You are here

Article Text

Article menu

Download PDFPDF

Health policy
Original research
Analysis of the patient-sharing network in hypertension management: a retrospective study in China
  1. Zhiwen Gong1,
  2. Ruilin Wang1,
  3. Huajie Hu1,
  4. Tao Huang1,
  5. Huangqianyu Li2,
  6. Sheng Han2,
  7. Luwen Shi1,2,
  8. Xiaodong Guan1,2
  1. 1Department of Pharmacy Administration and Clinical Pharmacy, Peking University, Beijing, China
  2. 2International Research Center for Medicinal Administration, Peking University, Beijing, China
  1. Correspondence to Dr Xiaodong Guan; guanxiaodong{at}pku.edu.cn

Abstract

Objective To explore the robustness of the patient-sharing network and validate the association between strength and persistence of physicians’ relationships in China.

Design, setting and participants We conducted a patient-sharing network analysis to describe the persistence of patient-sharing relationships and logistic regression to analyse factors associating with the persistence of patient-sharing relationships in the Yinzhou Health Information System from 1 January 2010 to 31 December 2018; all outpatient records that had a hypertension diagnosis were included in this study.

Outcome measures The persistence ratio was defined as the proportion of the patient-sharing relationships in a given year that continued to exist in the following year, the 1-, 2- and 3-year persistence to test the robustness of the findings.

Results This study included 3916 physicians from 42 public healthcare facilities in Yinzhou. The 1-year persistence ratio fluctuated around 80%, and the 3-year persistence ratio was around 60% over the study period. The strength of the relationship, tie characteristics and physician specialty were important factors associating with the persistence of the relationships. The persistence of the relationships increased significantly as the strength of the relationships increased (for relationships with strength ∈ [3, 5), OR=3.987, 95% CI 3.896 to 4.08; for relationships with strength ∈ [5, 7), OR=6.379, 95% CI 6.147 to 6.626; and for relationships with strength ∈ [7, 9), OR=8.373, 95% CI 7.941 to 8.829). Physicians from the same healthcare institution were more likely to form ties that persisted for at least 1 year compared with physicians from different institutions (OR=1.510, 95% CI 1.480 to 1.540).

Conclusions Our study showed that physicians frequently formed relationships with other physicians through sharing patients in Yinzhou, China, and these relationships had similar rates of persistence to studies conducted in developed countries, which indicated that findings of social network analyses conducted in developed countries still hold value in developing countries.

  • Social Interaction
  • Knowledge
  • Primary Care
  • China

Data availability statement

Data may be obtained from a third party and are not publicly available. Data for the present study are property of Center for Disease Control and Prevention of Ningbo. The data are available from these authorities, but restrictions apply.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

https://doi-org.ezproxy.u-pec.fr/10.1136/bmjopen-2024-093684

Statistics from Altmetric.com

    Request Permissions

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

    STRENGTHS AND LIMITATIONS OF THIS STUDY

    • We applied social network analysis methods to measure the structure and persistence of physician relationships in hypertension management.

    • We used a well-established regional electronic health information system to capture a comprehensive view of physician patient-sharing relationships across different levels of healthcare institutions.

    • The actual patient flow and dynamics of physician relationships were unable to be observed.

    • The association between the persistence of physician relationships and patient health outcomes was unable to be examined.

    • The results were subjected to unmeasured confounding bias due to limited data availability.

    Introduction

    The translation and diffusion of knowledge inevitably affect physicians’ prescribing behaviours, especially among physicians providing care to shared patients, with implications for patient health and healthcare utilisation.1 An increasing number of research have used social network as a tool to understand patient-sharing relationships of healthcare professionals.2 3 Such analyses can deepen understandings of the influences of knowledge exchange, either formal or informal, on the clinical practice of healthcare providers and identify pathways to improved quality of care and efficient use of healthcare resources.4 5

    Previous studies have explored the impact of social networks on the prevention and control of infectious diseases, including tuberculosis and malaria, and on knowledge diffusion in managing non-communicable diseases (NCDs) such as diabetes and hypertension.3 6–9 Patient-sharing networks among physicians signal formal and informal exchange of knowledge and information while providing care and can help identify and understand problems in medication adherence and therapeutic inertia, both of which are essential to managing NCDs.10–17 Published studies exploring patient-sharing networks were mainly from developed countries, which found that network-based interventions are effective methods to reduce healthcare costs and improve the quality of healthcare services.18–22 These studies have reported that dispersed physician network connection contributed to fragmentation of care and increased costs,18 while the intensive connections improved quality of care and clinical outcomes.20–22 The persistence (also known as stability, referring to the continuation of relationships from the previous year into the next) and strength (the number of shared patients between two physicians) of the patient-sharing relationships have been identified as important metrics in network-based interventions.23–25 Persistent patient-sharing relationships could enable physicians to foster trusting relationships with one another and help to create new referral loops and are thus facilitative to information exchange and coordinating care, which has been interpreted through theories of diffusion of innovation or social contagion.23 26 27 Studies have found that many factors might affect the persistence ratio of patient-sharing relationships, including tie characteristics, physician specialty, strength and when such relationships occurred.18 24 28 However, patient-sharing relationships and the effect they can exert on provider practices can vary across regions with differently structured healthcare systems. It is uncertain if examining these relationships could be a viable tool in studying health services in developing countries and how these relationships can vary from developed countries and regions.24 27 29 30

    Few studies in low- and middle-income countries have used network analysis to understand professional communication among healthcare providers. Before developing network-based interventions, we need to first understand the structure and persistence of physicians’ patient-sharing networks in developing countries and identify influencing factors and their mechanism of action.31 In China, hypertension is one of the most prevalent NCDs, with a high prevalence of 44.7% among adults aged 35–75 years and generally poorly managed.32 The management and control of hypertension typically requires collaboration across different healthcare institutions and healthcare providers, especially between various healthcare levels.33 34 Improved hypertension control has been reported in the well-connected physician professional environment,35–38 underscoring the significance of promoting the physician’s relationship in hypertension management. The patient-sharing network models were widely applied and validated methods to depict and measure these relationships among physicians in prior research.39 40 Given the knowledge gap in structure and persistence of physicians’ relationships in China, we conducted a social network analysis to describe the persistence of patient-sharing relationships of physicians managing patients with hypertension and measure the association between strength and persistence of physicians’ relationships in China, for providing insights for achieving better hypertension care coordination and disease control.

    Methods

    Study design

    In line with previous studies, the patient-sharing relationship between physicians (ie, two or more physicians providing care to the same patients) was recorded in and identified through reviewing records of outpatient visits, with the number of the shared patients between physicians representing the strength of the relationship.24 Patient-sharing represents exchanges of knowledge among physicians and could therefore be used to assess physicians’ coordination, a clear target for cost-saving and improving the quality of medical care.41 42 To understand the network structure and factors associating with patient-sharing relationships in Yinzhou district of Ningbo, a coastal city in south-east China, we built a theoretical model of Chinese physicians’ professional network based on patient-sharing relationships.

    Data sources

    We extracted data from the Yinzhou Health Information System (YHIS), which was established by the local health department in 2005. Since its inception, the database has achieved registration of over 98% of permanent residents (approximately 1.3 million) and all healthcare providers (5.8 thousand) in Yinzhou.43 44 The system collects and manages electronic medical records of residents and covered data including general characteristics, prescription and outpatient visit records. All information stored in the system has been de-identified to safeguard patient privacy; thus, the requirement of informed consent was exempted according to the national legislation and the institutional requirements. Ethical approval was obtained from the Peking University Institution Review Board (IRB00001052-22052).

    Study population

    We extracted all outpatient records from YHIS. Inclusion criteria were: (1) the patient was diagnosed with hypertension, shown by the corresponding International Classification of Diseases, Tenth Revision codes (I10, I11, I12, I13 and I15); (2) the patient’s hypertension diagnosis was between 1 January 1 2010 and 31 December 2018; and (3) Primary Care Physicians (PCPs) or specialists working in secondary and tertiary hospitals who treated adult hypertension patients (≥18 years old). Our exclusion criteria were: (1) outpatient records generated from patients not residing in Yinzhou district and (2) patient-sharing relationships not occurring in the same year (eg, a physician provided care to this patient in a year while the other physician only provided care to this patient in the other year).

    Network construction

    We constructed physician networks by identifying relationships between physicians if one patient had visited both of them within the same year. Specifically, we first constructed the bipartite network composed of physician-patient connections by extracting the outpatient visit records within a year and generated the adjacency matrix of bipartite network (figure 1A).40 Then, we constructed the physician-physician unipartite network by multiplying the adjacency matrix of bipartite network with its transpose.45 The elements in the matrix of unimodal network were the number of patients shared between two physicians, which represented the strength of their relationship (figure 1B). The threshold of the network was defined as the minimum strength of patient-sharing relationships needed to form a physician connection. For instance, ‘threshold=2’ indicated that two physicians would need to share at least two patients to be regarded as in a network. In the analysis, we did not apply a fixed threshold; instead, we tested multiple thresholds from 1 to 9 (the range was determined based on previous reports and validation) to identify stable patient-sharing relationships and reduce the impact of incidental connections that have a lower probability of knowledge exchange.27 39

    Figure 1
    Figure 1

    Schematic diagram of the patient-sharing relationships. Node, physicians in the network; tie, the connections between physicians in patients-sharing relationships; strength, the weight of the ties between the two nodes refers to the number in patients-sharing relationships. PCP, primary care physicians.

    Measurements and covariates

    The main outcome of the analysis was the persistence of the physician-physician connections in the patient-sharing network, defined as the physician relationship in a given year continued to exist in the following year.23 We used the persistence ratio, which refers to the proportion of remained relationships, to measure the degree of persistence. For instance, if 10 pairs of patient-sharing relationships were observed in 2008 and eight of them persisted to 2009, the 1year- persistence ratio would be 80%. Connections that persisted for two or more years were seen as long-term relationships that could diffuse knowledge more intensely and thus have a larger effect on the prescribing behaviours of the physicians involved.29

    Many factors have been reported in association with the persistence of patient-sharing relationships.24 29 We included the tie characteristics, physician specialty, strength of the relationship and when such relationships occurred as covariates in our analysis, based on previous literature, theoretical framework of diffusion and data availability. The tie characteristic was a binary variable reflecting whether the patient-sharing relationship happened within one or across multiple hospitals. A patient-sharing relationship was also classified according to specialties involved and assigned to a category of the following: ‘PCPs-PCPs’, ‘PCPs-Specialists’ and ‘Specialists-Specialists’. The year when the patient-sharing relationship occurred was defined as the year when the outpatient visit (signalling a patient-sharing relationship) happened. Based on previous literature, we assumed that the knowledge translation and diffusion across years and hospitals could promote coordination of PCPs and specialists and subsequently improve healthcare system efficiency. Detailed variable selection and definition are reported in table 1.

    View this table:
    Table 1

    Variable selection and definition in the analysis of patient-sharing relationship in Yinzhou

    Statistical analysis

    We conducted logistic regression to analyse factors associating with the persistence of patient-sharing relationships, with the 95% CI and P value reported for each OR.46 We applied logistic regression rather than exponential random graph models as we primarily focused on assessing the strength and persistence of the physicians’ relationships already identified, rather than examining the formation of the network. Strength ∈ [1, 3), “Tie characteristics” = “no”, “Physician specialty” = “PCP-PCP”, and the year of 2010 were set as reference group in the regression. To visualise the structure of the patient-sharing network, we chose the Fruchterman–Reingold algorithm, a spring-embedder method, to present the physician network, with two physicians having stronger patient-sharing relationships lying closer in the illustration.40 47 In the analysis of the persistence ratio of patient-sharing relationships, we conducted a sensitivity analysis based on thresholds and years of persistence of patient-sharing relationships; in the analysis of the factors of patient-sharing relationships, we conducted sensitivity analyses on the 1-, 2- and 3-year persistence to test the robustness of the findings. All analyses were performed using R 4.0.4, and a two-sided P value <0.05 was considered statistically significant.

    Patient and public involvement

    Patients and the public were not involved in this study.

    Results

    Study sample

    Our final sample included 27 267 hypertensive patients and 3916 physicians from 42 public healthcare facilities. The median and range of the number of patients, physicians, and public healthcare facilities each year are described in table 2. Overall, the median of physicians included according to the inclusion criteria was 1572 (range 1264–2106) from 2010 to 2018, including 68.1% (54.5%–77.5%) PCPs and 31.9% (22.5–45.5%) specialists. The detailed number of samples from 2010 to 2018 was reported in online supplemental etable 1.

    View this table:
    Table 2

    Number of sampled patient and physicians constructed physician network for medication therapy management of hypertensive patients in Yinzhou from 2010 to 2018

    Structure of patient-sharing network

    The patient-sharing network is visualised in online supplemental efigure 1. Overall, the median number of patient-sharing relationships (ie, physician ties) was 67 203 (range 36 543–89 463) from 2010 to 2018, while the median number of relationships per physician was 41.7 (28.9–44.4, table 3 and online supplemental etable 2). Among these ties, 70.8% (67.3%–74.1%) happened between physicians from different healthcare institutions (HCIs), while others were between physicians from the same HCI. The majority (55.1% (47.9%–60.9%)) of observed connections were between PCPs, 34.0% (30.9%–38.1%) were between PCPs and specialists and 9.9% (8.2%–14.4%) were between specialists. Most (68.9% (67.9%–75.3%)) observed connections had a strength ∈ [1, 3) (ie, the two physicians had one or two tie(s)); only 10.7% (9.5%–11.4%), 4.7% (3.6%–5.3%), and 2.7% (2.0%–3.0%) had a strength ∈ [3, 5) (ie, the two physicians had three or four ties), strength ∈ [5, 7) and strength ∈ [7, 9). There are many patient-sharing relationships that had a strength ≥9 (12.5% (9.5, 13.1)). The network characteristics at different thresholds were reported in the online supplemental etable 3. Overall, the network diameter from 2010 to 2018 ranged from 6 to 9 for thresholds from 1 to 9; the network density ranged from 0.18 to 0.55 for thresholds from 1 to 9; the clustering coefficient ranged from 0.33 to 0.45 for thresholds from 1 to 9.

    View this table:
    Table 3

    Characteristics of patient-sharing relationships for medication therapy management of hypertensive patients in Yinzhou from 2010 to 2018

    Persistence of patient-sharing network

    Changes in the persistence of patient-sharing relationships for different network thresholds are shown in figure 2. When the threshold equalled 1, relationships seemed random and displayed trends vastly different from those shown when other thresholds were applied. When the threshold was set at 3 (ie, only patient-sharing relationships with strength ≥ 3 were retained), the physician network demonstrated a gradually downward trend as compared with when the threshold was set at 1. Patient-sharing relationships showed similar patterns when the threshold was set at 3, 5, 7 or 9, with at least 80% of the relationships persisting after 1 year (threshold=3, 81.8%; threshold=5, 85.8%; threshold=7, 86.9%; and threshold=9, 87.4%) and at least 60% of the relationships persisting after 3 years (threshold=3, 60.4%; threshold=5, 64.4%; threshold=7, 66.1%; and threshold=9, 66.5%).

    Figure 2
    Figure 2

    The proportion of persistent ties for patient-sharing relationships for medication therapy management of hypertensive patients in Yinzhou from 2010 to 2018. (A–H) the proportion of persistent ties generated in 2010–2018, respectively.

    Factors associating with the persistence of patient-sharing network

    Results of the logistic analysis of the factors associating with the persistence of patient-sharing relationships are shown in table 4. We found the increase of strength of the relationships was associating with the increase in persistence of the relationships. Compared with the relationships that had a strength ∈ [1, 3), OR was 3.987 (95% CI 3.896 to 4.08); for relationships with a strength ∈ [3, 5) was 3.987 (95% CI 3.896 to 4.08); 6.379 (95% CI 6.147 to 6.626) for relationships with strength ∈ [5, 7); and 8.373 (95% CI 7.941 to 8.829) for relationships with strength ∈ [7, 9). Physicians from the same HCI were more likely to form connections that persisted for at least 1 year compared with physicians from different HCIs (OR=1.510, 95% CI 1.480 to 1.540). PCP-Specialist relationships and Specialist-Specialist relationships had lower rates of 1-year persistence compared with the PCP-PCP relationships (OR<1, p<0.001).

    View this table:
    Table 4

    Logistic regression of persistent ties for patient-sharing relationships for medication therapy management of hypertensive patients in Yinzhou from 2010 to 2018

    Sensitivity analysis of patient-sharing networks

    We conducted a sensitivity analysis using 2-year and 3-year ties. Results of the sensitivity analysis confirmed that the strength of the relationship, tie characteristics and physician specialty were important factors associating with the persistence of physician’s patient-sharing relationships (table 4).

    Discussion

    This study provides insights into how frequently physicians in China fostered relationships with each other through delivering care to a shared patient and how strong and persistent these relationships were. We found that over 80% of physician relationships formed through sharing patients persisted over a year, and 60% of these relationships persisted over 3 years, a result comparable to results from developed countries.29 The strength of the relationship, tie characteristics and physician specialty were important factors associating with the persistence of the physician’s patient-sharing relationships. To our knowledge, this study is the first to describe the structure and influencing factors of the patient-sharing network among physicians engaged in the management of NCDs in a developing country. These findings suggest that physicians’ patient-sharing networks can serve as a stable and viable target for future studies or interventions to promote care coordination and knowledge diffusion in China.

    We observed that physician relationships were more persistent as the relationship threshold increased, which aligned with observations from previous studies.23 39 This suggests that physicians with more shared patients may be more likely to form a more stable professional relationship and demonstrate greater relationship persistence, thereby exerting profound influence on knowledge diffusion within the network.24 Enhanced knowledge diffusion and information exchange strengthen the quality and coordination of healthcare services, as evidenced by reduced emergency room visits and lower medical costs for patients treated by physicians’ persistent connection.29 48 This effect may also extend across physicians from different hospitals to produce desirable patient outcomes, including lowered odds of readmissions and adverse events.39 49–51 Therefore, it is feasible to improve the relationship strength and improve healthcare quality by fostering physicians’ professional network and promoting regular physician communication among providers.

    In middle-income countries like China, the burden of cardiovascular diseases was more inequitably distributed than in high-income countries, which is often exacerbated by a severe under-resource of personnel with expertise in and infrastructure supporting the management of cardiovascular diseases.16 Developing interventions that use patient-sharing networks already present in physicians’ day-to-day practices and that strengthen the hierarchical medical system can be an effective and cost-saving approach to influence physicians’ behaviours, improve their coordination and promote patient outcomes. As patient-sharing relationships formed among physicians of different HCIs can diffuse knowledge and influence their prescribing behaviour, they can be used to promote the optimisation of patient treatment plans to reduce patient burden and improve clinical outcomes in managing NCDs like hypertension.13 18 48

    Our study confirmed that physicians both from primary care facilities were more likely to form and keep patient-sharing relationships, a result similar to a previous study.52 The finding implied that the less cohesive care coordination across different levels of facilities in the Chinese healthcare system53 may lead to suboptimal care continuity and disease control.54 These relationships among physicians across different levels of HCIs and specialties could lend insight into the barriers and promoters of an efficient healthcare system.55 They are conducive to the knowledge diffusion beyond one single HCI, which has the potential to spread clinical treatment experience to improve quality of care in primary healthcare facilities, where physicians normally have less clinical capacity.18 This is supported by a previous study which found that a high value of network statistics, reflecting global connectivity, is beneficial to medical cost savings.30 In contrast, hospitals with greater dispersion were associated with greater rates of readmission and lower rates of emergency department throughput.30 56 Patient-sharing relationships among physicians can be a clear target to develop network-based interventions to curb these unfavourable outcomes.

    The Chinese government launched a hierarchical medical system policy in 2014, aiming to alter patients’ healthcare-seeking behaviours. After its implementation, most patients are expected to first visit a primary care facility, which ideally should be the same facility to ensure consistency in care. This redirection of patient flow to primary care facilities may cause PCPs to share patients more frequently, and thus more physicians could have more possibility to form stronger patient-sharing relationships.54 We also found that the proportion of patient-sharing relationships formed between physicians practising at the same HCI was lower than that reported in developed countries.40 57 58 This may imply that HCIs in China have not established a harmonised patient referral system and thus patients frequently move between HCIs when certain needs remain unmet, which helped to form physician connections across HCIs.59 A previous study has found that PCPs have a central role in managing chronic diseases, and a hierarchical medical system can leverage the management of NCDs.42 Another study has confirmed the increased persistence of patient-sharing relationships across different healthcare levels since 2015, when the hierarchical medical system policy was implemented in China, which is attributed to the policy’s promotion of primary care physician’s centrality in disease management.54

    There are several limitations to this study. First, we established the social network of physicians based on a database from a single district in China; thus, our result may not be generalised to other areas in China with different physician network structures. Additionally, we were unable to observe the actual patient flow and the dynamics of these relationships from the retrospective data. A fuller landscape of the impact of knowledge diffusion through these relationships on physicians’ prescribing behaviours may only be gained through conducting qualitative studies in the future. Second, we limited the disease area to hypertension to represent the characteristics of the patient-sharing network of physicians managing NCDs. However, physician networks may be affected by differences in chronic diseases, such as patient characteristics, which were unable to be incorporated in this study. For instance, the differences in severity of the disease or comorbidities may lead to distinct patient visiting patterns; thus, our results should be interpreted within the specific context. Third, we are only constructing the network using the 1 year time frame to identify physicians’ relationships. Though there was a study observing that shorter time frames do not significantly affect the results,60 it is possible that the results could be different if we change the time frames in our setting. Future research should consider constructing networks over different time frames as data permit. Fourth, our association analysis may be subject to unmeasured confounding bias since we failed to include additional physician factors potentially associated with the persistence. For instance, factors such as physicians’ practising department, years in practice or professional title were not included, whereas it is possible that physicians may be more likely to establish connections with others who shared similar characteristics.61 Future research should consider incorporating relevant factors more comprehensively or applying methods such as instrumental variables to effectively control for potential confounding. Last, we have not examined the relationship between the persistence of physician relationships and patient health outcomes, which holds greater clinical and policy relevance. Future research should focus on this topic to offer novel insights for healthcare policy and practice, especially on leveraging physicians’ social networks to improve healthcare delivery.

    Conclusions

    We found that physicians frequently formed relationships with other physicians through sharing patients in Yinzhou, China. These relationships showed similar rates of persistence to studies conducted in developed countries, where network-based interventions have been proven effective in curbing some problems in healthcare delivery and patient outcomes. Future research and interventions to promote care coordination and knowledge diffusion can leverage these naturally occurring relationships and seek to understand mechanisms through which they can exert influences on healthcare providers’ practices and coordination with one another.

    Data availability statement

    Data may be obtained from a third party and are not publicly available. Data for the present study are property of Center for Disease Control and Prevention of Ningbo. The data are available from these authorities, but restrictions apply.

    Ethics statements

    Patient consent for publication

    Ethics approval

    Ethical approval was obtained from the Peking University Medical Ethics Committee (IRB00001052–22052). Requirement of informed consent for participation was exempted for this study according to the national legislation and the institutional requirements.

    References

    1. ↵
      1. Saghy E,
      2. Mulinari S,
      3. Ozieranski P
      . Drug company payments to General Practices in England: Cross-sectional and social network analysis. PLoS One 2021;16:e0261077. doi:10.1371/journal.pone.0261077
    2. ↵
      1. Jo W,
      2. Chang D,
      3. You M, et al
      . A social network analysis of the spread of COVID-19 in South Korea and policy implications. Sci Rep 2021;11:8581. doi:10.1038/s41598-021-87837-0
    3. ↵
      1. Valente TW,
      2. Pitts SR
      . An Appraisal of Social Network Theory and Analysis as Applied to Public Health: Challenges and Opportunities. Annu Rev Public Health 2017;38:103–18. doi:10.1146/annurev-publhealth-031816-044528
      OpenUrlPubMed
    4. ↵
      1. Shelley KD,
      2. Kamya C,
      3. Mpanya G, et al
      . Partnership and Participation-A Social Network Analysis of the 2017 Global Fund Application Process in the Democratic Republic of the Congo and Uganda. Ann Glob Health 2020;86:140. doi:10.5334/aogh.2961
    5. ↵
      1. Bae S-H,
      2. Nikolaev A,
      3. Seo JY, et al
      . Health care provider social network analysis: A systematic review. Nurs Outlook 2015;63:566–84. doi:10.1016/j.outlook.2015.05.006
      OpenUrlPubMed
    6. ↵
      1. Lopreite M,
      2. Puliga M,
      3. Riccaboni M, et al
      . A social network analysis of the organizations focusing on tuberculosis, malaria and pneumonia. Soc Sci Med 2021;278:S0277-9536(21)00272-0. doi:10.1016/j.socscimed.2021.113940
    7. ↵
      1. Ostovari M,
      2. Steele-Morris CJ,
      3. Griffin PM, et al
      . Data-driven modeling of diabetes care teams using social network analysis. J Am Med Inform Assoc 2019;26:911–9. doi:10.1093/jamia/ocz022
      OpenUrlPubMed
    8. ↵
      1. Hosseini Z,
      2. Veenstra G,
      3. Khan NA, et al
      . Social connections and hypertension in women and men: a population-based cross-sectional study of the Canadian Longitudinal Study on Aging. J Hypertens 2021;39:651–60. doi:10.1097/HJH.0000000000002688
      OpenUrlPubMed
    9. ↵
      1. Tusubira AK,
      2. Nalwadda CK,
      3. Akiteng AR, et al
      . Social Support for Self-Care: Patient Strategies for Managing Diabetes and Hypertension in Rural Uganda. Ann Glob Health 2021;87:86. doi:10.5334/aogh.3308
    10. ↵
      1. Ott C,
      2. Schmieder RE
      . Diagnosis and treatment of arterial hypertension 2021. Kidney Int 2022;101:36–46. doi:10.1016/j.kint.2021.09.026
      OpenUrlCrossRefPubMed
    11. ↵
      1. Marseille BR,
      2. Commodore-Mensah Y,
      3. Davidson PM, et al
      . Improving hypertension knowledge, medication adherence, and blood pressure control: A feasibility study. J Clin Nurs 2021;30:2960–7. doi:10.1111/jocn.15803
      OpenUrlPubMed
    12. ↵
      1. Chow CK,
      2. Atkins ER,
      3. Hillis GS, et al
      . Initial treatment with a single pill containing quadruple combination of quarter doses of blood pressure medicines versus standard dose monotherapy in patients with hypertension (QUARTET): a phase 3, randomised, double-blind, active-controlled trial. Lancet 2021;398:1043–52. doi:10.1016/S0140-6736(21)01922-X
      OpenUrlPubMed
    13. ↵
      1. Hu H,
      2. Zhang Y,
      3. Zhu D, et al
      . Physician patient-sharing relationships and healthcare costs and utilization in China: social network analysis based on health insurance data. Postgrad Med 2021;133:798–806. doi:10.1080/00325481.2021.1944650
      OpenUrlCrossRef
    14. ↵
      1. Yin H,
      2. Ma X,
      3. He Y, et al
      . Effect of an outpatient copayment scheme on health outcomes of hypertensive adults in a community-managed population in Xinjiang, China. PLoS One 2020;15:e0238980. doi:10.1371/journal.pone.0238980
    15. ↵
      1. Ruchman SG,
      2. Delong AK,
      3. Kamano JH, et al
      . Egocentric social network characteristics and cardiovascular risk among patients with hypertension or diabetes in western Kenya: a cross-sectional analysis from the BIGPIC trial. BMJ Open 2021;11:e049610. doi:10.1136/bmjopen-2021-049610
    16. ↵
      1. Timmis A,
      2. Townsend N,
      3. Gale CP, et al
      . European Society of Cardiology: Cardiovascular Disease Statistics 2019. Eur Heart J 2020;41:12–85. doi:10.1093/eurheartj/ehz859
      OpenUrlCrossRefPubMed
    17. ↵
      1. Rea F,
      2. Corrao G,
      3. Merlino L, et al
      . Initial Antihypertensive Treatment Strategies and Therapeutic Inertia. Hypertension 2018;72:846–53. doi:10.1161/HYPERTENSIONAHA.118.11308
      OpenUrl
    18. ↵
      1. Landon BE,
      2. Keating NL,
      3. Onnela JP, et al
      . Patient-Sharing Networks of Physicians and Health Care Utilization and Spending Among Medicare Beneficiaries. JAMA Intern Med 2018;178:66–73. doi:10.1001/jamainternmed.2017.5034
      OpenUrlPubMed
    19. ↵
      1. Mehrotra A,
      2. Adams JL,
      3. Thomas JW, et al
      . Cost Profiles: Should The Focus Be On Individual Physicians Or Physician Groups? Health Aff (Millwood) 2010;29:1532–8. doi:10.1377/hlthaff.2009.1091
      OpenUrlAbstract/FREE Full Text
    20. ↵
      1. Vlaanderen FP,
      2. de Man Y,
      3. Tanke MAC, et al
      . Density of Patient-Sharing Networks: Impact on the Value of Parkinson Care. Int J Health Policy Manag 2022;11:1132–9. doi:10.34172/ijhpm.2021.15
      OpenUrlPubMed
    21. ↵
      1. Stecher C
      . Physician network connections to specialists and HIV quality of care. Health Serv Res 2021;56:908–18. doi:10.1111/1475-6773.13628
      OpenUrlCrossRefPubMed
    22. ↵
      1. Davis J,
      2. Lim E,
      3. Taira DA, et al
      . Healthcare network analysis of patients with diabetes and their physicians. Am J Manag Care 2019;25:e192–7.
      OpenUrlPubMed
    23. ↵
      1. O’Hanlon CE,
      2. Whaley CM,
      3. Freund D
      . Medical Practice Consolidation and Physician Shared Patient Network Size, Strength, and Stability. Med Care 2019;57:680–7. doi:10.1097/MLR.0000000000001168
      OpenUrlPubMed
    24. ↵
      1. DuGoff EH,
      2. Fernandes-Taylor S,
      3. Weissman GE, et al
      . A scoping review of patient-sharing network studies using administrative data. Transl Behav Med 2018;8:598–625. doi:10.1093/tbm/ibx015
      OpenUrlCrossRefPubMed
    25. ↵
      1. Sen AP,
      2. Meiselbach MK,
      3. Anderson KE, et al
      . Physician Network Breadth and Plan Quality Ratings in Medicare Advantage. JAMA Health Forum 2021;2:e211816. doi:10.1001/jamahealthforum.2021.1816
    26. ↵
      1. Ndumele CD,
      2. Staiger B,
      3. Ross JS, et al
      . Network Optimization And The Continuity Of Physicians In Medicaid Managed Care. Health Aff (Millwood) 2018;37:929–35. doi:10.1377/hlthaff.2017.1410
      OpenUrlPubMed
    27. ↵
      1. Hu H,
      2. Yang Y,
      3. Zhang C, et al
      . Review of social networks of professionals in healthcare settings-where are we and what else is needed? Global Health 2021;17:139. doi:10.1186/s12992-021-00772-7
    28. ↵
      1. Rittenhouse DR,
      2. Shortell SM,
      3. Fisher ES
      . Primary care and accountable care--two essential elements of delivery-system reform. N Engl J Med 2009;361:2301–3. doi:10.1056/NEJMp0909327
      OpenUrlCrossRefPubMedWeb of Science
    29. ↵
      1. DuGoff EH,
      2. Cho J,
      3. Si Y, et al
      . Geographic Variations in Physician Relationships Over Time: Implications for Care Coordination. Med Care Res Rev 2018;75:586–611. doi:10.1177/1077558717697016
      OpenUrlPubMed
    30. ↵
      1. Geissler KH,
      2. Lubin B,
      3. Ericson KMM
      . The association between patient sharing network structure and healthcare costs. PLoS One 2020;15:e0234990. doi:10.1371/journal.pone.0234990
    31. ↵
      1. Sabot K,
      2. Wickremasinghe D,
      3. Blanchet K, et al
      . Use of social network analysis methods to study professional advice and performance among healthcare providers: a systematic review. Syst Rev 2017;6:208. doi:10.1186/s13643-017-0597-1
    32. ↵
      1. Lu J,
      2. Lu Y,
      3. Wang X, et al
      . Prevalence, awareness, treatment, and control of hypertension in China: data from 1·7 million adults in a population-based screening study (China PEACE Million Persons Project). Lancet 2017;390:2549–58. doi:10.1016/S0140-6736(17)32478-9
      OpenUrlCrossRefPubMed
    33. ↵
      1. Selby K,
      2. Michel M,
      3. Gildengorin G, et al
      . Disparities in Hypertension Control Across and Within Three Health Systems Participating in a Data-Sharing Collaborative. J Am Board Fam Med 2018;31:897–904. doi:10.3122/jabfm.2018.06.180166
      OpenUrlAbstract/FREE Full Text
    34. ↵
      1. Gao Q,
      2. Peng L,
      3. Min W, et al
      . Regularity of Clinical Visits and Medication Adherence of Patients with Hypertension or Diabetes in Rural Yunnan Province of China. Int J Environ Res Public Health 2020;17:9297. doi:10.3390/ijerph17249297
    35. ↵
      1. Vedanthan R,
      2. Ray M,
      3. Fuster V, et al
      . Hypertension Treatment Rates and Health Care Worker Density. Hypertension 2019;73:594–601. doi:10.1161/HYPERTENSIONAHA.118.11995
      OpenUrl
    36. ↵
      1. Zhou T,
      2. Wang Y,
      3. Zhang H, et al
      . Primary care institutional characteristics associated with hypertension awareness, treatment, and control in the China PEACE-Million Persons Project and primary health-care survey: a cross-sectional study. Lancet Glob Health 2023;11:e83–94. doi:10.1016/S2214-109X(22)00428-4
      OpenUrlPubMed
    37. ↵
      1. Lee J,
      2. Wilkens J,
      3. Meijer E, et al
      . Hypertension awareness, treatment, and control and their association with healthcare access in the middle-aged and older Indian population: A nationwide cohort study. PLoS Med 2022;19:e1003855. doi:10.1371/journal.pmed.1003855
    38. ↵
      1. Brunström M,
      2. Ng N,
      3. Dahlström J, et al
      . Association of Physician Education and Feedback on Hypertension Management With Patient Blood Pressure and Hypertension Control. JAMA Netw Open 2020;3:e1918625. doi:10.1001/jamanetworkopen.2019.18625
    39. ↵
      1. Barnett ML,
      2. Landon BE,
      3. O’Malley AJ, et al
      . Mapping physician networks with self-reported and administrative data. Health Serv Res 2011;46:1592–609. doi:10.1111/j.1475-6773.2011.01262.x
      OpenUrlCrossRefPubMedWeb of Science
    40. ↵
      1. Landon BE,
      2. Keating NL,
      3. Barnett ML, et al
      . Variation in patient-sharing networks of physicians across the United States. JAMA 2012;308:265–73. doi:10.1001/jama.2012.7615
      OpenUrlCrossRefPubMedWeb of Science
    41. ↵
      1. Evan Pollack C,
      2. Wang H,
      3. Bekelman JE, et al
      . Physician social networks and variation in rates of complications after radical prostatectomy. Value Health 2014;17:611–8. doi:10.1016/j.jval.2014.04.011
      OpenUrlPubMed
    42. ↵
      1. Barnett ML,
      2. Christakis NA,
      3. O’Malley J, et al
      . Physician patient-sharing networks and the cost and intensity of care in US hospitals. Med Care 2012;50:152–60. doi:10.1097/MLR.0b013e31822dcef7
      OpenUrlCrossRefPubMedWeb of Science
    43. ↵
      1. Li H,
      2. Zhao H,
      3. Lin H, et al
      . Utilization of Intravenous Ribavirin Among Reproductive Age Adults in 2010–2017: A Population-Based Study in the Yinzhou District, Ningbo City of China. Front Public Health 2021;9:678785. doi:10.3389/fpubh.2021.678785
      OpenUrlPubMed
    44. ↵
      1. Huang K,
      2. Tao S,
      3. Zhou X, et al
      . Incidence rates of health outcomes of interest among Chinese children exposed to selected vaccines in Yinzhou Electronic Health Records: A population-based retrospective cohort study. Vaccine (Auckl) 2020;38:3422–8. doi:10.1016/j.vaccine.2020.03.013
      OpenUrl
    45. ↵
      1. Landon BE,
      2. Onnela J-P,
      3. Keating NL, et al
      . Using Administrative Data to Identify Naturally Occurring Networks of Physicians. Med Care 2013;51:715–21:715. doi:10.1097/MLR.0b013e3182977991
      OpenUrlCrossRefPubMed
    46. ↵
      1. Schober P,
      2. Vetter TR
      . Logistic Regression in Medical Research. Anesth Analg 2021;132:365–6. doi:10.1213/ANE.0000000000005247
      OpenUrlCrossRefPubMed
    47. ↵
      1. Fruchterman TMJ,
      2. Reingold EM
      . Graph drawing by force‐directed placement. Softw Pract Exp 1991;21:1129–64. doi:10.1002/spe.4380211102
      OpenUrlCrossRef
    48. ↵
      1. Pollack CE,
      2. Weissman GE,
      3. Lemke KW, et al
      . Patient sharing among physicians and costs of care: a network analytic approach to care coordination using claims data. J Gen Intern Med 2013;28:459–65. doi:10.1007/s11606-012-2104-7
      OpenUrlCrossRefPubMed
    49. ↵
      1. Lomi A,
      2. Mascia D,
      3. Vu DQ, et al
      . Quality of care and interhospital collaboration: a study of patient transfers in Italy. Med Care 2014;52:407–14. doi:10.1097/MLR.0000000000000107
      OpenUrlPubMed
    50. ↵
      1. Pollack CE,
      2. Lemke KW,
      3. Roberts E, et al
      . Patient sharing and quality of care: measuring outcomes of care coordination using claims data. Med Care 2015;53:317–23. doi:10.1097/MLR.0000000000000319
      OpenUrlPubMed
    51. ↵
      1. Pollack CE,
      2. Frick KD,
      3. Herbert RJ, et al
      . It’s who you know: patient-sharing, quality, and costs of cancer survivorship care. J Cancer Surviv 2014;8:156–66. doi:10.1007/s11764-014-0349-3
      OpenUrlCrossRefPubMed
    52. ↵
      1. Lee BY,
      2. McGlone SM,
      3. Song Y, et al
      . Social network analysis of patient sharing among hospitals in Orange County, California. Am J Public Health 2011;101:707–13. doi:10.2105/AJPH.2010.202754
      OpenUrlCrossRefPubMedWeb of Science
    53. ↵
      1. Li X,
      2. Krumholz HM,
      3. Yip W, et al
      . Quality of primary health care in China: challenges and recommendations. Lancet 2020;395:1802–12. doi:10.1016/S0140-6736(20)30122-7
      OpenUrlCrossRefPubMed
    54. ↵
      1. Hu H,
      2. Wang R,
      3. Li H, et al
      . Effectiveness of hierarchical medical system policy: an interrupted time series analysis of a pilot scheme in China. Health Policy Plan 2023;38:609–19. doi:10.1093/heapol/czad018
      OpenUrlCrossRefPubMed
    55. ↵
      1. Robbins R,
      2. Seixas A,
      3. Schoenthaler A
      . The nature and scope of patient-sharing network research: a novel, important area for network science. Transl Behav Med 2018;8:626–8. doi:10.1093/tbm/iby052
      OpenUrlPubMed
    56. ↵
      1. Everson J,
      2. Adler-Milstein JR,
      3. Hollingsworth JM, et al
      . Dispersion in the hospital network of shared patients is associated with less efficient care. Health Care Manage Rev 2022;47:88–99. doi:10.1097/HMR.0000000000000295
      OpenUrlPubMed
    57. ↵
      1. Xu J,
      2. Powell-Jackson T,
      3. Mills A
      . Effectiveness of primary care gatekeeping: difference-in-differences evaluation of a pilot scheme in China. BMJ Glob Health 2020;5:e002792. doi:10.1136/bmjgh-2020-002792
    58. ↵
      1. Lu C,
      2. Zhang Z,
      3. Lan X
      . Impact of China’s referral reform on the equity and spatial accessibility of healthcare resources: A case study of Beijing. Soc Sci Med 2019;235:S0277-9536(19)30371-5. doi:10.1016/j.socscimed.2019.112386
    59. ↵
      1. Pylypchuk Y,
      2. Meyerhoefer CD,
      3. Encinosa W, et al
      . The role of electronic health record developers in hospital patient sharing. J Am Med Inform Assoc 2022;29:435–42. doi:10.1093/jamia/ocab263
      OpenUrlPubMed
    60. ↵
      1. Hietapakka L,
      2. Sinervo T,
      3. Väisänen V, et al
      . Patient-sharing networks among Finnish primary healthcare professionals taking care of patients with mental health or substance use problems: a register study. BMJ Open 2025;15:e089111. doi:10.1136/bmjopen-2024-089111
    61. ↵
      1. Donohue JM,
      2. Guclu H,
      3. Gellad WF, et al
      . Influence of peer networks on physician adoption of new drugs. PLoS One 2018;13:e0204826. doi:10.1371/journal.pone.0204826

    Footnotes

    • Contributors ZG: software, formal analysis, writing-original draft preparation. RW: software, formal analysis, writing-original draft preparation. HH: data curation, software, formal analysis, writing-original draft preparation. TH: data curation, validation, writing-review and editing. HL: writing-review and editing. SH: conceptualisation, supervision, writing-review and editing. LS: conceptualisation, supervision, writing-review and editing. XG: conceptualisation, supervision, writing-review and editing, funding acquisition. All authors gave final approval of the version to be published and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately. Guarantor: XG.

    • Funding This work was supported by the National Natural Science Foundation of China (grant number: 72074007). The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

    • Competing interests None declared.

    • Patient and public involvement Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.

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

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