G Model
ARTICLE IN PRESS
HEAP-3320; No. of Pages 9
Health Policy xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Health Policy journal homepage: www.elsevier.com/locate/healthpol
The effect of centralization of health care services on travel time and its equality Daisuke Kobayashi, Tetsuya Otsubo, Yuichi Imanaka ∗ Department of Healthcare Economics and Quality Management, Graduate School of Medicine, Kyoto University, Yoshida Konoe-cho, Sakyo-ku, Kyoto 606-8501, Japan
a r t i c l e
i n f o
Article history: Received 31 August 2013 Received in revised form 15 October 2014 Accepted 12 November 2014 Keywords: Health care access Travel time Regional variations Gini coefficient
a b s t r a c t Objectives: To analyze the regional variations in travel time between patient residences and medical facilities for the treatment of ischemic heart disease and breast cancer, and to simulate the effects of health care services centralization on travel time and equality of access. Methods: We used medical insurance claims data for inpatients and outpatients for the two target diseases that had been filed between September 2008 and May 2009 in Kyoto Prefecture, Japan. Using a geographical information system, patient travel times were calculated based on the driving distance between patient residences and hospitals via highways and toll roads. Locations of residences and hospital locations were identified using postal codes. We then conducted a simulation analysis of centralization of health care services to designated regional core hospitals. The simulated changes in potential spatial access to care were examined. Results: Inequalities in access to care were examined using Gini coefficients, which ranged from 0.4109 to 0.4574. Simulations of health care services centralization showed reduced travel time for most patients and overall improvements in equality of access, except in breast cancer outpatients. Conclusion: Our findings may contribute to the decision-making process in policies aimed at improving the potential spatial access to health care services. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Japan’s health care system underwent fundamental changes in the early 1960s with the introduction of two major policies: universal health insurance and free access to care. Universal health insurance ensures that all citizens are able to obtain necessary health care, while free access grants patients the freedom to seek care from any
∗ Corresponding author. Tel.: +81 75 753 4454; fax: +81 75 753 4455. E-mail addresses: [email protected] (D. Kobayashi), [email protected] (T. Otsubo), [email protected] (Y. Imanaka).
provider within the country. These policies were designed to ensure that the Japanese people would be able to obtain necessary health care wherever and whenever it was needed [1]. Since the implementation of these policies, Japan has consistently performed well in health metrics such as average life expectancy [2]. However, the circumstances surrounding the provision of health care have been changing, and various challenges have appeared in recent years. For example, investigations of regional variations in the Japanese health care system have revealed an uneven distribution of physicians [3], variations in locations of medical institutions and patient treatments [4], and disparities in mortality rates [5].
http://dx.doi.org/10.1016/j.healthpol.2014.11.008 0168-8510/© 2014 Elsevier Ireland Ltd. All rights reserved.
Please cite this article in press as: Kobayashi D, et al. The effect of centralization of health care services on travel time and its equality. Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.11.008
G Model HEAP-3320; No. of Pages 9
2
ARTICLE IN PRESS D. Kobayashi et al. / Health Policy xxx (2014) xxx–xxx
Several studies conducted in various countries have investigated the time taken for a patient to travel from their residence to a medical facility (hereinafter referred to as “travel time”). These studies have addressed travel times for patients with traumatic injuries [6], emergency conditions such as acute myocardial infarction [7], and cancer [8]. In Japan, a simulation analysis showed that the centralization of emergency services through the redistribution of emergency care beds and physicians would result in increases to patient travel times [9]. However, little is known about travel time for outpatient treatments requiring frequent visits or about the possible regional variations in patient travel time in Japan. Travel time should be minimized for patients with acute diseases that require prompt treatment (such as acute myocardial infarction) to ensure better prognoses. Similarly, patients requiring non-emergency care (such as outpatient treatments for cancer patients) may also be affected by lengthy travel times, as these can increase patient burdens and affect treatment adherence and efficacy. An analysis on patient travel times for both these types of care would be beneficial for supporting health policy decision-making processes. Both acute myocardial infarction and cancer are included in the “Five Diseases, Five Fields of Medicine” strategy in Japan, which forms the core of regional health policies at the prefectural level with the aim of providing comprehensive secondary and tertiary care within each region. Japan comprises 47 prefectures; these prefectures are analogous to the states of the U.S., and are the geographic units at which regional health programs are frequently implemented and managed. Cancer treatments have also been addressed by the Japanese Ministry of Health, Labour and Welfare’s Commission on the Promotion of Cancer Treatment Standardization [10], which aims to reduce regional disparities in different facets of health care provision (such as the availability of medical technologies) and standardize specialist treatments throughout the country. In the face of a seemingly inexorable increase in health care spending, the regionalization and centralization of health care services is frequently considered as a means to contain costs while improving the quality and efficiency of care [11]. The prefectures in Japan have been undertaking preparations for a shift toward the centralization of health care services to government-designated regional core hospitals. However, a consequence of this centralization policy is a reduction in the number of hospitals available to treat specific diseases, which may in turn increase inequalities in potential spatial access to care. In Kyoto Prefecture, located on the main island of Honshu, regional disparities in mean travel time have been reported in patients seeking inpatient and outpatient treatment for ischemic heart disease, breast cancer, and other diseases at the municipality (sub-prefectural geographic unit) level [12]. The designation of regional core hospitals and the centralization of health care services may potentially exacerbate problems of pre-existing inequalities in patient access. Investigating the possible effects of centralizing health care services on patient travel time and access is therefore necessary to predetermine the appropriateness of such a policy.
The objectives of this study were to analyze the regional variations in travel time between patient residences and medical facilities for the treatment of ischemic heart disease and breast cancer, as well as to simulate the possible effects of centralization of health care services to regional core hospitals on patient travel time and equality. Ischemic heart disease and breast cancer were selected as target diseases for this analysis because they represent conditions requiring acute care and non-emergency multidisciplinary care, respectively. 2. Methods 2.1. Data Data from a total of 5,854,918 anonymized medical reimbursement claims (hereinafter referred to as “claims data”) filed between September 2008 and May 2009 were obtained from the Kyoto branch of the Japan Health Insurance Association. The target diseases were identified using the relevant disease classification codes from the Japanese Social Insurance system (902: ischemic heart disease; 206: breast cancer). We identified a total of 60,280 claims for patients who had received inpatient or outpatient treatment for either ischemic heart disease or breast cancer during the study period. From these 60,280 claims, 456 claims exhibited data discrepancies where the total number of days of medical care and medical expenses was registered as zero; these claims were excluded from analysis. The remaining claims were then collated and examined in order to remove multiple claims by the same patients, resulting in a total of 5416 individual patients. Next, patients with a residential postal code outside of Kyoto Prefecture (i.e., individuals residing outside of Kyoto Prefecture) and those whose travel times were over 300 min (5 h) were also excluded. This cutoff of 300 min was based on calculations of the time required to travel from Kyoto Prefecture to any of the adjacent prefectures. The application of these exclusion criteria removed a further 953 individuals, leaving a total of 4463 patients in the final sample for analysis. Each individual patient had a single claims record per medical facility, regardless of how many times a patient visited a particular facility. Accordingly, this study was not able to conduct a weighted analysis according to the frequency of patient visits. 2.2. Calculation of patient travel times We first developed a table of postal codes and positional coordinates using the postal code list [13] released by the Japan Post Service Co., Ltd in conjunction with the (street) block-level location reference [14] released by Ministry of Land, Infrastructure, Transport and Tourism. In the interest of protecting personal information, patient and medical facility postal codes were truncated from the original seven digits to the first six digits. The positional coordinates of the 10 possible seven-digit postal codes matching each of the six-digit codes were then obtained from the table mentioned above, and the central points of each group of seven-digit codes were used as the assumed positional
Please cite this article in press as: Kobayashi D, et al. The effect of centralization of health care services on travel time and its equality. Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.11.008
G Model HEAP-3320; No. of Pages 9
ARTICLE IN PRESS D. Kobayashi et al. / Health Policy xxx (2014) xxx–xxx
coordinates of the six-digit postal codes. In this analysis, 4471 and 900 seven-digit codes were used for the locations of patient residences and medical facilities, respectively; 658 and 132 six-digit codes were used for the locations of patient residences and medical facilities, respectively. From the assumed positional coordinates based on postal codes, we calculated the locations of patient residences and the respective medical facilities using ArcGIS 9 (Esri Japan Corporation, Japan) and Market Planner GIS ver. 2.5 (PASCO Corporation, Japan). Patient travel times were calculated from the driving distance between their (assumed) residences and the respective medical facilities via highways and toll roads. The use of GIS to calculate travel times based on driving distance is a common approach in studies that address patient access to health care [15,16]. 2.3. Travel times: regional variations and related factors Kyoto prefecture is geographically divided into 26 municipalities, including Kyoto City. We calculated and compared travel times for inpatients and outpatients for both target diseases among the municipalities. The correlation between the number of medical facilities within inhabitable areas (land areas excluding mountainous areas, forested areas, and water bodies) and patient travel time for each municipality was assessed using Pearson’s correlation coefficient. Municipalities with fewer than 10 individual cases during the study period were excluded from comparison. 2.4. Simulation analysis of health care services centralization to regional core hospitals
regional core hospitals [18] and from the Kyoto Health Care Yorozu Net [19] website. In the centralization simulation, travel times were calculated based on the distance between each patient’s residence and the nearest medical facility that fulfilled government-stipulated criteria for the treatment of ischemic heart disease or cancer. The criteria for medical facilities to treat ischemic heart disease include the presence of full-time specialist staff; accreditation as training facilities by medical societies; and the capability to perform percutaneous coronary interventions 24 h a day, coronary artery bypass graft surgery, and comprehensive rehabilitation services. For medical facilities that would serve as regional core hospitals for cancer treatment, the criteria include the presence of full-time specialist staff; a minimum annual case volume of 1200 patients; and the capability to provide radiation therapy, chemotherapy, and palliative care. In contrast, actual travel times before centralization were calculated based on the distance between each patient’s residence and the actual hospital at which they were treated, regardless of whether the hospital fulfilled the criteria described above. Inequalities in travel time were analyzed using Gini and pseudo-Gini coefficients. A Gini coefficient is an index measuring inequalities, and is often used to evaluate economic disparities. The pseudo-Gini coefficient, which uses a calculation method similar to that of the Gini coefficient, was used to measure the inequalities among different levels of travel time. The Gini coefficient ranges from 0 to 1, with 1 representing maximal inequality. The Gini coefficient (G) was calculated using the following formula: 1 |yi − yj | 2n2 n
Because the types of medical care availability differ between hospitals and clinics, we limited the subjects for the simulation analysis to patients who were treated at hospitals. In Japan, clinics are small physician-owned facilities that are generally not equipped to handle emergency cases or provide specialized care, and were therefore excluded from the analysis. Regional core hospitals were designated for each disease (“refocusing”), and we calculated travel times under the assumption that patients would seek health care from the regional core hospital closest to their residence (“centralization”). Regional core hospitals were those identified by the Kyoto prefectural government as fulfilling the minimum requirements in manpower, facilities, and equipment needed to provide care for the respective diseases. The specific selection criteria are as follows: for ischemic heart disease inpatients, we selected medical facilities identified as fulfilling the minimum standards for acute-phase treatments for acute myocardial infarction that are publicly accessible in Kyoto Prefecture [17]; for ischemic heart disease outpatients, we selected medical facilities identified as fulfilling the minimum standards in recovery-phase treatments for acute myocardial infarction that are publicly accessible in Kyoto Prefecture [17]; for breast cancer inpatients and outpatients, we selected hospitals identified as able to provide radiation therapy and chemotherapy for breast cancer from the Kyoto prefectural government’s online resource for
3
G=
n
i=1 j=1
where n is the total number of patients, y1 , y2 , . . . yn represents the different levels of patient travel time, is the average travel time, and |yi − yj | is the average of the absolute value of the differences in travel time. In addition, Gini coefficients can be calculated using Lorenz curves, which are graphical representations of the equality of a distribution using cumulative proportions of travel time and cumulative numbers of individuals for inpatients and outpatients of each of the target diseases. The pseudo-Gini coefficient was calculated by stratifying travel time into 10 different categories, with each category consisting of a 10-min interval. We then determined the differences between actual equality of access and the simulated conditions, and conducted a contribution analysis for each time interval by calculating which categories of travel time contributed to changes in inequality and to what degree. The pseudoGini coefficient (G) was calculated using the following formula: k
G = −
(yi · wi + xi · vi )
i=1
wi = 2 − (2X¯ i − x¯ i ),
vi = 2Y¯ i − y¯ i
Please cite this article in press as: Kobayashi D, et al. The effect of centralization of health care services on travel time and its equality. Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.11.008
G Model HEAP-3320; No. of Pages 9
ARTICLE IN PRESS
1134 24.0 77 0.4574 0.4433 13.5 13 0.3885 0.3512 68 21.7 21 0.4286 0.4096 15.1 15 0.4222 0.3890
Post-simulation Post-simulation
3069 19.0 104 0.4432 0.4228
Actual Actual
Actual
Inpatient Outpatient Inpatient
Table 1 shows the actual average travel times and Gini coefficients for inpatients and outpatients for each disease, as well as the results of the centralization simulation analysis. For inpatients with ischemic heart disease, the actual average travel time was 19.1 min and the Gini coefficient was 0.4109. The simulation of health care service centralization to regional core hospitals resulted in the reduction of average travel time to 12.3 min, and the Gini coefficient showed improvements to equality of access by dropping to 0.3486. In outpatients with ischemic heart disease, the actual average travel time was 19.0 min and the Gini coefficient was 0.4432; in the centralization simulation, these figures fell to 15.1 min and 0.4222, respectively, indicating improvements to both travel time and equality of access. For inpatients with breast cancer, the actual average travel time was 21.7 min and the Gini coefficient was 0.4286; in the centralization simulation, these figures fell to 13.5 min and 0.3885, respectively, indicating improvements to both travel time and equality of access. For outpatients with breast cancer, the actual average travel time was 24.0 min and the Gini coefficient was 0.4574; in the centralization simulation, these figures were 18.3 min and 0.4612. Here,
Breast cancer
3.2. Simulation analysis of health care services centralization to regional core hospitals
Ischemic heart disease
We calculated and compared the average patient travel times for each municipality in Kyoto Prefecture using data from 192 inpatients and 3069 outpatients with ischemic heart disease, and 68 inpatients and 1134 outpatients with breast cancer. For breast cancer inpatients, there was only one municipality with data from 10 or more people, which precluded inter-municipality comparisons. The minimum average travel times among the municipalities were 10.9 min for ischemic heart disease inpatients, 9.8 min for ischemic heart disease outpatients, and 9.3 min for breast cancer outpatients. The maximum average travel times among the municipalities were 20.4 min for ischemic heart disease inpatients, 43.8 min for ischemic heart disease outpatients, and 71.4 min for breast cancer outpatients. A negative correlation was observed between travel time for outpatients and the number of medical facilities within inhabitable areas per municipality.
Table 1 Actual and post-simulation results of travel time and equality of access for inpatients and outpatients (ischemic heart disease and breast cancer).
3.1. Travel times: regional variations and related factors
12.3 22 0.3486 0.2997
3. Results
192 19.1 31 0.4109 0.3870
Post-simulation
This study was approved (Registration number: E-1023) by the Ethics Committee of Kyoto University Graduate School and Faculty of Medicine.
Number of patients Average travel time (min) Number of hospitals (per six-digit postal code) Gini coefficient Pseudo-Gini coefficient
2.5. Ethical considerations
Actual
Outpatient
Post-simulation
where xi represents the ratio of patients within each 10min travel time category to all patients in the dataset, given by x1 , x2 . . . xk ; Xi is the cumulative sum of xi ; and Yi is the cumulative proportion of travel times, given by yi . Calculations were made with the maximum value of k set at 10.
18.3 18 0.4612 0.4374
D. Kobayashi et al. / Health Policy xxx (2014) xxx–xxx
4
Please cite this article in press as: Kobayashi D, et al. The effect of centralization of health care services on travel time and its equality. Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.11.008
G Model HEAP-3320; No. of Pages 9
ARTICLE IN PRESS
−0.0059
−0.1316 −0.0699 0.0197 0.0367 0.0432 0.0346 0.0266 0.0147 0.0065 0.0136 0.9 −1.9 −6.3 −10.0 −13.1 −21.1 −22.4 −38.2 −21.7 −32.0 7.4 12.5 18.3 24.4 32.5 33.0 41.0 37.4 63.9 91.5
−0.0583
6.5 14.4 24.6 34.4 45.6 54.0 63.4 75.5 85.6 123.5 −0.3128 −0.1527 0.1479 0.0635 0.0523 0.0557 0.0290 0.0293 0.0294 0.0000 1.0 −0.7 −7.0 −14.5 −21.7 −47.3 −6.5 −67.8 −78.3 − 7.4 12.9 17.3 20.7 19.2 8.1 54.8 5.0 9.5 – −0.0338
6.4 13.6 24.3 35.2 41.0 55.3 61.3 72.8 87.8 – −0.1099 −0.1244 0.0726 0.0743 0.0173 0.0140 0.0071 −0.0067 0.0051 0.0169 2.3 −0.7 −7.0 −10.5 −17.6 −23.9 −27.0 −28.8 −37.8 −66.9 8.6 13.7 17.5 23.8 27.0 30.1 37.8 46.2 46.9 47.0 −0.0873
6.3 14.4 24.5 34.3 44.6 54.0 64.9 75.1 84.7 113.9 −0.2643 −0.1828 0.1602 0.0793 0.0386 0.0403 0.0000 −0.0003 0.0103 0.0313 −0.1 −2.6 −9.0 −13.7 −13.6 −37.3 – 0.0 −75.9 −98.7 6.5 11.7 15.1 20.6 30.0 18.1 – 72.2 9.3 4.5 Difference in pseudo-Gini coefficient
6.6 14.3 24.2 34.3 43.6 55.4 – 72.2 85.2 103.2 0–9 10–19 20–29 30–39 40–49 50–59 60–69 70–79 80–89 90 and over
Difference Contribution Actual Postsimulation Difference Contribution Actual Postsimulation Difference Contribution Actual Postsimulation Actual Postsimulation
Outpatient Inpatient Inpatient
Outpatient
Breast cancer Ischemic heart disease
Acute diseases (such as ischemic heart disease) requiring immediate treatment would benefit from an even distribution of emergency medical facilities that ensures quick access for the majority of residents. In contrast, diseases such as breast cancer generally do not require emergency care, but instead require a high level of multidisciplinary specialist treatment; for such diseases, the consolidation of specialized medical staff and treatment facilities may support effective continuous treatment. Even if these specialized treatment facilities are located in areas where travel time is somewhat longer for a proportion of patients, treatment may still be most efficient if provided at these core facilities. Here, we have analyzed the regional variations in potential spatial access [20] to health care services using travel time and equality of access in Kyoto Prefecture, Japan. We then investigated changes to potential spatial access to care through a simulation analysis of health care services centralization to designated regional core hospitals. It is intuitive to think that the centralization of health care services would actually prolong travel times for a large proportion of the population, as such a policy would entail a reduction in the number of hospitals that currently provide specialized care. Our findings of a negative
Average travel time (min)
4. Discussion
Table 2 Average travel times and pseudo-Gini coefficients for various travel time categories for inpatients and outpatients (ischemic heart disease and breast cancer).
centralization appeared to improve travel times, but had little effect on inequality. Fig. 1 shows Lorenz curves of both the actual conditions and post-simulation results for inpatients and outpatients for the two diseases. The post-simulation Lorenz curves for inpatients in both diseases were observed to move closer to the line of equal distribution, indicating an improvement in equality. However, only a slight improvement in equality of access was observed in ischemic heart disease outpatients, and equality of access in breast cancer outpatients was relatively unchanged. We observed that more than half of all outpatients for both diseases would experience a reduction in travel times as a result of centralization: 1192 ischemic heart disease outpatients (53.5%) and 401 breast cancer outpatients (61.2%) had post-simulation travel times that were shorter than their actual travel times. However, 500 ischemic heart disease outpatients (22.5%) and 75 breast cancer outpatients (11.5%) showed increases in post-simulation travel times; municipalities with a high proportion of these patients were generally located in the northern region of the prefecture, which consists of more rural areas. The results of the contribution analysis of the variations in pseudo-Gini coefficients are presented in Table 2. In the 0–9 min travel time category, the simulation analysis showed no changes or slightly increased average travel times. However, for all longer travel time categories, the simulations resulted in either little change or reductions in average travel times. Although the categories under 20 min showed post-simulated trends toward improvements in equality of access, the categories within 20 and 39 min for ischemic heart disease and the categories within 20 and 59 min for breast cancer showed trends toward reductions in equality of access.
5
Difference Contribution
D. Kobayashi et al. / Health Policy xxx (2014) xxx–xxx
Please cite this article in press as: Kobayashi D, et al. The effect of centralization of health care services on travel time and its equality. Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.11.008
G Model HEAP-3320; No. of Pages 9
6
ARTICLE IN PRESS D. Kobayashi et al. / Health Policy xxx (2014) xxx–xxx
Fig. 1. Lorenz curves of actual conditions and post-simulation conditions of the centralization of health care services for inpatients and outpatients. (A) Ischemic heart disease (inpatients). (B) Breast cancer (inpatients). (C) Ischemic heart disease (outpatients). (D) Breast cancer (outpatients).
correlation between patient travel time and the available number of medical facilities would also appear to support this view. However, our simulations showed that centralization would result in an overall reduction in average travel times for inpatients and outpatients for both target diseases, despite the simulated decrease in the number of hospitals. Furthermore, the number of individuals with reduced travel times as a result of centralization vastly outnumbered the individuals with protracted travel times. The Gini coefficients showed that equality of access was either unaffected or improved due to centralization. The contribution analysis using the calculated pseudo-Gini coefficients indicated that the centralization of health care services to regional core hospitals may also improve the overall levels of equality of access, or at the very least, would not result in more inequalities. Our analysis therefore indicates that the centralization of health care services to the designated regional core hospitals in Kyoto Prefecture would generally improve travel times, with little or no adverse effects on equality of access. It should be noted that the actual travel times prior to the simulation analysis reflect the availability of free access to care in Japan, wherein patients are able to seek health care at any provider within the country. Under this system of free access, patients may base their choices not only on the proximity of a medical facility, but also take into account perceived levels of care and hospital reputations. Accordingly, some patients may choose to travel further from their residences to seek care. In contrast, the simulation analysis was based on the assumption that patients would seek care at the nearest regional core hospital, and
this may have influenced a reduction in post-simulation travel times for most patients. Individuals whose travel times increased in the simulation analysis should, however, also be examined. For example, centralization would result in a substantial proportion of patients residing in Kyotango city whose travel times would be protracted by 10 min; this city is located in the rural Tango area in the northern part of the prefecture. A lack of a regional core hospital in this area for acute myocardial infarction implies that centralization of health care would greatly affect residents, as they would have to travel further to seek treatment for ischemic heart disease. This problem is likely exacerbated by the absence of an efficient transport network, as well as the lack of a secondary care facility in the adjacent prefecture, which shares a border with the Tango area. As patients living near prefectural borders may seek health care from outside their prefecture of residence, future policies should incorporate the cooperation of neighboring prefectures to ensure that regional health programs extend beyond prefectural borders. The centralization of health care services to regional core hospitals may also have a positive impact on the quality of medical treatments. When comparing the pre- and post-simulation numbers of breast cancer patients treated at the designated regional core hospitals, all regional core hospitals showed a post-simulation increase in the number of visiting patients, with the exception of hospitals in Kyoto City. The establishment of regional core hospitals (Fig. 2) would result in the centralization of medical resources such as physicians, other specialized staff, and equipment. Together with an increase in patient volume,
Please cite this article in press as: Kobayashi D, et al. The effect of centralization of health care services on travel time and its equality. Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.11.008
G Model HEAP-3320; No. of Pages 9
ARTICLE IN PRESS D. Kobayashi et al. / Health Policy xxx (2014) xxx–xxx
7
Fig. 2. Distribution of regional core hospitals visited by breast cancer patients. Left: current distribution of hospitals. Right: simulated distribution of regional core hospitals.
these factors may support the improvement of efficiency and quality of care [21]. However, there is a risk that the centralization of health care services may overload the current capacities of the designated regional core hospitals, and there is a need to ensure that these hospitals are sufficiently staffed and equipped to manage a possible surge in patient volume. This study was unable to include a weighted analysis according to the frequency of patient visits. However, for patients with diseases requiring regular frequent visits, such as for radiation therapy or chemotherapy for breast cancer, a small increase in travel time may result in a substantial increase in the sense of burden for the patient, thereby creating a barrier to patient access to care. Although we were unable to examine this effect, policy development would benefit from analyses of various diseases and treatment methods that also take into account the visitation frequency of individual patients. In this study, we examined potential spatial access using patient travel times (driving times) based on postal codes. A previous study has estimated that the average error in distance when using postal codes to approximate patient addresses is less than 1 km [22], which supports the use of this methodology. Previous studies have noted differences in spatial access among low-population and high-population areas [23,24]. In order to take into account these differences, we utilized a combination of travel time and equality coefficients to analyze potential spatial access to care. A study by Boscoe et al. [25] revealed negligible differences between straight-line distance and travel distance or time, except in specific cases due to the presence of geographical physical barriers such as shorelines or rural areas with extremely low population densities. However, the majority of cases in this study were located in relatively urbanized areas, which would not be expected to have many such barriers.
If regional health programs are to improve the provision and availability of comprehensive treatment within a region, it would also be prudent to conduct simulations of the effects of new regional core hospitals in locations where there are currently none; the introduction of these hospitals would act as facilitators to health care in rural areas. Alternatively, in regions currently without a regional core hospital, a simulation analysis can be conducted by designating the hospital with the highest volume of patients as a nominal regional core hospital. If such a simulation shows that equality is not adversely affected by the centralization of health care services to this hospital, it is possible that the construction of a new regional core hospital would not be necessary. In this way, policy-makers may be better informed when examining possible facilitators to access to health care. 4.1. Limitations Target diseases were identified using the disease classification codes for Japanese Social Insurance. An intrinsic limitation of these codes is that acute and non-acute episodes cannot be distinguished from each other. However, inpatients would most likely comprise acute cases, whereas outpatients would generally be non-acute cases. Another limitation of this study stems from the fact that patient residences and locations of medical facilities were based on representative positional coordinates obtained from truncated postal codes. Despite determining that the conversion of these postal codes did not deviate from the original positional coordinates of each municipality, there would still be a certain degree of geographical divergence from the actual locations of residences and medical facilities. Next, this study was limited to claims data from the Kyoto branch of the Japan Health Insurance Association, and therefore did not include the data of patients belonging
Please cite this article in press as: Kobayashi D, et al. The effect of centralization of health care services on travel time and its equality. Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.11.008
G Model HEAP-3320; No. of Pages 9
ARTICLE IN PRESS D. Kobayashi et al. / Health Policy xxx (2014) xxx–xxx
8
to other health insurance agencies. The sample population in this study covers approximately 32% of the overall Kyoto Prefecture population. However, there is little reason to think that there would be substantial differences in patient visitation patterns among the different types of insurance, as patient payments are determined at the same rates regardless of insurance types, and patients are free to choose where they obtain health care. Also, our calculation of travel time did not take into account stop signs, traffic lights, or the impact of public transports; and the actual travel times may therefore be slightly different from the reported times. Finally, our study did not compare the travel times to the hospitals nearest to each patient before and after centralization. Instead, we compared precentralization travel times to the hospitals actually visited with post-centralization times to the nearest hospital. Therefore, our conclusions should be interpreted in the context of this limitation. 5. Conclusions In this study, we shed light on regional variations in patient travel times and equality of access in inpatient and outpatient care for ischemic heart diseases and breast cancer. A simulation analysis of centralization to regional core hospitals demonstrated that average patient travel times could be reduced with no detrimental effects on equality of access to care. Our findings address the potential spatial access to care for two different types of diseases, and the methodology presented here may support the fair and effective allocation of medical resources. Competing interests The authors declare no conflicts of interest in this study. Funding sources This study was funded by a Grant-in-Aid for Scientific Research from the Ministry of Health, Labour and Welfare (H22-SEISAKU-IPPAN-028) and the Japan Society for the Promotion of Science ((A)25253033). Acknowledgments We would like to express our gratitude toward the Kyoto branch of the Japan Health Insurance Association, the Kyoto Prefectural Government, and the cooperation of many other individuals who supported this study. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.healthpol.2014.11.008. References [1] Ministry of Health, Labour and Welfare, Japan. Service guide. Health Policy Bureau; 2013. http://www.mhlw.go.jp/english/org/ pamphlet/dl/pamphlet-about mhlw.pdf [accessed 06.03.14].
[2] Ministry of Health, Labour and Welfare, Japan. Abridged life tables for Japan 2012, http://www.mhlw.go.jp/english/database/ db-hw/lifetb12/index.html [accessed 06.03.14]. [3] Ministry of Health, Labour and Welfare. Survey of actual number of required doctors; 2010. http://www.mhlw.go.jp/bunya/iryou/ other/iryou01.html [accessed 06.03.14] [in Japanese]. [4] Kyoto Prefecture Medical System Research Council. Kyoto prefecture medical system research council Report. Kyoto; 2010 [in Japanese]. [5] Hattori M, Fujita M, Nakamura K, Ioka A. Estimation of the effects of centralization of cancer treatment on mortality reduction by in Fukui Prefecture. Japanese Journal of Public Health 2010;57(4):263–70 [in Japanese with English abstract]. [6] Branas CC, MacKenzie EJ, Williams JC, Schwab CW, Teter HM, Flanigan MC, et al. Access to trauma centers in the United States. The Journal of the American Medical Association 2005;293(21): 2626–33. [7] Nallamothu BK, Bates ER, Wang Y, Bradley EH, Krumholz HM. Driving times and distances to hospitals with percutaneous coronary intervention in the United States: implications for prehospital triage of patients with ST-elevation myocardial infarction. Circulation 2006;113(9):1189–95. [8] Baird G, Flynn R, Baxter G, Donnelly M, Lawrence J. Travel time and cancer care: an example of the inverse care law? Rural Remote Health 2008;8(4):1003. [9] Yamada Y, Kawahara K, Kawaguchi Y. Study of an emergency medical service system vis-à-vis ambulance transport access time—examination by simulating consolidating emergency medical services in Ibaraki prefecture. Journal of Japan Society for Healthcare Administration 2009;46(3):167–76 [in Japanese with English abstract]. [10] Ministry of Health, Labour and Welfare. Investigative report of the commission into the promotion of cancer treatment standardization; 2005. http://www.mhlw.go.jp/shingi/2005/04/s0419-6a.html [accessed 06.03.14] [in Japanese]. [11] Church J, Barker P. Regionalization of health services in Canada: a critical perspective. International Journal of Health Services 1998;28(3):467–86. [12] Kobayashi D, Otsubo T, Imanaka Y. Regional variations in healthcare systems through the analysis of patient transit times for various diseases. In: Proceedings of the 31st joint conference on medical informatics, vol. 31. 2011. p. 470–3 [in Japanese with English abstract]. [13] Japan Post Service Co., Ltd. Postal code data download, http://www. post.japanpost.jp/zipcode/dl/kogaki.html [accessed 06.03.14] [in Japanese]. [14] Ministry of Land, Infrastructure, Transport and Tourism. Block-level location reference download service, http://nlftp.mlit.go.jp/isj/ index.html [accessed 06.03.14] [in Japanese]. [15] Matsumoto M, Kashima S, Ogawa T, Takeuchi K. Do rural and remote areas really have limited accessibility to health care? Geographic analysis of dialysis patients in Hiroshima, Japan. Rural Remote Health 2013;13(3):2507. [16] Tracy R, Nam I, Gruca TS. The influence of visiting consultant clinics on measures of access to cancer care: evidence from the state of Iowa. Health Services Research 2013;48(October (5)):1719–29. [17] Kyoto Prefectural Government. Kyoto prefecture health care plan: medical facilities that provide medical functions for acute myocardial infarction, http://www.pref.kyoto.jp/drkyoto/condition/ file/condition 10 5.pdf [accessed 06.03.14] [in Japanese]. [18] Kyoto Prefectural Government. Kyoto prefecture information related to cancer treatments. Treatment information of core hospitals, collaborative hospitals and progressive hospitals within the prefecture, http://www.pref.kyoto.jp/gan/shinryou.html [accessed 06.03.14] [in Japanese]. [19] Kyoto Prefectural Government. Kyoto health care Yorozu net, http://www.mfis.pref.kyoto.lg.jp/ [accessed 06.03.14] [in Japanese]. [20] Khan AA. An integrated approach to measuring potential spatial access to health care services. Socio-Economic Planning Sciences 1992;26(4):275–87. [21] Gooiker GA, van Gijn W, Post PN, van de Velde CJ, Tollenaar RA, Wouters MW. A systematic review and meta-analysis of the volume-outcome relationship in the surgical treatment of breast cancer. Are breast cancer patients better of with a high volume provider? European Journal of Surgical Oncology 2010;36(Suppl. 1): S27–35. [22] Kunisawa S, Kobayashi D, Lee J, Otsubo T, Ikai H, Yokota C, et al. Factors associated with the administration of tissue plasminogen activator for acute ischemic stroke. Journal of Stroke and Cerebrovascular Diseases 2014;23(4):724–31.
Please cite this article in press as: Kobayashi D, et al. The effect of centralization of health care services on travel time and its equality. Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.11.008
G Model HEAP-3320; No. of Pages 9
ARTICLE IN PRESS D. Kobayashi et al. / Health Policy xxx (2014) xxx–xxx
[23] Schuurman N, Fiedler RS, Grzybowski SC, Grund D. Defining rational hospital catchments for non-urban areas based on travel-time. International Journal of Health Geographics 2006;5:43. [24] Hanlon N, Halseth G. The greying of resource communities in northern British Columbia: implications for health care delivery
9
in already-underserviced communities. The Canadian Geographer 2005;49(1):1–24. [25] Boscoe FP, Henry KA, Zdeb MS. A nationwide comparison of driving distance versus straight-line distance to hospitals. Professional Geographer 2012;64(2):188–96.
Please cite this article in press as: Kobayashi D, et al. The effect of centralization of health care services on travel time and its equality. Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.11.008
ARTICLE IN PRESS
HEAP-3320; No. of Pages 9
Health Policy xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Health Policy journal homepage: www.elsevier.com/locate/healthpol
The effect of centralization of health care services on travel time and its equality Daisuke Kobayashi, Tetsuya Otsubo, Yuichi Imanaka ∗ Department of Healthcare Economics and Quality Management, Graduate School of Medicine, Kyoto University, Yoshida Konoe-cho, Sakyo-ku, Kyoto 606-8501, Japan
a r t i c l e
i n f o
Article history: Received 31 August 2013 Received in revised form 15 October 2014 Accepted 12 November 2014 Keywords: Health care access Travel time Regional variations Gini coefficient
a b s t r a c t Objectives: To analyze the regional variations in travel time between patient residences and medical facilities for the treatment of ischemic heart disease and breast cancer, and to simulate the effects of health care services centralization on travel time and equality of access. Methods: We used medical insurance claims data for inpatients and outpatients for the two target diseases that had been filed between September 2008 and May 2009 in Kyoto Prefecture, Japan. Using a geographical information system, patient travel times were calculated based on the driving distance between patient residences and hospitals via highways and toll roads. Locations of residences and hospital locations were identified using postal codes. We then conducted a simulation analysis of centralization of health care services to designated regional core hospitals. The simulated changes in potential spatial access to care were examined. Results: Inequalities in access to care were examined using Gini coefficients, which ranged from 0.4109 to 0.4574. Simulations of health care services centralization showed reduced travel time for most patients and overall improvements in equality of access, except in breast cancer outpatients. Conclusion: Our findings may contribute to the decision-making process in policies aimed at improving the potential spatial access to health care services. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Japan’s health care system underwent fundamental changes in the early 1960s with the introduction of two major policies: universal health insurance and free access to care. Universal health insurance ensures that all citizens are able to obtain necessary health care, while free access grants patients the freedom to seek care from any
∗ Corresponding author. Tel.: +81 75 753 4454; fax: +81 75 753 4455. E-mail addresses: [email protected] (D. Kobayashi), [email protected] (T. Otsubo), [email protected] (Y. Imanaka).
provider within the country. These policies were designed to ensure that the Japanese people would be able to obtain necessary health care wherever and whenever it was needed [1]. Since the implementation of these policies, Japan has consistently performed well in health metrics such as average life expectancy [2]. However, the circumstances surrounding the provision of health care have been changing, and various challenges have appeared in recent years. For example, investigations of regional variations in the Japanese health care system have revealed an uneven distribution of physicians [3], variations in locations of medical institutions and patient treatments [4], and disparities in mortality rates [5].
http://dx.doi.org/10.1016/j.healthpol.2014.11.008 0168-8510/© 2014 Elsevier Ireland Ltd. All rights reserved.
Please cite this article in press as: Kobayashi D, et al. The effect of centralization of health care services on travel time and its equality. Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.11.008
G Model HEAP-3320; No. of Pages 9
2
ARTICLE IN PRESS D. Kobayashi et al. / Health Policy xxx (2014) xxx–xxx
Several studies conducted in various countries have investigated the time taken for a patient to travel from their residence to a medical facility (hereinafter referred to as “travel time”). These studies have addressed travel times for patients with traumatic injuries [6], emergency conditions such as acute myocardial infarction [7], and cancer [8]. In Japan, a simulation analysis showed that the centralization of emergency services through the redistribution of emergency care beds and physicians would result in increases to patient travel times [9]. However, little is known about travel time for outpatient treatments requiring frequent visits or about the possible regional variations in patient travel time in Japan. Travel time should be minimized for patients with acute diseases that require prompt treatment (such as acute myocardial infarction) to ensure better prognoses. Similarly, patients requiring non-emergency care (such as outpatient treatments for cancer patients) may also be affected by lengthy travel times, as these can increase patient burdens and affect treatment adherence and efficacy. An analysis on patient travel times for both these types of care would be beneficial for supporting health policy decision-making processes. Both acute myocardial infarction and cancer are included in the “Five Diseases, Five Fields of Medicine” strategy in Japan, which forms the core of regional health policies at the prefectural level with the aim of providing comprehensive secondary and tertiary care within each region. Japan comprises 47 prefectures; these prefectures are analogous to the states of the U.S., and are the geographic units at which regional health programs are frequently implemented and managed. Cancer treatments have also been addressed by the Japanese Ministry of Health, Labour and Welfare’s Commission on the Promotion of Cancer Treatment Standardization [10], which aims to reduce regional disparities in different facets of health care provision (such as the availability of medical technologies) and standardize specialist treatments throughout the country. In the face of a seemingly inexorable increase in health care spending, the regionalization and centralization of health care services is frequently considered as a means to contain costs while improving the quality and efficiency of care [11]. The prefectures in Japan have been undertaking preparations for a shift toward the centralization of health care services to government-designated regional core hospitals. However, a consequence of this centralization policy is a reduction in the number of hospitals available to treat specific diseases, which may in turn increase inequalities in potential spatial access to care. In Kyoto Prefecture, located on the main island of Honshu, regional disparities in mean travel time have been reported in patients seeking inpatient and outpatient treatment for ischemic heart disease, breast cancer, and other diseases at the municipality (sub-prefectural geographic unit) level [12]. The designation of regional core hospitals and the centralization of health care services may potentially exacerbate problems of pre-existing inequalities in patient access. Investigating the possible effects of centralizing health care services on patient travel time and access is therefore necessary to predetermine the appropriateness of such a policy.
The objectives of this study were to analyze the regional variations in travel time between patient residences and medical facilities for the treatment of ischemic heart disease and breast cancer, as well as to simulate the possible effects of centralization of health care services to regional core hospitals on patient travel time and equality. Ischemic heart disease and breast cancer were selected as target diseases for this analysis because they represent conditions requiring acute care and non-emergency multidisciplinary care, respectively. 2. Methods 2.1. Data Data from a total of 5,854,918 anonymized medical reimbursement claims (hereinafter referred to as “claims data”) filed between September 2008 and May 2009 were obtained from the Kyoto branch of the Japan Health Insurance Association. The target diseases were identified using the relevant disease classification codes from the Japanese Social Insurance system (902: ischemic heart disease; 206: breast cancer). We identified a total of 60,280 claims for patients who had received inpatient or outpatient treatment for either ischemic heart disease or breast cancer during the study period. From these 60,280 claims, 456 claims exhibited data discrepancies where the total number of days of medical care and medical expenses was registered as zero; these claims were excluded from analysis. The remaining claims were then collated and examined in order to remove multiple claims by the same patients, resulting in a total of 5416 individual patients. Next, patients with a residential postal code outside of Kyoto Prefecture (i.e., individuals residing outside of Kyoto Prefecture) and those whose travel times were over 300 min (5 h) were also excluded. This cutoff of 300 min was based on calculations of the time required to travel from Kyoto Prefecture to any of the adjacent prefectures. The application of these exclusion criteria removed a further 953 individuals, leaving a total of 4463 patients in the final sample for analysis. Each individual patient had a single claims record per medical facility, regardless of how many times a patient visited a particular facility. Accordingly, this study was not able to conduct a weighted analysis according to the frequency of patient visits. 2.2. Calculation of patient travel times We first developed a table of postal codes and positional coordinates using the postal code list [13] released by the Japan Post Service Co., Ltd in conjunction with the (street) block-level location reference [14] released by Ministry of Land, Infrastructure, Transport and Tourism. In the interest of protecting personal information, patient and medical facility postal codes were truncated from the original seven digits to the first six digits. The positional coordinates of the 10 possible seven-digit postal codes matching each of the six-digit codes were then obtained from the table mentioned above, and the central points of each group of seven-digit codes were used as the assumed positional
Please cite this article in press as: Kobayashi D, et al. The effect of centralization of health care services on travel time and its equality. Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.11.008
G Model HEAP-3320; No. of Pages 9
ARTICLE IN PRESS D. Kobayashi et al. / Health Policy xxx (2014) xxx–xxx
coordinates of the six-digit postal codes. In this analysis, 4471 and 900 seven-digit codes were used for the locations of patient residences and medical facilities, respectively; 658 and 132 six-digit codes were used for the locations of patient residences and medical facilities, respectively. From the assumed positional coordinates based on postal codes, we calculated the locations of patient residences and the respective medical facilities using ArcGIS 9 (Esri Japan Corporation, Japan) and Market Planner GIS ver. 2.5 (PASCO Corporation, Japan). Patient travel times were calculated from the driving distance between their (assumed) residences and the respective medical facilities via highways and toll roads. The use of GIS to calculate travel times based on driving distance is a common approach in studies that address patient access to health care [15,16]. 2.3. Travel times: regional variations and related factors Kyoto prefecture is geographically divided into 26 municipalities, including Kyoto City. We calculated and compared travel times for inpatients and outpatients for both target diseases among the municipalities. The correlation between the number of medical facilities within inhabitable areas (land areas excluding mountainous areas, forested areas, and water bodies) and patient travel time for each municipality was assessed using Pearson’s correlation coefficient. Municipalities with fewer than 10 individual cases during the study period were excluded from comparison. 2.4. Simulation analysis of health care services centralization to regional core hospitals
regional core hospitals [18] and from the Kyoto Health Care Yorozu Net [19] website. In the centralization simulation, travel times were calculated based on the distance between each patient’s residence and the nearest medical facility that fulfilled government-stipulated criteria for the treatment of ischemic heart disease or cancer. The criteria for medical facilities to treat ischemic heart disease include the presence of full-time specialist staff; accreditation as training facilities by medical societies; and the capability to perform percutaneous coronary interventions 24 h a day, coronary artery bypass graft surgery, and comprehensive rehabilitation services. For medical facilities that would serve as regional core hospitals for cancer treatment, the criteria include the presence of full-time specialist staff; a minimum annual case volume of 1200 patients; and the capability to provide radiation therapy, chemotherapy, and palliative care. In contrast, actual travel times before centralization were calculated based on the distance between each patient’s residence and the actual hospital at which they were treated, regardless of whether the hospital fulfilled the criteria described above. Inequalities in travel time were analyzed using Gini and pseudo-Gini coefficients. A Gini coefficient is an index measuring inequalities, and is often used to evaluate economic disparities. The pseudo-Gini coefficient, which uses a calculation method similar to that of the Gini coefficient, was used to measure the inequalities among different levels of travel time. The Gini coefficient ranges from 0 to 1, with 1 representing maximal inequality. The Gini coefficient (G) was calculated using the following formula: 1 |yi − yj | 2n2 n
Because the types of medical care availability differ between hospitals and clinics, we limited the subjects for the simulation analysis to patients who were treated at hospitals. In Japan, clinics are small physician-owned facilities that are generally not equipped to handle emergency cases or provide specialized care, and were therefore excluded from the analysis. Regional core hospitals were designated for each disease (“refocusing”), and we calculated travel times under the assumption that patients would seek health care from the regional core hospital closest to their residence (“centralization”). Regional core hospitals were those identified by the Kyoto prefectural government as fulfilling the minimum requirements in manpower, facilities, and equipment needed to provide care for the respective diseases. The specific selection criteria are as follows: for ischemic heart disease inpatients, we selected medical facilities identified as fulfilling the minimum standards for acute-phase treatments for acute myocardial infarction that are publicly accessible in Kyoto Prefecture [17]; for ischemic heart disease outpatients, we selected medical facilities identified as fulfilling the minimum standards in recovery-phase treatments for acute myocardial infarction that are publicly accessible in Kyoto Prefecture [17]; for breast cancer inpatients and outpatients, we selected hospitals identified as able to provide radiation therapy and chemotherapy for breast cancer from the Kyoto prefectural government’s online resource for
3
G=
n
i=1 j=1
where n is the total number of patients, y1 , y2 , . . . yn represents the different levels of patient travel time, is the average travel time, and |yi − yj | is the average of the absolute value of the differences in travel time. In addition, Gini coefficients can be calculated using Lorenz curves, which are graphical representations of the equality of a distribution using cumulative proportions of travel time and cumulative numbers of individuals for inpatients and outpatients of each of the target diseases. The pseudo-Gini coefficient was calculated by stratifying travel time into 10 different categories, with each category consisting of a 10-min interval. We then determined the differences between actual equality of access and the simulated conditions, and conducted a contribution analysis for each time interval by calculating which categories of travel time contributed to changes in inequality and to what degree. The pseudoGini coefficient (G) was calculated using the following formula: k
G = −
(yi · wi + xi · vi )
i=1
wi = 2 − (2X¯ i − x¯ i ),
vi = 2Y¯ i − y¯ i
Please cite this article in press as: Kobayashi D, et al. The effect of centralization of health care services on travel time and its equality. Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.11.008
G Model HEAP-3320; No. of Pages 9
ARTICLE IN PRESS
1134 24.0 77 0.4574 0.4433 13.5 13 0.3885 0.3512 68 21.7 21 0.4286 0.4096 15.1 15 0.4222 0.3890
Post-simulation Post-simulation
3069 19.0 104 0.4432 0.4228
Actual Actual
Actual
Inpatient Outpatient Inpatient
Table 1 shows the actual average travel times and Gini coefficients for inpatients and outpatients for each disease, as well as the results of the centralization simulation analysis. For inpatients with ischemic heart disease, the actual average travel time was 19.1 min and the Gini coefficient was 0.4109. The simulation of health care service centralization to regional core hospitals resulted in the reduction of average travel time to 12.3 min, and the Gini coefficient showed improvements to equality of access by dropping to 0.3486. In outpatients with ischemic heart disease, the actual average travel time was 19.0 min and the Gini coefficient was 0.4432; in the centralization simulation, these figures fell to 15.1 min and 0.4222, respectively, indicating improvements to both travel time and equality of access. For inpatients with breast cancer, the actual average travel time was 21.7 min and the Gini coefficient was 0.4286; in the centralization simulation, these figures fell to 13.5 min and 0.3885, respectively, indicating improvements to both travel time and equality of access. For outpatients with breast cancer, the actual average travel time was 24.0 min and the Gini coefficient was 0.4574; in the centralization simulation, these figures were 18.3 min and 0.4612. Here,
Breast cancer
3.2. Simulation analysis of health care services centralization to regional core hospitals
Ischemic heart disease
We calculated and compared the average patient travel times for each municipality in Kyoto Prefecture using data from 192 inpatients and 3069 outpatients with ischemic heart disease, and 68 inpatients and 1134 outpatients with breast cancer. For breast cancer inpatients, there was only one municipality with data from 10 or more people, which precluded inter-municipality comparisons. The minimum average travel times among the municipalities were 10.9 min for ischemic heart disease inpatients, 9.8 min for ischemic heart disease outpatients, and 9.3 min for breast cancer outpatients. The maximum average travel times among the municipalities were 20.4 min for ischemic heart disease inpatients, 43.8 min for ischemic heart disease outpatients, and 71.4 min for breast cancer outpatients. A negative correlation was observed between travel time for outpatients and the number of medical facilities within inhabitable areas per municipality.
Table 1 Actual and post-simulation results of travel time and equality of access for inpatients and outpatients (ischemic heart disease and breast cancer).
3.1. Travel times: regional variations and related factors
12.3 22 0.3486 0.2997
3. Results
192 19.1 31 0.4109 0.3870
Post-simulation
This study was approved (Registration number: E-1023) by the Ethics Committee of Kyoto University Graduate School and Faculty of Medicine.
Number of patients Average travel time (min) Number of hospitals (per six-digit postal code) Gini coefficient Pseudo-Gini coefficient
2.5. Ethical considerations
Actual
Outpatient
Post-simulation
where xi represents the ratio of patients within each 10min travel time category to all patients in the dataset, given by x1 , x2 . . . xk ; Xi is the cumulative sum of xi ; and Yi is the cumulative proportion of travel times, given by yi . Calculations were made with the maximum value of k set at 10.
18.3 18 0.4612 0.4374
D. Kobayashi et al. / Health Policy xxx (2014) xxx–xxx
4
Please cite this article in press as: Kobayashi D, et al. The effect of centralization of health care services on travel time and its equality. Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.11.008
G Model HEAP-3320; No. of Pages 9
ARTICLE IN PRESS
−0.0059
−0.1316 −0.0699 0.0197 0.0367 0.0432 0.0346 0.0266 0.0147 0.0065 0.0136 0.9 −1.9 −6.3 −10.0 −13.1 −21.1 −22.4 −38.2 −21.7 −32.0 7.4 12.5 18.3 24.4 32.5 33.0 41.0 37.4 63.9 91.5
−0.0583
6.5 14.4 24.6 34.4 45.6 54.0 63.4 75.5 85.6 123.5 −0.3128 −0.1527 0.1479 0.0635 0.0523 0.0557 0.0290 0.0293 0.0294 0.0000 1.0 −0.7 −7.0 −14.5 −21.7 −47.3 −6.5 −67.8 −78.3 − 7.4 12.9 17.3 20.7 19.2 8.1 54.8 5.0 9.5 – −0.0338
6.4 13.6 24.3 35.2 41.0 55.3 61.3 72.8 87.8 – −0.1099 −0.1244 0.0726 0.0743 0.0173 0.0140 0.0071 −0.0067 0.0051 0.0169 2.3 −0.7 −7.0 −10.5 −17.6 −23.9 −27.0 −28.8 −37.8 −66.9 8.6 13.7 17.5 23.8 27.0 30.1 37.8 46.2 46.9 47.0 −0.0873
6.3 14.4 24.5 34.3 44.6 54.0 64.9 75.1 84.7 113.9 −0.2643 −0.1828 0.1602 0.0793 0.0386 0.0403 0.0000 −0.0003 0.0103 0.0313 −0.1 −2.6 −9.0 −13.7 −13.6 −37.3 – 0.0 −75.9 −98.7 6.5 11.7 15.1 20.6 30.0 18.1 – 72.2 9.3 4.5 Difference in pseudo-Gini coefficient
6.6 14.3 24.2 34.3 43.6 55.4 – 72.2 85.2 103.2 0–9 10–19 20–29 30–39 40–49 50–59 60–69 70–79 80–89 90 and over
Difference Contribution Actual Postsimulation Difference Contribution Actual Postsimulation Difference Contribution Actual Postsimulation Actual Postsimulation
Outpatient Inpatient Inpatient
Outpatient
Breast cancer Ischemic heart disease
Acute diseases (such as ischemic heart disease) requiring immediate treatment would benefit from an even distribution of emergency medical facilities that ensures quick access for the majority of residents. In contrast, diseases such as breast cancer generally do not require emergency care, but instead require a high level of multidisciplinary specialist treatment; for such diseases, the consolidation of specialized medical staff and treatment facilities may support effective continuous treatment. Even if these specialized treatment facilities are located in areas where travel time is somewhat longer for a proportion of patients, treatment may still be most efficient if provided at these core facilities. Here, we have analyzed the regional variations in potential spatial access [20] to health care services using travel time and equality of access in Kyoto Prefecture, Japan. We then investigated changes to potential spatial access to care through a simulation analysis of health care services centralization to designated regional core hospitals. It is intuitive to think that the centralization of health care services would actually prolong travel times for a large proportion of the population, as such a policy would entail a reduction in the number of hospitals that currently provide specialized care. Our findings of a negative
Average travel time (min)
4. Discussion
Table 2 Average travel times and pseudo-Gini coefficients for various travel time categories for inpatients and outpatients (ischemic heart disease and breast cancer).
centralization appeared to improve travel times, but had little effect on inequality. Fig. 1 shows Lorenz curves of both the actual conditions and post-simulation results for inpatients and outpatients for the two diseases. The post-simulation Lorenz curves for inpatients in both diseases were observed to move closer to the line of equal distribution, indicating an improvement in equality. However, only a slight improvement in equality of access was observed in ischemic heart disease outpatients, and equality of access in breast cancer outpatients was relatively unchanged. We observed that more than half of all outpatients for both diseases would experience a reduction in travel times as a result of centralization: 1192 ischemic heart disease outpatients (53.5%) and 401 breast cancer outpatients (61.2%) had post-simulation travel times that were shorter than their actual travel times. However, 500 ischemic heart disease outpatients (22.5%) and 75 breast cancer outpatients (11.5%) showed increases in post-simulation travel times; municipalities with a high proportion of these patients were generally located in the northern region of the prefecture, which consists of more rural areas. The results of the contribution analysis of the variations in pseudo-Gini coefficients are presented in Table 2. In the 0–9 min travel time category, the simulation analysis showed no changes or slightly increased average travel times. However, for all longer travel time categories, the simulations resulted in either little change or reductions in average travel times. Although the categories under 20 min showed post-simulated trends toward improvements in equality of access, the categories within 20 and 39 min for ischemic heart disease and the categories within 20 and 59 min for breast cancer showed trends toward reductions in equality of access.
5
Difference Contribution
D. Kobayashi et al. / Health Policy xxx (2014) xxx–xxx
Please cite this article in press as: Kobayashi D, et al. The effect of centralization of health care services on travel time and its equality. Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.11.008
G Model HEAP-3320; No. of Pages 9
6
ARTICLE IN PRESS D. Kobayashi et al. / Health Policy xxx (2014) xxx–xxx
Fig. 1. Lorenz curves of actual conditions and post-simulation conditions of the centralization of health care services for inpatients and outpatients. (A) Ischemic heart disease (inpatients). (B) Breast cancer (inpatients). (C) Ischemic heart disease (outpatients). (D) Breast cancer (outpatients).
correlation between patient travel time and the available number of medical facilities would also appear to support this view. However, our simulations showed that centralization would result in an overall reduction in average travel times for inpatients and outpatients for both target diseases, despite the simulated decrease in the number of hospitals. Furthermore, the number of individuals with reduced travel times as a result of centralization vastly outnumbered the individuals with protracted travel times. The Gini coefficients showed that equality of access was either unaffected or improved due to centralization. The contribution analysis using the calculated pseudo-Gini coefficients indicated that the centralization of health care services to regional core hospitals may also improve the overall levels of equality of access, or at the very least, would not result in more inequalities. Our analysis therefore indicates that the centralization of health care services to the designated regional core hospitals in Kyoto Prefecture would generally improve travel times, with little or no adverse effects on equality of access. It should be noted that the actual travel times prior to the simulation analysis reflect the availability of free access to care in Japan, wherein patients are able to seek health care at any provider within the country. Under this system of free access, patients may base their choices not only on the proximity of a medical facility, but also take into account perceived levels of care and hospital reputations. Accordingly, some patients may choose to travel further from their residences to seek care. In contrast, the simulation analysis was based on the assumption that patients would seek care at the nearest regional core hospital, and
this may have influenced a reduction in post-simulation travel times for most patients. Individuals whose travel times increased in the simulation analysis should, however, also be examined. For example, centralization would result in a substantial proportion of patients residing in Kyotango city whose travel times would be protracted by 10 min; this city is located in the rural Tango area in the northern part of the prefecture. A lack of a regional core hospital in this area for acute myocardial infarction implies that centralization of health care would greatly affect residents, as they would have to travel further to seek treatment for ischemic heart disease. This problem is likely exacerbated by the absence of an efficient transport network, as well as the lack of a secondary care facility in the adjacent prefecture, which shares a border with the Tango area. As patients living near prefectural borders may seek health care from outside their prefecture of residence, future policies should incorporate the cooperation of neighboring prefectures to ensure that regional health programs extend beyond prefectural borders. The centralization of health care services to regional core hospitals may also have a positive impact on the quality of medical treatments. When comparing the pre- and post-simulation numbers of breast cancer patients treated at the designated regional core hospitals, all regional core hospitals showed a post-simulation increase in the number of visiting patients, with the exception of hospitals in Kyoto City. The establishment of regional core hospitals (Fig. 2) would result in the centralization of medical resources such as physicians, other specialized staff, and equipment. Together with an increase in patient volume,
Please cite this article in press as: Kobayashi D, et al. The effect of centralization of health care services on travel time and its equality. Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.11.008
G Model HEAP-3320; No. of Pages 9
ARTICLE IN PRESS D. Kobayashi et al. / Health Policy xxx (2014) xxx–xxx
7
Fig. 2. Distribution of regional core hospitals visited by breast cancer patients. Left: current distribution of hospitals. Right: simulated distribution of regional core hospitals.
these factors may support the improvement of efficiency and quality of care [21]. However, there is a risk that the centralization of health care services may overload the current capacities of the designated regional core hospitals, and there is a need to ensure that these hospitals are sufficiently staffed and equipped to manage a possible surge in patient volume. This study was unable to include a weighted analysis according to the frequency of patient visits. However, for patients with diseases requiring regular frequent visits, such as for radiation therapy or chemotherapy for breast cancer, a small increase in travel time may result in a substantial increase in the sense of burden for the patient, thereby creating a barrier to patient access to care. Although we were unable to examine this effect, policy development would benefit from analyses of various diseases and treatment methods that also take into account the visitation frequency of individual patients. In this study, we examined potential spatial access using patient travel times (driving times) based on postal codes. A previous study has estimated that the average error in distance when using postal codes to approximate patient addresses is less than 1 km [22], which supports the use of this methodology. Previous studies have noted differences in spatial access among low-population and high-population areas [23,24]. In order to take into account these differences, we utilized a combination of travel time and equality coefficients to analyze potential spatial access to care. A study by Boscoe et al. [25] revealed negligible differences between straight-line distance and travel distance or time, except in specific cases due to the presence of geographical physical barriers such as shorelines or rural areas with extremely low population densities. However, the majority of cases in this study were located in relatively urbanized areas, which would not be expected to have many such barriers.
If regional health programs are to improve the provision and availability of comprehensive treatment within a region, it would also be prudent to conduct simulations of the effects of new regional core hospitals in locations where there are currently none; the introduction of these hospitals would act as facilitators to health care in rural areas. Alternatively, in regions currently without a regional core hospital, a simulation analysis can be conducted by designating the hospital with the highest volume of patients as a nominal regional core hospital. If such a simulation shows that equality is not adversely affected by the centralization of health care services to this hospital, it is possible that the construction of a new regional core hospital would not be necessary. In this way, policy-makers may be better informed when examining possible facilitators to access to health care. 4.1. Limitations Target diseases were identified using the disease classification codes for Japanese Social Insurance. An intrinsic limitation of these codes is that acute and non-acute episodes cannot be distinguished from each other. However, inpatients would most likely comprise acute cases, whereas outpatients would generally be non-acute cases. Another limitation of this study stems from the fact that patient residences and locations of medical facilities were based on representative positional coordinates obtained from truncated postal codes. Despite determining that the conversion of these postal codes did not deviate from the original positional coordinates of each municipality, there would still be a certain degree of geographical divergence from the actual locations of residences and medical facilities. Next, this study was limited to claims data from the Kyoto branch of the Japan Health Insurance Association, and therefore did not include the data of patients belonging
Please cite this article in press as: Kobayashi D, et al. The effect of centralization of health care services on travel time and its equality. Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.11.008
G Model HEAP-3320; No. of Pages 9
ARTICLE IN PRESS D. Kobayashi et al. / Health Policy xxx (2014) xxx–xxx
8
to other health insurance agencies. The sample population in this study covers approximately 32% of the overall Kyoto Prefecture population. However, there is little reason to think that there would be substantial differences in patient visitation patterns among the different types of insurance, as patient payments are determined at the same rates regardless of insurance types, and patients are free to choose where they obtain health care. Also, our calculation of travel time did not take into account stop signs, traffic lights, or the impact of public transports; and the actual travel times may therefore be slightly different from the reported times. Finally, our study did not compare the travel times to the hospitals nearest to each patient before and after centralization. Instead, we compared precentralization travel times to the hospitals actually visited with post-centralization times to the nearest hospital. Therefore, our conclusions should be interpreted in the context of this limitation. 5. Conclusions In this study, we shed light on regional variations in patient travel times and equality of access in inpatient and outpatient care for ischemic heart diseases and breast cancer. A simulation analysis of centralization to regional core hospitals demonstrated that average patient travel times could be reduced with no detrimental effects on equality of access to care. Our findings address the potential spatial access to care for two different types of diseases, and the methodology presented here may support the fair and effective allocation of medical resources. Competing interests The authors declare no conflicts of interest in this study. Funding sources This study was funded by a Grant-in-Aid for Scientific Research from the Ministry of Health, Labour and Welfare (H22-SEISAKU-IPPAN-028) and the Japan Society for the Promotion of Science ((A)25253033). Acknowledgments We would like to express our gratitude toward the Kyoto branch of the Japan Health Insurance Association, the Kyoto Prefectural Government, and the cooperation of many other individuals who supported this study. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.healthpol.2014.11.008. References [1] Ministry of Health, Labour and Welfare, Japan. Service guide. Health Policy Bureau; 2013. http://www.mhlw.go.jp/english/org/ pamphlet/dl/pamphlet-about mhlw.pdf [accessed 06.03.14].
[2] Ministry of Health, Labour and Welfare, Japan. Abridged life tables for Japan 2012, http://www.mhlw.go.jp/english/database/ db-hw/lifetb12/index.html [accessed 06.03.14]. [3] Ministry of Health, Labour and Welfare. Survey of actual number of required doctors; 2010. http://www.mhlw.go.jp/bunya/iryou/ other/iryou01.html [accessed 06.03.14] [in Japanese]. [4] Kyoto Prefecture Medical System Research Council. Kyoto prefecture medical system research council Report. Kyoto; 2010 [in Japanese]. [5] Hattori M, Fujita M, Nakamura K, Ioka A. Estimation of the effects of centralization of cancer treatment on mortality reduction by in Fukui Prefecture. Japanese Journal of Public Health 2010;57(4):263–70 [in Japanese with English abstract]. [6] Branas CC, MacKenzie EJ, Williams JC, Schwab CW, Teter HM, Flanigan MC, et al. Access to trauma centers in the United States. The Journal of the American Medical Association 2005;293(21): 2626–33. [7] Nallamothu BK, Bates ER, Wang Y, Bradley EH, Krumholz HM. Driving times and distances to hospitals with percutaneous coronary intervention in the United States: implications for prehospital triage of patients with ST-elevation myocardial infarction. Circulation 2006;113(9):1189–95. [8] Baird G, Flynn R, Baxter G, Donnelly M, Lawrence J. Travel time and cancer care: an example of the inverse care law? Rural Remote Health 2008;8(4):1003. [9] Yamada Y, Kawahara K, Kawaguchi Y. Study of an emergency medical service system vis-à-vis ambulance transport access time—examination by simulating consolidating emergency medical services in Ibaraki prefecture. Journal of Japan Society for Healthcare Administration 2009;46(3):167–76 [in Japanese with English abstract]. [10] Ministry of Health, Labour and Welfare. Investigative report of the commission into the promotion of cancer treatment standardization; 2005. http://www.mhlw.go.jp/shingi/2005/04/s0419-6a.html [accessed 06.03.14] [in Japanese]. [11] Church J, Barker P. Regionalization of health services in Canada: a critical perspective. International Journal of Health Services 1998;28(3):467–86. [12] Kobayashi D, Otsubo T, Imanaka Y. Regional variations in healthcare systems through the analysis of patient transit times for various diseases. In: Proceedings of the 31st joint conference on medical informatics, vol. 31. 2011. p. 470–3 [in Japanese with English abstract]. [13] Japan Post Service Co., Ltd. Postal code data download, http://www. post.japanpost.jp/zipcode/dl/kogaki.html [accessed 06.03.14] [in Japanese]. [14] Ministry of Land, Infrastructure, Transport and Tourism. Block-level location reference download service, http://nlftp.mlit.go.jp/isj/ index.html [accessed 06.03.14] [in Japanese]. [15] Matsumoto M, Kashima S, Ogawa T, Takeuchi K. Do rural and remote areas really have limited accessibility to health care? Geographic analysis of dialysis patients in Hiroshima, Japan. Rural Remote Health 2013;13(3):2507. [16] Tracy R, Nam I, Gruca TS. The influence of visiting consultant clinics on measures of access to cancer care: evidence from the state of Iowa. Health Services Research 2013;48(October (5)):1719–29. [17] Kyoto Prefectural Government. Kyoto prefecture health care plan: medical facilities that provide medical functions for acute myocardial infarction, http://www.pref.kyoto.jp/drkyoto/condition/ file/condition 10 5.pdf [accessed 06.03.14] [in Japanese]. [18] Kyoto Prefectural Government. Kyoto prefecture information related to cancer treatments. Treatment information of core hospitals, collaborative hospitals and progressive hospitals within the prefecture, http://www.pref.kyoto.jp/gan/shinryou.html [accessed 06.03.14] [in Japanese]. [19] Kyoto Prefectural Government. Kyoto health care Yorozu net, http://www.mfis.pref.kyoto.lg.jp/ [accessed 06.03.14] [in Japanese]. [20] Khan AA. An integrated approach to measuring potential spatial access to health care services. Socio-Economic Planning Sciences 1992;26(4):275–87. [21] Gooiker GA, van Gijn W, Post PN, van de Velde CJ, Tollenaar RA, Wouters MW. A systematic review and meta-analysis of the volume-outcome relationship in the surgical treatment of breast cancer. Are breast cancer patients better of with a high volume provider? European Journal of Surgical Oncology 2010;36(Suppl. 1): S27–35. [22] Kunisawa S, Kobayashi D, Lee J, Otsubo T, Ikai H, Yokota C, et al. Factors associated with the administration of tissue plasminogen activator for acute ischemic stroke. Journal of Stroke and Cerebrovascular Diseases 2014;23(4):724–31.
Please cite this article in press as: Kobayashi D, et al. The effect of centralization of health care services on travel time and its equality. Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.11.008
G Model HEAP-3320; No. of Pages 9
ARTICLE IN PRESS D. Kobayashi et al. / Health Policy xxx (2014) xxx–xxx
[23] Schuurman N, Fiedler RS, Grzybowski SC, Grund D. Defining rational hospital catchments for non-urban areas based on travel-time. International Journal of Health Geographics 2006;5:43. [24] Hanlon N, Halseth G. The greying of resource communities in northern British Columbia: implications for health care delivery
9
in already-underserviced communities. The Canadian Geographer 2005;49(1):1–24. [25] Boscoe FP, Henry KA, Zdeb MS. A nationwide comparison of driving distance versus straight-line distance to hospitals. Professional Geographer 2012;64(2):188–96.
Please cite this article in press as: Kobayashi D, et al. The effect of centralization of health care services on travel time and its equality. Health Policy (2014), http://dx.doi.org/10.1016/j.healthpol.2014.11.008
