing long-term care and providing acute care with reduced LOS and bonus payments. By contrast, the majority of public hospitals opted to provide acute care with reduced LOS in line with public targets. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction The average length of stay (LOS) has been one of the most significant indicators of hospital management performance over the last four decades [1,2]. While some authors have pointed out the complex and diverse implications of LOS for hospital care [3,4], others have argued that reducing LOS leads to reduction in costs and improvements in effectiveness, or outcomes, of hospital care [5,6], thereby resulting in greater efficiency at the hospital level [7,8], as well as at the general level of the health system of the country as a whole [9]. The average LOS of Japanese hospital patients is much longer than those of other developed countries (Fig. 1a) [10]; however, since the 1980s, it has been remarkably reduced.
∗ Corresponding author. Tel.: +81 298 53 5904; fax: +81 298 53 5904. E-mail address: [email protected] (M. Kondo). 0168-8510/$ – see front matter © 2014 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.healthpol.2014.01.002
The cornerstone of Japanese health reform over the last four decades was an advocacy initiated by the head of the Insurance Bureau of the Ministry of Health and Welfare in 1983 that stated that ever-rising medical expenses would ruin the country [11]. Expansion policies in health care were switched to contraction and cost-containment policies in the mid-1980s, when a variety of measures were implemented to reduce LOS in the belief that shorter LOS would result in cost containment [12]. Many factors affect LOS trends, such as medical advancement [5,13], physician’s attitude [14], patient’s behaviour [15], socioeconomic factors [16,17], and payment system to hospitals [6,18]. However, it is generally believed that in the case of Japan, the governmental regulations and inducements inevitably moved the country’s hospitals toward reducing LOS [19–21]. Yet, while the average LOS as a whole has been successfully reduced, how or when each hospital responded to the policy and whether variations in behaviour exist among diverse hospitals have not been elucidated.
N. Kato et al. / Health Policy 115 (2014) 180–188
181
Fig. 1. Paths of average length of hospital stays, 1971–2008.
In stark contrast to the uniformity of the payment system by social insurers and patients, Japan’s hospital sector is characterised by a diversity of ownership types [22,23]: ownership by government at various levels from the national to the municipal levels, as well as by physicians and their families and by organisations such as farmers’ associations, charity bodies, and social insurers. In the United States, hospital ownership has been found to influence reductions in LOS [24,25]. However, no study to date has looked at the relationship between hospital ownership and reductions in LOS in Japan. The aim of this study was to demonstrate the different responses to cost-containment policies associated with LOS according to hospital ownership type in Japan.
2. Materials and methods To demonstrate the difference in responses to costcontainment policies related to LOS among Japanese hospitals according to ownership type, we carried out quantitative analysis of LOS, qualitative analysis of the policies, and interpretive analysis of the correspondences between the policies and LOS. First, to demonstrate distinct responses to the policies, we analysed paths of LOS according to ownership types by using government data
published between 1971 and 2008. Second, we conducted a chronological analysis of health-care policy based on the literature to examine the assumption that no LOSrelated policy existed targeting hospitals of any particular type of ownership; i.e., all hospitals were exposed to the same policy regardless of ownership type. Finally, we interpreted the correspondences between the costcontainment policies and the paths of LOS on the basis of the results of our qualitative and quantitative analyses.
2.1. Quantitative analysis of LOS 2.1.1. Data Government published data on LOS are summarised according to the legislative category of hospital beds. Five types of hospital beds are defined by the current Medical Care Law: general beds, long-term care beds, psychiatric beds, infectious disease beds, and tuberculosis beds, although long-term care beds were not distinguished from general beds until 2000. In this study, we focus on “ordinary beds,” defined as hospital beds other than psychiatric beds, infectious disease beds, or tuberculosis beds, to observe the LOS between 1971 and 2008. Although we include longterm care beds as ordinary beds in our analysis tracing
182
N. Kato et al. / Health Policy 115 (2014) 180–188
back to 1971, long-term care beds are evidently associated with LOS. However, given the new categorisation in 2001 of long-term care beds reflecting the governmental policy of reducing LOS by differentiating between acuteand chronic-care beds, including both types of beds allows us to pay close attention to health-care provision policies drafted in the Medical Care Law in the following analyses and discussions. “Hospital Report” [26], a publication based on a complete count survey of hospitals in Japan by the Ministry of Health, Labour and Welfare (MHLW), reports the average LOS of ordinary beds by ownership type annually since 1982, and the values can be traced back to 1971 by implementing the same calculations with published figures. As shown in Table 1, 27 small classifications of ownership types are currently defined by the Statistics and Information Department (SID) of the MHLW. Although several revisions have been made since 1971, we define 15 ownership-type groups for our analysis from the 16 small classifications of the SID. These 16 classifications can be traced back to 1971, and each share of all hospital beds was more than 1% in 2008. Although the National Hospital Organizations were separated from the MHLW in 2004, in our analysis, we merge them for our observations from 1971 to 2008. Table 1 also shows shares of hospitals and beds included in our analysis – 95.3% and 94.5%, respectively – in 2008. It must be difficult for international readers to imagine the nature of these ownership-type groups, since each is embedded in the culture and history of Japan’s health system. Although the ownerships have undergone many transformations over the last 40 years, basically, they can be simplified as follows: government ownerships by (1) the national government and (2) national universities; (3) the workman’s compensation scheme is an incorporated administrative agency controlled by the MHLW, which operates its own hospitals; (4) prefecture and (5) municipality hospitals are government ownerships; (6) the Red Cross in Japan operates its own hospitals; (7) the Saiseikai Imperial Gift Foundation is a non-governmental organisation established to help the poor, including by the provision of health care; (8) the Kouseiren Agricultural Cooperative is a farmers’ cooperation; (9) Zensharen is a social insurer for workers at small-to-midsize firms; (10) Kyosaikumiai is a large group of social insurers for public servants; (11) public interest corporations are non-profit organisations; (12) the private university corporation is a higher education intuition with hospitals; (13) the business enterprise consists of some long-standing private companies that own hospitals for their employees and that usually accept patients from the general public as well; (14) the medical cooperation is a form of privately owned hospital stipulated in the Medical Care Law; and (15) hospitals run by individual physicians are also a form of private ownership. In principle, every hospital is open to the general public. The SID classification locates these categories in the spectrum of public to private ownership to some extent. The ownership-type groups for our analysis from (1) to (8) are labelled public hospitals by the Medical Care Law. A more detailed difference between business enterprises
and medical corporations should be noted here, since the Medical Care Law in Japan prohibits hospitals from making profits. Therefore, for-profit companies cannot own hospitals today, while some long-standing industrial companies maintain the non-profit nature of their hospitals by also engaging in businesses other than hospital management. In contrast, medical corporations are deemed similar to forprofit corporations in the sense that they are established by direct investments from private shareholders [21]. Family ownership is one of the typical types of medical corporations. The spectrum of public to private ownership is also indicated in Table 1. Japanese ownership types are not compatible with conventional ownership types in the literature – public, non-profit private, and for-profit private – because of their high level of diversity, the legal ban on profit-making, and the entrepreneurial behaviour that is occasionally observed [21]. 2.1.2. Statistical analysis We examined the paths of LOS according to these 15 ownership-type groups from 1971 to 2008 with the assumption that LOS is predominantly changed by the supply side, i.e., the hospital side, for example, through revisions in discharging policy and target case-mix. On the demand side, i.e., the patient side, the hospital-ownership type is assumed to be irrelevant to the patients’ seeking care, owing to strictly standardised, comprehensive social insurance benefit packages [21]. We also assume that the effect of demographic changes, such as the increase in the percentage of the elderly population (aged 65 years and over) from 7.2% to 22.1%, or of epidemiologic changes, such as the replacement of the leading cause of death from stroke to cancer over the last 40 years, are independent from ownership types, although they might have influenced the overall LOS of both public and private hospitals over the study period. Cluster analysis of the paths of LOS was performed to classify ownership type groups into ownership clusters. IBM SPSS Statistics 20 was used for this analysis. Optional clustering methods including average linkage between/within groups, single linkage, complete linkage, the centroid method, the median method, and Ward’s method were applied against the squared Euclidean distance between the clusters to construct dendrograms for exploratory scrutiny [27]. One dendrogram was selected in light of the spectrum of public to private ownership. Furthermore, we conducted joinpoint analysis [28] to detect the inflection points of the paths by ownership clusters. The joinpoint regression technique is useful for delineating changes in trend data, especially when the number of change points is unknown. The Joinpoint Regression Program (version 3.4.2; Research and Applications Branch, National Cancer Institute 2009) was used for the analysis. 2.2. Qualitative analysis of policy We deliberately searched for literature that systematically allows us to identify health-care policies relevant
N. Kato et al. / Health Policy 115 (2014) 180–188
183
Table 1 Ownership-type groups. Ownership type groups for analysis
SID’s small classification
Public
(1) National government
(01) MHLW (02) National Hospital Organizations (03) National University Corporations (04) Japan Health, Labour and Welfare Organizations (06) Prefectures (07) Municipalities (09) Japanese Red Cross (10) Saiseikai Imperial Gift Foundation
1971
In-between
Private
(2) National university (3) Workman’s compensation scheme (4) Prefecture (5) Municipality (6) Red Cross (7) Saiseikai Imperial Gift Foundation (8) Agricultural Cooperative (Kouseiren) (9) Social insurer for workers at small-to-midsized firms (Zensharen) (10) Social insurer for public servants (Kyosaikumiai) (11) Public interest corporation (12) Private university corporation (13) Business enterprise (14) Medical corporation (15) Individual physician Others: excluded from our analysis
Number of beds†
Number of hospitals*
Spectrum of public to private ownership
2008
1971
2008
215
165
40,635
60,395
45 34
48 34
18,590 10,524
30,734 13,422
232 742 99 71
231 725 92 81
40,701 57,202 88,053 151,038 24,846 36,308 11,900 22,336
(12) National Welfare Federations
119
113
20,896
34,252
(14) All-Japan Federation of Social Insurance Associations
54
52
10,236
14,050
(18) Mutual Aid Associations and their Federations (20) Public Interest Corporations (22) Private University Corporations
52
45
10,670
14,335
250 44
330 105
30,136 15,587
66,242 52,729
128 1,667 2,801 358
69 4,831 436 366
13,130 13,713 113,873 582,475 118,942 32,849 45,314 68,704
6,911
7,723
614,033
(25) Companies (21) Medical Corporations (27) Individuals Other type of national governments (05) Other type of public organizations (08, 11, 13) Other type of social insurance bodies (15, 16, 17, 19) Other type of medical corporation (23, 24, 25, 26)
Total
1,250,784
SID: Statistics and Information Department; MHLW: Ministry of Health, Labour and Welfare. * Number of general hospitals excluding psychiatric hospitals and tuberculosis sanatoriums. † Number of ordinary beds.
to LOS between 1971 and 2008 to construct a chronological table of related policies. The PubMed database and Japana Centra Revuo Medicina database (a Japanese medical literature database) were searched with combinations of relevant terms such as policy, hospital management, LOS, cost containment, and hospital ownership. Additionally, we electronically or manually scanned the library of the Japanese National Institute of Public Health. As a result, we mainly referred to “The chronological table of the medical insurance system and revisions of the fee-schedule,” an appendix included in the “Estimates of national medical care expenditure 2008” [29]. This appendix gives a comprehensive overview of Japan’s health-care policy in the context of cost containment from which we could retrieve LOS-related policies. We also referred to the “Medical fee point book” [30] for details of financing policies and to the “Annual statistical report of national health conditions” [31] for details of health-care provision policies. Policies were identified and selected on the basis of their theoretical relevance to LOS, and their specificity to hospital ownership was scrutinised. We included investigator triangulation in this process: two of the authors each constructed a chronological table independently, and the four authors then collated the differences to finalise the table.
2.3. Interpretive analysis of the relationship between cost-containment policies and LOS Finally, to integrate the former results of the qualitative and quantitative analyses, we explored the correspondences between the cost-containment policies and the paths of LOS by each ownership cluster. 3. Results 3.1. Paths of LOS according to hospital ownership Fig. 2 shows the results of the hierarchical cluster analysis in the form of a dendrogram using single-linkage. Three clusters were formed from the 15 ownership-type groups. Twelve ownership-type groups formed a large cluster: from (2) national universities to (13) business enterprises. Two ownership type groups— (14) medical corporations and (15) individual physicians—formed a small cluster. And different from any of these, (1) the national government formed the third cluster. We labelled these three clusters as the public cluster, the private cluster, and the national government cluster on the basis of the spectrum of public to private ownership types that underlay the groupings. The ownership-type groups from (9) to (13) were located
184
N. Kato et al. / Health Policy 115 (2014) 180–188
0
Rescaled Distance Cluster Combination 5 10 15 20
25
5) Municipality
Public cluster
8) Agricultural Cooperative (Kouseiren) 12) Private university corporation 9) Social insurer for workers at small-to-mid-sized firms (Zensharen) 13) Business enterprise 6) Red Cross 4) Prefecture 7) Saiseikai Imperial Gift Foundation 10) Social insurer for public servants (Kyosaikumiai) 2) National university 11) Public interest corporation 3) Workman’s compensation scheme 14) Medical corporation
Private cluster
15) Individual physician National government cluster
1) National government
Fig. 2. Results of hierarchical cluster analysis: dendrogram using single linkage.
in the middle of the spectrum but were found to be similar to the groups labelled as public by the Medical Care Law, i.e., (2) to (8). This means that, as anticipated, ownership types from public to private affect LOS. The joinpoint regression analysis was applied for the paths of LOS by ownership clusters; the results are shown in Fig. 1b. The LOS of the public cluster was reduced from 32.4 days in 1974 to 17.2 days in 2008. The statistically significant joinpoints in 1980, 1985, and 1996 indicate that there was a change in the trend of LOS reduction in those years. The increased LOS of the private cluster was reduced from 53.1 days in 1986 to 43.1 days in 2008. The joinpoints in 1986 and 1999 indicate statistically significant changes. The increased LOS of the national government cluster was also reduced from 56.4 days in 1979 to 26.5 days in 2008, with statistically significant joinpoints in 1979 and 1998. These joinpoints are also shown in Fig. 3. 3.2. LOS-relevant cost-containment policies Fig. 3 illustrates the results of the qualitative analysis of the cost-containment policies As regards our assumption that there was no LOS-related policy targeting hospitals of any particular type of ownership, two policies were found targeting (1) the national government and (14) medical corporations. The former policy consists of the reorganisation plan of national hospitals in 1986, its revision in 1999, and the subsequent transformation of national hospitals into independent agencies in 2004. The purpose of this reorganisation was to compress the national hospital network and improve its efficiency. Although the objective of reducing LOS was not specified in the reorganisation plans, emphasis was placed on shifting from a service-mix to acute care, thus implying shorter LOS. In
1986, 138 sanatorium-type hospitals, which imply long LOS, operated in partnership with 98 general hospitals. By 2004, the reorganisation saw the complete abolishment of sanatorium-type hospitals, and the national hospital network became a network of 144 general hospitals through mergers and privatisations. The latter policy was made to create a subtype of medical corporation under the Medical Care Law in 2007, and we consider this policy as not meant to reduce LOS directly. Moreover, its impact could not be observed in our analysis within the time frame studied. Among the policies toward all hospital-ownership types, two tendencies can be observed. One is the differentiation of ordinary beds into acute-care and chronic-care beds, as discussed in the Methods section. This differentiation led to the introduction of long-term care insurance in 2000 and to the distinct category of long-term care beds under the Medical Care Law. Another tendency is the provision of financial incentives to hospitals that decided to adopt short LOS/acute care. The combination of these two types of policies was expected to lead to a reduction in the LOS of ordinary beds as a whole. Since the 1980s, hospital owners and managers have been repeatedly faced with management decision-making between short LOS/acute care and long LOS/chronic care of their ordinary beds, and as a result, their choice of direction toward short LOS/acute care reduced the LOS. These policies were motivated as a consequence of the removal of user charges for the elderly in 1973, which was one of the attainments of the health-care expansion policies. This removal of user charges for the elderly led to “social hospitalisation,” i.e., the use of hospitals as providers of geriatric nursing care offered free of charge [32]. Many ordinary beds became substitutes for nursing
N. Kato et al. / Health Policy 115 (2014) 180–188
185
Fig. 3. Chronological table of health-care policy in Japan related to LOS.
home beds, which prolonged the LOS during the 1970s (Fig. 1a) [20]. In the early 1980s, government policy was geared toward cost containment. In 1981, bonus payment levels for achieving shorter LOS, which had been set in the fee schedule since 1958, were drastically raised. In 1983, free medical care for the elderly came to an end, and the government allowed hospitals that provided such care to continue to serve as providers of “social hospitalisation” by converting to “hospitals for the elderly,” with small numbers of qualified staff and subject to independent fee schedules with no bonus payments for shorter LOS. In 1985, the Medical Care Law set a ceiling limit on the number of ordinary beds in a health district in the belief that an excess
number of beds prolonged LOS (i.e., Roemer’s law) [33]. Building new hospitals or wards, i.e., creating more beds, was banned in areas where there were already sufficient beds. From the late 1980s to the millennium, bonus payments for shorter LOS were continuously reinforced. In 1992, the Medical Care Law defined hospital wards for long-term care, and beds in these wards were subcategorised under ordinary beds, while in 1998, bonuses or penalties in the form of reimbursements depending on the average LOS at the ward level were clearly itemized in the medical fee schedule. In 2000, the long-term care insurance system was established to separate geriatric nursing care from medical insurance and to support social care for the elderly. In
186
N. Kato et al. / Health Policy 115 (2014) 180–188
2001, the Medical Care Law clearly stated and formally categorised long-term care beds as beds for long LOS/chronic care (not geriatric care), and since then, ordinary beds have been legally separated into general beds (short LOS/acute care) and long-term care beds (long LOS/chronic care). In 2003, a prospective payment system (Diagnosis Procedure Combination: DPC) was introduced to “acute-care hospitals” such as teaching hospitals or large hospitals, replacing the “fee for service” payment system. This is a diagnosisrelated groups (DRG) -like payment system designed to give financial incentives to hospitals to reduce their LOS [34]. It is generally believed that the DPC has also moved these hospitals toward reducing LOS [35]. In 2006, the government started to encourage the conversion of hospitals with beds for long-term care into long-term care institutions such as nursing homes. 3.3. Correspondences between LOS and health-care policies Finally, we examined the correspondences between health-care policies (Fig. 3) and the paths of LOS by each ownership cluster (Fig. 1b). We compared the paths of the public and private clusters before examining the national government cluster, because no ownership-specific policy has been imposed on these two ownership clusters. The paths of LOS of the public cluster and private cluster hospitals differ from each other. In 1971, both had similar LOS of about 30 days, but by 2008, their LOSs were 17 days and 43 days, respectively. The path of the public cluster starts with a gentle upward slope and then begins to drop from 1981, while that of the private cluster starts with a steep upward slope and begins to drop from 1987. Both downward slopes from the mid-1980s to the 1990s are almost parallel, but the LOS of the public cluster is about 23 days shorter than that of the private cluster. While the public cluster’s rate of reduction continued through 2008, that of the private cluster decelerated from 2000, thus widening the gap in the average LOS between the two clusters during the 2000s. Regarding the response to the policies, the private cluster was more responsive to the expansion policy at the early stage than was the public cluster, and we may presume that “social hospitalisation” involved mainly privately owned hospitals. In contrast, the public cluster promptly responded to the cost-containment policy around 1980, reflecting the fact that the owners/managers of the hospitals in this cluster shared the public objectives and interests. Of interest is the fact that the attainment target LOS for bonus payments set in the official fee schedule was 25 days in 2000, 21 days in 2002, and 14 days in 2006. And the path of the public cluster shows that hospitals in this cluster have consistently pursued these targets, while their similar path to those of other developed countries (Fig. 1) may simply reflect their interest in the advancement of cosmopolitan medicine or international trends in health care, such as day surgery, which are of common interest among developed countries. The setting of the ceiling limit on ordinary beds in 1985 was a counter policy to increasing ordinary beds that would have served for “social hospitalisation,” which appeared to
be occurring more in the private cluster. The private cluster responded to this policy by reducing the LOS of existing beds. The gap in the LOS between these two clusters reflects the fact that more hospitals in the private cluster chose to move towards long-term care, such as “hospitals for the elderly” in 1983, “hospitals empowered for long-term care” in 1990, and “wards for long-term care” in 1992. These are different types of geriatric nursing care hospitals/wards classified by the MHLW. The different categorisation of long-term care beds in 2001 seems to have moved private hospitals in the direction of a new category for their ordinary beds, which explains the deceleration in the rate of reduction of LOS. The path of LOS of the national government cluster starts with the longest LOS, 48 days in 1971, and then reduces to 27 days in 2008, which is midway between those of the public and private clusters. It starts with an upward slope and then begins to drop from 1980. Its downward slope is almost similar to that of the private cluster during the mid-1980s to the 1990s. This rate of reduction accelerates after 1999, approaching the public cluster’s LOS in the 2000s. The national government cluster seems to have responded similarly well to the expansion policy at the early stage as the private cluster. However, similar to the public cluster, it also promptly responded to the costcontainment policy around 1980. Its downward slope since the mid-1980s reflects its response to the reorganisation plan of national hospitals rather than to the cost-containment policies. And the acceleration in the rate of LOS reduction after 1999 may be presumed to be a response to the further revision of the reorganisation plan. 4. Discussion The changing average LOS of Japanese hospitals over the last four decades differs from those of other developed countries (Fig. 1a). It is generally believed that the steep decline seen since the 1980s has been a distinct response to cost-containment policies targeted at reducing LOS [19–21]. However, Japan’s hospital sector is characterised by its diversity of ownership types [22,23], and we have revealed the diverse paths of average LOS according to ownership-type groups that is behind this decline. Our hierarchical cluster analysis produced a cluster of public hospitals and a cluster of private hospitals, while the national government cluster has a path of its own (Fig. 2). Our joinpoint regression analyses produced contrasting differences among the three clusters (Fig. 1b). In particular, the analysis of the private cluster is different from those of the public or national government clusters, or the paths seen in other developed countries. Our review of health-care cost-containment policies in reducing LOS identified an absence of regulations and inducements targeting particular ownership-type groups except for a series of reorganisation policies for national government hospitals and a recent reform of medical corporations (Fig. 3). Arguably, the former may explain why national government hospitals have their own path, for it is natural that the MHLW enforces direct control over its own hospitals. The reduction in LOS was implicitly emphasised
N. Kato et al. / Health Policy 115 (2014) 180–188
and enforced in the reorganisation plans. The effect of the latter, a recent reform of medical corporations implemented in 2007, could not be observed within our time frame from 1971 to 2008. We found two main tendencies of the cost-containment policies: separation of acute care and chronic care and financial incentives to encourage reduced LOS (Fig. 3). Our analyses have several shortcomings. Our analysis focuses on the category of “ordinary beds,” which include beds for chronic care, since it was the only dataset available to examine ownership differences over the last 40 years. However, this makes our interpretation ambiguous. A further study separating LOS for long-term care from LOS for short-term care since 2001 might reveal a different and detailed pattern according to hospital ownership types. Furthermore, our categorisation of hospital ownership types does not reflect hospital networks under single ownership. Private acute-care hospitals and private chronic-care hospitals may coordinate with each other. The factors not examined were demographic and epidemiologic transitions on the demand side, or only a brief mention of numbers of hospitals and beds on the supply side of health care, although these factors are significant determinants of LOS. And our study is embedded in the Japanese context of policies in reducing LOS, so its relevance to health system performance or hospital management was not examined. Nevertheless, the Japanese context enables us to observe LOS over the long term and its relevance to health-care policies, which makes this case study unique and worthwhile.
5. Conclusion In conclusion, by combining quantitative analysis of LOS and qualitative analysis of relevant policies, we determined that the responses of privately and publicly owned hospitals to cost-containment policies for reducing LOS over the last 40 years in Japan were different. Privately owned hospitals made deliberate decisions between acute-care beds and chronic-care beds in response to the incentives provided by the policies, whereas the majority of public hospitals opted to provide acute care with reduced LOS, in line with public policy. Thus, the creation of a new ownership type of social medical corporation under the recent reform in 2007 has the rationale of fundamentally trying to draw a different response from private hospitals by encouraging them to change their ownership type. It allows medical corporations to become social medical corporations with tax privileges, eligibility for subsidies for community care provision, and so forth, by accepting in turn certain conditions such as giving up family ownership. Among developed countries, the changing role of the state or the recognition that private hospitals are more efficient than public ones has led to a trend towards hospital privatisation [36]. However, this case study suggests that policy makers may need to take a long-term view of the regulation of private hospitals because private hospitals are less likely than public hospitals to be in line with public targets in the long run.
187
References [1] Brian E. Foundation for medical care control of hospital utilization: CHAP—a PSRO prototype. New England Journal of Medicine 1973;288:878–82. [2] White HL, Glazier RH. Do hospitalist physicians improve the quality of inpatient care delivery? A systematic review of process, efficiency and outcome measures. BMC Medicine 2011;9:58. [3] Boaz RF. Utilization review and containment for hospital utilization: some implications of providing care in the “most appropriate setting”. Medical Care 1979;17:315–30. [4] Rotter T, Kugler J, Koch R, Gothe H, Twork S, van Oostrum JM, Steyerberg EW. A systematic review and meta-analysis of the effects of clinical pathways on length of stay, hospital costs and patient outcomes. BMC Health Services Research 2008;8:265. [5] Sloan FA, Valvona J. Why has hospital length of stay declined? An evaluation of alternative theories. Social Science and Medicine 1986;22:63–73. [6] Clarke A. Why are we trying to reduce length of stay? Evaluation of the costs and benefits of reducing time in hospital must start from the objectives that govern change. Quality in Health Care 1996;5: 172–9. [7] Lagoe RJ, Westert GP, Kendrick K, Morreale G, Mnich S. Managing hospital length of stay reduction: a multihospital approach. Health Care Management Review 2005;30:82–92. [8] Hussey PS, de Vries H, Romley J, Wang MC, Chen SS, Shekelle PG, McGlynn EA. A systematic review of health care efficiency measures. Health Services Research 2009;44:784–805. [9] Hall CJ, Peel NM, Comans TA, Gray LC, Scuffham PA. Can postacute care programmes for older people reduce overall costs in the health system? A case study using the Australian Transition Care Programme. Health & Social Care in the Community 2012;20:97– 102. [10] OECD. OECD health data 2010. Paris: OECD; 2010. [11] Yoshimura H. Personal view on trend and response toward medical expense. Kenkohoken 1983;37:28–32. [12] Ikegami N, Campbell JC. Medical care in Japan. New England Journal of Medicine 1995;333:1295–9. [13] Mitsutake N, Nishimura Y, Sakamoto S, Takahashi T. The impact of decreasing hospital average length of stay on clinical outcome. Hospital Administration 2002;39:5–12. [14] Louis DZ, Taroni F, Yuen EJ, Umesato Y, Gonnella JS. Patterns of hospital care and physician perspectives from an Italian, Japanese, and USA hospital. American Journal of Medical Quality 1996;11: 123–34. [15] Muramatsu N, Liang J. Hospital length of stay in the United States and Japan: a case study of myocardial infarction patients. International Journal of Health Services 1999;29:189–209. [16] Ogata A, Kono K, Yamamoto K, Nishikawa M, Deguchi Y, Kagehira K. Analysis of average length of hospital stay, bed occupancy rate and social structure at prefectural level. Hospital Administration 1983;20:231–41. [17] Egawa H. Average length of stay in hospitals. Journal of University of Occupational and Environmental Health 1984;6:1–14. [18] Weiner SL, Maxwell JH, Sapolsky HM, Dunn DL, Hsiao WC. Economic incentives and organizational realities: managing hospitals under DRGs. Milbank Quarterly 1987;65:463–87. [19] Fujita Y. The administrative management methods in medical policy since 1975 in Japan. Review of Population and Social Policy 1996;5:73–96. [20] Campbell JC, Ikegami N. The art of balance in health policy: maintaining Japan’s low-cost, egalitarian system. Cambridge: Cambridge University Press; 1998. [21] Tatara K, Okamoto E. Japan: health system review. Health Systems in Transition 2009;11:1–164. [22] Roemer MI. National health systems of the world. Oxford: Oxford University Press; 1991. [23] Yoshikawa A, Bhattacharya J, Vogt WB. Health economics of Japan: patients, doctors and hospitals under a universal health insurance system. Tokyo: University of Tokyo Press; 1996. [24] Freund D, Shachtman RH, Ruffin M, Quade D. Analysis of length-ofstay differences between investor-owned and voluntary hospitals. Inquiry 1985;22:33–44. [25] Younis MZ. Length of hospital stay of Medicare patients in the postprospective-payment-system era. Journal of Health Care Finance 2004;31:23–30. [26] Statistics and Information Department. Ministry of health labour and welfare. Hospital report. Health and Welfare Statistics Association: Tokyo; 1973–2010.
188
N. Kato et al. / Health Policy 115 (2014) 180–188
[27] Everitt BS, Dunn G. Applied multivariate data analysis, 2nd ed. London: Wiley; 2001. [28] Kim HJ, Fay MP, Feuer EJ, Midthune DN. Permutation tests for joinpoint regression with applications to cancer rates. Statistics in Medicine 2000;19:335–51. [29] Statistics Information Department. Ministry of Health Labour and Welfare. Estimates of national medical care expenditure 2008. Health and Welfare Statistics Association; 2010. [30] Social Insurance Institute. Medical fee point book. Tokyo: Social Insurance Institute; 1970–2008. [31] Health and Welfare Statistics Association. Annual statistical report of national health conditions. Tokyo: Health and Welfare Statistics Association; 1971–2010. [32] Idezuki Y. Long live the health care system in Japan. Bioscience Trends 2008;2:50–2.
[33] Ginsburg PB, Koretz DM. Bed availability and hospital utilization: estimates of the “Roemer effect”. Health Care Financing Review 1983;5:87–92. [34] Hamada H, Sekimoto M, Imanaka Y. Effects of the per diem prospective payment system with DRG-like grouping system (DPC/PDPS) on resource usage and healthcare quality in Japan. Health Policy 2012;107:194–201. [35] Anderson G, Ikegami N. How can Japan’s DPC inpatient hospital payment system be strengthened? Washington, DC: Lessons from the U.S. Medicare Prospective System. Center for Strategic and International Studies; 2011. [36] Schmid A, Cacace M, Götze R, Rothgang H. Explaining health care system change: problem pressure and the emergence of “hybrid” health care systems. Journal of Health Politics Policy and Law 2010;35:455–86.
Contents lists available at ScienceDirect
Health Policy journal homepage: www.elsevier.com/locate/healthpol
Length of hospital stay in Japan 1971–2008: Hospital ownership and cost-containment policies Naoko Kato a,b , Masahide Kondo c,∗ , Ichiro Okubo c , Toshihiko Hasegawa d a b c d
International University of Health and Welfare, Department of Social Services and Healthcare Management, Otawara, Tochigi, Japan University of Tsukuba, Graduate School of Comprehensive Human Sciences, Tsukuba, Ibaraki, Japan University of Tsukuba, Faculty of Medicine, Tsukuba, Ibaraki, Japan Nippon Medical School, Department of Health Service Administration, Bunkyo, Tokyo, Japan
a r t i c l e
i n f o
Article history: Received 28 February 2013 Received in revised form 19 December 2013 Accepted 2 January 2014
Keywords: Cost containment Hospital behaviour Length of stay Private hospital Public hospital
a b s t r a c t The average length of stay (LOS) is considered one of the most significant indicators of hospital management. The steep decline in the average LOS among Japanese hospitals since the 1980s is considered to be due to cost-containment policies directed at reducing LOS. Japan’s hospital sector is characterised by a diversity of ownership types. We took advantage of this context to examine different hospital behaviours associated with ownerships types. Analysing government data published from 1971 to 2008 for the effect of a series of costcontainment policies aimed at reducing LOS revealed distinctly different paths behind the declines in LOS between privately owned and publicly owned hospitals. In the earlier years, private hospitals focused on providing long-term care to the elderly, while in the later years, they made a choice between providing long-term care and providing acute care with reduced LOS and bonus payments. By contrast, the majority of public hospitals opted to provide acute care with reduced LOS in line with public targets. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction The average length of stay (LOS) has been one of the most significant indicators of hospital management performance over the last four decades [1,2]. While some authors have pointed out the complex and diverse implications of LOS for hospital care [3,4], others have argued that reducing LOS leads to reduction in costs and improvements in effectiveness, or outcomes, of hospital care [5,6], thereby resulting in greater efficiency at the hospital level [7,8], as well as at the general level of the health system of the country as a whole [9]. The average LOS of Japanese hospital patients is much longer than those of other developed countries (Fig. 1a) [10]; however, since the 1980s, it has been remarkably reduced.
∗ Corresponding author. Tel.: +81 298 53 5904; fax: +81 298 53 5904. E-mail address: [email protected] (M. Kondo). 0168-8510/$ – see front matter © 2014 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.healthpol.2014.01.002
The cornerstone of Japanese health reform over the last four decades was an advocacy initiated by the head of the Insurance Bureau of the Ministry of Health and Welfare in 1983 that stated that ever-rising medical expenses would ruin the country [11]. Expansion policies in health care were switched to contraction and cost-containment policies in the mid-1980s, when a variety of measures were implemented to reduce LOS in the belief that shorter LOS would result in cost containment [12]. Many factors affect LOS trends, such as medical advancement [5,13], physician’s attitude [14], patient’s behaviour [15], socioeconomic factors [16,17], and payment system to hospitals [6,18]. However, it is generally believed that in the case of Japan, the governmental regulations and inducements inevitably moved the country’s hospitals toward reducing LOS [19–21]. Yet, while the average LOS as a whole has been successfully reduced, how or when each hospital responded to the policy and whether variations in behaviour exist among diverse hospitals have not been elucidated.
N. Kato et al. / Health Policy 115 (2014) 180–188
181
Fig. 1. Paths of average length of hospital stays, 1971–2008.
In stark contrast to the uniformity of the payment system by social insurers and patients, Japan’s hospital sector is characterised by a diversity of ownership types [22,23]: ownership by government at various levels from the national to the municipal levels, as well as by physicians and their families and by organisations such as farmers’ associations, charity bodies, and social insurers. In the United States, hospital ownership has been found to influence reductions in LOS [24,25]. However, no study to date has looked at the relationship between hospital ownership and reductions in LOS in Japan. The aim of this study was to demonstrate the different responses to cost-containment policies associated with LOS according to hospital ownership type in Japan.
2. Materials and methods To demonstrate the difference in responses to costcontainment policies related to LOS among Japanese hospitals according to ownership type, we carried out quantitative analysis of LOS, qualitative analysis of the policies, and interpretive analysis of the correspondences between the policies and LOS. First, to demonstrate distinct responses to the policies, we analysed paths of LOS according to ownership types by using government data
published between 1971 and 2008. Second, we conducted a chronological analysis of health-care policy based on the literature to examine the assumption that no LOSrelated policy existed targeting hospitals of any particular type of ownership; i.e., all hospitals were exposed to the same policy regardless of ownership type. Finally, we interpreted the correspondences between the costcontainment policies and the paths of LOS on the basis of the results of our qualitative and quantitative analyses.
2.1. Quantitative analysis of LOS 2.1.1. Data Government published data on LOS are summarised according to the legislative category of hospital beds. Five types of hospital beds are defined by the current Medical Care Law: general beds, long-term care beds, psychiatric beds, infectious disease beds, and tuberculosis beds, although long-term care beds were not distinguished from general beds until 2000. In this study, we focus on “ordinary beds,” defined as hospital beds other than psychiatric beds, infectious disease beds, or tuberculosis beds, to observe the LOS between 1971 and 2008. Although we include longterm care beds as ordinary beds in our analysis tracing
182
N. Kato et al. / Health Policy 115 (2014) 180–188
back to 1971, long-term care beds are evidently associated with LOS. However, given the new categorisation in 2001 of long-term care beds reflecting the governmental policy of reducing LOS by differentiating between acuteand chronic-care beds, including both types of beds allows us to pay close attention to health-care provision policies drafted in the Medical Care Law in the following analyses and discussions. “Hospital Report” [26], a publication based on a complete count survey of hospitals in Japan by the Ministry of Health, Labour and Welfare (MHLW), reports the average LOS of ordinary beds by ownership type annually since 1982, and the values can be traced back to 1971 by implementing the same calculations with published figures. As shown in Table 1, 27 small classifications of ownership types are currently defined by the Statistics and Information Department (SID) of the MHLW. Although several revisions have been made since 1971, we define 15 ownership-type groups for our analysis from the 16 small classifications of the SID. These 16 classifications can be traced back to 1971, and each share of all hospital beds was more than 1% in 2008. Although the National Hospital Organizations were separated from the MHLW in 2004, in our analysis, we merge them for our observations from 1971 to 2008. Table 1 also shows shares of hospitals and beds included in our analysis – 95.3% and 94.5%, respectively – in 2008. It must be difficult for international readers to imagine the nature of these ownership-type groups, since each is embedded in the culture and history of Japan’s health system. Although the ownerships have undergone many transformations over the last 40 years, basically, they can be simplified as follows: government ownerships by (1) the national government and (2) national universities; (3) the workman’s compensation scheme is an incorporated administrative agency controlled by the MHLW, which operates its own hospitals; (4) prefecture and (5) municipality hospitals are government ownerships; (6) the Red Cross in Japan operates its own hospitals; (7) the Saiseikai Imperial Gift Foundation is a non-governmental organisation established to help the poor, including by the provision of health care; (8) the Kouseiren Agricultural Cooperative is a farmers’ cooperation; (9) Zensharen is a social insurer for workers at small-to-midsize firms; (10) Kyosaikumiai is a large group of social insurers for public servants; (11) public interest corporations are non-profit organisations; (12) the private university corporation is a higher education intuition with hospitals; (13) the business enterprise consists of some long-standing private companies that own hospitals for their employees and that usually accept patients from the general public as well; (14) the medical cooperation is a form of privately owned hospital stipulated in the Medical Care Law; and (15) hospitals run by individual physicians are also a form of private ownership. In principle, every hospital is open to the general public. The SID classification locates these categories in the spectrum of public to private ownership to some extent. The ownership-type groups for our analysis from (1) to (8) are labelled public hospitals by the Medical Care Law. A more detailed difference between business enterprises
and medical corporations should be noted here, since the Medical Care Law in Japan prohibits hospitals from making profits. Therefore, for-profit companies cannot own hospitals today, while some long-standing industrial companies maintain the non-profit nature of their hospitals by also engaging in businesses other than hospital management. In contrast, medical corporations are deemed similar to forprofit corporations in the sense that they are established by direct investments from private shareholders [21]. Family ownership is one of the typical types of medical corporations. The spectrum of public to private ownership is also indicated in Table 1. Japanese ownership types are not compatible with conventional ownership types in the literature – public, non-profit private, and for-profit private – because of their high level of diversity, the legal ban on profit-making, and the entrepreneurial behaviour that is occasionally observed [21]. 2.1.2. Statistical analysis We examined the paths of LOS according to these 15 ownership-type groups from 1971 to 2008 with the assumption that LOS is predominantly changed by the supply side, i.e., the hospital side, for example, through revisions in discharging policy and target case-mix. On the demand side, i.e., the patient side, the hospital-ownership type is assumed to be irrelevant to the patients’ seeking care, owing to strictly standardised, comprehensive social insurance benefit packages [21]. We also assume that the effect of demographic changes, such as the increase in the percentage of the elderly population (aged 65 years and over) from 7.2% to 22.1%, or of epidemiologic changes, such as the replacement of the leading cause of death from stroke to cancer over the last 40 years, are independent from ownership types, although they might have influenced the overall LOS of both public and private hospitals over the study period. Cluster analysis of the paths of LOS was performed to classify ownership type groups into ownership clusters. IBM SPSS Statistics 20 was used for this analysis. Optional clustering methods including average linkage between/within groups, single linkage, complete linkage, the centroid method, the median method, and Ward’s method were applied against the squared Euclidean distance between the clusters to construct dendrograms for exploratory scrutiny [27]. One dendrogram was selected in light of the spectrum of public to private ownership. Furthermore, we conducted joinpoint analysis [28] to detect the inflection points of the paths by ownership clusters. The joinpoint regression technique is useful for delineating changes in trend data, especially when the number of change points is unknown. The Joinpoint Regression Program (version 3.4.2; Research and Applications Branch, National Cancer Institute 2009) was used for the analysis. 2.2. Qualitative analysis of policy We deliberately searched for literature that systematically allows us to identify health-care policies relevant
N. Kato et al. / Health Policy 115 (2014) 180–188
183
Table 1 Ownership-type groups. Ownership type groups for analysis
SID’s small classification
Public
(1) National government
(01) MHLW (02) National Hospital Organizations (03) National University Corporations (04) Japan Health, Labour and Welfare Organizations (06) Prefectures (07) Municipalities (09) Japanese Red Cross (10) Saiseikai Imperial Gift Foundation
1971
In-between
Private
(2) National university (3) Workman’s compensation scheme (4) Prefecture (5) Municipality (6) Red Cross (7) Saiseikai Imperial Gift Foundation (8) Agricultural Cooperative (Kouseiren) (9) Social insurer for workers at small-to-midsized firms (Zensharen) (10) Social insurer for public servants (Kyosaikumiai) (11) Public interest corporation (12) Private university corporation (13) Business enterprise (14) Medical corporation (15) Individual physician Others: excluded from our analysis
Number of beds†
Number of hospitals*
Spectrum of public to private ownership
2008
1971
2008
215
165
40,635
60,395
45 34
48 34
18,590 10,524
30,734 13,422
232 742 99 71
231 725 92 81
40,701 57,202 88,053 151,038 24,846 36,308 11,900 22,336
(12) National Welfare Federations
119
113
20,896
34,252
(14) All-Japan Federation of Social Insurance Associations
54
52
10,236
14,050
(18) Mutual Aid Associations and their Federations (20) Public Interest Corporations (22) Private University Corporations
52
45
10,670
14,335
250 44
330 105
30,136 15,587
66,242 52,729
128 1,667 2,801 358
69 4,831 436 366
13,130 13,713 113,873 582,475 118,942 32,849 45,314 68,704
6,911
7,723
614,033
(25) Companies (21) Medical Corporations (27) Individuals Other type of national governments (05) Other type of public organizations (08, 11, 13) Other type of social insurance bodies (15, 16, 17, 19) Other type of medical corporation (23, 24, 25, 26)
Total
1,250,784
SID: Statistics and Information Department; MHLW: Ministry of Health, Labour and Welfare. * Number of general hospitals excluding psychiatric hospitals and tuberculosis sanatoriums. † Number of ordinary beds.
to LOS between 1971 and 2008 to construct a chronological table of related policies. The PubMed database and Japana Centra Revuo Medicina database (a Japanese medical literature database) were searched with combinations of relevant terms such as policy, hospital management, LOS, cost containment, and hospital ownership. Additionally, we electronically or manually scanned the library of the Japanese National Institute of Public Health. As a result, we mainly referred to “The chronological table of the medical insurance system and revisions of the fee-schedule,” an appendix included in the “Estimates of national medical care expenditure 2008” [29]. This appendix gives a comprehensive overview of Japan’s health-care policy in the context of cost containment from which we could retrieve LOS-related policies. We also referred to the “Medical fee point book” [30] for details of financing policies and to the “Annual statistical report of national health conditions” [31] for details of health-care provision policies. Policies were identified and selected on the basis of their theoretical relevance to LOS, and their specificity to hospital ownership was scrutinised. We included investigator triangulation in this process: two of the authors each constructed a chronological table independently, and the four authors then collated the differences to finalise the table.
2.3. Interpretive analysis of the relationship between cost-containment policies and LOS Finally, to integrate the former results of the qualitative and quantitative analyses, we explored the correspondences between the cost-containment policies and the paths of LOS by each ownership cluster. 3. Results 3.1. Paths of LOS according to hospital ownership Fig. 2 shows the results of the hierarchical cluster analysis in the form of a dendrogram using single-linkage. Three clusters were formed from the 15 ownership-type groups. Twelve ownership-type groups formed a large cluster: from (2) national universities to (13) business enterprises. Two ownership type groups— (14) medical corporations and (15) individual physicians—formed a small cluster. And different from any of these, (1) the national government formed the third cluster. We labelled these three clusters as the public cluster, the private cluster, and the national government cluster on the basis of the spectrum of public to private ownership types that underlay the groupings. The ownership-type groups from (9) to (13) were located
184
N. Kato et al. / Health Policy 115 (2014) 180–188
0
Rescaled Distance Cluster Combination 5 10 15 20
25
5) Municipality
Public cluster
8) Agricultural Cooperative (Kouseiren) 12) Private university corporation 9) Social insurer for workers at small-to-mid-sized firms (Zensharen) 13) Business enterprise 6) Red Cross 4) Prefecture 7) Saiseikai Imperial Gift Foundation 10) Social insurer for public servants (Kyosaikumiai) 2) National university 11) Public interest corporation 3) Workman’s compensation scheme 14) Medical corporation
Private cluster
15) Individual physician National government cluster
1) National government
Fig. 2. Results of hierarchical cluster analysis: dendrogram using single linkage.
in the middle of the spectrum but were found to be similar to the groups labelled as public by the Medical Care Law, i.e., (2) to (8). This means that, as anticipated, ownership types from public to private affect LOS. The joinpoint regression analysis was applied for the paths of LOS by ownership clusters; the results are shown in Fig. 1b. The LOS of the public cluster was reduced from 32.4 days in 1974 to 17.2 days in 2008. The statistically significant joinpoints in 1980, 1985, and 1996 indicate that there was a change in the trend of LOS reduction in those years. The increased LOS of the private cluster was reduced from 53.1 days in 1986 to 43.1 days in 2008. The joinpoints in 1986 and 1999 indicate statistically significant changes. The increased LOS of the national government cluster was also reduced from 56.4 days in 1979 to 26.5 days in 2008, with statistically significant joinpoints in 1979 and 1998. These joinpoints are also shown in Fig. 3. 3.2. LOS-relevant cost-containment policies Fig. 3 illustrates the results of the qualitative analysis of the cost-containment policies As regards our assumption that there was no LOS-related policy targeting hospitals of any particular type of ownership, two policies were found targeting (1) the national government and (14) medical corporations. The former policy consists of the reorganisation plan of national hospitals in 1986, its revision in 1999, and the subsequent transformation of national hospitals into independent agencies in 2004. The purpose of this reorganisation was to compress the national hospital network and improve its efficiency. Although the objective of reducing LOS was not specified in the reorganisation plans, emphasis was placed on shifting from a service-mix to acute care, thus implying shorter LOS. In
1986, 138 sanatorium-type hospitals, which imply long LOS, operated in partnership with 98 general hospitals. By 2004, the reorganisation saw the complete abolishment of sanatorium-type hospitals, and the national hospital network became a network of 144 general hospitals through mergers and privatisations. The latter policy was made to create a subtype of medical corporation under the Medical Care Law in 2007, and we consider this policy as not meant to reduce LOS directly. Moreover, its impact could not be observed in our analysis within the time frame studied. Among the policies toward all hospital-ownership types, two tendencies can be observed. One is the differentiation of ordinary beds into acute-care and chronic-care beds, as discussed in the Methods section. This differentiation led to the introduction of long-term care insurance in 2000 and to the distinct category of long-term care beds under the Medical Care Law. Another tendency is the provision of financial incentives to hospitals that decided to adopt short LOS/acute care. The combination of these two types of policies was expected to lead to a reduction in the LOS of ordinary beds as a whole. Since the 1980s, hospital owners and managers have been repeatedly faced with management decision-making between short LOS/acute care and long LOS/chronic care of their ordinary beds, and as a result, their choice of direction toward short LOS/acute care reduced the LOS. These policies were motivated as a consequence of the removal of user charges for the elderly in 1973, which was one of the attainments of the health-care expansion policies. This removal of user charges for the elderly led to “social hospitalisation,” i.e., the use of hospitals as providers of geriatric nursing care offered free of charge [32]. Many ordinary beds became substitutes for nursing
N. Kato et al. / Health Policy 115 (2014) 180–188
185
Fig. 3. Chronological table of health-care policy in Japan related to LOS.
home beds, which prolonged the LOS during the 1970s (Fig. 1a) [20]. In the early 1980s, government policy was geared toward cost containment. In 1981, bonus payment levels for achieving shorter LOS, which had been set in the fee schedule since 1958, were drastically raised. In 1983, free medical care for the elderly came to an end, and the government allowed hospitals that provided such care to continue to serve as providers of “social hospitalisation” by converting to “hospitals for the elderly,” with small numbers of qualified staff and subject to independent fee schedules with no bonus payments for shorter LOS. In 1985, the Medical Care Law set a ceiling limit on the number of ordinary beds in a health district in the belief that an excess
number of beds prolonged LOS (i.e., Roemer’s law) [33]. Building new hospitals or wards, i.e., creating more beds, was banned in areas where there were already sufficient beds. From the late 1980s to the millennium, bonus payments for shorter LOS were continuously reinforced. In 1992, the Medical Care Law defined hospital wards for long-term care, and beds in these wards were subcategorised under ordinary beds, while in 1998, bonuses or penalties in the form of reimbursements depending on the average LOS at the ward level were clearly itemized in the medical fee schedule. In 2000, the long-term care insurance system was established to separate geriatric nursing care from medical insurance and to support social care for the elderly. In
186
N. Kato et al. / Health Policy 115 (2014) 180–188
2001, the Medical Care Law clearly stated and formally categorised long-term care beds as beds for long LOS/chronic care (not geriatric care), and since then, ordinary beds have been legally separated into general beds (short LOS/acute care) and long-term care beds (long LOS/chronic care). In 2003, a prospective payment system (Diagnosis Procedure Combination: DPC) was introduced to “acute-care hospitals” such as teaching hospitals or large hospitals, replacing the “fee for service” payment system. This is a diagnosisrelated groups (DRG) -like payment system designed to give financial incentives to hospitals to reduce their LOS [34]. It is generally believed that the DPC has also moved these hospitals toward reducing LOS [35]. In 2006, the government started to encourage the conversion of hospitals with beds for long-term care into long-term care institutions such as nursing homes. 3.3. Correspondences between LOS and health-care policies Finally, we examined the correspondences between health-care policies (Fig. 3) and the paths of LOS by each ownership cluster (Fig. 1b). We compared the paths of the public and private clusters before examining the national government cluster, because no ownership-specific policy has been imposed on these two ownership clusters. The paths of LOS of the public cluster and private cluster hospitals differ from each other. In 1971, both had similar LOS of about 30 days, but by 2008, their LOSs were 17 days and 43 days, respectively. The path of the public cluster starts with a gentle upward slope and then begins to drop from 1981, while that of the private cluster starts with a steep upward slope and begins to drop from 1987. Both downward slopes from the mid-1980s to the 1990s are almost parallel, but the LOS of the public cluster is about 23 days shorter than that of the private cluster. While the public cluster’s rate of reduction continued through 2008, that of the private cluster decelerated from 2000, thus widening the gap in the average LOS between the two clusters during the 2000s. Regarding the response to the policies, the private cluster was more responsive to the expansion policy at the early stage than was the public cluster, and we may presume that “social hospitalisation” involved mainly privately owned hospitals. In contrast, the public cluster promptly responded to the cost-containment policy around 1980, reflecting the fact that the owners/managers of the hospitals in this cluster shared the public objectives and interests. Of interest is the fact that the attainment target LOS for bonus payments set in the official fee schedule was 25 days in 2000, 21 days in 2002, and 14 days in 2006. And the path of the public cluster shows that hospitals in this cluster have consistently pursued these targets, while their similar path to those of other developed countries (Fig. 1) may simply reflect their interest in the advancement of cosmopolitan medicine or international trends in health care, such as day surgery, which are of common interest among developed countries. The setting of the ceiling limit on ordinary beds in 1985 was a counter policy to increasing ordinary beds that would have served for “social hospitalisation,” which appeared to
be occurring more in the private cluster. The private cluster responded to this policy by reducing the LOS of existing beds. The gap in the LOS between these two clusters reflects the fact that more hospitals in the private cluster chose to move towards long-term care, such as “hospitals for the elderly” in 1983, “hospitals empowered for long-term care” in 1990, and “wards for long-term care” in 1992. These are different types of geriatric nursing care hospitals/wards classified by the MHLW. The different categorisation of long-term care beds in 2001 seems to have moved private hospitals in the direction of a new category for their ordinary beds, which explains the deceleration in the rate of reduction of LOS. The path of LOS of the national government cluster starts with the longest LOS, 48 days in 1971, and then reduces to 27 days in 2008, which is midway between those of the public and private clusters. It starts with an upward slope and then begins to drop from 1980. Its downward slope is almost similar to that of the private cluster during the mid-1980s to the 1990s. This rate of reduction accelerates after 1999, approaching the public cluster’s LOS in the 2000s. The national government cluster seems to have responded similarly well to the expansion policy at the early stage as the private cluster. However, similar to the public cluster, it also promptly responded to the costcontainment policy around 1980. Its downward slope since the mid-1980s reflects its response to the reorganisation plan of national hospitals rather than to the cost-containment policies. And the acceleration in the rate of LOS reduction after 1999 may be presumed to be a response to the further revision of the reorganisation plan. 4. Discussion The changing average LOS of Japanese hospitals over the last four decades differs from those of other developed countries (Fig. 1a). It is generally believed that the steep decline seen since the 1980s has been a distinct response to cost-containment policies targeted at reducing LOS [19–21]. However, Japan’s hospital sector is characterised by its diversity of ownership types [22,23], and we have revealed the diverse paths of average LOS according to ownership-type groups that is behind this decline. Our hierarchical cluster analysis produced a cluster of public hospitals and a cluster of private hospitals, while the national government cluster has a path of its own (Fig. 2). Our joinpoint regression analyses produced contrasting differences among the three clusters (Fig. 1b). In particular, the analysis of the private cluster is different from those of the public or national government clusters, or the paths seen in other developed countries. Our review of health-care cost-containment policies in reducing LOS identified an absence of regulations and inducements targeting particular ownership-type groups except for a series of reorganisation policies for national government hospitals and a recent reform of medical corporations (Fig. 3). Arguably, the former may explain why national government hospitals have their own path, for it is natural that the MHLW enforces direct control over its own hospitals. The reduction in LOS was implicitly emphasised
N. Kato et al. / Health Policy 115 (2014) 180–188
and enforced in the reorganisation plans. The effect of the latter, a recent reform of medical corporations implemented in 2007, could not be observed within our time frame from 1971 to 2008. We found two main tendencies of the cost-containment policies: separation of acute care and chronic care and financial incentives to encourage reduced LOS (Fig. 3). Our analyses have several shortcomings. Our analysis focuses on the category of “ordinary beds,” which include beds for chronic care, since it was the only dataset available to examine ownership differences over the last 40 years. However, this makes our interpretation ambiguous. A further study separating LOS for long-term care from LOS for short-term care since 2001 might reveal a different and detailed pattern according to hospital ownership types. Furthermore, our categorisation of hospital ownership types does not reflect hospital networks under single ownership. Private acute-care hospitals and private chronic-care hospitals may coordinate with each other. The factors not examined were demographic and epidemiologic transitions on the demand side, or only a brief mention of numbers of hospitals and beds on the supply side of health care, although these factors are significant determinants of LOS. And our study is embedded in the Japanese context of policies in reducing LOS, so its relevance to health system performance or hospital management was not examined. Nevertheless, the Japanese context enables us to observe LOS over the long term and its relevance to health-care policies, which makes this case study unique and worthwhile.
5. Conclusion In conclusion, by combining quantitative analysis of LOS and qualitative analysis of relevant policies, we determined that the responses of privately and publicly owned hospitals to cost-containment policies for reducing LOS over the last 40 years in Japan were different. Privately owned hospitals made deliberate decisions between acute-care beds and chronic-care beds in response to the incentives provided by the policies, whereas the majority of public hospitals opted to provide acute care with reduced LOS, in line with public policy. Thus, the creation of a new ownership type of social medical corporation under the recent reform in 2007 has the rationale of fundamentally trying to draw a different response from private hospitals by encouraging them to change their ownership type. It allows medical corporations to become social medical corporations with tax privileges, eligibility for subsidies for community care provision, and so forth, by accepting in turn certain conditions such as giving up family ownership. Among developed countries, the changing role of the state or the recognition that private hospitals are more efficient than public ones has led to a trend towards hospital privatisation [36]. However, this case study suggests that policy makers may need to take a long-term view of the regulation of private hospitals because private hospitals are less likely than public hospitals to be in line with public targets in the long run.
187
References [1] Brian E. Foundation for medical care control of hospital utilization: CHAP—a PSRO prototype. New England Journal of Medicine 1973;288:878–82. [2] White HL, Glazier RH. Do hospitalist physicians improve the quality of inpatient care delivery? A systematic review of process, efficiency and outcome measures. BMC Medicine 2011;9:58. [3] Boaz RF. Utilization review and containment for hospital utilization: some implications of providing care in the “most appropriate setting”. Medical Care 1979;17:315–30. [4] Rotter T, Kugler J, Koch R, Gothe H, Twork S, van Oostrum JM, Steyerberg EW. A systematic review and meta-analysis of the effects of clinical pathways on length of stay, hospital costs and patient outcomes. BMC Health Services Research 2008;8:265. [5] Sloan FA, Valvona J. Why has hospital length of stay declined? An evaluation of alternative theories. Social Science and Medicine 1986;22:63–73. [6] Clarke A. Why are we trying to reduce length of stay? Evaluation of the costs and benefits of reducing time in hospital must start from the objectives that govern change. Quality in Health Care 1996;5: 172–9. [7] Lagoe RJ, Westert GP, Kendrick K, Morreale G, Mnich S. Managing hospital length of stay reduction: a multihospital approach. Health Care Management Review 2005;30:82–92. [8] Hussey PS, de Vries H, Romley J, Wang MC, Chen SS, Shekelle PG, McGlynn EA. A systematic review of health care efficiency measures. Health Services Research 2009;44:784–805. [9] Hall CJ, Peel NM, Comans TA, Gray LC, Scuffham PA. Can postacute care programmes for older people reduce overall costs in the health system? A case study using the Australian Transition Care Programme. Health & Social Care in the Community 2012;20:97– 102. [10] OECD. OECD health data 2010. Paris: OECD; 2010. [11] Yoshimura H. Personal view on trend and response toward medical expense. Kenkohoken 1983;37:28–32. [12] Ikegami N, Campbell JC. Medical care in Japan. New England Journal of Medicine 1995;333:1295–9. [13] Mitsutake N, Nishimura Y, Sakamoto S, Takahashi T. The impact of decreasing hospital average length of stay on clinical outcome. Hospital Administration 2002;39:5–12. [14] Louis DZ, Taroni F, Yuen EJ, Umesato Y, Gonnella JS. Patterns of hospital care and physician perspectives from an Italian, Japanese, and USA hospital. American Journal of Medical Quality 1996;11: 123–34. [15] Muramatsu N, Liang J. Hospital length of stay in the United States and Japan: a case study of myocardial infarction patients. International Journal of Health Services 1999;29:189–209. [16] Ogata A, Kono K, Yamamoto K, Nishikawa M, Deguchi Y, Kagehira K. Analysis of average length of hospital stay, bed occupancy rate and social structure at prefectural level. Hospital Administration 1983;20:231–41. [17] Egawa H. Average length of stay in hospitals. Journal of University of Occupational and Environmental Health 1984;6:1–14. [18] Weiner SL, Maxwell JH, Sapolsky HM, Dunn DL, Hsiao WC. Economic incentives and organizational realities: managing hospitals under DRGs. Milbank Quarterly 1987;65:463–87. [19] Fujita Y. The administrative management methods in medical policy since 1975 in Japan. Review of Population and Social Policy 1996;5:73–96. [20] Campbell JC, Ikegami N. The art of balance in health policy: maintaining Japan’s low-cost, egalitarian system. Cambridge: Cambridge University Press; 1998. [21] Tatara K, Okamoto E. Japan: health system review. Health Systems in Transition 2009;11:1–164. [22] Roemer MI. National health systems of the world. Oxford: Oxford University Press; 1991. [23] Yoshikawa A, Bhattacharya J, Vogt WB. Health economics of Japan: patients, doctors and hospitals under a universal health insurance system. Tokyo: University of Tokyo Press; 1996. [24] Freund D, Shachtman RH, Ruffin M, Quade D. Analysis of length-ofstay differences between investor-owned and voluntary hospitals. Inquiry 1985;22:33–44. [25] Younis MZ. Length of hospital stay of Medicare patients in the postprospective-payment-system era. Journal of Health Care Finance 2004;31:23–30. [26] Statistics and Information Department. Ministry of health labour and welfare. Hospital report. Health and Welfare Statistics Association: Tokyo; 1973–2010.
188
N. Kato et al. / Health Policy 115 (2014) 180–188
[27] Everitt BS, Dunn G. Applied multivariate data analysis, 2nd ed. London: Wiley; 2001. [28] Kim HJ, Fay MP, Feuer EJ, Midthune DN. Permutation tests for joinpoint regression with applications to cancer rates. Statistics in Medicine 2000;19:335–51. [29] Statistics Information Department. Ministry of Health Labour and Welfare. Estimates of national medical care expenditure 2008. Health and Welfare Statistics Association; 2010. [30] Social Insurance Institute. Medical fee point book. Tokyo: Social Insurance Institute; 1970–2008. [31] Health and Welfare Statistics Association. Annual statistical report of national health conditions. Tokyo: Health and Welfare Statistics Association; 1971–2010. [32] Idezuki Y. Long live the health care system in Japan. Bioscience Trends 2008;2:50–2.
[33] Ginsburg PB, Koretz DM. Bed availability and hospital utilization: estimates of the “Roemer effect”. Health Care Financing Review 1983;5:87–92. [34] Hamada H, Sekimoto M, Imanaka Y. Effects of the per diem prospective payment system with DRG-like grouping system (DPC/PDPS) on resource usage and healthcare quality in Japan. Health Policy 2012;107:194–201. [35] Anderson G, Ikegami N. How can Japan’s DPC inpatient hospital payment system be strengthened? Washington, DC: Lessons from the U.S. Medicare Prospective System. Center for Strategic and International Studies; 2011. [36] Schmid A, Cacace M, Götze R, Rothgang H. Explaining health care system change: problem pressure and the emergence of “hybrid” health care systems. Journal of Health Politics Policy and Law 2010;35:455–86.
