Participants and procedure

We conducted a cross-sectional, web-based survey from October 2018 to February 2019 at Seoul National University, the largest public university in South Korea. Participants were recruited from the entire students, staff, and faculty of the university (n = 35,026). We sent potential participants invitation e-mails explaining the purpose and an outline of the survey. A web link to the survey was posted on the university’s intranet, websites, and social media. We consecutively enrolled participants who voluntarily responded to and self-administered the survey. Inclusion criteria were those aged ≥18 years and who agreed to answer the survey. Exclusion criteria were individuals who did not give consent and those with missing data on more than 80% of the main outcome measurement. The study was approved by the Institutional Review Board of Seoul National University College of Medicine/Seoul National University Hospital (Seoul, South Korea; IRB number, H-1807-062-957). All participants provided informed consent online before participation in the survey.

Measures

Participants were asked to compare, based on their past experiences, the qualities of primary care provided by 3 different types of healthcare facilities: the university health service, community clinics, and hospitals. A university health service is a primary care model on campus dealing with medical care and health promotion of university students and employees. The university health service in this university implements a wide range of activities: primary medical consultation for acute and chronic diseases, mental health care, communicable disease prevention, immunizations, health screening, and health education. Family physicians play the central role in these activities, partly supported by some specialty care.

Participants’ perceptions of care quality was assessed using the Korean Primary Care Assessment Tool (K-PCAT),[29] which measures key attributes of primary care related to effective care organization and delivery at the population level.[30] In the current study, the K-PCAT was modified into 7 domains with 18 subordinate questions (S1 and S2 Files). The 7 domains consisted of: first contact (accessibility of care at the time of need and its utilization towards solving a new health problem), comprehensiveness (a range of services encompassing common problems in the population), continuity (the longitudinal use of a regular source of care over time, focusing on the long-term health of a person, rather than the short-term duration of the disease), coordination (the role of coordinating other specialist services or resources that patients may need), personalized care (a holistic approach that considers physical, mental, and spiritual well-being), family/community orientation (the inclusion of family health concerns in decision-making and the provider’s knowledge of community health needs), and trust/satisfaction.[6,27,30–32] Responses were obtained using a 5-point Likert scale (5 = strongly agree, 4 = agree, 3 = neutral, 2 = disagree, 1 = strongly disagree). Subtotal scores for each domain were yielded by adding up the subordinate item scores. A total K-PCAT score was calculated by summing the subtotal scores of the 7 domains.

As correlated variables, we obtained information on various sociodemographic characteristics and healthcare utilization. Sociodemographic variables included age, sex, job (student/staff/faculty), income level (low/middle/high), self-perceived health (good/fair/poor), presence of medical doctor in one’s family (yes/no), and comorbid medical conditions. Comorbidity was self-reported and categorized into 3 groups: none, only acute conditions, and chronic conditions. Chronic condition was defined as the presence of any of the following: hypertension, diabetes, hyperlipidemia, heart disease, hyperuricemia, chronic viral hepatitis, arthritis, cancer, depression, and anxiety disorder. Variables on healthcare utilization included having a regular doctor (yes/no), the number of ambulatory care uses per year (0–3, 4–6, 7–12, ≥13), and medical expenses per year (out-of-pocket payment only). We asked if anyone in their family regularly visited a doctor for chronic conditions (yes/no).

Statistical analysis

Characteristics of the participants were summarized as numbers and percentages for categorical variables and means ± standard deviations (SDs) for continuous variables. We compared the K-PCAT scores between the healthcare facilities and examined the factors associated with the differences. We also evaluated correlates of the total K-PCAT score for each care facility. General linear models were used to calculate regression coefficients (βs) and standard errors (SEs) with adjustment for sociodemographic and healthcare utilization variables. Missing item values of K-PCAT were treated as zero. In sensitivity analyses, participants with any missing value for the K-PCAT items were excluded from the analyses. Statistical tests were 2-sided and P < .05 was considered significant. All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA).

Comparison of K-PCAT scores across facilities

The K-PCAT total score was highest for the university health service (61.0 ± 15.9) and lowest for hospitals (48.1 ± 14.5), with significant differences between facilities (P < .001) (Table 2). Strengths in the key attributes of primary care varied across the facilities (Fig 1). The university health service received the highest scores for first contact, comprehensiveness, personalized care, and family/community orientation; community clinics for continuity; and hospitals for coordination and trust/satisfaction. Conversely, the university health service was rated lowest for continuity; community clinics for coordination, family/community orientation, and trust/satisfaction; and hospitals for first contact, comprehensiveness, continuity, and personalized care. In sensitivity analyses excluding participants with missing data on any item of the K-PCAT, results were similar: the K-PCAT total scores were 63.8(± 12.9), 57.1(± 10.5), and 51.2(± 11.0) for the university health service (n = 4,772), community clinics (n = 4,879), and hospitals (n = 4,333), respectively, with significant differences between facilities (P < .001).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Comparison of patient-perceived primary care qualities across healthcare facilities by domains of K-PCAT.

K-PCAT = Korean Primary Care Assessment Tool. Each domain score was calculated as the mean of subordinate item scores. Each item score ranged from 1 (low) to 5 (high).

https://doi.org/10.1371/journal.pone.0230034.g001

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Comparison of the K-PCAT scores between different types of healthcare facilities.

https://doi.org/10.1371/journal.pone.0230034.t002

Correlates of the differences in K-PCAT total scores between facilities

We examined the factors significantly associated with the differences in the K-PCAT total scores between the facilities (Table 3). Participants who rated higher scores for the university health service than hospitals were older, students, and those with low income levels, having no regular doctor, using ambulatory care service more frequently, and spending less on medical expenses. Individuals who rated higher for community clinics than hospitals were women and those in good health, with low income levels, using ambulatory care service more frequently, and spending less on medical expenses. Correlates of higher scores for the university health service than community clinics were old age, men, students, having chronic medical conditions, having no regular doctor, and using ambulatory care services more frequently. Correlates of higher scores for the community-based primary care facilities (both the university health service and community clinics) than hospitals included good self-perceived health, low income levels, frequent ambulatory care use, and low medical expenses. Participants with low income levels rated the university health service the highest, and hospitals the lowest. We additionally investigated the correlates of the K-PCAT total scores for each facility (S1 Table). Participants with high income levels rated hospitals higher than those with low income levels. Frequent ambulatory care users perceived the university health service and community clinics more highly and hospitals lower than did less frequent users.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Factors associated with the differences in K-PCAT total scores between healthcare facilities.

https://doi.org/10.1371/journal.pone.0230034.t003

Perspectives of primary care

The concept of primary care is defined in Korea as “a healthcare service first encountered by people. Primary care physicians resolve most of the common health needs of the population, and see patients in the context of family and community with continuous doctor-patient relationships, coordinating healthcare resources appropriately.”[33] The perspective of primary care is built on the principles of clinical efficiency, cost-effectiveness, and equitable care for the population.[34] Previous studies demonstrated that specialist-dominated health systems enforce inequity in access, whereas primary-care-oriented systems tend to be more pro-poor, equitable, and accessible.[6,10,11,35] Broadening access to primary care can reduce demand for expensive, specialist-led hospital care. In our study, individuals with low income levels, using ambulatory care services more frequently, but spending less on medical expenses rated primary care facilities higher than hospitals. These results suggest that providing high-quality primary care can expand healthcare utilization for the low-income population at lower cost. In particular, the university health service was better assessed among students, individuals with chronic medical conditions, and those without a regular doctor. Therefore, it may fulfill the unmet needs of these vulnerable groups in a university, enhance their accessibility, expand service utilization, and reduce health inequalities.

A setting-based, people-centered primary care model

Setting-based (e.g., schools and workplaces) health promotion has been emphasized by the WHO.[36] People-centeredness is one of the core principles of primary care: through delivery points embedded in communities, health systems can better respond to people’s needs.[2,7] Previous studies indicated that a higher level of population health was achieved when primary care providers served a well-defined population in a designated area, knowing their target population better.[4,30] In the present study, the university health service received the highest user-perceived score and had strengths in many attributes of primary care. A university health service is usually the most accessible primary care point on campus; care providers on campus best apprehend the university’s health issues, resources, and physical and sociocultural environments within a campus. Therefore, it can serve as an excellent model of setting-based, people-centered primary care. The main financing of the university health service, where this study was conducted, is from the university funding with some revenue through reimbursement by the National Health Insurance and out-of-pocket payments. Thus, it has a mixed financing of public, self-governing, and non-profit models. Similar to our results, an alternative non-profit primary care service was evaluated more highly in the performance of primary care than public or private/for-profit facilities in Korea and other countries [34,35]

Continuity of care, a sustained partnership between patients and physicians, is a key element of high-quality primary care.[37] A doctor knowing a patient’s history and values can provide more suitable care.[38] Many previous studies indicated that continuity of care can improve patient adherence and self-management, improved health outcomes, and lower unnecessary healthcare utilization and costs.[38–40] We found that the university health service needs to be improved in continuity of care. Because this weakness might be attributed to the properties of group practice, it could be overcome by establishing a regular doctor system.

Primary care services of community clinics

Community clinics received higher user assessments than hospitals; however, their score was lower than that of the university health service. Specifically, community clinics had the lowest scores for care coordination, family/community orientation, and trust/satisfaction. These findings can be explained by the following reasons, mostly linked to the fundamental problems of the Korean health system. First, primary care services of community clinics are heterogeneous. Many of community doctors in Korea are specialists and sub-specialties rather than general physicians (e.g., family doctors).[16,23,24] Since the range and qualifications of primary care providers have never been explicitly defined at the national level, all medical doctors can run a private clinic and practice as the first-contact point, regardless of their specialties.[16,20,22] Prior studies revealed that most specialists had no correct understanding of the roles of primary care and tended to underuse preventive care, focusing on curative and specialty care,[7,16,23] whereas general physicians were more likely to provide comprehensive and family/community-oriented care.[6,20,35]

Second, the private sector, aiming to maximize profits, dominates the health system in the absence of clear distinctions between the roles of hospitals and community clinics. Despite the inherent characteristic of health services as public commodities, these circumstances have shaped the health system to resemble a laissez-faire market, consequently leading to excessive competition between care providers.[16,34] Health services, including primary care, are highly fragmented and overlapping with little coordination between providers, even in the National Health Insurance system.[21–24] A survey of 466 general internists reported that coordination and comprehensiveness were most vulnerable attributes of primary care in Korea; one third of the physicians were reluctant to refer their patients for fear of losing them.[18]

Third, the fee-for-service payment system, a dominant reimbursement schedule in Korea, is unfavorable for achieving the key attributes of primary care (e.g., comprehensiveness, coordination).[16,18] As most services for patient education are not reimbursed, there is little financial incentive for doctors to focus on disease prevention and health promotion.[18,22]

Lastly, an overwhelming majority (around 94%) of community clinics are run by solo practitioners.[17] Hence, they have strength in continuity of care as found in our results, but tend to be limited in manpower and infrastructure to ensure quality of care.[22]

Hospital outpatient care

Hospital outpatient services were rated lowest in the performance of primary care of all facilities with many important primary care attributes unfulfilled. Since the original mission of hospitals is the provision of specialty care for complicated or rare cases, not comprehensive and integrated primary care, our results seem to be inevitable. Although hospitals were highly assessed in care coordination, this might imply that hospital patients tended to be easily referred from a specialist to other specialists within the same hospital once patients entered the hospital system,[17] possibly leading to fragmented care and over-utilization of specialty care.

Globally, the importance of providing healthcare at the right time and place, without compromising on care quality, is being emphasized.[3] Unnecessary utilization of hospital outpatient care is one of the main contributors of escalating healthcare costs.[25] Substituting hospital outpatient care with primary care is recommended as a solution to accomplish efficient and cost-effective healthcare delivery.[1,3] Evidence supports this perspective, indicating shorter waiting lists, shorter clinic waiting times, and higher patient satisfaction after the substitution interventions.[25]

Nonetheless, we found that the levels of trust and user satisfaction were highest for hospitals. This user view might be partly linked to the belief that good facilities and equipment of large hospitals guarantee good qualities of care.[13,23] Similar user misperception and distorted health-seeking behaviors were also reported in China; community individuals preferred to seek care in second or tertiary hospitals rather than primary care facilities, although the latter provide more accessible and affordable care. An important driver for this health-seeking behavior was trust in doctors in tertiary hospitals.[13] Therefore, efforts to raise public trust in primary care physicians and facilities are essential.

Strengths and limitations

Our study has several strengths. We enrolled a large population in a university setting. To the best of our knowledge, studies of primary care in a university setting are rare worldwide. Only 2 universities in Korea, including the university where this study was conducted, have accredited primary care services on campus covered by the National Health Insurance System. Because Korean citizens can freely choose and visit multiple healthcare facilities without regulation,[16] we were able to compare the real-world patient experiences of primary care services in the community population. To obtain information that was as unbiased as possible, the questionnaire was self-administered and anonymity was ensured. Relatively homogeneous participants exposed to similar geographic and sociocultural environments were sampled, which might have reduced the possibility of unmeasured confounding.

This study has following limitations. First, the cross-sectional design of the study limits causal and temporal inferences. Second, we evaluated user experiences and perception rather than health outcomes. Third, because we did not specify particular clinics, hospitals, and specialties, heterogeneity within the comparison groups might have affected our results. However, it might also have helped us to aggregate various user experiences in the real world. Fourth, we were not able to rule out sampling bias. Because our sample was recruited from a university, this might have influenced the positive results for the university service. Although we did not restrict our sample to the users of the university health service to decrease such bias, our sample size (16.9% of the study population) and the non-random sampling method could have limited its representativeness. Fifth, covariate information was obtained by participants’ self-report, thus we cannot exclude the possibility of misclassification. Lastly, our findings, derived from a single university, may not be generalizable to the entire Korean population or applicable to different health systems.