Associations between the number of benzodiazepine receptor agonists prescriptions and social determinants of health, including social capital: an ecological study

Abstract

Background

The widespread prescribing of benzodiazepine receptor agonists (BZRA) is a public health problem worldwide. However, few studies have explored the relationship between the number of BZRA prescriptions and social determinants of health (SDH). Furthermore, to our knowledge, no reports have investigated the relationship between BZRA prescribing and social capital. Therefore, our objective was to conduct a prefecture-level ecological study focused on the number of BZRA prescriptions and SDH, including social capital.

Methods

We drew data for the number of BZRA prescriptions for April 2018 to March 2019 from The National Database of Health Insurance Claims and Specific Health Checkups of Japan Open Data. For the SDH, we obtained data on the relative poverty rate, Gini coefficient, unemployment rate, post-high school education advancement rate, number of single-parent households per 1,000 households, single-person household rate, and social capital scores from publicly available internet sources and previous report. We calculated the standardized claim ratio (SCR) for the number of BZRA prescriptions by prefecture. We then investigated the relationship between the SCR for BZRA and SDH through an ecological study using Spearman’s rank correlation coefficients and multivariate regression analysis.

Results

The relative poverty rate, Gini coefficient, unemployment rate, number of single-parent households per 1,000 households, and single-person household rate were positively and significantly correlated with the SCR for BZRA. In addition, the social participation domain of social capital was significantly negatively correlated with the SCR for BZRA. In the multivariate analysis, the relative poverty rate and single-person household rate remained independently associated with higher BZRA utilization. However, the association between social participation and BZRA use was no longer significant after adjusting for these socioeconomic factors.

Conclusions

This ecological study suggests that regional BZRA prescribing is associated with socioeconomic deprivation and household instability. While further individual-level research is needed to clarify the underlying mechanisms, our findings highlight the importance of addressing social determinants to manage regional disparities in BZRA use.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12889-026-27682-w.

Introduction

The widespread prescribing of benzodiazepine receptor agonists (BZRA), which have sedative, anxiolytic, and hypnotic effects, has become a public health problem in many countries [1–3]. While BZRAs are effective for their sedative, anxiolytic, and hypnotic effects, they are associated with adverse drug reactions. These include dose-dependent side effects such as impaired concentration, somnolence, and oversedation [4–6]. Furthermore, chronic use of BZRAs significantly increases the risks of physical dependence, cognitive decline, and falls [4, 7, 8]. Despite these identified negative health effects, BZRA remain widely prescribed worldwide [4, 9–11].

To address the extensive prescribing of BZRA, it is essential to understand the socio-demographic factors associated with BZRA prescriptions. Various studies have explored factors related to BZRA prescribing from a socioeconomic perspective. In the United States and European countries, ecological studies based on region- or practice-level data reported that lower socioeconomic status was associated with higher rates of hypnotics prescribed, including BZRA [12–14]. Furthermore, several studies from those countries based on individual-level data showed that socioeconomic status, including unemployment, low income, and living alone, were risk factors for long-term BZRA prescription [15–17]. Similarly, an ecological study conducted in Japan that used region-level claim data showed there were regional differences in the number of anxiolytics and hypnotics prescriptions [18]. In terms of socioeconomic characteristics, that study examined welfare receipt and taxable income, of which welfare receipt was reported to be significantly associated with the prescription of anxiolytics and hypnotics [18].

Socioeconomic factors influence human health and are recognized as social determinants of health (SDH). SDH include factors such as education, unemployment, work, income, early childhood development, race, gender, living conditions, health services, and social capital [19]. These factors are reported to be associated with comprehensive health effects, including mental well-being and the use of health-related resources [20–22]. To our knowledge, no reports are available worldwide that investigated the relationship between the number of BZRA prescriptions and social capital, using either group- or individual-level data. While some structural SDH, such as poverty and education levels, are persistent and require long-term systemic change [19], social capital, a factor that can be fostered through community-oriented interventions, may offer a more practical target for public health strategies. Identifying how BZRA prescribing patterns are associated with such social factors is a crucial first step toward optimizing prescribing practices.

Therefore, the aim of this ecological study was to further examine the relationship between the number of BZRA prescriptions and SDH in Japan, and examine the hypothesis that social capital has an impact on BZRA prescribing.

Method

This ecological study involved a secondary analysis of publicly available government statistics and research reports. We first collected prefectural-level data on BZRA prescriptions and various SDH variables, including the relative poverty rate, Gini coefficient, unemployment rate, post-high school enrollment rate, single-person household rate, number of single-parent households per 1,000 households, and social capital scores. To account for regional demographic differences, we calculated the standardized claim ratio (SCR) for BZRA prescriptions. After conducting descriptive analyses to confirm the geographic distribution of prescriptions and SDH variables, we evaluated the associations between BZRA prescription rates and these SDH variables.

SCR for BZRA

To obtain data for the number of BZD and ZD prescriptions, we used national prescription data compiled from April 2018 to March 2019 that were drawn from The National Database of Health Insurance Claims and Specific Health Checkups of Japan (NDB) Open Data [23]. The NDB Open Data details the top 100 prescribing volumes by drug efficacy category, but data for products with fewer than 1,000 prescriptions are truncated and not published.

Following the previous study [18], we calculated and standardized the number of BZRA prescriptions. While the Anatomical Therapeutic Chemical / Defined Daily Dose (ATC/DDD) methodology is widely used to measure absolute drug consumption, we employed the SCR to focus on regional disparities relative to the national average. This approach is particularly suitable for NDB Open Data due to the aforementioned truncation rule. While such data constraints can lead to a systematic underestimation of absolute volume in DDD-based metrics, the SCR remains more robust as it compares regional data against a national baseline that is subject to the same truncation rules. We used prescription data for BZD and ZD, as they are designated as hypnotics and anxiolytics by Japan’s Ministry of Health, Labour and Welfare. We summed out-of-hospital, in-hospital, and hospitalization prescription data to obtain the total number of prescriptions. Next, we calculated diazepam-equivalent doses (mg) for the total number of drugs taken by gender-age, and the total number of drugs taken by prefecture. Due to data limitations, only published data were analyzed [18]. Since it was not possible to obtain prescription data by sex-age group for each prefecture, we adjusted the sex-age distribution among prefectures, using the sex-age specific population data from the Basic Resident Register Survey [24].

Based on the adjusted data, we calculated the SCRs for BZD, ZD, and BZRA prescriptions. The SCR offers a comparison between the actual number of claims and the predicted number of claims obtained by applying the nationwide claim incidence rate by sex and age group to the target area. A SCR higher (smaller) than 100 indicates a more (less) frequent level of medical care utilization relative to the whole of Japan, even after adjusting for sex and age. The formula used for calculating SCR is shown below [25].

SCR = (Σ number of claims by age group / Σ population by age group × nationwide appearance rate of claims by age group) × 100.

Social capital

To determine the social capital score for each prefecture, we used the results from a joint research report by Shiga University and the Economic and Social Research Institute of the Cabinet Office [26], which was available on the Internet. This joint research report presented the results of a 2015 questionnaire survey conducted on a population-proportional basis in municipalities in each prefecture. The survey comprised three domains: interaction, trust, and social participation. The question items for each domain were listed in the joint research report [26] (p. 108). Answers to Questions 1–4 were measured on Likert scales ranging from four to five points, reflecting the intensity or frequency of interaction. For example, Question 1 assesses the intensity of neighborhood relationships on a 4-point scale (1 = “Consulting or cooperating,” 2 = “Daily chatting,” 3 = " Greetings only,” 4 = “None”), whereas Question 4 assesses the frequency of interaction with relatives on a 5-point scale (1 = “Daily,” 2 = “Frequent,” 3 = “Sometimes,” 4 = “Rarely,” and 5 = “Never”). The full list of question items and their specific response options are provided in Supplemental Appendix 1.

Following the scoring methodology of the original report [26], we dichotomized these responses to calculate the percentage of positive social capital indicators. Specifically, “positive answers” were defined as ratings of 1 or 2 for Questions 1–4. For Questions 6 and 7 (generalized and neighborhood trust), only a rating of 1 (“Most people can be trusted”) was defined as positive, consistent with the original report’s protocol. Binary items (Questions 5, 8, and 9) were scored as positive for “yes” responses. The domain score for each prefecture was calculated by summing the positive response rates (%) for all items within the domain and dividing the sum by the number of items.

An English translation of the question items for the three social capital domains (Interaction, Trust, and Social participation) is presented below.

< Interaction>

• Question 1: What kind of relationship do you have with your neighbors?

• Question 2: Please select the answer that best describes the number of friends you have.

• Question 3: How often do you usually socialize with friends and acquaintances outside of school or work?

• Question 4: How often do you usually interact with relatives?

• Question 5: Are you currently engaged in sports, hobbies, or recreational activities (e.g., sports, arts and culture activities, lifelong learning)?

< Trust>

• Question 6: Generally speaking, would you say that people can be trusted?

• Question 7: Can you trust the people you meet while traveling or in an unfamiliar place?

< Social participation>

• Question 8: Do you currently participate in community activities (e.g., residents’ association, neighborhood association, women’s association, senior citizens’ association, youth association, kids club)?

• Question 9: Are you currently involved in volunteers, non-profit organizations, or civic activities (e.g., regional revitalization, welfare for older people or those with disabilities, local childcare support, sports instruction, beautification, prevention of crime or disaster, environment, international cooperation activities)?

Other SDH variables

Because there are no government statistics in Japan that disclose relative poverty rates for each prefecture, we used data for relative poverty rates as reported in a previous study for our analysis [27]. We used the Gini coefficient as a measure of income inequality, which was obtained from the 2019 National Household Structure Survey [28]. The unemployment rate was obtained from the 2015 social demographic system [29]. The post-high school graduation advancement rate was based on prefectures statistics in 2018 [30]. The number of single-parent households per 1,000 households was calculated by summing the number of father-child households and the number of mother-child households drawn from the 2015 Census, dividing the sum by the total number of households, and multiplying the result by 1000 [31]. The single-person household rate was calculated by dividing the number of one-person households obtained from the 2018 Housing and Land Census by the total number of households [32].

Statistical analysis

We performed descriptive analyses of SDH variables and SCR, and graphically presented the distribution of BZRA prescriptions. Using Spearman’s rank correlation, we conducted univariable analyses to examine the associations between SDH variables and SCRs for BZRA, BZD, and ZD in each prefecture.

Next, multivariable linear regression (ecological regression) was performed using the BZRA prescription rate per prefecture as the outcome variable. To avoid overfitting and ensure model stability given the limited sample size (n = 47), the number of explanatory variables in each model was limited to three.

Based on previous studies and our univariable analysis [15–17], we selected the relative poverty rate as an economic indicator and the single-person household rate as a household structural indicator to serve as common covariates. To examine the independent association of each social capital domain, we constructed the following three analytical models:

• Model 1: Relative poverty rate + Single-person household rate + Social capital (Interaction)

• Model 2: Relative poverty rate + Single-person household rate + Social capital (Trust)

• Model 3: Relative poverty rate + Single-person household rate + Social capital (Participation)

The variance inflation factor (VIF) was used to assess multicollinearity, with a VIF > 10 defined as indicating multicollinearity. All statistical analyses were conducted using R (version 4.5.2), and a p-value of < 0.05 was considered statistically significant.

Results

Table 1 lists the generic names of BZRAs included in this study. We analyzed 20 types of BZDs and three types of ZDs that were available in the NDB Open Data. Table 2 shows the characteristics of SDH variables, with minimum, maximum, mean, and standard deviation.

Table 1.

Generic names of drugs included in the analysis

BZRA	BZD	Estazolam	Oxazolam	Tofisopam
		Flurazepam	Cloxazolam	Ethyl loflazepate
		Nitrazepam	Diazepam	Quazepam
		Flunitrazepam	Fludiazepam	Rilmazafone
		Triazolam	Bromazepam	Clotiazepam
		Brotizolam	Lorazepam	Etizolam
		Lormetazepam	Alprazolam
	ZD	Zolpidem	Zopiclone	Eszopiclone

Table 2.

Characteristics of the prefectural level SDH variables

	Minimum	Maximum	Mean	SD
Relative poverty rate	10.20	22.18	12.89	1.36
Gini coefficient	0.59	0.73	0.65	0.04
Unemployment rate	2.90	6.30	4.22	0.66
Post-high school graduation advancement rate	39.20	66.50	51.09	6.59
Number of single-parent households per 1,000 households	10.10	29.00	17.13	3.26
Single-person household rate	25.70	47.00	31.71	4.10
Social capital domain: Interaction	31.20	39.70	35.02	2.22
Social capital domain: Trust	6.20	17.70	11.80	2.15
Social capital domain: Social participation	10.20	25.70	17.36	3.54

Figure 1 presents the geographic distributions of SCRs for BZRA, BZD, and ZD at the prefecture level. The distribution patterns for BZRA and BZD were similar; both showed the highest SCRs (> 130) in Hokkaido and exhibited concentrations in the Tohoku and southwestern regions. While areas with high SCRs for ZD generally aligned with those for BZD, some regional differences were observed. For instance, the SCR for ZD was highest (> 130) in Nagasaki Prefecture. Furthermore, unlike BZD, the SCR for ZD showed no concentration in the Tohoku region, although some clustering was observed in western Japan.

Fig. 1.

Prefectural level distribution of SCR for BZRA, BZD, and ZD

Figure 2 shows scatter plots demonstrating the correlation between the SCR for BZRA and SDH variables, and Table 3 presents the Spearman’s rank correlation test results. The relative poverty rate, Gini coefficient, unemployment rate, number of single-parent households per 1,000 households, and single-person household rate were positively and significantly correlated with the SCRs for BZRA and BZD. In addition, evaluation of social capital showed a significant negative correlation for the social participation domain. The relative poverty rate, Gini coefficient, number of single-parent households per 1,000 households, and single-person household rate were positively and significantly correlated with the SCR for ZD, but unemployment rate and social participation were not.

Fig. 2.

Prefectural level correlation between SCR for BZRA and SDH variables

Table 3.

Result of prefectural level Spearman’s rank correlation test between the number of drug prescriptions and SDH

	SCR for BZRA		SCR for BZD		SCR for ZD
	r	p-value	r	p-value	r	p-value
Relative poverty rate	0.33	0.022	0.30	0.043	0.39	0.006
Gini coefficient	0.42	0.003	0.37	0.010	0.41	0.004
Unemployment rate	0.34	0.020	0.30	0.040	0.28	0.060
Post-high school graduation advancement rate	-0.27	0.070	-0.27	0.070	-0.12	0.440
Number of single-parent households per 1,000 households	0.33	0.020	0.29	0.040	0.36	0.010
Single-person household rate	0.43	0.002	0.34	0.020	0.60	< 0.001
Social capital domain: Interaction	-0.13	0.380	-0.18	0.220	0.06	0.700
Social capital domain: Trust	-0.03	0.870	-0.04	0.810	0.07	0.620
Social capital domain: Social participation	-0.31	0.040	-0.32	0.030	-0.19	0.210

Table 4 presents the results of the multivariable linear regression analysis. Across all models (Models 1–3), both the relative poverty rate and single-person household rate were significantly and independently associated with SCR for BZRA. In contrast, social participation, which showed a significant correlation in the univariable analysis, lost its statistical significance after including these socioeconomic factors in Model 3. For all models, the VIF for each variable was less than 2, indicating that there was no significant multicollinearity affecting the estimates.

Table 4.

Multivariable linear regression analysis of SCR for BZRA including social capital and socioeconomic factors: Unstandardized coefficient (95%CI)

	Model 1	Model 2	Model 3
Relative poverty rate	1.54 ( 0.38–2.71 )*	1.57 ( 0.43–2.71 )**	1.50 ( 0.32–2.68 )*
Single-person household rate	0.95 ( 0.10–1.79 )*	0.93 ( 0.22–1.64 )*	0.87 ( 0.03–1.70 )*
Social capital domain: Interaction	0.04 ( -1.49–1.58 )	−	−
Social capital domain: Trust	−	-0.36 ( -1.70–0.97 )	−
Social capital domain: Social participation	−	−	-0.15 ( -1.15–0.85 )

** p < 0.01, * p < 0.05

Model 1: Adjusted for relative poverty rate, single-person household rate, and social capital domain (interaction)

Model 2: Adjusted for relative poverty rate, single-person household rate, and social capital domain (trust)

Model 3: Adjusted for relative poverty rate, single-person household rate, and social capital domain (social participation)

Discussion

This ecological study using prefecture-level data showed there was regional diversity in the number of BZRA prescriptions in Japan. In the univariable analysis, the number of BZRA prescriptions was positively correlated with the relative poverty rate, Gini coefficient, unemployment rate, number of single-parent households per 1,000 households, and single-person household rate. In addition, social participation, which is a domain of social capital, had a negative correlation with the number of BZRA prescriptions. These findings regarding socioeconomic deprivation are consistent with previous ecological studies conducted in Japan and Europe, which reported that regional factors such as lower income and higher poverty rates are significant predictors of psychotropic drug distribution [12, 18, 33]. Notably, while social participation initially showed a negative correlation, this association was no longer significant in the multivariate analysis after adjusting for other socioeconomic factors. This suggests that the influence of social participation on BZRA prescribing may be closely intertwined with, or mediated by, broader structural vulnerabilities such as poverty and living alone.

We observed a correlation between the number of BZRA prescriptions and socioeconomic status at the prefecture level in Japan. However, our finding that the number of prescriptions was associated with SDH does not immediately lead to the conclusion that there is an association between high BZRA use and SDH. That is, we also need to consider the regional diversity in the prevalence of insomnia and psychiatric disorders. A national epidemiological survey reported no regional differences in prevalence [34, 35], though those studies divided Japan into only a few large blocks and might have missed finer regional diversity. The observed correlations suggest that regional BZRA prescribing volumes are deeply intertwined with a complex web of socioeconomic vulnerabilities, a finding consistent with a growing body of evidence linking structural social determinants to psychotropic drug utilization [12–18]. Specifically, poverty and income inequality (the Gini coefficient) may diminish social cohesion and increase chronic psychosocial stress across the community [36, 37]. Additionally, unemployment, which is closely linked to economic deprivation, has been suggested to increase the use of psychotropic medications by intensifying individual and collective psychological distress [38]. Furthermore, household instability, such as high rates of single-parent and single-person households, often leads to a lack of informal social support, which has been associated with a higher psychotropic use [39, 40]. Therefore, while the possibility of regional variation in disease prevalence cannot be entirely excluded, the diversity in socioeconomic status observed in this study (Table 2) suggests that the variation in BZRA prescription volume may reflect, among other factors, potential regional disparities in high BZRA use. This interpretation is consistent with previous evidence indicating that low socioeconomic status, including low income, unemployment, and social isolation, is significantly related to higher BZRA prescribing [17, 41, 42].

Regarding social capital, our univariable analysis demonstrated a significant negative correlation between social participation and the number of BZRA prescriptions. This finding suggests that social participation may serve as a marker of supportive community contexts, which could be associated with lower reliance on BZRAs, although this cannot be directly tested in our ecological design [43]. While social participation is conceptually distinct from social isolation, the two may share similar underlying mechanisms in their influence on health behaviors. For instance, it has been observed that socially isolated individuals often exhibit higher health risk behaviors, including the long-term use of psychotropic drugs [17, 43–45]. However, in our multivariable models, the independent association of social participation was attenuated after adjusting for relative poverty and single-person household rates. This suggests an intricate interplay between these factors rather than a simple unidirectional pathway. It is possible that socioeconomic disadvantages, such as poverty, living alone, and low social participation, function as interconnected elements of a broader social context that collectively contributes to increased BZRA demand. While our ecological design cannot disentangle specific causal pathways—such as whether social participation acts as a mediator or is confounded by structural socioeconomic factors—the results suggest that BZRA prescribing patterns are shaped by a cluster of interrelated social risks.

In this study, the “interaction” domain of social capital was not associated with the number of BZRA prescriptions in either the univariable or multivariable models. At first glance, the interaction and social participation appear similar. However, in terms of the items assessing interaction and social participation in the present study, the former measured people’s connections within close communities (e.g., relatives and friends), whereas the latter measured people’s connections in broader communities (e.g., neighborhood associations, religious groups, and volunteer organizations) [46]. Therefore, our results suggested that social participation in the community may have better effect on reducing the number of prescriptions compared with interaction.

We also found that the “trust” domain of social capital was not associated with the number of BZRA prescriptions in either univariable or multivariable analyses. Trust may play different roles in health problems in Western countries and Japan [47]. In the present study, the item covering trust included trust in “out-group” people, which is termed “generalized trust” [48]. Generalized trust has been associated with health problems in Western countries [48–50]. The mechanism linking generalized trust and health outcomes is that trust is considered a by-product of increased social participation [49]. In the Japanese cultural context, however, this relationship may differ. Japan is a relatively collectivist society with intense group connections, and human relationships are based on mutual assurance between group members rather than mutual trust [47]. This differing role of trust in the Japanese cultural context may have influenced the lack of significant differences observed in this study.

To our knowledge, this is the first study to report the relationship between social capital and the number of BZRA prescriptions. The major advantage of this study was that the findings based on NDB Open Data can be generalized to a wide population in Japan because they represent the trends in Japan as a whole. However, the present study also had several limitations. First, as this study was an ecological study and not an analysis based on individual data, causality cannot be concluded. It should be noted that this study could not evaluate the clinical appropriateness of individual BZRA prescriptions. However, in the Japanese clinical context, there has been significant public health concern regarding prolonged BZRA use and polypharmacy, leading to national efforts to optimize prescribing patterns [51]. While recognizing this limitation, identifying regional social vulnerabilities associated with higher BZRA utilization may provide a valuable foundation for developing targeted community-oriented health policies. Therefore, further individual-level studies (e.g., cohort studies) are needed to explore these mechanisms. Second, NDB Open Data provides the top 100 prescriptions by drug efficacy category, meaning prescribing data are not available for products with low prescriptions. Therefore, the total number of prescriptions might have been underestimated. However, because the number of prescriptions that ware not included in the available data was small, we believe the effect would also be small. Third, the social capital scores in this study were based on a report from joint research by Shiga University and the Cabinet Office [26]. In this analysis, we focused on three distinct domains such as interaction, trust, and participation rather than an aggregate score. This decision was informed by the perspective that social capital is a multidimensional construct comprising different elements, such as structural and cognitive components, which may relate to health behaviors in different ways [52]. While we were unable to confirm the formal validation of the specific questionnaire used in the original report, these data provided a unique opportunity to examine these diverse domains across all 47 prefectures. By exploring these domains separately, we aimed to provide a descriptive foundation for regional health social contexts. Fourth, this study utilized data from 2016 to 2018. While these data are not the most recent, socioeconomic determinants and social capital are structural factors that tend to remain stable over several years [19]. Furthermore, these findings capture regional variations just before BZRA-related policy measures were further intensified in 2018, following their gradual introduction since 2012 [51]. Identifying the relationship between SDH and BZRA use at this stage is essential for future studies to evaluate the long-term impact of these national interventions on reducing regional disparities.

Conclusions

This ecological study suggests that BZRA prescribing patterns in Japan are associated with a complex cluster of social determinants of health, particularly socioeconomic deprivation and household instability. Although the independent association of social participation was not sustained after adjusting for other factors, our findings highlight the importance of considering the multifaceted social context, which includes both structural vulnerabilities and community engagement, when addressing regional BZRA utilization. Future studies using individual-level data are warranted to further clarify these interrelationships and to inform public health strategies for optimizing BZRA prescribing.

Abbreviations

BZRA

Benzodiazepine receptor agonists

BZD

Benzodiazepines

ZD

Z drug

SDH

Social determinants of health

NDB

National Database of Health Insurance Claims and Specific Health Checkups of Japan

SCR

Standardized claim ratio

ATC/DDD

Anatomical Therapeutic Chemical / Defined Daily Dose

Authors’ contributions

SZ, TA, YS, and MM conceptualized and designed this study. SZ wrote the first draft of the manuscript. SZ obtained data and conducted the statistical analyses. SZ prepared the tables and figures. All authors critically revised the manuscript and approved the submitted version.

Funding

We received no funding to conduct this research.

Data availability

Data used in this study are available from the websites of Japanese government statistics, the Ministry of Health, Labour and Welfare, and the Cabinet Office. Prescription data for prefectures were obtained from NDB Open data (Available from: https://www.mhlw.go.jp/stf/seisakunitsuite/bunya/0000177221_00008.html; cited 1 December, 2023). Population data were obtained from the survey of population, demographics, and household number based on the basic resident register (Available from: https://www.e-stat.go.jp/stat-search/files?page=1&toukei=00200241&tstat=000001039591; cited 1 December, 2023). Data for SDH in prefectures were obtained from e-Stat general counter for government statistics (Available from: https://www.e-stat.go.jp; cited 1 December, 2023). Data for social capital were obtained from Shiga University/Cabinet Office Economic and Social Research Institute joint research report in 2016 (Available from: https://www.esri.cao.go.jp/jp/esri/prj/hou/hou075/hou075.html; cited 1 December, 2023).

Declarations

Ethics approval and consent to participate

This study did not use individual data but analysed aggregate data published in government statistics and research. Therefore, the Jikei University Ethics Committee waived the need for ethics approval, including informed consent to participate from all study participants (approval number 34–452(11609)). Ethical approval is not applicable to this study.

Consent for publication

Not applicable.

Competing interests

SZ and YS declare that they do not have any conflicts of interest relating to this study. MM and TA received lecture fees and lecture travel fees from the Centre for Family Medicine Development of the Japanese Health and Welfare Co-operative Federation. MM and TA are advisers for the Centre for Family Medicine Development Practice-Based Research Network. MM is a program director of the Jikei Clinical Research Program for Primary-care. MM’s son-in-law worked at IQVIA Services Japan K.K., which is a contract research organization and a contract sales organization. MM’s son-in-law works at Syneos Health Clinical K.K. which is a contract research organization and a contract sales organization. MM and TA are board members of the Japan Primary Care Association.