Article Text

Download PDFPDF
Prevalence of type 2 diabetes by age, sex and geographical area among two million public assistance recipients in Japan: a cross-sectional study using a nationally representative claims database
  1. Tami Sengoku1,
  2. Tatsuro Ishizaki2,
  3. Yoshihito Goto1,
  4. Tomohide Iwao3,
  5. Shosuke Ohtera3,4,
  6. Michi Sakai1,5,
  7. Genta Kato3,
  8. Takeo Nakayama1,
  9. Yoshimitsu Takahashi1
  1. 1 Department of Health Informatics, Kyoto University School of Public Health, Kyoto, Japan
  2. 2 Human Care Research Team, Tokyo Metropolitan Institute of Gerontology, Itabashi-ku, Tokyo, Japan
  3. 3 Kyoto University Hospital, Kyoto, Japan
  4. 4 National Institute of Public Health, Wako, Japan
  5. 5 Comprehensive Unit for Health Economic Evidence Review and Decision Support, Ritsumeikan University, Kyoto, Japan
  1. Correspondence to Dr Tami Sengoku, School of Public Health, Kyoto University, Kyoto, Japan; sengoku.tami.5{at}gmail.com

Abstract

Background Recognising the importance of the social determinants of health, the Japanese government introduced a health management support programme targeted at type 2 diabetes (T2D) for public assistance recipients (PAR) in 2018. However, evidence of the T2D prevalence among PAR is lacking. We aimed to estimate T2D prevalence by age and sex among PAR, compared with the prevalence among health insurance enrollees (HIE). Additionally, regional differences in T2D prevalence among PAR were examined.

Methods This was a cross-sectional study using 1-month health insurance claims of both PAR and HIE. The Fact-finding Survey data on Medical Assistance and the National Database of Health Insurance Claims data were used. T2D prevalence among PAR and HIE were assessed by age and sex, respectively. Moreover, to examine regional differences in T2D prevalence of inpatients and outpatients among PAR, T2D crude prevalence and age-standardised prevalence were calculated by prefecture. Multilevel logistic regression analysis was also conducted at the city level.

Results T2D crude prevalence was 7.7% in PAR (inpatients and outpatients). Among outpatients, the prevalence was 7.5% in PAR and 4.1% in HIE, respectively. The mean crude prevalence and age-standardised prevalence of T2D (inpatients and outpatients) among 47 prefectures were 7.8% and 3.9%, respectively. In the city-level analysis, the OR for the prevalence of T2D by region ranged from 0.31 to 1.51.

Conclusion The prevalence of T2D among PAR was higher than HIE and there were regional differences in the prevalence of PAR. Measures to prevent the progression of diabetes among PAR by region are needed.

  • health inequalities
  • morbidity
  • nutritional sciences
  • social inequalities

Data availability statement

No data are available. The raw data of the FSMA and NDB-SD were not shared. Application for MHLW should be required to obtain the data.

Statistics from Altmetric.com

Introduction

Social and economic environments are listed as determinants of health by the WHO1 and many countries have adopted various measures2–4 to combat them. In recent years, it has been pointed out that income disparities are expanding in Japan as well and health inequalities are widening due to socioeconomic conditions.5 The proportion of the total population on public assistance has grown from 0.7% in 1995 to 1.6% in 2019. Therefore, in Japan, improvement plans that address social inequalities in health have been developed.6 7

The Public Assistance Act8 aims for the state to guarantee a minimum standard of living by providing the necessary public assistance according to the level of poverty. Individuals suffering from poverty apply for public assistance through welfare offices. Their utilisation of assets, utilisation of abilities and precedence of support are assessed and if they are approved, they become public assistance recipients (PAR). The public assistance system is composed of support for livelihood, education, housing, medical, long-term care, maternity care, occupational aid and funeral assistance. PAR can receive any or all of these supports according to their needs as long as their application is accepted by the welfare office. The medical assistance system that provides medical services is 100% financed by the government because the services are important in enabling individuals to maintain and improve their health status, quality of life and life expectancy. Japan has universal health coverage and almost all citizens have some kind of public medical insurance; meanwhile, recipients of medical assistance receive the same level of medical care as those with health insurance. The medical assistance system contributes to the health of PAR, although approximately 80% of PAR receive medical assistance and medical care costs are increasing. In a preliminary report, PAR had a higher incidence of non-communicable diseases such as type 2 diabetes (T2D) than health insurance enrollees (HIE).9 Progression of diabetes causes microvascular complications and presents as major obstacles to independent living such as dialysis treatment due to diabetic nephropathy, blindness from diabetic retinopathy and leg amputation from diabetic neuropathy. Also, diabetes, hypertension and dyslipidaemia comorbidities are risk factors of heart disease and stroke.10 11

The Ministry of Health, Labour and Welfare (MHLW), Japan, focuses on the prevention of onset and progression of non-communicable diseases among PAR. The MHLW started a health management support programme to reduce non-communicable diseases, especially T2D.9 The MHLW instructed local governments and welfare offices to prepare and enact a localised plan with a plan-do-check-action (PDCA) cycle. However, information related to the overall current status of T2D among PAR is limited to reports from certain geographical areas or is based only on the results of medical check-ups. The actual nationwide prevalence in Japan and regional differences are also unclear. Therefore, we aimed to describe the T2D prevalence by age and sex among PAR in Japan. To consider the significance of T2D prevalence, we calculated the prevalence among HIE for comparison. Also, we examined regional differences in T2D prevalence among PAR.

Methods

Design

A cross-sectional study using health insurance claims data for both medical assistance and public medical insurance.

Data source 1: PAR

The Fact-finding Survey on Medical Assistance (FSMA) data were used to obtain the number of patients.12 13 The National Survey on Public Assistance Recipients (NAPSR) data were used to obtain the number of PAR. The prevalence of recipients was calculated using the number from the FSMA as the numerator and the number from the NAPSR as the denominator.

The FSMA is conducted every year by the MHLW. FSMA, a general statistical survey based on the Statistics Act, aims to understand the treatment details of medical assistance recipients under the Public Assistance Act. All claims data of medical assistance recipients are assessed every year in June (for medical treatments in April and May) and the survey includes claims for outpatients, inpatients, diagnosis procedure combination (DPC) inpatients, prescription dispensing and dental treatments. The present study used data on medical treatments for May related to inpatients, outpatients and prescription dispensing. As for personal identification, each anonymised data entry was given an ID number created from the number of municipal welfare offices and the ‘ID2’,14 which was generated based on the person’s name, date of birth and sex.

NAPSR, a general statistical survey based on the Statistics Act, aims to understand the status of households that are current and past recipients of public assistance based on the Public Assistance Act. Annual surveys, the baseline survey and the individual survey are conducted every year in July in addition to monthly overview surveys. Since the monthly surveys report total numbers, the total number of PAR for May by sex and age was calculated by multiplying the sex and age ratios reported in the annual survey by the total number of recipients in May.

Data source 2: HIE

We used the sampling dataset of the National Database of Health Insurance Claims and Specific Health Checkups of Japan (NDB)15 16 to obtain the number of patients. The NDB is a database that collects almost 100% of the digitised health insurance claims data nationwide. In Japan, all persons are required to be covered by public health insurance and it is possible to assess the status of nearly all medical treatments provided to HIE. The MHLW provides a NDB sampling dataset (NDB-SD)17 that is created by extracting 10% of health insurance claims for inpatients and DPC inpatients and 1% of claims for outpatients over a 1-month period from the NDB. The NDB-SD is random sampling data and it was not possible to connect the inpatient and outpatient claims data of the same individual. Therefore, outpatient and prescription dispensing claims were used.

The baseline survey on health insurance data was used to obtain the number of HIE. The baseline survey on health insurance, compiled by the MHLW’s Health Insurance Bureau’s Actuarial Research Division, is a collection of data such as the application of different systems and surveys of the status of its services in each health insurance system. Since HIE are reported as a total number by age group, the numbers for each sex were calculated by the rate of population numbers by sex and age group shown in the National Livelihood Survey report.

The prevalence of HIE was calculated using the number from the NDB-SD as the numerator and the number from the baseline survey on health insurance as the denominator.

Target period

One month’s worth of data of medical treatments for May in 2015, 2016 and 2017 from FSMA data and medical treatments for April 2015 from the NDB-SD were used. Data from the July annual surveys in 2015, 2016 and 2017 and the monthly surveys in May from the NAPSR were used along with the data from the baseline survey on health insurance and the 2015 annual National Livelihood Survey data.

Variables

Items related to sex, age group, geographical area, disease code and medication were obtained from the health insurance claims data. For area, data for prefecture (47 areas) and city level (112 geographical areas) were examined. At the city level, 20 government ordinance-designated cities and 45 core cities were examined and all others were aggregated at the prefectural level because publication of results analysed by smaller categories was restricted by the MHLW.

Definition of disease

Diabetes, hypertension, dyslipidaemia and diabetes-related renal complications, ophthalmic complications and neurological complications were defined based on the 2013 version of the International Statistical Classification of Diseases and Related Health Problems (ICD-10) code. Also, those not receiving medication for diabetes, hypertension and dyslipidaemia were excluded to avoid including individuals whose disease was not clinically confirmed. Drugs were defined using classification code 87 of the Japan Standard Commodity Classification (online supplemental appendix 1).18 Moreover, type 1 diabetes was excluded from the analysis and T2D was targeted.

Supplemental material

Statistical analysis

Among PAR, T2D prevalence and the proportion of comorbidity (hypertension and dyslipidaemia) and complications (diabetic nephropathy, retinopathy and polyneuropathy) were determined by sex and by age group. The data used were inpatient and outpatient claims data in May 2015, 2016 and 2017. Among HIE, prevalence and proportion were determined by sex and age group. The data used were outpatient claims data in April 2015. In the comparison between PAR and HIE, the population composition by age group was different (figure 1), so the T2D prevalence was age-standardised with a 1985 Japanese model population.19 The data used for comparison were both outpatient claims data in May 2015 and April 2015, respectively. The proportion of comorbidity and complications was calculated with T2D patients as the denominator.

Figure 1

Japanese population, public assistance recipient (PAR) population and diabetes population of PARs by age group.

Additionally, regional differences in T2D prevalence at the prefecture (2015, 2016, 2017) and city levels (in 2015) were determined among PAR. For prefecture, T2D prevalence of crude and age-standardised prevalence was examined. The prevalence in each prefecture was classified into five categories: (A) higher than the mean value of 47 prefectures plus the SD; (B) higher than the mean value and lower than A; (C) mean value; (D) lower than the mean value and higher than E and (E) lower than the mean value minus the SD. For the city-level analysis, multilevel logistic regression analysis with a random intercept was performed using T2D as the outcome variable, sex and age group as the fixed-effect parameters and the welfare office region as the random-effect parameter. The target participants of the analysis were those aged 30 years and older. The number of people who did not consult healthcare facilities was derived from the number of PAR in the NAPSR for the corresponding month (by sex, age group and geographical area). The HIE aged 40 years and older required health check-ups focusing on visceral fat obesity in Japan. However, a previous report found that the impact of diabetes on coronary artery disease was markedly greater in men aged 31–40 years compared with those aged 41–60 years.20 Thus, we targeted PAR aged 30 years and older.

The t-test was used to test for differences and the χ2 test was used to test for independence. The significance levels of the tests were set at 0.05. Statistical analysis was performed using R V.3.4.3.

Ethical considerations

The raw data of the FSMA was obtained after receiving approval from the MHLW in accordance with Article 33 of the Statistics Act.21 The raw data of the NDB-SD were obtained after receiving approval from the MHLW according to the ‘Guidelines for the provision of NDB’.22 The data for the NAPSR, national livelihood surveys and baseline surveys on health insurance were collected using online aggregated data.

Results

The number of PAR in 2015, 2016 and 2017 was 2 161 442 (49.5% men; mean age, 56 years), 2 148 282 (49.5%; 56 years) and 2 130 482 (49.6%; 56) and the T2D crude prevalence in inpatients and outpatients was 7.7%, 7.9% and 8.3% (figures 1 and 2), respectively. The T2D prevalence tended to be higher with survey data and was highest (11.5%, 11.6%, 11.9%) among people in their 60s. The prevalence in men (8.6%, 8.8%, 9.2%) was statistically significantly higher (<0.001) compared with that in women (6.8%, 7.0%, 7.3%) (figure 3; online supplemental appendix 2).

Supplemental material

Figure 2

Flow diagram of public assistance recipients in the study population.

Figure 3

Type 2 diabetes prevalence among public assistance recipients.

In 2015, for comorbidities, the proportion with hypertension was 66.8% (65.7% of men and 68.2% of women), dyslipidaemia was 47.7% (42.3% and 54.3%, respectively) and both hypertension and dyslipidaemia was 35.5% (31.6% and 40.4%, respectively). The proportion of hypertension was higher with age in both men and women until their 70s and there was a slight decrease in their 80s. As for dyslipidaemia, the comorbid proportion peaked in their 40s among men and decreased thereafter and among women, it continued to be higher until their 60s. The comorbid proportion of both hypertension and dyslipidaemia was higher with age, peaking in their 60s in men and their 70s in women. The proportion of complications was as follows: renal complications 15.9% (16.9% and 14.8%, respectively), ophthalmic complications 16.4% (17.7% and 14.8%, respectively) and neurological complications 12.5% (14.1% and 10.6%, respectively) (see online supplemental appendix 3 for age-specific details).

Supplemental material

The mean crude prevalence and age-standardised prevalence of PAR among 47 prefectures in 2015 were 7.8% (SD; 1.5%) and 3.9% (0.8%), respectively. There were no notable characteristics when the correlation was examined by plotting T2D prevalence (crude and age standardised) on the y-axis and old-age index (the proportion of the population 65 years and older to the population between 15 and 64 years), public assistance proportion (the proportion of the population receiving public assistance to the population of each prefecture) and prefectural population on each x-axis (online supplemental appendix 4). The T2D prevalence in each prefecture was classified into five categories and the results showed there was no considerable difference in the categories of each prefecture for 3 years in both crude and age-standardised prevalence (figure 4).

Supplemental material

Figure 4

Age-standardised and crude prevalence of type 2 diabetic public assistance recipients by prefecture.

When examined by city level (112 geographical areas), the crude prevalence of patients with T2D was 166 335 (9.1%) out of a total of 1 830 162 people. There was a fivefold difference (13.1%/2.6%) in diabetes prevalence in the crude prevalence by city level (online supplemental appendix 5). The multilevel logistic regression analysis showed that the SD for the geographical area was 0.27 (exp(−0.27)=0.76, exp(0.27)=1.31) and the OR by city level was within the range of 0.31 and 1.51 (table 1).

Supplemental material

Table 1

Results of multilevel analysis of the prevalence of patients with type 2 diabetes by city for sex and age group

A comparison of PAR and HIE among outpatients in 2015 showed the T2D crude prevalence of outpatients was 7.5% in PAR (8.3% in men and 6.6% in women) and 4.1% among HIE (4.8% and 3.3%, respectively) (see online supplemental appendix 6 for age-specific details). In other words, the prevalence was higher in PAR. The age-standardised prevalence was 3.8% in PAR (4.3% and 3.3%, respectively) and 2.3% in HIE (3% and 1.7%, respectively) (p<0.01). As for the proportion of comorbidities, 67.4% of PAR and 65.2% of HIE had comorbid hypertension. Similarly, the proportion with comorbid dyslipidaemia was 48.5% and 47.9%, respectively, and the proportion of those with both dyslipidaemia and hypertension was 36.2% and 34.2%, respectively. In other words, the proportion of comorbidity was higher among PAR. The proportion of renal, ophthalmic and neurological complications was 16.2%, 16.7% and 12.7%, respectively, among PAR and 15.1%, 13.1% and 6.8%, respectively, among HIE. The proportion with complications was higher among PAR (see online supplemental appendix 7 for age-specific details).

Supplemental material

Supplemental material

Discussion

We revealed through health insurance claims that the prevalence of those receiving drug treatment for T2D was 7.7% (8.6% in men, 6.8% in women) among inpatients and outpatients. In outpatients, the prevalence was 7.5% (age-standardised prevalence; 3.8%) and 4.1% (2.3%) among PAR and HIE, respectively. Diabetes prevalence among the general Japanese population has been reported by the MHLW’s ‘National Health and Nutrition Survey’. The survey includes interviews regarding diabetes and the presence or absence of medication and it is estimated that the number of diabetics and prediabetics patients in Japan is 10 million and the age-standardised prevalence of diabetes is 12.1% (16.3% for men and 9.3% for women in 2016).23 A community-based cohort study reported that the prevalence of age-normalised diabetes was 20.8% in men and 11.2% in women.24 Furthermore, a study using medical claim data from regional HIE reported a diabetes prevalence of 9.6%.25 There is a limitation in understanding diabetes prevalence based on interview surveys reporting estimates and regional surveys; therefore, the comparison prevalence of diabetes among HIE and PAR nationwide is unknown. In this study, we determined the prevalence by using health insurance claims.

The T2D prevalence was higher among PAR and the proportion of comorbid hypertension, dyslipidaemia and diabetic renal, ophthalmic and neurological complications was also higher in PAR compared with HIE. By sex, the prevalence in men was higher than in women, as in previous reports. The high proportion of diabetes comorbidities and complications among PAR could be due to the fact that life difficulties harm their health. Complications greatly affect the rise in healthcare expenditure. Suzuki et al 25 reported that, compared with having no complication, necessary expenditure rises 1.7 and 1.6 times with renal complications and ophthalmic complications, respectively, and measures to address the prevention of complications are also needed. Funakoshi et al 26 reported that low socioeconomic status is related to the high prevalence of T2D complications among young adults. Their study suggested that to decrease diabetes complications, interventions that target both general risk factors and social determinants of health are needed. Various complications develop when diabetes progresses and therefore, continued support in health management is needed to delay the progression. Among individuals who were first diagnosed with diabetes at a check-up, those who started to visit the hospital once a month had a 10–14 times higher chance of improving haemoglobin A1c levels by 1% or more at the following year’s check-up compared with those who were untreated.27 The recommendation to receive periodic medical check-ups and follow-ups among PAR may prevent the progression of the disease.

In the examination of T2D prevalence by geographical area, the city-level analysis suggested that there are regional differences. By prefecture, the old-age index, regional population and the number of PAR were considered as factors; however, as none of these factors alone was the cause of the regional difference, multiple factors may have been associated.28 29 Minagawa and Saito30 pointed at regional welfare spending as a factor related to healthy life expectancy in PAR. Nomura et al 31 suggested that the performance of regional health variations, including the subnational health system, may also be related to behavioural factors, psychosocial factors and social factors. Furthermore, Nishioka et al 32 reported that unemployment and living alone might be additional risk factors for diabetes among younger PAR.

The progression of diabetes not only decreases the quality of life but is also a great burden on the medical economy. In Japan, local governments cover one quarter of the finances for public assistance programmes and rising healthcare costs per patient are a heavy burden on the financial resources. The amount of health expenditure paid on behalf of PAR as medical assistance accounted for 4.2% of total healthcare expenditure in 2014.33 To limit further increase in healthcare spending, the government promotes health management programmes based on data for people availing of public assistance. Our method of research using health insurance claims will provide the fundamental methodological procedures to conduct a ‘data health plan’ which aims to enable efficient and effective health activities based on the PDCA cycle for the MHLW and local governments. Japan seemed to have successful universal health coverage and public assistance system as a safety net. However, Japan is struggling with the escalation of medical costs and health disparities.31 The potential sources of fundamental health status among PAR urgently need assessment. This empirical evidence in Japan would be helpful to achieve no poverty, good health and reduced inequalities as sustainable development goals.

We acknowledge that there are several limitations of this study. First, the results were obtained through a cross-sectional study using 1-month data of health insurance claims. Thus, the prevalence could be underestimated because some patients with diabetes may not visit healthcare facilities every month. According to a report using an electronic database of Japanese community pharmacy chains, the average number of prescription days per prescription for outpatients with chronic diseases was metformin hydrochloride 38.1±15 days and sitagliptin phosphate hydrate 36.7±15.5 days34. According to a 2019 survey report on hypoglycemic drugs by the MHLW, about 70% were within 30 prescription days.35 The proportion of prescriptions written for ≥31 days could be very small in Japan. However, the difference between prescription days for PAR and HIE should be considered in future research. Second, the reverse causality may have biased the findings. Patients with T2D among PAR include cases of both patients derived from poverty (T2D onset after PAR) and people who become PAR because of T2D (PAR after T2D onset). The high diabetes prevalence among PAR could be due to the inclusion of people receiving public assistance due to diseases and life difficulties that may harm their health. Third, FSMA does not include some patients, whose other welfare services are prioritised over the public assistance programme. For example, if recipients are receiving other welfare services such as disability assistance, claims on their medical services are not reflected in medical assistance survey data. Also, some PAR (2%–3%) are still enrolled in employee health insurance36 and very few of the individuals who have received medical treatment under employee health insurance are included in the NDB-SD. These facts may have biased the findings and there is a possibility that the number of patients T2D in PAR is underestimated.

Conclusion

The prevalence of T2D and comorbidity proportion among PAR was higher compared with public insurance enrollees. Furthermore, there was a regional difference in the prevalence. Measures by region to prevent the progression of diabetes among PAR are needed.

What is already known on this subject

  • It is well known that social and economic environments are determinants of health. Public assistance recipients (PAR) had a higher incidence of non-communicable diseases such as type 2 diabetes (T2D) than health insurance enrollees (HIE) in a preliminary report; however, the evidence is limited. Since the Japanese government has started a health management support programme especially targeted at T2D among PAR, local governments and welfare offices should prepare and enact a localised plan.

What this study adds

  • Type 2 diabetes (T2D) receiving medical treatment comprised 7.5% of public assistance recipients (PAR) and 4.1% of health insurance enrollees (HIE), respectively. There was a regional difference in the prevalence of PAR. The prevalence of T2D among PAR was higher than HIE and measures to prevent the progression of diabetes depending on the region are needed.

Data availability statement

No data are available. The raw data of the FSMA and NDB-SD were not shared. Application for MHLW should be required to obtain the data.

Ethics statements

Patient consent for publication

References

Footnotes

  • Contributors YT conceived and designed the study, acquired the funds and supervised the project. TS, YG and YT contributed to data acquisition. TIwao contributed to formal data preparation. TS carried out the analysis. TS, TIshizaki, YG, SO, MS, GK, TN and YT contributed to the interpretation of the results. TS and YT wrote the manuscript. YT is the guarantor. All authors provided substantial critical work on the manuscript and approved the final version of the manuscript.

  • Funding This work was supported by the Japan Society for the Promotion of Science KAKENHI (16K15372 and 20H01594) and by the Ministry of Health, Labour and Welfare in Japan (H28 Tokubetsu-Shitei-031 and H29-Seisaku-Shitei-007).

  • Disclaimer The findings and conclusions of this article are solely the responsibility of the authors and do not represent the official views of the Japanese government.

  • Map disclaimer The depiction of boundaries on this map does not imply the expression of any opinion whatsoever on the part of BMJ (or any member of its group) concerning the legal status of any country, territory, jurisdiction or area or of its authorities. This map is provided without any warranty of any kind, either express or implied.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.