Continuity of primary care and avoidable hospitalization in a young population with asthma: a population-based cohort study Authors: Iyn-Hyang Lee, Sangwan Kim, Eunjung Choo, Eun Jin Jang, Nam Kyung Je Published: 2024 ABSTRACT Existing evidence regarding the impact of continuity of care (COC) in asthma patients is limited, and its quality is low to moderate. This study aimed to investigate associations between relational COC within primary care and asthma-related hospitalizations in children using a robust methodology. This study is a population-based cohort study that utilized a national claims database in South Korea. The study comprised 136,296 individuals under 20 years old who were newly diagnosed with asthma between 2015 and 2016. These were classified into high, medium, and low continuity groups based on the continuity of care index (COCI). The primary outcome measure was the incidence of asthma-related hospitalizations. During a two-year follow-up period, 10,922 patients (8.01%) were hospitalized: 2520 (5.59%) in the high-continuity group, 3188 (6.98%) in the medium-continuity group, and 3867 (8.48%) in the low-continuity group. After adjusting for confounding covariates, patients in the low- and medium-continuity groups exhibited significantly higher risks of hospital admission (adjusted hazard ratio (aHR) = 1.30 [95% confidence interval = 1.24–1.37] and aHR = 1.14 [1.08–1.20], than those in the high-continuity group. Sensitivity and subgroup analyses findings were consistent. In a young population with asthma increased continuity of primary care was associated with less hospitalization.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-024-78057-3. Introduction Continuity of care (COC) is a widely acknowledged fundamental component of primary care1–3, representing a cohesive sequence of healthcare interactions that are aligned with patients’ medical and personal needs. COC is comprised of three interrelated components, namely, informational, management, and relational. Informational continuity involves the exchange of medical histories among current and previous healthcare providers. Management continuity entails coordinating and harmonizing care according to shared treatment plans across providers, and relational continuity refers to establishing an enduring therapeutic bond between patients and their healthcare providers2–5. Of these, relational continuity is of greatest importance when managing chronic conditions, as the establishment of a prolonged doctor-patient relationship can facilitate communication, enhance care provider understanding of patients’ medical backgrounds2, improve the management of chronic conditions, and encourage the establishment of long-term disease-monitoring mechanisms6. Recently, the significance of COC in chronic conditions has attracted research interest7–10. Research to date shows that maintaining strong, ongoing relationships between patients and healthcare providers is associated with a lower likelihood of premature death in individuals with ambulatory care-sensitive conditions (ACSCs)11. Such relationships can also help prevent health deterioration in patients with dyslipidemia12, decrease the need for healthcare services, such as hospitalizations and emergency room visits12–14, and lower medical costs for those with chronic illnesses12,15. Although evidence across various diseases is accumulating, disease specific data remains limited. Asthma is an ACSC, and effective management and treatment in an ambulatory setting can prevent hospital admissions16,17. In the case of asthma, a recent systematic review showed that existing evidence on the effect of COC in asthma patients had low to moderate certainty11. In addition, previous COC studies have been criticized for methodological shortcomings18,19, highlighting the need for additional research. In this study, national claims data were used to assess the relationship between COC in primary care and hospital admissions using a robust study design. Specifically, we concentrated on younger patients aged 0–19 years, who particularly benefit from stable COC-based management of their conditions throughout life. We hypothesized that patients with higher relational care continuity with their primary care physician would achieve better clinical outcomes at lower costs than those without such continuity. Methods Study design This population-based cohort study was conducted and reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline20. Data sources Anonymized national insurance claims data between 2014 and 2020 provided by the Korea National Health Insurance Service (KNHIS) were analyzed. The KNHIS database contains details of all claims made by Korean residents covered by National Health Insurance (NHI) and Medical Aid (MedAid)21. KNHIS claims data include de-identified patient demographic information, diagnoses, and details of all medical services provided and medications dispensed22. Study timeframe Figure 1 demonstrates the study period, which ranged from 2014 to 2020. The index date was defined as the earliest date of the claims made by patients diagnosed with asthma in 2015 or 2016. The pre-index year was defined as the 12-month period before the index date. The exposure period was defined as the 2 years following the index date, and the outcome period referred to time after the end of the exposure period. Patients were followed from the end of the exposure period until outcomes were determined or the data follow-up was lost, whichever occurred first.Fig. 1Defined study periods. Defined study periods. Study population Patients diagnosed with asthma in 2015 or 2016, who were less than 20 years old on the index date, were eligible for enrollment. A broad age range was chosen to study asthma management across the pediatric and adolescent spectrum, including transitional care. While a broad age range was included, eligibility for enrollment required a minimum of four ambulatory visits during the exposure period and at least one prescription for an inhaler or patch for treating asthma on the index date to exclude possible differential diagnoses19,23,24. Asthma was identified based on the International Classification of Diseases, 10th revision (ICD-10) using diagnosis codes J45-46 (Supplementary Table 1). To ensure the inclusion of only recently diagnosed patients, individuals diagnosed with asthma during one year before the index date (the pre-index year) were excluded. Those who visited tertiary facilities or experienced hospitalizations or ED visits during the exposure period were also excluded. After exclusions, the final analytic cohort contained 136,296 unique patients. Measuring continuity as exposure Continuity of primary care is conceptualized as the extent to which a patient’s visits are concentrated among primary care doctors. Primary care is defined differently across countries25. In the present study, primary care was defined as community-based care, including outpatient care, provided by facilities other than tertiary hospitals. The South Korean healthcare system is characterized by the limited gatekeeping function of primary care26. Patients enjoy universal coverage and are required to provide a mandatory referral document only for tertiary hospitals. In this context, patients place considerable importance on the proximity to medical facilities. Thus, if patients are located within the catchment area of a semi-sized hospital, they may receive ambulatory services at that facility. The Bice and Boxerman Continuity of Care Index (COCI) was used to measure COC23. COCI is a dispersion index, and a useful metric for patients who might visit different providers. COCI is predominantly employed for analyzing claims data27 and is suggested in regions such as South Korea, where patients have considerable autonomy in selecting their doctor as mentioned above. Bice and Boxerman COCIs were calculated using the following formula23,27.\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text{COCI}=\frac{{\sum }_{i=1}^{M}{n}_{i}^{2}-N}{N(N-1)}$$\end{document}Where N is the total number of visits made by a patient to a doctor, ni is the number of patient visits to provider i, and M is the number of potentially available providers. COCI has a value of unity (1) if a patient always visits one specific health provider and a value of 0 if the patient visits different providers at each visit. Outcome measures The primary outcome measure was the incidence of asthma-related hospital admissions during the outcome period because ACSC hospitalization rates can serve as an indicator of primary care quality. Secondary outcomes were the incidences of asthma-related emergency department (ED) visits, systemic corticosteroid uses, and asthma-related costs. Systemic corticosteroid use was defined as corticosteroid use in tablet or injection form and was investigated to evaluate asthma exacerbation. Asthma-related medical costs were defined as average annual medical costs, that is, the sum of the costs of claims identified using ICD-10 diagnosis codes J45-46 during the 4-year study period. Covariates Covariates included individual characteristics such as sex and age. NHI contributions were classified as high, moderate, or low and used as proxies of patients’ economic circumstances. The KNHI has two types of health programs: NHI and MedAid. About 97% of the population is covered by NHI and 3% by MedAid28. Locations of residences on index dates were classified as large urban (metropolitan cities), small urban (other cities), or rural. Large and small urban and rural areas were classified based on population densities. Elixhauser comorbidity indices (ECIs) were computed as proxies of patients’ health statuses29 based on diagnoses obtained from outpatient and inpatient records during the pre-index year. In addition, we considered whether patients had a diagnosis of allergic rhinitis, atopic dermatitis, hypertension, diabetes, dyslipidemia, cancer, osteoarthritis or rheumatoid arthritis during the pre-index year, or were being prescribed systemic corticosteroids during the exposure period. Frequent doctor visits were defined as outpatient clinic visits ranking at the 90th percentile or higher during the exposure period. Disability status was defined as yes or no on index dates. For reference purposes, we also collected average annual asthma-related medical costs during the exposure period. All ICD-10 and medication codes used for data analysis are reported in Supplementary Table 1. Configuration of the continuity cohort Interim analysis was conducted to assess the distribution of continuity scores for primary care doctor visits during the exposure period. To the best of our knowledge, there is no generally accepted cut-off value for high or low COCI. Thus, we categorized patients based on our analysis and assigned those with a continuity score of ≥ 0.67 during the exposure period to the high-continuity group (the HC group), those with scores of < 0.67 and ≥ 0.4 to the medium-continuity group (the MC group), and those with scores of < 0.4 to the low-continuity group (the LC group). These groups contained 45,056 (33.1%), 45,650 (33.5%), and 45,590 (33.5%) patients, respectively (Fig. 2).Fig. 2Continuity score distribution. COCI continuity of care index, ED emergency department. Continuity score distribution. COCI continuity of care index, ED emergency department. Statistical analysis Baseline characteristics and healthcare utilization were summarized using means, standard deviations or medians, and interquartile range (IQR) for continuous variables and frequencies and proportions for categorical variables. Inter-group comparisons of continuous and categorical variables were performed using the Kruskal–Wallis test or the Chi-squared test, respectively. Adjusted hazard ratios (aHRs) and 95% confidence intervals (CIs) for the effects of COC on hospital admission and ED visits due to asthma were assessed using a Cox proportional hazard regression model with adjustments for baseline characteristics (age, sex, insurance contribution level, insurance program, urbanization level of residence, disability, ECI score, coexistence of major comorbidities, frequent doctor visitor, and systemic corticosteroid use). Systemic corticosteroid use was assessed using a logistic regression model, and results are presented as odds ratios (ORs) and 95% CIs. Asthma-associated costs were estimated using a gamma distribution and a log link function, and results are reported as exponentiated coefficients. Basic features during the exposure period, including age, sex, insurance contribution level, insurance program, urbanization level of residence, disability, ECI score, major comorbidities, frequent doctor visitor or not, systemic corticosteroid use and annual average medical costs, were adjusted for as covariates. Individuals with missing information about insurance contributions or residence location were not excluded, and missing values were considered an independent category in the statistical models. Three continuity groups were compared using Kaplan–Meier plots. Restricted cubic spline (RCS) plots were used to access nonlinear relationships between continuous COCI and outcomes30,31. Optimal cubic functions for the adjusted HRs of each 3rd COCI percentile were estimated using COCI = 1 as the reference32. For sensitivity analysis, COCIs were calculated every two years following the index date to construct longitudinal data, which were then analyzed using the generalized estimating equations (GEE) method. Subgroup analyses were performed on basic characteristics. The analysis was conducted using SAS Ver. 9.4 (SAS Institute Inc, Cary, NC, USA), and statistical significance was accepted for p values < 0.05. Ethics approval and consent to participate This study was performed in accordance with the Declaration of Helsinki and was reviewed beforehand by the Institutional Review Board of Yeungnam University (Approval No. 7002016-E-2022-002-01). The requirement for participant consent was waived because the study was conducted using anonymous claims data provided for research purposes by the KNHIS. Results Baseline characteristics of the study population The selection process is shown in Fig. 3. A total of 136,296 patients were eligible for analysis.Fig. 3Selection of the study patients. Selection of the study patients. Table 1 shows the baseline characteristics of study population. Mean age was 3.6 years (SD 3.1), and 79.7% were younger than 6. Patients were more likely to come from households paying high or moderate insurance contributions (38.3% and 46.1% respectively) and living in large urban areas (55.6%). The majority were National Health Insurance (NHI) beneficiaries (99.1%). Based on the ECI scores, the majority (83.6%) of patients had no major diseases besides asthma. Comorbidities not included in ECI configuration; allergic rhinitis was present in 86.6%. Approximately a third of the study patients (33.3%) were prescribed systemic corticosteroid agents during the exposure period. A comparison of the three continuity groups showed that patients in the HC group were more likely to be older, living in large urban areas, and frequent visitors to healthcare providers.Table 1Baseline characteristics of study population.Variable, n (%)All patientsN = 136,296Continuity groupLowN = 45,590MediumN = 45,650HighN = 45,056p-valueAgeMedian (IQR)3 (1, 5)2 (1, 4)3 (1, 5)3 (2, 6)< 0.0001Younger than 6108,687 (79.7)40,106 (88.0)36,710 (80.4)31,871 (70.7)< 0.00016–younger than 1224,006 (17.6)5075 (11.1)8010 (17.6)10,921 (24.2)12–younger than 203603 (2.6)409 (0.9)930 (2.0)2264 (5.0)SexWomen66,341 (48.7)21,968 (48.2)22,147 (48.5)22,226 (49.3)0.0019Insurance programNHI135,090 (99.1)45,299 (99.4)45,275 (99.2)44,516 (98.8)< 0.0001MedAid1206 (0.9)291 (0.6)375 (0.8)540 (1.2)Insurance contributionsaHigh52,185 (38.3)16,903 (37.1)17,513 (38.4)17,769 (39.4)< 0.0001Moderate62,769 (46.1)21,499 (47.2)21,010 (46.0)20,260 (45.0)Low18,394 (13.5)5880 (12.9)6215 (13.6)6299 (14.0)Urbanization level of residenceLarge urban area75,812 (55.6)23,713 (52.0)25,420 (55.7)26,679 (59.2)< 0.0001Small urban area46,515 (34.1)16,773 (36.8)15,542 (34.1)14,200 (31.5)Rural area13,969 (10.3)5104 (11.2)4688 (10.3)4177 (9.3)DisabilityYes88 (0.06)24 (0.05)26 (0.06)38 (0.08)0.1259Elixhauser comorbidity indexMean ± SD1.20 ± 0.501.21 ± 0.511.20 ± 0.501.19 ± 0.49< 0.00011113,945 (83.6)37,716 (82.7)38,137 (83.5)38,092 (84.5)< 0.0001218,263 (13.4)6394 (14.0)6144 (13.5)5725 (12.7)3+4088 (3.0)1480 (3.3)1369 (3.0)1239 (2.8)ComorbidityAllergic rhinitis118,000 (86.6)39,824 (87.4)39,622 (86.8)38,554 (85.6)< 0.0001Atopic dermatitis34,304 (25.2)12,356 (27.1)11,630 (25.5)10,318 (22.9)< 0.0001Cancer91 (0.07)20 (0.04)30 (0.07)41 (0.09)0.0229Diabetes mellitus268 (0.2)69 (0.15)95 (0.2)104 (0.2)0.0207Dyslipidemia1152 (0.9)352 (0.8)398 (0.9)402 (0.9)0.1063Hypertension100 (0.07)28 (0.06)35 (0.08)37 (0.08)0.4900Osteoarthritis437 (0.3)130 (0.3)132 (0.3)175 (0.4)0.0079Rheumatoid arthritis60 (0.04)14 (0.03)19 (0.04)27 (0.06)0.1061Systemic corticosteroids during the exposure periodYes45,411 (33.3)14,639 (32.1)14,887 (32.6)15,885 (35.3)< 0.0001Frequent visitors90b14,698 (10.8)4377 (9.6)4994 (10.9)5327 (11.8) < 0.0001IQR interquartile range, MedAid Medical Aid, NHI National Health Insurance, SD standard deviation. Note: The p-values were calculated by Kruskal–Wallis tests for continuous variables and by Chi-squared tests for categorical variables.aInformation for insurance contributions was missing for 2948 (2.2%) patients in all patients; 1308 (2.9) in the low-continuity group; 912 (2.0%) in the medium-continuity group; 728 (1.6%) in the high-continuity group.b‘Frequent visitors90’ are patients who were in the 90th percentile or higher based on the number of visits during the exposure period. Table 1Baseline characteristics of study population.Variable, n (%)All patientsN = 136,296Continuity groupLowN = 45,590MediumN = 45,650HighN = 45,056p-valueAgeMedian (IQR)3 (1, 5)2 (1, 4)3 (1, 5)3 (2, 6)< 0.0001Younger than 6108,687 (79.7)40,106 (88.0)36,710 (80.4)31,871 (70.7)< 0.00016–younger than 1224,006 (17.6)5075 (11.1)8010 (17.6)10,921 (24.2)12–younger than 203603 (2.6)409 (0.9)930 (2.0)2264 (5.0)SexWomen66,341 (48.7)21,968 (48.2)22,147 (48.5)22,226 (49.3)0.0019Insurance programNHI135,090 (99.1)45,299 (99.4)45,275 (99.2)44,516 (98.8)< 0.0001MedAid1206 (0.9)291 (0.6)375 (0.8)540 (1.2)Insurance contributionsaHigh52,185 (38.3)16,903 (37.1)17,513 (38.4)17,769 (39.4)< 0.0001Moderate62,769 (46.1)21,499 (47.2)21,010 (46.0)20,260 (45.0)Low18,394 (13.5)5880 (12.9)6215 (13.6)6299 (14.0)Urbanization level of residenceLarge urban area75,812 (55.6)23,713 (52.0)25,420 (55.7)26,679 (59.2)< 0.0001Small urban area46,515 (34.1)16,773 (36.8)15,542 (34.1)14,200 (31.5)Rural area13,969 (10.3)5104 (11.2)4688 (10.3)4177 (9.3)DisabilityYes88 (0.06)24 (0.05)26 (0.06)38 (0.08)0.1259Elixhauser comorbidity indexMean ± SD1.20 ± 0.501.21 ± 0.511.20 ± 0.501.19 ± 0.49< 0.00011113,945 (83.6)37,716 (82.7)38,137 (83.5)38,092 (84.5)< 0.0001218,263 (13.4)6394 (14.0)6144 (13.5)5725 (12.7)3+4088 (3.0)1480 (3.3)1369 (3.0)1239 (2.8)ComorbidityAllergic rhinitis118,000 (86.6)39,824 (87.4)39,622 (86.8)38,554 (85.6)< 0.0001Atopic dermatitis34,304 (25.2)12,356 (27.1)11,630 (25.5)10,318 (22.9)< 0.0001Cancer91 (0.07)20 (0.04)30 (0.07)41 (0.09)0.0229Diabetes mellitus268 (0.2)69 (0.15)95 (0.2)104 (0.2)0.0207Dyslipidemia1152 (0.9)352 (0.8)398 (0.9)402 (0.9)0.1063Hypertension100 (0.07)28 (0.06)35 (0.08)37 (0.08)0.4900Osteoarthritis437 (0.3)130 (0.3)132 (0.3)175 (0.4)0.0079Rheumatoid arthritis60 (0.04)14 (0.03)19 (0.04)27 (0.06)0.1061Systemic corticosteroids during the exposure periodYes45,411 (33.3)14,639 (32.1)14,887 (32.6)15,885 (35.3)< 0.0001Frequent visitors90b14,698 (10.8)4377 (9.6)4994 (10.9)5327 (11.8) < 0.0001IQR interquartile range, MedAid Medical Aid, NHI National Health Insurance, SD standard deviation. Note: The p-values were calculated by Kruskal–Wallis tests for continuous variables and by Chi-squared tests for categorical variables.aInformation for insurance contributions was missing for 2948 (2.2%) patients in all patients; 1308 (2.9) in the low-continuity group; 912 (2.0%) in the medium-continuity group; 728 (1.6%) in the high-continuity group.b‘Frequent visitors90’ are patients who were in the 90th percentile or higher based on the number of visits during the exposure period. Baseline characteristics of study population. IQR interquartile range, MedAid Medical Aid, NHI National Health Insurance, SD standard deviation. Note: The p-values were calculated by Kruskal–Wallis tests for continuous variables and by Chi-squared tests for categorical variables. aInformation for insurance contributions was missing for 2948 (2.2%) patients in all patients; 1308 (2.9) in the low-continuity group; 912 (2.0%) in the medium-continuity group; 728 (1.6%) in the high-continuity group. b‘Frequent visitors90’ are patients who were in the 90th percentile or higher based on the number of visits during the exposure period. Changes in continuity of care The COCI of the study patients was 0.56 during the exposure period and 0.51 during the outcome period (Table 2). The corresponding COCIs of the HC, MC, and LC groups decreased from 0.91 to 0.80, 0.51 to 0.46, and 0.26 to 0.27, respectively.Table 2Continuity of care and annual health services utilization.VariableAll patientsN = 136,296Continuity groupLowN = 45,590MediumN = 45,650HighN = 45,056Exposure period COCIMean ± SD0.56 ± 0.280.26 ± 0.080.51 ± 0.080.91 ± 0.11Median (IQR)0.50 (0.30, 0.80)0.27 (0.20, 0.33)0.50 (0.45, 0.58)1.00 (0.80, 1.00) Number of ambulatory visitsMean ± SD14.5 ± 11.814.4 ± 10.614.5 ± 11.714.8 ± 12.9Median (IQR)11 (6, 19)11 (7, 19)11 (6, 19)10 (6, 19) Asthma-related costs (1000 KRWc)Total103 ± 185117 ± 214102 ± 16890 ± 168Out-of-pocket payment20 ± 2421 ± 2619 ± 2318 ± 23Public expenditure83 ± 16395 ± 19082 ± 14872 ± 147Exposure and outcome period COCIMean ± SD0.51 ± 0.270.27 ± 0.110.46 ± 0.140.80 ± 0.21Median (IQR)0.45 (0.29, 0.69)0.25 (0.19, 0.33)0.46 (0.36, 0.56)0.84 (0.64, 1.00) Number of ambulatory visitsMean ± SD21.5 ± 18.621.9 ± 16.921.4 ± 18.621.1 ± 20.1Median (IQR)16 (9, 28)17 (10, 29)16 (8, 28)15 (8, 27) Asthma-related costs (1000 KRWa)Total898 ± 2954979 ± 4231882 ± 1243833 ± 2590Out-of-pocket payment190 ± 302200 ± 435189 ± 167182 ± 238Public expenditure708 ± 2695779 ± 3813692 ± 1120651 ± 2444COCI continuity of care index, IQR interquartile range, KRW Korean Won, SD standard deviation.a1 US dollar = 1300 KRW in January 2024. Table 2Continuity of care and annual health services utilization.VariableAll patientsN = 136,296Continuity groupLowN = 45,590MediumN = 45,650HighN = 45,056Exposure period COCIMean ± SD0.56 ± 0.280.26 ± 0.080.51 ± 0.080.91 ± 0.11Median (IQR)0.50 (0.30, 0.80)0.27 (0.20, 0.33)0.50 (0.45, 0.58)1.00 (0.80, 1.00) Number of ambulatory visitsMean ± SD14.5 ± 11.814.4 ± 10.614.5 ± 11.714.8 ± 12.9Median (IQR)11 (6, 19)11 (7, 19)11 (6, 19)10 (6, 19) Asthma-related costs (1000 KRWc)Total103 ± 185117 ± 214102 ± 16890 ± 168Out-of-pocket payment20 ± 2421 ± 2619 ± 2318 ± 23Public expenditure83 ± 16395 ± 19082 ± 14872 ± 147Exposure and outcome period COCIMean ± SD0.51 ± 0.270.27 ± 0.110.46 ± 0.140.80 ± 0.21Median (IQR)0.45 (0.29, 0.69)0.25 (0.19, 0.33)0.46 (0.36, 0.56)0.84 (0.64, 1.00) Number of ambulatory visitsMean ± SD21.5 ± 18.621.9 ± 16.921.4 ± 18.621.1 ± 20.1Median (IQR)16 (9, 28)17 (10, 29)16 (8, 28)15 (8, 27) Asthma-related costs (1000 KRWa)Total898 ± 2954979 ± 4231882 ± 1243833 ± 2590Out-of-pocket payment190 ± 302200 ± 435189 ± 167182 ± 238Public expenditure708 ± 2695779 ± 3813692 ± 1120651 ± 2444COCI continuity of care index, IQR interquartile range, KRW Korean Won, SD standard deviation.a1 US dollar = 1300 KRW in January 2024. Continuity of care and annual health services utilization. COCI continuity of care index, IQR interquartile range, KRW Korean Won, SD standard deviation. a1 US dollar = 1300 KRW in January 2024. Risk of hospitalization or emergency department visit A total of 9575 patients (7.03%) were hospitalized during the 2-year outcome period. Hospitalization rates in the HC, MC, and LC groups were 5.59, 6.98, and 8.48%, respectively (Table 3). After adjusting for confounding covariates, the MC and LC groups showed significantly higher hospitalization risks than the HC group (adjusted HR [aHR] = 1.14 [95% confidence interval = 1.08–1.20] and 1.30 [1.24–1.37], respectively). Figure 4 demonstrates the probabilities of patients avoiding hospital admission during the outcome period. Sensitivity analysis revealed similar risks of hospital admission for the MC (OR = 1.34 [1.27–1.41]) and LC (1.27 [1.20–1.33]) groups. In terms of ED visits, no significant difference was observed between the MC and HC groups, whereas the LC group had a significantly higher ED visit risk than the HC group (aHR = 1.32 [1.10–1.59]). Inverse relationships were observed between COCIs and the aHRs of hospital admission and ED visit risks (Fig. 5).Table 3Descriptive summary of study outcomes and association of continuity of primary care with study outcomes.OutcomeContinuity groupMeasure of association(ref. High)Unadjusted model(95% CI)Adjusted modela(95% CI)LowN = 45,590MediumN = 45,650HighN = 45,056Hospitalization, n (%)3867 (8.48)3188 (6.98)2520 (5.59)HRLow1.53 (1.46–1.61)§1.30 (1.24–1.37)§Medium1.25 (1.19–1.32)§1.14 (1.08–1.20)§ED visit, n (%)281 (0.62)222 (0.49)194 (0.43)HRLow1.45 (1.21–1.74)§1.32 (1.10–1.59)†Medium1.13 (0.94–1.38)1.07 (0.88–1.30)Systemic corticosteroids, n (%)2673 (5.86)2370 (5.19)2088 (4.63)ORLow1.28 (1.21–1.36)§1.13 (1.06–1.20)§Medium1.13 (1.06–1.20)§1.03 (0.97–1.10)Asthma-related costs, mean ± SD (1000 KRWb)979 ± 4231882 ± 1243833 ± 2590expβLow1.18 (1.16–1.19)§1.09 (1.08–1.09)§Medium1.06 (1.05–1.07)§1.03 (1.02–1.04)§Sensitivity analysis using the generalized estimating equations method Hospitalization–––ORLow1.37(1.30–1.44)§1.27 (1.20–1.33)§Medium1.48 (1.41–1.56)§1.34 (1.27–1.41)§ ED visit–––ORLow1.34 (1.11–1.60)§–Medium1.53 (1.27–1.85)§–CI confidential interval, ED Emergency department, exp exponential, HR hazard ratio, KRW Korean Won, OR odds ratio, ref reference, SD standard deviation.aThe adjusted HR was analyzed after adjusting for covariates including sex, age, health insurance program, level of insurance contributions, urbanization level of residence, disability, Elixhauser score, systemic use of corticosteroids, co-existence of allergic rhinitis, and whether the individual was a frequent doctor visitor or not. 'Frequent doctor visitors’ refers to individuals who ranked in the 90th percentile or higher based on the number of visits during the exposure period. In the analysis of medical costs, the annual medical cost during the exposure period was included as a covariate.b1 US dollar = 1300 KRW in January 2024.*p < 0.05; †p < 0.01; §p < 0.001.Fig. 4Event-free probability for different levels of primary care continuity. (a) Hospital admission, (b) Emergency department (ED) visits. HC high continuity, LC low continuity, MC medium continuity.Fig. 5Changes in the risks of asthma-related event by level of continuity of primary care. (a) Relationship between adjusted HRs of hospital admission and continuity of primary care. (b) Relationship between adjusted HRs of ED visits and continuity of primary care. *Optimal cubic functions for the adjusted HRs of each 3rd COCI percentile were estimated using COCI = 1 as the reference. COCI continuity of care index, ED emergency department, HC high continuity, HR hazard ratio, LC low continuity, MC medium continuity. Table 3Descriptive summary of study outcomes and association of continuity of primary care with study outcomes.OutcomeContinuity groupMeasure of association(ref. High)Unadjusted model(95% CI)Adjusted modela(95% CI)LowN = 45,590MediumN = 45,650HighN = 45,056Hospitalization, n (%)3867 (8.48)3188 (6.98)2520 (5.59)HRLow1.53 (1.46–1.61)§1.30 (1.24–1.37)§Medium1.25 (1.19–1.32)§1.14 (1.08–1.20)§ED visit, n (%)281 (0.62)222 (0.49)194 (0.43)HRLow1.45 (1.21–1.74)§1.32 (1.10–1.59)†Medium1.13 (0.94–1.38)1.07 (0.88–1.30)Systemic corticosteroids, n (%)2673 (5.86)2370 (5.19)2088 (4.63)ORLow1.28 (1.21–1.36)§1.13 (1.06–1.20)§Medium1.13 (1.06–1.20)§1.03 (0.97–1.10)Asthma-related costs, mean ± SD (1000 KRWb)979 ± 4231882 ± 1243833 ± 2590expβLow1.18 (1.16–1.19)§1.09 (1.08–1.09)§Medium1.06 (1.05–1.07)§1.03 (1.02–1.04)§Sensitivity analysis using the generalized estimating equations method Hospitalization–––ORLow1.37(1.30–1.44)§1.27 (1.20–1.33)§Medium1.48 (1.41–1.56)§1.34 (1.27–1.41)§ ED visit–––ORLow1.34 (1.11–1.60)§–Medium1.53 (1.27–1.85)§–CI confidential interval, ED Emergency department, exp exponential, HR hazard ratio, KRW Korean Won, OR odds ratio, ref reference, SD standard deviation.aThe adjusted HR was analyzed after adjusting for covariates including sex, age, health insurance program, level of insurance contributions, urbanization level of residence, disability, Elixhauser score, systemic use of corticosteroids, co-existence of allergic rhinitis, and whether the individual was a frequent doctor visitor or not. 'Frequent doctor visitors’ refers to individuals who ranked in the 90th percentile or higher based on the number of visits during the exposure period. In the analysis of medical costs, the annual medical cost during the exposure period was included as a covariate.b1 US dollar = 1300 KRW in January 2024.*p < 0.05; †p < 0.01; §p < 0.001. Descriptive summary of study outcomes and association of continuity of primary care with study outcomes. CI confidential interval, ED Emergency department, exp exponential, HR hazard ratio, KRW Korean Won, OR odds ratio, ref reference, SD standard deviation. aThe adjusted HR was analyzed after adjusting for covariates including sex, age, health insurance program, level of insurance contributions, urbanization level of residence, disability, Elixhauser score, systemic use of corticosteroids, co-existence of allergic rhinitis, and whether the individual was a frequent doctor visitor or not. 'Frequent doctor visitors’ refers to individuals who ranked in the 90th percentile or higher based on the number of visits during the exposure period. In the analysis of medical costs, the annual medical cost during the exposure period was included as a covariate. b1 US dollar = 1300 KRW in January 2024. *p < 0.05; †p < 0.01; §p < 0.001. Event-free probability for different levels of primary care continuity. (a) Hospital admission, (b) Emergency department (ED) visits. HC high continuity, LC low continuity, MC medium continuity. Changes in the risks of asthma-related event by level of continuity of primary care. (a) Relationship between adjusted HRs of hospital admission and continuity of primary care. (b) Relationship between adjusted HRs of ED visits and continuity of primary care. *Optimal cubic functions for the adjusted HRs of each 3rd COCI percentile were estimated using COCI = 1 as the reference. COCI continuity of care index, ED emergency department, HC high continuity, HR hazard ratio, LC low continuity, MC medium continuity. Risk of systemic corticosteroid use Systemic corticosteroid use was significantly higher in the MC and LC groups than in the HC group (Table 3). After adjusting for covariates, the LC group had a 1.13-fold higher risk of being prescribed systemic corticosteroids than the HC group (adjusted OR = 1.13 [1.06–1.20]), whereas no significant difference was observed between the MC and HC groups. Asthma-related medical costs Mean annual Asthma-related medical costs were significantly higher in both of the MC and LC groups compared to the HC group (Table 3). After adjusting for major covariates, gamma regression modeling analysis showed that patients in the LC and MC groups spent approximately 1.09 [1.08–1.09] and 1.03 [1.02–1.04] times more, respectively, than patients in the HC group. Subgroup analysis Primary analysis trends of hospital admission risks by levels of primary care continuity were similar for all subgroups (Supplementary Fig. 1). Age-based subgroup analysis produced some notable results. From ages 12 to 19, the risk of hospital admission was significantly higher in the MC and LC groups than in the HC group (2.17 [1.06–4.44], 3.41 [1.57–7.45], respectively). This difference was twice as large for in the MC group and three times as large for in the LC group as in children up to 11 years old. Discussion Nationwide claims data were analyzed to investigate the association between continuity of primary care and clinical and economic outcomes in young asthma patients. Study patients with low-level COCs with primary care doctors were 1.3 times more likely to be hospitalized or visit an emergency department and incurred 1.1 times higher healthcare costs than those with high-level COCs. In addition, the likelihood of receiving systemic corticosteroids due to asthma exacerbation was 1.13 times greater in the LC group than in the HC group. Patients in the MC group also had significantly higher risks across all study outcomes than patients in the HC group but lower risks than patients in the LC group. These results were consistent throughout the subgroup analyses, underscoring the affirmative association between high relational continuity of primary care and improved health outcomes and reduced healthcare costs, which concurred with the findings of previous studies11. Sensitivity analysis confirmed these findings for hospital admissions, but the same was not true for ED visits, likely due to their low overall incidence. However, overall results of subgroup and sensitivity analyses were consistent in terms of their directions, which indicated that higher continuity of care was more beneficial than lower continuity of care across a range of different outcomes. Utilizing data from extensive national-level sources increases the trustworthiness of research findings, and the consistency observed between our primary, sensitivity, and subgroup analyses enhanced our confidence in the results obtained. In addition, this study visualizes relationships between COC within primary care and the risks of hospitalization and emergency department visits. Furthermore, we investigated several comorbidities as potential confounders in the analysis to address limitations previously mentioned in a systematic review by Lytsy et al.11. Nonetheless, several limitations should be considered when interpreting our results. Firstly, the COC measurement used in the analysis is related to patient visit patterns and concentration of visits to primary care doctors; however, the extent to which it accurately reflects the interpersonal aspects of continuity in patient-provider relationships remains unclear5,27,33. Secondly, we utilized claims data for the analysis, which inherently introduces shortcomings; for example, we were unable to incorporate objective parameters indicative of patients’ health statuses (e.g., laboratory results) in the analysis. However, the use of systemic corticosteroids was included as a secondary outcome measure to assess asthma exacerbation. Another important consideration is that the data period does not encompass the patient’s entire lifetime but only a specific timeframe. Consequently, the limitations of the operational definition must be taken into account, particularly the assumption that a patient is considered to have a first diagnosis if no asthma diagnosis was recorded in the previous year. Thirdly, our dataset lacked detailed information on specific asthma treatments, such as the use of inhaled corticosteroids (ICS) or short-acting beta agonists (SABA) alone. This is a valuable aspect to address in future research. Fourthly, we were unable to assess the nature of personal friendliness between doctors and patients or their parents, which are likely to be related to decisions regarding future visits. Lastly, care should be taken when generalizing our results because the primary care facilities involved in the study reflect the Korean situation. However, our results are consistent with those of previous studies despite being derived from a different healthcare setting. In this regard, our analysis may improve the external validity of this theme. The study patients consisted predominantly of children under the age of 6, who accounted for approximately 80% of the study patients. Furthermore, a larger proportion of these children than of those aged ≥ 6 were allocated to the LC group. We propose two explanations for this result. Firstly, the health status of young children often exhibits considerable volatility34, which prompts parents and caregivers to respond more sensitively. Secondly, young parents raising children may encounter housing instability35, resulting in frequent relocations and diminished ongoing care from primary care doctors. The most significant difference in hospitalization risk was observed between the HC and LC groups for patients aged ≥ 12, despite the perceived illness state stability observed in this age group. Additional studies are required to determine whether this phenomenon is observed in other countries. It is argued an environment like that in South Korea, where patients have almost complete freedom to choose their doctors, promotes market competition, improves quality of care, and helps control healthcare costs36. However, the results of this study, like those of other continuity studies, suggest that free choice tends to decrease continuity and potentially negatively affect patient outcomes. Time is needed to establish a sense of familiarity between a doctor and patient and enable the doctor to understand the disease and patient’s preferences and behavior, which distinguishes healthcare provision from goods or services where immediate judgment is possible. The results of our study provide valuable insights for policymakers and primary healthcare providers. We advise that primary care practitioners continue to strive to nurture continuous relationships with their patients, despite the growing complexities of patient-doctor interactions37,38 and advocate that societal support be provided where the tradition of the family doctor is upheld. Conversely, in healthcare systems that diverge from the family doctor model, such as in South Korea, initiatives are required that focus on facilitating a steadfast bond between patients and their primary care doctors. For example, reimbursement models could be developed to reward primary care practices for maintaining high levels of relational continuity. These models could include incentives for practices that achieve certain benchmarks in patient satisfaction surveys related to continuity or reduced hospitalization rates for chronic conditions. The UK’s Quality and Outcomes Framework provides an example of a system that rewards practices for meeting specific quality indicators and includes indicators related to chronic disease management. Conclusions Based on a well-designed cohort study, the study findings suggest that improving continuity of care in young asthma patients may lead to better health outcomes and reduced healthcare utilization. This advises primary care practitioners to continue striving to nurture continuous relationships with their patients, and society should make efforts to maintain the relevant system. Significant variations in hospitalization risk were observed between the high and low continuity groups among patients aged 12 and above, suggesting the need for further research to determine whether this phenomenon is specific to certain regions or universal. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2