https://orcid.org/0000-0003-4613-2159
Figures
Abstract
Background
Previous studies have demonstrated geographical disparities regarding the quality of care for acute myocardial infarction (AMI). The aim of this study was two-fold: first, to calculate the proportion of patients with AMI who received primary percutaneous coronary interventions (pPCIs) by secondary medical areas (SMAs), which provide general inpatient care, as a quality indicator (QI) of the process of AMI practice. Second, to identify patterns in their trajectories and to investigate the factors related to regional differences in their trajectories.
Methods
We included patients hospitalized with AMI between April 2014 and March 2020 from the national health insurance claims database in Japan and calculated the proportion of pPCIs across 335 SMAs and fiscal years. Using these proportions, we conducted group-based trajectory modeling to identify groups that shared similar trajectories of the proportions. In addition, we investigated area-level factors that were associated with the different trajectories.
Results
The median (interquartile range) proportions of pPCIs by SMAs were 63.5% (52.9% to 70.5%) in FY 2014 and 69.6% (63.3% to 74.2%) in FY 2020. Four groups, named low to low (LL; n = 48), low to middle (LM; n = 16), middle to middle (MM; n = 68), and high to high (HH; n = 208), were identified from our trajectory analysis. The HH and MM groups had higher population densities and higher numbers of physicians and cardiologists per capita than the LL and LM groups. The LL and LM groups had similar numbers of physicians per capita, but the number of cardiologists per capita in the LM group increased over the years of the study compared with the LL group.
Citation: Itoshima H, Shin J-h, Sasaki N, Goto E, Kunisawa S, Imanaka Y (2024) Regional variations in primary percutaneous coronary intervention for acute myocardial infarction patients: A trajectory analysis using the national claims database in Japan. PLoS ONE 19(10): e0312248. https://doi.org/10.1371/journal.pone.0312248
Editor: Karam R. Motawea, Alexandria University, EGYPT
Received: June 26, 2024; Accepted: October 3, 2024; Published: October 22, 2024
Copyright: © 2024 Itoshima et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The data underlying this study are owned by the Ministry of Health, Labour and Welfare and are not publicly available due to licensing restrictions. Access to these data can be requested from the Ministry of Health, Labour and Welfare, subject to reasonable request and approval. For inquiries, please contact the Office of Research Promotion, General Affairs and Planning Division, Kyoto University (E-mail: kikaku06@mail2.adm.kyoto-u.ac.jp; Tel: +81-75-753-9301).
Funding: The current study was supported by the JPMH21IA1005 grant from the Ministry of Health, Labour and Welfare (MHLW) and the JSPS KAKENHI Grant Number 23H00448 awarded to Yuichi Imanaka. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Health equity is one of the most important concerns in public health. The World Health Organization states several health equity policy action areas and indicators to ensure access to health services of equally good quality [1]. According to Donabedian’s model, the quality of health care is divided into three components: structure, process, and outcome [2]. Process indicators refer to what health care providers do for patients. These include adherence to standard treatments advised in clinical practice guidelines, such as door-to-balloon time for patients requiring primary percutaneous coronary intervention (pPCI) and the use of antiplatelet drugs at arrival for patients with acute myocardial infarction (AMI) patients [3].
Previous studies have demonstrated geographical variation regarding the quality of care for AMI and other acute diseases [3–8]. For example, the use of beta-blockers after AMI is recommended, but the proportion of beta-blocker usage was low and geographical variation was observed [4]. pPCI is a favorable treatment strategy for patients with ST-segment elevation myocardial infarction (STEMI). The American College of Cardiology Foundation/American Heart Association guidelines 2013 for STEMI recommends that pPCI should be conducted within 90 min of arrival at the hospital [9]. The Japanese circulation society 2018 Guideline on Diagnosis and Treatment of Acute Coronary Syndrome recommends a door-to-device time shorter than 90 min as the minimum acceptable time, and total ischemic time should be as short as possible [10]. The time until pPCI (door-to-balloon time) is related to patient outcomes, and the provision of rapid treatment is one of the easiest quality of care indices to measure for AMI patients [11]. Although, it can be inferred that it is preferable for the status of pPCI to be fair across geographic regions, based on previous papers mentioned above. Not all AMI patients are eligible for pPCI, but it was performed in 72.5% of patients (83,658/115,407) in Japan [12]; however, there have been few studies on the geographic variation in pPCI for AMI patients.
The aim of this study was two-fold: first, to calculate the proportion of patients with AMI who received pPCI by secondary medical areas (SMAs), which provide general inpatient care, as a quality indicator (QI) of the process of AMI practice. Second, to identify patterns in their trajectories and investigate the factors related to regional differences in their trajectories.
Materials and methods
Data source
We conducted this study using the National Database of Health Insurance Claims and Specific Health Checkups of Japan (NDB) administrated by the Japanese Ministry of Health, Labor and Welfare (MHLW) since 2008. Japan had a universal health coverage system since the 1960s [13], and all health insurance claims data have been collected by MHLW since 2009. Therefore, this large database includes the health insurance claims data of nearly all (≥90% in 2014) medical practices such as procedures, examinations, drug prescriptions, etc. [14, 15]. Since 2012, more than 1.6 billion claim records have been registered in the NDB annually, and MHLW has given permission to use this data for research and health policy-making [14–16]. The NDB contains age, sex, diagnoses, date of consultation, date of admission, procedure, medication, and health checkup data. We used this database from January 4 2021 to July 4 2021, and didn’t access to information that could identify individual participants during or after data collection.
Study population
We included AMI patients who were hospitalized between April 1, 2014, and March 30, 2020, from the NDB. The definition of AMI patients was that they had been diagnosed with AMI (ICD-10 code: I21.x) and received blood tests for cardiac biomarkers (troponin I, etc.) or an echocardiogram or ultrasonic echocardiography or coronary angiography or chest X-ray based on the guidelines in Japan [17].
Regional units: Secondary medical areas in Japan
Japan consists of 47 prefectures and 1718 municipalities. The medical provision system is based on the medical care plan, which is made by prefectural governments [18, 19]. The Japanese medical system has a three-level hierarchy: primary medical areas, secondary medical areas, and tertiary medical areas. Primary medical areas provide primary care at a municipal level, secondary medical areas (SMAs) provide general inpatient care at several municipal levels, and tertiary medical areas are responsible for tertiary medical care at a prefectural level. SMAs have been restructured since 2014, with 335 units in June 2022. The data for SMAs before the reorganization were adjusted to those after the reorganization, and the data for 335 SMAs were used for this analysis [18, 19].
Outcomes
We calculated the proportion of pPCIs (within one day after admission) for AMI patients in each SMA (adjusted by a shrinkage estimate). We supposed that the number of pPCIs based on actual measurements was unstable due to wide variation among the AMI patients of different SMAs. To compensate for this assumption, a shrinkage estimation was introduced by taking into account the population of the SMA and converging it to the average of each prefecture. This is the same method used to calculate the English indices of deprivation (the IoD) [20].
We calculated the shrinkage estimate for pPCI for AMI patients based on the methods used for the IoD [20].
The logit SMAj for an event occurrence rj for a population nj in a jth SMA is as follows:
The estimated standard error sj is
The logit Pref in a prefecture to which jth SMA belongs is calculated using a population n and event r in a prefecture
The “shrinkage” logit in a jth SMA with low event occurrence is a weighted average.
wj is a weight given to an SMA j where the event occurrence is low, and (1 − wj) is a weight given to a prefecture to which it belongs. wj is calculated using standard error sj and variance t2.
Where t2 is the inter-SMAs variance for the k lower-layer SMAs in a prefecture, calculated as:
Statistical analysis
Trajectory analysis.
In recent studies, more attention has been given to the analysis of longitudinal data, typically using multilevel models. These model results estimate the average trajectory of a population and the variation of individual-level trajectories around this average [21, 22]. However, the identification of several typical trajectories, rather than just one average trajectory, and more detailed modeling of these individual variations, may be a more useful approach to exploring common subgroups [21–25]. These methods can be classified into:
- Growth mixture models
- Latent class growth analysis (as group-based trajectory models)
- Longitudinal latent class analysis
Several published articles examine the association between socioeconomic factors at an individual level and regional differences using trajectory model analysis [26–28]. We conducted a group-based trajectory model analysis for separating clusters according to their similarity in terms of their time courses of the proportion of pPCIs in each SMA using the latrend package (version 1.4.1) for R [29]. The optimal number of clusters was decided by Bayesian information criterion (BIC), and the mean absolute error of the fitted trajectories was weighted by cluster assignment probability (WMAE) [29]. The fitted function was selected as linear or quadratic, spline based on the trajectory of the proportion of pPCIs. A two-sided P-value <0.05 was considered statistically significant, and all analyses were conducted using R 4.1.2 (R Foundation for Statistical Computing, Vienna, Austria). All the methods were carried out in accordance with relevant Institutional guidelines and regulations.
Post-hoc analysis.
To examine geographic and socioeconomic, health care factors associated with regional differences among groups of varied pPCI trajectories, the following parameters were used as area-level factors. Geographic and socioeconomic factors included the population, the proportion of the population over 65 years of age, population density, unemployment rate, and per capita taxable income. Health care factors included: the number of total medical doctors per 100,000 population, cardiologists per 100,000 population, dentists per 100,000 population, pharmacists per 100,000 population, general hospitals per 100,000 population, emergency hospitals per 100,000 population and general hospital beds per 100,000 population [19, 28]. These variables were collected from the population census in 2015 in Japan and the 2016 survey of physicians, dentists, and pharmacists, and the 2016 survey of medical institutions from the MHWL [30–32]. Regarding healthcare-related factors, the trends in the number of medical doctors, cardiologists, general hospitals, and emergency hospitals between 2014 and 2018 were also obtained from the 2014 and 2018 surveys of physician, dentists, and pharmacists, and the 2014 and 2018 surveys of medical institutions from the MHWL [33–36]. We displayed these group characteristic variables using a median in each group. The definition of a general hospitals included all hospitals except psychiatric hospitals, and the definition of emergency hospitals included hospitals that provided any emergency medical services at the primary, secondary, or tertiary level [33, 34].
Ethical statement
The present study was approved by the Kyoto University Graduate School of Faculty of Medicine, Ethics Committee (approval number: R2215). Informed consent was not required because the data was anonymous, in accordance with the Ethical Guidelines for Medical and Health Research Involving Human Subjects, as stipulated by the Japanese Government.
Results
Overall, 469,289 cases with an AMI diagnosis were included in this study, and during the study period, there were 322,779 pPCI procedures for AMI patients among the 355 SMAs. The time trend in the shrinkage estimated proportion of pPCIs for AMI patients showed large variation in each SMA (Fig 1).
The trend on proportion of pPCI for AMI patients in 335 secondary medical areas between 2014 and 2020. These lines demonstrate the trends in the proportion of pPCI in each of the SMAs. The graph shows variation in the percentage of pPCI in each SMA.
In Fig 1, it can be seen that linear and quadratic terms did not fit well. Therefore, the spline function was used. We also calculated the optimal number of trajectories between 1 to 5 by BIC and WMAE, then determined that a suitable number of trajectories was 4, based on BIC and WMAE (Fig 2). Model adequacy was as follows: Weighted residual sum of squares (WRSS) was 17.593, WMAE was 0.062 and BIC was -4398.594. The 335 SMAs were classified into four groups: group A had 208 SMAs (62.0%), group B had 68 SMAs (20.0%), group C had 16 SMAs (5.0%), and group D had 43 SMAs (13.0%). Our model also classified similar trajectories well (Fig 3). Each group was named based on its trajectory: group A was the high to high group (HH group), group B was the middle to middle group (MM group), group C was the low to middle group (LM group), and group D was the low to low group (LL group). S1 Fig indicated that each secondary medical area belongs to which group.
Our analysis calculated the optimal number of trajectories between 1 to 5 by Bayesian information criterion (BIC) and the mean absolute error of the fitted trajectories was weighted by cluster assignment probability (WMAE), then determined that a suitable number of trajectories was 4, based on BIC and WMAE.
The 335 secondary medical areas (SMAs) were classified into four groups: group A had 208 SMAs (62.0%), group B had 68 SMAs (20.0%), group C had 16 SMAs (5.0%), and group D had 43 SMAs (13.0%).
The characteristics of each group, such as its population, population density, unemployment rate, and the number of physicians per 100,000 capita, are reported in Table 1. Groups A and B had a greater population, population density, and medical professions than groups C and D. Supplementary figures demonstrate the distribution of the population, the proportion of the population over 65 years of age, the total number of medical doctors per 100,000 population, cardiologists per 100,000 population, general hospitals per 100,000 population, and general hospital beds per 100,000 population in each group (S2–S7 Figs).
The trends in the number of physicians, cardiologists, general hospitals, and emergency hospitals between 2014 and 2018 are shown in Fig 4. Each group had no change regarding the number of hospitals; however, cardiologists increased only in the LM group (4 to 6 per 100,000 population).
Each group had no change regarding the number of hospitals; however, cardiologists increased only in the LM group (4 to 6 per 100,000 population). * All value in Y axis is median.
Discussion
We found geographical differences in the proportion of pPCIs for AMI patients and their trends over time in Japanese SMAs between 2014 and 2020. Furthermore, our trajectory modeling analysis displayed four groups of SMAs using group-based trajectory analysis of the trends in the proportion of pPCIs for AMI patients and assessed whether area-level factors related to this distribution. The groups were classified into those with a medium to high proportion of pPCIs over time (the HH and the MM groups), those with a low proportion over time (the LL group), and those initially with a low but gradually increasing proportion over time (the LM group). In terms of geographic, socioeconomic, and healthcare-related factors, the HH and MM groups were those with large populations and abundant healthcare resources, while the LL and LM groups were those with small populations and few healthcare resources. Compared to the LL group, the total number of physicians remained the same in the LM group, but the number of cardiologists increased over time in the LM group.
Strengths and weaknesses of the study
Regional disparities in the proportion of AMI patients who received pPCIs were assessed in this study form the NDB, which contains all medical services covered by the National Health Insurance Scheme, including pPCIs. Therefore, our study could analyze national-level regional differences in this topic using complete longitudinal data. The time to balloon is a quality indicator of pPCI; therefore, it might be desirable to complete pPCI procedures in a region close to the patient’s home [37, 38]. The regional difference in the proportion of AMI patients who received a pPCI indicated regional disparities in the quality of health care for AMI patients. There were several studies discussing quality indicators related to AMI, including pPCI and regional differences [4–7, 37–39]. Several cross-sectional studies have discussed socioeconomic factors, health care resources, and healthcare performance regarding AMI practice [40–42]. This study examined changes in quality indicators over seven years and identified a group that initially had a low proportion of pPCIs, which improved over time (the LM group).
Also, we found a difference in the trajectory of the proportion of pPCIs among lower groups at the beginning of the study (the LL group and LM group), which might be related to the number of cardiologists increasing between 2014 and 2018 in the LM group. Studies examining the density of specialists and outcomes in a surgical field have shown that colorectal cancer mortality is higher in areas with fewer colorectal specialists and prostate cancer mortality is lower in areas with radiation oncologists than in areas without [43, 44]. However, few studies have examined the relationship between an increase in cardiologists and improvements in the quality of health care regarding AMI practice. The reason behind the increasing number of cardiologists in the LM group was unclear from our data. One possible explanation of this phenomenon might occur the centralization of cardiologists in the LM group. Park et al. indicated that the centralization of cardiologists at a hospital level is associated with lower in-hospital mortality; therefore, the centralization of cardiologists might occur in this group [45]. However, the group-based trajectory model does not assume intra-group varieties, we could not discuss the varieties among SMAs in the LM group. This study had several limitations. First, our study only evaluated the proportion of pPCI procedures in SMAs, which is a process indicator for pPCI, but could not evaluate other process indicators such as the administration of aspirin before pPCI or, the proportion of pPCIs with GFR documentation [37–39]. Although outside the scope of this study, other process measures are also important for assessing the quality of care for AMI. Further implementation research is needed on these other process measures. Second, due to using the claim database, the diagnosis of AMI had the risk of misclassification. We minimized this possibility by defining the diagnosis of AMI and combing examinations for AMI based on the guidelines. Third, the study examined the change in the number of cardiologists, which could be a time-dependent confounding factor. We did not take this point into account; therefore, it is unclear whether the increased proportion of pPCIs in the LM group was influenced by the number of cardiologists. As mentioned earlier, possible reasons for the increase in cardiologists in the LM group compared to the LL group, such as hospital centralization [45], are unclear from the present data. Further analysis using more recent data is required to elucidate this observation. We also used geographic, socioeconomic, and healthcare factors (that could be associated with regional differences) to examine the results of our trajectory analysis. However, we were unable to examine factors such as the educational environment at the regional level or the presence of healthcare policies that affect the number of physicians. Therefore, such factors may be residual confounders.
Meaning of the study: Possible explanations and implications for clinicians and policymakers
This study has shown regional disparities in the proportion of pPCIs for AMI patients. Trajectory analysis revealed that the proportion of pPCIs was stable over time in three groups and increased yearly in one group. Policymakers should consider such regional differences in the trajectory of this proportion to plan a medical system for AMI practice. This might help to assign the right medical resources in each region equally.
Unanswered questions and future research
Our study primarily assessed regional disparities in the proportion of pPCIs that were related to process indicators. Therefore, further research is required to examine the association between the proportion of pPCIs and patient outcome indicators such as cardiovascular death.
Conclusion
The trajectory of the proportion of pPCIs for AMI patients identified groups of SMAs. Among the four groups, the LM group showed an increasing trend in the proportion of pPCIs, whereas the three other groups showed relatively stable trends. The LM group revealed that the number of cardiologists had increased over time in its regions.
Supporting information
S1 Fig. The regional variation of trajectory on the proportion of pPCI for AMI patients.
Group A (high to high) indicated that the proportion of pPCI for AMI patients was high from 2014 to 2020, group B (middle to middle) indicated that the proportion of pPCI for AMI patients was moderate from 2014 to 2020, group C (low to middle) indicated that the proportion of pPCI for AMI patients had increased from 2014 to 2020, and group D (low to low) indicated that the proportion of pPCI for AMI patients was low from 2014 to 2020.
https://doi.org/10.1371/journal.pone.0312248.s001
(TIF)
S2 Fig. The population distribution in each group.
https://doi.org/10.1371/journal.pone.0312248.s002
(TIF)
S3 Fig. The distribution of the proportion of the population aged 65 years and older by each group.
https://doi.org/10.1371/journal.pone.0312248.s003
(TIF)
S4 Fig. The distribution of the total number of medical doctors per 100,000 population in each group.
https://doi.org/10.1371/journal.pone.0312248.s004
(TIF)
S5 Fig. The distribution of the number of cardiologists per 100,000 population in each group.
https://doi.org/10.1371/journal.pone.0312248.s005
(TIF)
S6 Fig. The distribution of the number of general hospitals per 100,000 population in each group.
https://doi.org/10.1371/journal.pone.0312248.s006
(TIF)
S7 Fig. The distribution of the number of general hospital beds per 100,000 population in each group.
https://doi.org/10.1371/journal.pone.0312248.s007
(TIF)
References
- 1.
World Health Organization. Regional Office for Europe. Health Equity Policy Tool. A framework to track policies for increasing health equity in the WHO European Region. [Cited 2022. June 10]. https://www.euro.who.int/__data/assets/pdf_file/0003/410457/FINAL-20190812-h1905-policy-tool-en.pdf.
- 2. DONABEDIAN A. Evaluating the Quality of Medical Care. The Milbank Quarterly. 2005 Nov 9;83(4):691–729. Available from: pmid:16279964
- 3. He Y, Wolf RE, Normand SLT. Assessing geographical variations in hospital processes of care using multilevel item response models. Health Services and Outcomes Research Methodology. 2010 Sep 17;10(3–4):111–33. Available from: pmid:33981181
- 4. Krumholz HM, Radford MJ, Wang Y, Chen J, Heiat A, Marciniak TA. National Use and Effectiveness of β-Blockers for the Treatment of Elderly Patients After Acute Myocardial Infarction. JAMA. 1998 Aug 19;280(7):623. Available from: pmid:9718054
- 5. O’Connor GT, Quinton HB, Traven ND, Ramunno LD, Dodds TA, Marciniak TA, et al. Geographic Variation in the Treatment of Acute Myocardial Infarction. JAMA. 1999 Feb 17;281(7):627. Available from: pmid:10029124
- 6. Jencks SF, Cuerdon T, Burwen DR, Fleming B, Houck PM, Kussmaul AE, et al. Quality of Medical Care Delivered to Medicare Beneficiaries. JAMA. 2000 Oct 4;284(13):1670. Available from: pmid:11015797
- 7. Kaul P, Peterson ED. The Cardiovascular World Is Definitely Not Flat. Circulation. 2007 Jan 16;115(2):158–60. Available from: pmid:17228010
- 8. Beatty AL, Truong M, Schopfer DW, Shen H, Bachmann JM, Whooley MA. Geographic Variation in Cardiac Rehabilitation Participation in Medicare and Veterans Affairs Populations. Circulation. 2018 May;137(18):1899–908. Available from: pmid:29305529
- 9. O’Gara PT, Kushner FG, Ascheim DD, Casey DE, Chung MK, de Lemos JA, et al. 2013 ACCF/AHA Guideline for the Management of ST-Elevation Myocardial Infarction: Executive Summary. Journal of the American College of Cardiology. 2013 Jan;61(4):485–510. Available from: pmid:23256913
- 10. Kimura K, Kimura T, Ishihara M, Nakagawa Y, Nakao K, Miyauchi K, et al. JCS 2018 Guideline on Diagnosis and Treatment of Acute Coronary Syndrome. Circulation Journal. 2019 Apr 25;83(5):1085–196. Available from: pmid:30930428
- 11. Park J, Choi KH, Lee JM, Kim HK, Hwang D, Rhee T, et al. Prognostic Implications of Door-to-Balloon Time and Onset-to-Door Time on Mortality in Patients With ST-Segment–Elevation Myocardial Infarction Treated With Primary Percutaneous Coronary Intervention. Journal of the American Heart Association. 2019 May 7;8(9). Available from: pmid:31041869
- 12. Uemura S, Okamoto H, Nakai M, Nishimura K, Miyamoto Y, Yasuda S, et al. Primary Percutaneous Coronary Intervention in Elderly Patients With Acute Myocardial Infarction—An Analysis From a Japanese Nationwide Claim-Based Database—. Circulation Journal. 2019 May 24;83(6):1229–38. Available from: pmid:31019165
- 13.
Sakamoto H, Rahman M, Nomura S, Okamoto E, Koike S, et al. Japan health system review. World Health Organization. Regional Office for South-East Asia; 2018. [Cited 2022 June 18]. https://apps.who.int/iris/handle/10665/259941.
- 14. Matsuda S, Fujimori K. The Claim Database in Japan. Asian Pacific Journal of Disease Management. 2014;6(3–4):55–9. Available from:
- 15. Kato G. History of the secondary use of National Database of Health Insurance Claims and Specific Health Checkups of Japan (NDB). Trans Jpn Soc Med Biol Eng. 2017;55:143–150. Available from:
- 16. Hirose N, Ishimaru M, Morita K, Yasunaga H. A review of studies using the Japanese National Database of Health Insurance Claims and Specific Health Checkups. Annals of Clinical Epidemiology. 2020;2(1):13–26. Available from:
- 17.
Guidelines for the management of patients with ST-elevation acute myocardial infarction (JCS 2013). The Japanese Circulation Society; 2013. [Cited 2022 June 18]. http://saigaiin.sakura.ne.jp/sblo_files/saigaiin/image/320E382ACE382A4E38389E383A9E382A4E383B3.pdf.
- 18.
Ministry of Health, Labour, and Welfare, Japan. Regional medical plan. [Cited 2022 June 18]. https://www.mhlw.go.jp/stf/seisakunitsuite/bunya/kenkou_iryou/iryou/iryou_keikaku/index.html.
- 19. Watanabe S, Shin J ho, Goto E, Kunisawa S, Imanaka Y. Factors Associated With Regional Differences in Healthcare Quality for Patients With Acute Myocardial Infarction in Japan. medRxiv [Preprint]. 2022 [Cited 2022 June 18]. Available from: http://dx.doi.org/10.1101/2022.05.20.22275402
- 20.
The English Indices of Deprivation 2019. Technical report. Ministry of Housing, Communities and Local Government; September 2019. [Cited 2022 June 5]. https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/833951/IoD2019_Technical_Report.pdf.
- 21. Nagin DS, Odgers CL. Group-Based Trajectory Modeling in Clinical Research. Annual Review of Clinical Psychology. 2010 Mar 1;6(1):109–38. Available from: pmid:20192788
- 22. Herle M, Micali N, Abdulkadir M, Loos R, Bryant-Waugh R, Hübel C, et al. Identifying typical trajectories in longitudinal data: modelling strategies and interpretations. European Journal of Epidemiology. 2020 Mar;35(3):205–22. Available from: pmid:32140937
- 23. Nguena Nguefack HL, Pagé MG, Katz J, Choinière M, Vanasse A, Dorais M, et al. Trajectory Modelling Techniques Useful to Epidemiological Research: A Comparative Narrative Review of Approaches. Clinical Epidemiology. 2020 Oct;Volume 12:1205–22. Available from: pmid:33154677
- 24. Den Teuling NGP, Pauws SC, van den Heuvel ER. A comparison of methods for clustering longitudinal data with slowly changing trends. Communications in Statistics—Simulation and Computation. 2021 Jan 19;52(3):621–48. Available from:
- 25. Serra L, Farrants K, Alexanderson K, Ubalde M, Lallukka T. Trajectory analyses in insurance medicine studies: Examples and key methodological aspects and pitfalls. Mockridge J, editor. PLOS ONE. 2022 Feb 11;17(2):e0263810. Available from: pmid:35148351
- 26. Kingston A, Davies K, Collerton J, Robinson L, Duncan R, Kirkwood TBL, et al. The enduring effect of education-socioeconomic differences in disability trajectories from age 85 years in the Newcastle 85+ Study. Archives of Gerontology and Geriatrics. 2015 May;60(3):405–11. Available from: pmid:25747850
- 27. Curtis S, Cunningham N, Pearce J, Congdon P, Cherrie M, Atkinson S. Trajectories in mental health and socio-spatial conditions in a time of economic recovery and austerity: A longitudinal study in England 2011–17. Social Science & Medicine. 2021 Feb;270:113654. Available from: pmid:33445118
- 28. Satokangas M, Lumme S, Arffman M, Keskimäki I. Trajectory modelling of ambulatory care sensitive conditions in Finland in 1996–2013: assessing the development of equity in primary health care through clustering of geographic areas–an observational retrospective study. BMC Health Services Research. 2019 Sep 4;19(1). Available from: pmid:31484530
- 29.
Den Teuling N. Demonstration of latrend package. [Cited 2022 June 5]. https://mran.microsoft.com/snapshot/2021-03-14/web/packages/latrend/vignettes/demo.html.
- 30.
Population census 2015. [Cited 2022 July 9]. https://www.e-stat.go.jp/en/stat-search/files?page=1&layout=datalist&toukei=00200521&tstat=000001080615&cycle=0&tclass1=000001089055&tclass2=000001089056.
- 31.
Survey 2016 of medical facilities. Ministry of Health, Labour, and Welfare. [Cited 2022 July 9]. https://www.e-stat.go.jp/stat-search/files?page=1&layout=datalist&toukei=00450021&tstat=000001030908&cycle=7&year=20160&month=0&tclass1=000001106456&tclass2=000001106459.
- 32.
Survey 2016 of physician, dentists and pharmacists. Ministry of Health, Labour, and Welfare. [Cited 2022 July 9]. https://www.e-stat.go.jp/stat-search/files?page=1&toukei=00450026&tstat=000001030962&year=20160.
- 33.
Survey 2014 of physician, dentists and pharmacists. Ministry of Health, Labour, and Welfare. [Cited 2022 July 9]. https://www.e-stat.go.jp/stat-search/files?page=1&toukei=00450026&tstat=000001030962&year=20140.
- 34.
Survey 2018 of physician, dentists and pharmacists. Ministry of Health, Labour, and Welfare. [Cited 2022 July 9]. https://www.e-stat.go.jp/stat-search/files?page=1&layout=datalist&toukei=00450026&tstat=000001135683&cycle=7&tclass1=000001135684&tclass2=000001135687&tclass3val=0.
- 35.
Survey 2014 of medical facilities. Ministry of Health, Labour, and Welfare. [Cited 2022 July 9]. https://www.e-stat.go.jp/stat-search/files?page=1&layout=datalist&toukei=00450021&tstat=000001030908&cycle=7&year=20140&tclass1=000001077195&tclass2=000001077198&stat_infid=000031336380&tclass3val=0.
- 36.
Survey 2018 of medical facilities. Ministry of Health, Labour, and Welfare. [Cited 2022 July 9]. https://www.e-stat.go.jp/dbview?sid=0003404862.
- 37. Khare RK, Mark Courtney D, Kang R, Adams JG, Feinglass J. The Relationship Between the Emergent Primary Percutaneous Coronary Intervention Quality Measure and Inpatient Myocardial Infarction Mortality. Academic Emergency Medicine. 2010 Jul 29;17(8):793–800. Available from: pmid:20670315
- 38. Chui PW, Parzynski CS, Nallamothu BK, Masoudi FA, Krumholz HM, Curtis JP. Hospital Performance on Percutaneous Coronary Intervention Process and Outcomes Measures. Journal of the American Heart Association. 2017 May 5;6(5). Available from: pmid:28446493
- 39. Ko DT, Wijeysundera HC, Zhu X, Richards J, Tu JV. Canadian quality indicators for percutaneous coronary interventions. Canadian Journal of Cardiology. 2008 Dec;24(12):899–903. Available from: pmid:19052669
- 40. Matetic A, Bharadwaj A, Mohamed MO, Chugh Y, Chugh S, Minissian M, et al. Socioeconomic Status and Differences in the Management and Outcomes of 6.6 Million US Patients With Acute Myocardial Infarction. The American Journal of Cardiology. 2020 Aug;129:10–8. Available from: pmid:32576369
- 41. Wadhera RK, Bhatt DL, Kind AJH, Song Y, Williams KA, Maddox TM, et al. Association of Outpatient Practice-Level Socioeconomic Disadvantage With Quality of Care and Outcomes Among Older Adults With Coronary Artery Disease. Circulation: Cardiovascular Quality and Outcomes. 2020 Apr;13(4). Available from: pmid:32228065
- 42. Daley S, Kajendrakumar B, Nandhakumar S, Personett C, Sholes M, Thapa S, et al. County-Level Socioeconomic Status Adjustment of Acute Myocardial Infarction Mortality Hospital Performance Measure in the U.S. Healthcare. 2021 Oct 22;9(11):1424. Available from: pmid:34828471
- 43. Aneja S, Yu JB. The impact of county-level radiation oncologist density on prostate cancer mortality in the United States. Prostate Cancer and Prostatic Diseases. 2012 Jul 24;15(4):391–6. Available from: pmid:22824828
- 44. Albarrak J, Firouzbakht A, Peixoto RD, Ho MY, Cheung WY. Correlation between County-Level Surgeon Density and Mortality from Colorectal Cancer. Journal of Gastrointestinal Cancer. 2016 May 25;47(4):389–95. Available from: pmid:27221330
- 45. Park S, Lee J, Ikai H, Otsubo T, Imanaka Y. Decentralization and centralization of healthcare resources: Investigating the associations of hospital competition and number of cardiologists per hospital with mortality and resource utilization in Japan. Health Policy. 2013 Nov;113(1–2):100–9. Available from: pmid:23830562