Medical expenditure in Japan: an analysis with administrative data from a Citizen s Health Insurance plan Medical Spending across Developed World Institute for Fiscal Studies March 27 & 28 2015 Yoko Ibuka, Tohoku University Stacey Chen, Academia Sinica 1 Overview 1. Institutional Background 2. Basic trend in healthcare expenditure 3. Medical expenditure with data from a Citizen’s Health Insurance plan 2 INSTITUTIONAL BACKGROUND : JAPANESE HEALTH INSURANCE SYSTEM 3 Overview • The national universal health insurance program(the national health insurance, NHI) started in 1961 – The first government-­‐involved health insurance was introduced in 1922 and gradually expanded in both its eligibilities and benefits • There are about 3500 plans – Having different risk structures and financial status – Roughly half employee-­‐based and half community-­‐based 4 Healthcare financing • Insurance Premiums: 48.6% ( in 2011) – Employee-­‐based plans A set percentage of wages, with the employer contributing at least half (55% on average) – Community-­‐based plans Each municipality has its own method of setting contributions, which are roughly based half on income (sometimes also assets), and half on a flat amount per enrollee (sometimes also per household) Subsidies: 38.4% – Subsidies from the government – Cross-­‐subsidizations across health insurance plans exist • Patients’ out of pocket: 12.3% • 5 Money flows in payment system Patients Medical care services Coinsurance Healthcare Providers Health insurance premium Insurers Medical claim bills Payment for services Review & Reimburse ment Services Medical claim bills after review Payment for bills Source: Ministry of Finance, Labour and Welfare, Japanese Health Insurance System http://www.mhlw.go.jp/stf/seisakunitsuite/bunya/kenkou_iryou/iryouhoken/iryouhoken01/index.html6 Health insurers 1 • Individuals have no choice of plans – All aged 75 years or above are covered by Late Elders’ Health Insurance (LEHI) – Employers have to enroll their employees: Society-­‐ Managed Health Insurance (SMHI) and National Health Insurance Associations (NHIA) – Dependents (except those older than 75 years) are covered by the plan of the head of the household – All others are covered by Citizen’s Health Insurance (CHI) 7 Health insurers 2 Employee-based insurance SMHI NHIA Community-based insurance CHI LEHI SocietyNational Health Citizen's Health managed Health Insurance Insurance Insurance Associations Eligibility Employees at Employees at Self-employed, large firms, small and part-time government medium workers, those officers, their enterprises, their retired dependents dependents Enrollment (thousand) 31,000 35,000 38,000 (proportion) 26% 29% 32% Number of health plans 1500 1 1900 Medical (proportion) 15% 15% 29% spending Late Elders’ Health Insurance All aged 75 years or older 16,000 13% 47 41% Source: Ministry of Finance, Labour and Welfare, Japanese Health Insurance System (in Japanese) http://www.mhlw.go.jp/stf/seisakunitsuite/bunya/kenkou_iryou/iryouhoken/iryouhoken01/index.html 8 Insurance benefits • Services covered include prescription drugs and dentistry • Single fee schedule is applied to all plans – The government committee revises the fee schedule every two years • Coinsurance are age-­‐dependent and the same across plans • Extra coverage for catastrophic expenses: those with monthly out-­‐of-­‐pocket payment of 40,000 JPY (=400 USD) for 70+ and 80,100 JPY (= 810 USD) for those under 70 – For many, 1% coinsurance for medical costs above the payment values – Can be more or less generous depending on income level 9 Coinsurance as of 2015 Age 75 or over 70 - 75 6 - 70 0 - 6 Those with income above a threshold value* All other 10% 20%§ 30%** 30% 20% * Monthly salary earnings excluding bonus payments more than 280 thousand JPY. Adjustments are made for those who have income other than salary. ** 20% until October 2006 § 10% until April 2014 10 Other financial sources for medical care for patients • Municipal government can provide subsidies for medical care for infants and children to reduce out-­‐of-­‐pocket payment – Eligibility and coverage are determined by each municipal government • Some purchase a voluntary private health insurance on top of the NHI • Both are independent of the NHI and individuals who qualify the subsidies or who purchase insurance receive reimbursement after they make payment – They do not appear as the source of medical spending in NHI statistics and counted as a part of out-­‐of-­‐pocket payment 11 Long-term care insurance • Long-­‐term care insurance (LTCI) started in 2000 • The insurance is independent of NHI • Spending for long-­‐term care accounts for 7.7% of the total social security payment while spending for medical care accounts for 32.1% in 2011 12 BASIC TREND IN HEALTHCARE EXPENDITURE: NATIONAL STATISTICS 13 2011 2008 (%) 2.45 2005 262 2002 3.45 1999 762 1996 4.45 1993 1262 1990 5.45 1987 1762 1984 6.45 1981 2262 1978 7.45 1975 2762 1972 8.45 1969 3262 1966 9.45 1963 3762 1960 Spending per person(2010 USD) Per capita medical spending and medical spending as percentage of GDP, 1960-2012 Year Spending per person in 2010 USD Percent of GDP 14 Per capita medical spending by age group population vs. study sample all 14000 12000 2010 US dollar 10000 8000 2000 6000 2010 2012 4000 Sample 2000 0 0~9 10~19 20~29 30~39 40~49 50~59 60~69 70~79 80 or beyond Age ( years) Note: Medical spending in 2000, 2010 and 2012 are from the national statistics. 15 Per capita medical spending by age group population vs. study sample Men 16000 14000 2010 US dollar 12000 10000 8000 2010 2012 6000 Sample 4000 2000 0 0~9 10~19 20~29 30~39 40~49 50~59 60~69 70~79 80 or beyond Age (years) Note: Medical spending in 2000, 2010 and 2012 are from the national statistics. 16 Per capita medical spending by age group population vs. study sample Women 12000 2010 US dollar 10000 8000 6000 2010 2012 4000 Sample 2000 0 0~9 10~19 20~29 30~39 40~49 50~59 60~69 70~79 80 or beyond Age (years) Note: Medical spending in 2000, 2010 and 2012 are from the national statistics. 17 Share of medical spending, by source of funding, 1960-2012 100 90 80 (%) 70 60 50 40 30 20 10 0 1960 1970 1980 1990 2000 2010 2012 Central govt Local govt Employers & Insurers Insured Out-of-pocket Other 18 Share of medical spending, by the type of care, 1970-2012 100 90 80 70 (%) 60 50 40 30 20 10 0 1970 Inpatient care 1980 Outpatient care 1990 2000 Dental care 2010 Prescription drugs 2012 Other Note: Prescription drugs are included as a part of outpatient care in 1970. 19 MEDICAL SPENDING WITH DATA FROM A CITIZEN’S HEALTH INSURANCE PLAN 20 Study sample 1 • Study period: 2006-­‐2011 • Coverage: All CHI beneficiaries who resides in city X during the sample period • Medical claim records are obtained from a CHI plan of municipality X in the north-­‐east region of Japan – All medical claim records during the sample period are compiled into the individual level and summed over 12 months to produce a yearly panel dataset on medical spending – The dataset is matched with the list of all beneficiaries and thus it includes those who did not seek for care • The panel data on medical spending is merged with household income data from tax records 21 Study sample 2 • Medical spending – Include: Inpatient and outpatient services, prescription medicine, dental care – Exclude: • long term care, over-­‐the-­‐counter drugs • A part of out-­‐of-­‐pocket expenditure may be reimbursed by voluntary private health insurance or municipal-­‐level government subsidies (for children) • Year is defined as the fiscal year (April to March next year) 22 Discontinuity in eligibility • In April 2008, the late elders’ health insurance (LEHI) was introduced • Data includes only those under 75 after 2008 Age 0-­‐74 75 + 2006, 2007 2008-­‐2011 CHI CHI CHI LEHI 23 Comparison 1: Mean expenditure Population, CHI(all), CHI(City X) Population Mean expenditure Enrollment Thousand persons 2006 2007 2008 2009 2010 2011 127,747 127,687 127,566 127,445 127,708 127,567 CHI, all Mean expenditure Enrollment Thousand 2010 USD persons 2,937 3,023 3,036 3,192 3,329 3,450 51,268 50,724 39,492 39,098 38,769 38,313 CHI, city X Enrollment Thousand 2010 USD persons 2,811 2,984 3,036 3,172 3,297 3,419 12.71 13.12 9.69 9.63 10.17 10.61 Mean expenditure All <74 years >75 years 2010 USD 2010 USD 2010 USD 6,217 6,145 3,728 3,810 3,837 3,854 4,441 4,273 3,728 3,810 3,837 3,854 11,020 11,003 NA NA NA NA Note: CHI covered all the age group until 2008, and only individuals aged under 75 years as the Late Elders’ Health Insurance started in 2008. 24 Comparison 2: Age distribution Population, CHI(all), CHI(city X) 100 (%) 90 80 80 or above 70 70-79 60 60-69 50 50-59 40-49 40 30-39 30 20-29 20 10-19 10 0-9 0 Population (2011) CHI (2011) CHI, City X (2009-2011) CHI, City X (2006-2008) 25 Medical spending, by payers (%) (A) All period All 21.9 78.2 Men 22.7 76.9 Women 21.1 78.9 All 16.8 83.1 Men 17.4 82.8 Women 16.2 83.6 (C) 2008-2011 (Under 75) All Out-of-pocket payment 27.5 National Health Insurance 72.5 Men 28.2 71.8 Women 26.8 73.1 Out-of-pocket payment National Health Insurance (B) 2006, 2007 (All age) Out-of-pocket payment National Health Insurance Note: CHI covered all the age group until 2008, and only individuals aged under 75 years as the Late Elders’ Health Insurance started in 2008. 26 Medical care spending, by type of care (%) (A) All period Medical care Inpatient care Outpatient care Dental care Prescription drugs All Men 50.2 26.8 4.2 18.8 (B) 2006, 2007 (All age) All Men Women Medical care Inpatient care 55.1 59.7 51.4 Outpatient care 25.0 23.3 26.1 Dental care 3.3 3.0 3.5 Prescription drugs 16.8 13.9 18.9 54.9 25.6 3.8 15.8 Women 46.1 27.8 4.6 21.2 (C) 2008-2011 (Under 75) All Men Women Medical care Inpatient care 45.1 50.1 40.5 Outpatient care 28.9 27.6 29.9 Dental care 5.3 4.6 5.8 Prescription drugs 21.1 17.7 23.8 Note: CHI covered all the age group until 2008, and only individuals aged under 75 years as the Late Elders’ Health Insurance started in 2008. 27 Concentration of spending (A) All period Spending Percentile Percentage of the enrollees Percentage of spending Average spending per enrollee (in 2010 USD) 100% 5% 5% 20% 20% 50% 100% 51% 14% 22% 9% 4% 4,738 48,470 13,491 5,205 2,156 334 Everyone 95%-100% 90-95% 70-90% 50-70% 0-50% (B) 2006, 2007 (All age) Spending Percentile Everyone 95%-100% 90-95% 70-90% 50-70% 0-50% Percentage Percentage of Average of the spending spending per enrollees enrollee in 2010 USD) 100% 100% 6,181 5% 46% 57,328 5% 16% 20,369 20% 23% 7,193 20% 10% 2,998 50% 4% 512 (C) 2008-2011 (Under 75) Spending Percentile Everyone 95%-100% 90-95% 70-90% 50-70% 0-50% Percentage Percentage Average of the of spending spending per enrollees enrollee in 2010 USD) 100% 5% 5% 20% 20% 50% 100% 53% 12% 22% 9% 3% 3,809 40,637 9,479 4,128 1,738 259 28 Mean income, by income quintile 50000 45000 2010 US dollar 40000 35000 30000 All 25000 Men 20000 Women 15000 10000 5000 0 Everyone Bottom Fourth Third Second Top 29 Mean medical expenditure, by expenditure quintile 16000 14000 2010 US dollar 12000 10000 All 8000 Men Women 6000 4000 2000 0 Everyone Bottom Fourth Third Second Top 30 Mean medical expenditure, by income quintile 4500 4000 2010 US dollar 3500 3000 2500 All 2000 Men Women 1500 1000 500 0 Everyone Bo<om Fourth Third Second Top 31 Correlation in medical spending, over time (A) Spending in Levels All Men Women t 1 1 1 t+1 0.58 0.58 0.57 t+2 0.49 0.49 0.49 (B) Spending in Logs All Men Women t 1 1 1 t+1 0.73 0.76 0.69 t+2 0.62 0.67 0.59 32 Total medical expenditure transition matrix (A) One year transition Quintile Current Year Bottom Fourth Third Second Top Bottom 60.4 25.0 7.7 2.2 5.0 Fourth 26.9 44.6 24.3 4.7 2.6 Next year Third 8.2 21.7 40.0 22.8 6.4 Second 2.3 5.2 21.2 46.9 22.1 Top 2.1 3.5 6.8 23.4 63.9 (B) Two year transition Quintile Two years time Current Year Bottom Fourth Third Second Top Bottom 54.1 24.9 11.4 5.5 4.1 Fourth 27.5 38.8 20.3 8.4 5.0 Third 10.9 24.1 34.1 20.4 10.6 Second 3.9 5.4 19.1 43.2 28.4 Top 9.7 3.8 7.3 16.5 62.7 Note: Panel A is produced with data limited to those who remained at least for three periods to keep consistency in sample in the two panels. 33 Concentration over 1, 2, and 3 years Gini coefficient on medical spending Percentage spent by top 1% of spenders Percentage spent by top 10% of spenders Total medical spending averaged over: 1 year 2 years 3 years 0.77 0.73 0.71 19% 17% 17% 65% 61% 59% 34 .6 .4 .2 0 Probability .8 1 Average medical spending over 1, 2, and 3 years 100 500 1000 10000 100000 300000 Average Medical Spending (2010 USD) 1 year 2 year 3 year 35 Identifying the last year of life 1 • The data does not show the reason why they left the CHI • Possible reasons 1. Death 2. Move to the Late Elders’ Health Insurance 3. Move to other health insurance plan • Our Strategy 1. Identify those who left the CHI during the study period 2. We dropped those who moved to the Late Elders’ Health Insurance based on the timing (in 2008 April) or their age (after April 2008) 3. We dropped those who left the CHI at the same timing as their dependents (i.e. those who are likely to change health insurance plan) 36 Identifying the last year of life 2 • Two types of errors 1. Those who are actually not dead are treated as the dead • Individuals who changed a health insurance plan by him/herself alone, including those who are single, are treated as being dead 2. Those who are dead are treated as those alive • Individuals who died at the age of 75 are treated as those are alive • Approximately 10% of the total sample fall in the definition of death • Higher number compared to the mortality rate • Error 1 is more serious? à Underestimate of medical spending in the last year of life 37 Medical spending in three groups Death Exit other No exit 3000 2010 USD 2500 2000 1500 1000 500 0 -36 -34 -32 -30 -28 -26 -24 -22 -20 -18 -16 -14 -12 -10 -8 Months before the last month -6 -4 -2 Note: “Death” are defined in the way explained in the text. “Exit other” showed those who left the CHI for presumably different reasons other than death. “No exit” are those 38 who remained CHI throughout the study period. Medical spending in the last year of life (A) Aggregate medical spending and mortality Personal Health Care Expenditure Mean spending per person (2010 USD) Aggregate spending (billions) Mortality Deaths (millions) Popoulation Over 65 population 3,503 489 8,250 252 1.19 1.07 Sources: Vital Statistics of Japan 2010 for the number of deaths; National Medical Care Expenditure 2010 for mean and aggregate spending. 39 Medical spending in the last year of life (B) Medical spending in the last year of life All age Over 65 As a percentage of Mean Mean As a percentage spending, of aggregate Hashimoto aggregate Last years of life spending , from data et al., 2010 study sample spending study sample spending Year of death 16,974 4.59 28,095 11.94 22,354 2nd to last 11,084 3.00 13,902 5.91 NA 3rd to last 9,338 2.53 11,853 5.04 NA Sum of last 3 years 37,396 10.12 53,850 22.88 NA Note: The share of medical spending in the last year of life was calculated by multiplying the mean spending from the data by the number of deaths per year divided by the total medical spending. The number of deaths and total medical spending were from the national statistics. Hashimoto et al., 2010 does not include medical spending for dentistry.40 Cumulative spending in 2010 USD Cumulative spending in the last year of life 20000 18000 16000 14000 12000 10000 8000 6000 4000 2000 0 41 Concluding remarks • We demonstrated macro and micro structure of medical spending using an administrative data from a Citizen’s Health Insurance plan in Japan • The study sample – represents a greater proportion of those aged 60 years or above than the population and the entire CHI – demonstrates consistent age-­‐specific mean medical spending as the national statistics – shows a greater average expenditure per person than the national statistics, largely attributed to the age distribution of the sample • Medical spending is – disproportionately distributed with top 5% individuals accounting for 50% of total spending – not strongly associated with household income level – highly persistent overtime • Medical spending in the last year of life shows concentration in the last years of life (although caution should be taken to interpret the results) 42 Acknowledgements • We acknowledge financial support from the Japan Society for the Promotion of Science #24330097 (PI: Nobuyuki Izumida) and #26780168 (PI: Yoko Ibuka) • We are grateful to Dr. Yui Otsu for discussion 43