Projections of psychiatrists’ distribution for patients in Japan: a utilization-based approach

Background Depopulation accompanied by population aging is a major public health concern in Japan. Although adequate allocation of mental healthcare resources is needed, there have been few studies on the impact of population change on the supply–demand balance for mental illness in Japan. The aim of this study is to predict psychiatrists' distribution for patients with mental illness via a utilization-based approach. Methods We set patients with schizophrenia, mood disorders, vascular dementia or Alzheimer’s disease as study subjects and conducted analyses for 2015, 2025, 2035, and 2045 across all prefectures. Moreover, we evaluated the regional maldistribution of demand and supply by calculating the number of psychiatrists per patient, Gini coefficients (GC), and Herfindahl–Hirschman Index (HHI). Results The mean number of psychiatrists per patient for patients with schizophrenia, mood disorders, vascular dementia, and Alzheimer’s disease in 2025, 2035, and 2045 was significantly lower than in 2015. For all of the abovementioned diseases, both the GC and HHI will increase until 2045. Conclusion If psychiatrists are allocated at the current population-to-psychiatrist ratio, the shortage of psychiatrists will continue to worsen in the future. To overcome this inequity, policy makers should make plans to shift responsibilities from psychiatrists to other mental health workers and to ensure the adequate geographical allocation of healthcare resources.


Introduction
The increasing number of patients with mood disorders and Alzheimer's disease has increased the demand for psychiatrists in Japan [1]. In 2013, the Ministry of Health, Labour, and Welfare (MHLW) designated mental illness as the fifth priority disease for the national medical service, and all prefectures in Japan were required to start regional medical care planning for mental illness [2]. Optimizing the balance between supply and demand for the mental healthcare system is a public health issue, and psychiatrists are an essential human resource for the system.
Healthcare systems in Japan are facing the problems of depopulation accompanied by population aging. Based on 2015 national census data, the National Institute of Population and Social Security Research (IPSS) predicted that the Japanese population will decrease from 127 to 106 million by 2045 [3]. In the same analysis, the IPSS also indicated that the elderly population (aged 65 years and over) will increase by 15.7%, while the young population (aged 0 to 14 years) will decrease by 28.6%. Because the age of onset differs by disease, changes in population structure could lead to different utilization patterns Open Access *Correspondence: nsuga3@dokkyomed.ac.jp Department of Psychiatry, Dokkyo Medical University School of Medicine, 880 Kitakobayashi, Mibu, Tochigi 321-0293, Japan of healthcare services for each mental illness. Although adequate allocation of mental healthcare resources is needed, there have been few studies concerning the impact of population change on the supply-demand balance with respect to mental illness in Japan.
In this study, we employed a utilization-based approach in which current or target rates of healthcare system utilization are multiplied by future population estimates to estimate the demand of mental illness. This approach has been widely used in Organisation for Economic Cooperation and Development (OECD) member countries [4,5]. There have been other approaches to estimate the demand for healthcare workforces. A service-based approach or a task-based approach could be used for the same purpose [6]. The former approach is based on the estimation of the shifting needs of an organization that are required to operate effectively. The estimation requires data on the burden of disease, epidemiological changes, hospital bed-to-staff ratios, and the expected budget for staff salaries. On the other hand, the estimation of need in a task-based approach is founded on the tasks a typical professional can undertake in a given time period. Both approaches could be useful for healthcare workforce planning at relatively local level, but effectively accomplish this planning at the national level requires an enormous amount of data.
The aim of this study is to predict psychiatrists' distribution for patients with mental illness and to predict the future healthcare supply-demand balance. The projections of the availability of human resources in the mental healthcare system could support policy decision-making. To the best of our knowledge, this study is the first report on the projection of psychiatrists' distribution for patients with mental illness in Japan.

Analytical parameters
Data on the number of psychiatrists per prefecture were obtained from the 2016 Survey of Physicians, Dentists, and Pharmacists (SPDP) on the MHLW website [7]. In addition, we obtained data on population projections until 2045 from the IPSS [3]. These projections took the count from the 2015 Population Census as the base population [8]. The utilization rate per 100,000 population per prefecture was obtained from the 2017 Patient Survey on the MHLW website [1]. Based on the Statistics Act, Article 2, the MHLW conducted the Patient Survey to obtain basic data needed for the development of health policies by identifying age, sex, diagnosis according to the International Classification of Diseases, tenth revision (ICD- 10), and condition at time of survey for each patient. The survey covers inpatient and outpatient treatment in medical facilities, including 6395 hospitals and 5526 clinics. The survey report provides estimates of the utilization rates broken down by sex, 5-year age groups and ICD-10 diagnoses. We set patients with schizophrenia, mood disorders, vascular dementia, and Alzheimer's disease as the study subjects, as those are the leading mental disorders associated with relatively higher loss of disability-adjusted life-years (DALYs) in Japan [9]. The requirement for written informed consent was waived by the Ethics Committee since the study involved record review only.
To project the number of future psychiatrists, we assumed that the psychiatrist-to-population ratio in Japan would be constant from 2015 to 2045. Although the current distribution of psychiatrists in Japan is not adequate, we assumed that attractiveness of urban areas to psychiatrists would not change [10]. We calculated the number of psychiatrists per population based on the 2016 SPDP and the 2015 Population Census for each prefecture. Following the abovementioned assumption, the estimation of the number of psychiatrists in 2015, 2025, 2035 and 2045 was based on the psychiatrist-to-population ratio and population projections in Japan.
In a utilization-based approach, future demand is calculated by multiplying the future population by the utilization rate for each disease. First, we obtained the utilization rate for each disease, by age and sex, as variables from the 2017 Patient Survey. Population by age and sex in the future was based on population projections from 2015 to 2045. We then multiplied these variables for estimates of the future number of patients as the criterion for demand in each prefecture. For each disease, we calculated the number of psychiatrists per patient in Japan.
We employed the Gini coefficient (GC) as an indicator of the distribution of psychiatrists to aid in the evaluation of inequity in human resources by prefecture. In this study, Lorenz curves are drawn by plotting the cumulative proportion of psychiatrists on the vertical axis and the cumulative proportion of the estimated number of patients on the horizontal axis in ascending order by psychiatrists per patient across all prefectures. After that, we calculated the GCs based on the Lorenz curves. The GC is traditionally used to analyze the distribution of income and wealth and has a theoretical range from 0 (perfect evenness) to 1 (maximum possible unevenness). It provides a standardized value to reflect the relative unevenness of a distribution. In this study, higher values of the GC indicated higher levels of human resource (psychiatrists) inequality experienced by patients with mental illnesses among prefectures.
The Herfindahl-Hirschman Index (HHI), which has been widely used to evaluate mergers and acquisitions, was adopted as an indicator of patient concentration. In this study, the HHI for each disease is calculated as the sum of squared patient shares (percentages) across all prefectures. It approaches zero when a market is occupied by a large number of competitors of relatively equal size and reaches its maximum of 10,000 points when there is a market monopoly. The HHI was interpreted as the concentration of patients with mental illness to estimate future demand transfer. In this study, higher values of the HHI indicated higher concentrations of patients with mental illnesses among prefectures.
In Tables 1 and 2, we ordered the prefectures according to the identification codes (JIS X 0401) from the Japanese Industrial Standards Committee.

Statistical analysis
Because the Shapiro-Wilk test did not confirm the normality of the data distribution, the Wilcoxon Signed-Ranks Test with the Bonferroni correction was employed for comparisons between 2015 and other time points. A value of p < 0.05 was considered significant. The data analysis was performed using R for Windows, Version 3.6.3 (The R Foundation for Statistical Computing, Vienna, Austria) [11]. Table 1 displays forecasts of the number of psychiatrists and patients with mental illness in each prefecture. The mean numbers of psychiatrists and patients with schizophrenia and mood disorders in 2025, 2035, and 2045 are significantly lower than those in 2015. In each prefecture, excluding Tokyo, the number of psychiatrists is forecasted to decrease. Similarly, the number of patients with schizophrenia or mood disorders in each prefecture, excluding Tokyo and Okinawa, will decrease by 2045. On the other hand, the mean numbers of patients with vascular dementia and Alzheimer's disease at the abovementioned three time points will be significantly higher than those in 2015. In all prefectures, the number of patients with vascular dementia or Alzheimer's disease is projected to increase by 2045. Figure 1 shows the relationship between population growth rate and patient growth rate from 2015 to 2045 in each prefecture. We also summarized the number of psychiatrists per patient ( Table 2). The mean number of psychiatrists per patient for patients with schizophrenia, mood disorders, vascular dementia, and Alzheimer's disease at the abovementioned three time points is projected to be significantly lower than in 2015.

Results
The GC and HHI for each mental illness are shown in Figs. 2 and 3, respectively. The results show that both the GC and HHI for the four mental illnesses will increase.

Discussion
In this study, we predicted psychiatrists' distribution for patients with mental illness in Japan. On the supply side, the mean numbers of psychiatrists in 2025, 2035, and 2045 are significantly lower than those in 2015. On the demand side, in line with depopulation, the mean numbers of patients with schizophrenia and mood disorders are significantly lower than those in 2015. However, regarding vascular dementia and Alzheimer's disease, the mean numbers of patients with these diseases at the abovementioned three time points are significantly higher than those in 2015. For all of the abovementioned diseases, the HHI will consistently increase from 2015 to 2045. Regarding the supply-demand balance, the mean number of psychiatrists per patient for patients with schizophrenia, mood disorders, vascular dementia, and Alzheimer's disease at the abovementioned three time points is significantly lower than in 2015. For all of the abovementioned diseases, the GC will consistently increase from 2015 to 2045.
In Japan, the shortage of physicians, including psychiatrists, has recently become a serious public health issue [12,13]. Several studies have indicated that the cause of this shortage is related not only to the absolute number of physicians but also to their maldistribution [14,15]. Regarding the mental healthcare system in Japan, the absolute number of psychiatrists increased from 1996 to 2012, while the GC based on the number of physicians per population did not change during the same period [16]. Because the population decline has continued to accelerate since the population peaked at 128 million in 2008 [17], we could not predict the future demand-supply balance and equality based on this short observation period. Furthermore, different patterns of healthcare services utilization for each mental illness were not considered in the analysis, and changes in the population structure might not be consistent with the utilization patterns of patients. A study from the US [18], in which the population is predicted to increase in the future, indicated that a shortage of psychiatrists per population will occur despite the increasing number of psychiatrists. Apart from mental illness, Ishikawa and colleagues forecasted the distribution of physicians for patients with acute myocardial infarction, cerebral stroke, and all medical care in Hokkaido [5]. Their results indicated that the GCs for the abovementioned three conditions will decrease from 2015 to 2035, while the HHIs will increase in Hokkaido.
Our results indicate that the change in disease structure with the increase in patients with dementia and decrease in those with schizophrenia and mood disorders will continue until 2045. Unlike the overall trend, the predicted number of patients with schizophrenia or mood disorders   will worsen in the future. To overcome this inequity, policy makers should make plans for not only the adequate geographical allocation of healthcare resources, but also the shifting of responsibilities from psychiatrists to other mental health workers. The use of information and communication technologies (ICTs) for the delivery of health services to rural communities and improved productivity of psychiatrists with more effective interventions would also ameriolate the inequity.
Several limitations of this study should be acknowledged. First, our study focuses on the number of psychiatrists as the supply side of the mental healthcare system. However, human resources in the healthcare system consist of not only psychiatrists but also nurses and other health care professionals. Furthermore, the accessibility, number and performance of medical facilities are also important factors for the supply side of the system. Analysis of supply and demand in view of these various factors is important for carrying out a more detailed analysis that will be useful for supporting policy formulation. Increasing data collection on relevant values will minimize the limitations in this area. Second, we estimated the number of psychiatrists using population projections until 2045 and psychiatrists' distribution in 2015. Our results indicate that the shortage of psychiatrists will continue to worsen if psychiatrists are allocated at the current population-to-psychiatrist ratio. However, the age distribution, retirement patterns, and future supply of psychiatrists could affect the future number of psychiatrists. Further updating research is needed to predict the number of psychiatrists for forecasting the supply-demand balance accurately. Third, our results are limited by the fact that the utilization-based approach is based on several assumptions, as with other modeling methods. The utilization-based approach could result in an over-estimation of the demand, particularly in service areas open to supply-induced demand (for instance private psychiatry services) or areas where best practices are poorly implemented. The assumption of this approach is that patients' behavior will not change during the forecast period. Several factors, such as innovations in preventive medicine, screening and treatment, changes in medical care preferences, and changes in the capacity of the population to pay for services, could affect the behaviors of patients with mental illness. Although this analysis is based on a fixed value for the utilization rate, future research with newer rates would enable us to provide more accurate results.
In conclusion, this study forecasts the psychiatrists' distribution for patients with mental illness to analyze the healthcare supply-demand balance based on a utilization-based approach. While the number of patients with schizophrenia or mood disorders in each prefecture, excluding Tokyo and Okinawa, will decrease by 2045, the number with Alzheimer's disease or vascular dementia in all prefectures is projected to increase. For the four mental illness estimated considered, the difference between prefectures in the minimum and maximum number of psychiatrists per patient were approximately 2-folds or more in 2015. As long as psychiatrists are allocated at the current population-to-psychiatrist ratio, the shortage of psychiatrists will continue to worsen in the future. To overcome this inequity, it is necessary to discuss incentives for medical services in rural area, or mandatory requirement of practice in rural areas for psychiatrists who become board-certified psychiatrist. Although this analysis is based on a fixed value for the utilization rate, future research with frequent model updating would yield more accurate results.