Equality and social determinants of spatial accessibility, availability, and affordability to primary health care in Hong Kong, a descriptive study from the perspective of spatial analysis

Ideally, primary health care (PHC) should bring healthcare as close as possible to where people live and work, and constitutes the first element of an ongoing healthcare process. PHC is widely perceived to be the backbone of a national health care system [1, 2], with evidence demonstrating that higher quality primary care is associated with better health quality outcomes, higher patient satisfaction, and lower health-related inequity at lower cost [3‐6]. As such, many countries are strengthening community-based primary care, focusing on prevention and quality improvement in disease management [7‐9].

In Hong Kong, family doctors are the main primary care service providers to residents [10]. Strengthening the role of primary care is a key element of the ongoing reform strategy in Hong Kong [11, 12]. In 2008, the Government set out a comprehensive package of proposals for reforming the healthcare system, including enhancing primary care development. A year later, in 2009, it published the "Primary Care Development in Hong Kong: Strategy Document", setting out the major strategies and pathways of action for improving the delivery of primary care services. Then in 2017, Hong Kong set up the District Health Centre (DHC) as a pilot to strengthen medical-social collaboration and public-private partnership [12]. Whilst this is a worthy aspiration, it is no easy task to achieve because the current primary care system in Hong Kong remains provider-dominated and fragmented [13‐15]. In Hong Kong, 70% of primary care is provided by the private sector, while the other 30% is mainly provided by 73 government-run general outpatient clinics (GOPCs) [16‐18]. GOPCs provide service to all the population but target the poor and the elderly. Primary care is more frequently managed privately, due to greater accessibility compared with public care [19]. However, elderly people, particularly those with lower incomes, tend to go to the publicly funded general out-patient clinics for their primary care, as the majority cannot afford private healthcare service [20]. Thus, affordability is said to be an impediment to receiving primary care in Hong Kong. Introduced in 2009, the Elderly Health Care Voucher Scheme provides an annual allowance in the form of vouchers [21] for citizens aged over 65 years to use on private outpatient services, which restructured health financial equitably for elderly patients to seek private providers [22, 23].

Accessibility of healthcare is a broad and flexible concept. Scholars have proposed a classic understanding of “access”, defining the measurement dimensions of health accessibility as availability, accessibility, accommodation, affordability, and acceptability [24]. But the dimensions of access to healthcare are not always applied as they were originally conceptualized, and have been modified variously in subsequent studies [25‐28]. Access to primary health care (PHC) is considered a fundamental right and an important facilitator of overall population health [29, 30]. Ensuring equal access to primary care for those in equal need has long been of concern to public health policy makers, service providers, researchers, and consumers alike [31‐33]. Spatial access to healthcare refers to the geographic distribution of physicians and patients. It constitutes a measure of healthcare delivery and encompasses both accessibility and availability: travel costs paid by patients as well as the adequacy of the physician supply at the point of care [34‐39]. This article focuses on the measurement of potential spatial accessibility and availability, exploring spatial affordability, while ignoring aspatial considerations like acceptability or accommodation.

An accurate assessment of PHC is essential to understand and address existing inequalities, thus narrowing the gaps in access to PHC [40, 41]. Methods measuring spatial access can be roughly classified into the following categories: provider-to-population ratios (PPR); distance to nearest provider; and gravity models [26]. Many new techniques have been introduced based on the gravity model, of which the Two-Step Floating Catchment Area (2SFCA) method has gained the most traction [42]. Based on this, several modifications and extensions have been suggested [43‐46]. In this paper, distance to nearest provider and 2SFCA methods were used to assess the accessibility, availability, and affordability of PHC in Hong Kong.

It is necessary to analyze the factors that influence spatial access to PHC, as they are essential for conditioning equitable access to strengthen Hong Kong’s primary care system [47]. Linear regression analysis is used to develop a regression model and test whether independent variables influence the dependent variable. Ordinary least squares (OLS) is one of the most frequently used methods to estimate the parameters of linear regression models [48], with several assumptions that must be fulfilled. In the context of spatial data, OLS is not appropriate, because spatial dependence is one of the main consequences of using geographical data [49]. Spatial dependence occurs when the accessibility present in a particular suburb is determined both by its own characteristics as well as by the characteristics of nearby properties. Spatial regression models can thus deal with the spatial dependence issue. Many researchers have applied spatial regression models to analyze social problems [50‐52]. In this paper, we use spatial regression model to analyze the social factors affecting access to primary health care.

In existing literature, many studies have analyzed or evaluated the spatial accessibility or availability of PHC services [53‐57]. However, few articles incorporate the other aspatial factors, like affordability into the measurement of spatial access. Additionally, researches focusing on Hong Kong mainly discuss the delivery system of primary care [58, 59] and reflect on social factors influencing the utilization of primary care [19, 59]. By contrast, few studies provide a comprehensive map of spatial access to primary health care in Hong Kong or analyze the factors affecting potential access to PHC. This study takes Hong Kong as the study area, integrating accessibility, availability, and affordability of primary health care in the perspective of spatial view, modeling the equality of accessing PHC, using a spatial regression model to better understand the social determinants affecting access to it in Hong Kong.

The SARAR (Spatial AutoRegressive with additional AutoRegressive error structure) model follows an autoregressive process, which is indicated by the presence of dependence relationship among a set of observations or spatial units [66, 67]. It is written as

Where Y is an n-by-1 vector of response variables, X is an n-by-p matrix of explanatory variables; ρ is spatial lag coefficient parameter that will determine the strength of the spatial autoregression on the response variable and |ρ| < 1; λ is a spatial error parameter that will determine the strength of the spatial autoregression on μ and |λ| < 1; μ is an n-by-1 vector of error terms assumed to have autocorrelation; β is a p-by-1 vector regression coefficients; and W, M are standardized spatial weights matrix (n×n) ( ∑_jw_ij = 1, ∑_jm_ij = 1 ) where n is the number of observation. Each observation of the spatial-lag variable is a weighted average of the values of the dependent variable observed for the other cross-sectional units. The matrix containing the weights is known as the spatial-weighting matrix. It is a quantification of the spatial relationships that exist among the observation location. All modelling processes were calculated by R programming using “spdep” packages. The SARAR model has two types of spatial regression model.

To identify whether there is autocorrelation in the values of the dependent variable or in its errors, a Lagrange Multiplier (LM) test and a Robust Lagrange Multiplier (Robust-LM) test were conducted. Goodness-of-fit statistics such as the R-squared, Log likelihood, AIC and BIC can be used to estimate the fitting degree of regressions.

From the analysis of time cost in accessing the nearest PHC facility, we deduced that in Hong Kong, the maximum walking time for residents to the nearest PHC facility is over 1 hour, while the average walking time is around 5.21 min (SD ±6.55min). Figure 1 shows the travel time map, while Fig. 2 shows the statistical chart of access time to public and private PHC and the coverage of the population. Among 1,622 LSBGs, over half LSBGs with 66% of residents can reach a PHC institution within 5 minutes: 435 LSBGs with 21% of residents can reach a PHC institution in 5-10 minutes; 148 LSBGs with 9% of residents can reach a PHC institution in 10-15 minutes; 70 LSBGs with 4% of residents can reach a PHC institution in 15-30 minutes; 28 LSBGs with 1% of residents can reach a PHC institution in 30-60 minutes; and 5 LSBGs with 0.1% of residents can reach a PHC institution in over 60 minutes.

In summary, 95% of residents in Hong Kong can reach a PHC institution within 15 minutes, 99% of residents can reach a PHC institution within 30 minutes, and almost all residents can reach a PHC institution within 60 minutes. There is a discrepancy between the contribution of public sector and private sector to the accessibility of PHC. For public PHC institutions, 56% of residents can reach one within 15 minutes, 90% of residents can reach one within 30 minutes, and 98% of residents can reach one within 60 minutes. For private PHC institutions, 95% of residents can reach one within 15 minutes, 99% of residents can reach one within 30 minutes, and almost all residents can reach one within 60 minutes.

The 2SFCA method was used to express he availability of PHC within the threshold range of 15 minutes walking distance with the index of A_i. From the analysis of availability, it was determined that in Hong Kong the maximum value of A_i is over 10, and the average is around 2.06 (SD ±1.44), as shown in Fig. 3. Figure 4 presents the statistical chart of availability to public and private PHC and the coverage of the population. It reveals that the Ai value of 72 LSBGs with 4% of residents is less than 0.5, and the Ai value of 217 LSBGs with 14% of residents is greater than 0.5 and less than 1. This means that in a month, 17% of residents cannot get PHC within the threshold range of 15 minutes walking distance. On the other hand, 491 LSBGs with 34% of residents have an Ai value between 1 and 2; 386 LSBGs with 28% of residents have an Ai value between 2 and 3; 118 LSBGs with 10% of residents have an Ai value between 3 and 4; 74 LSBGs with 6% of residents have an Ai value between 4 and 5; and 74 LSBGs with 4% of residents have an Ai value greater than 5.

In summary, 83% of residents in Hong Kong can receive PHC service within the threshold range of 15 minutes walking distance in a month (Ai>1). Among them, 48% of the residents are able to access PHC supplies that are greater than twice their needs (Ai>2); 21% of the residents to PHC supplies that are more than three times their needs (Ai>3); a 4% of the residents have access to PHC supplies that are more than five times their needs (Ai>5). There is a discrepancy between the contribution of public and private sectors to the availability of PHC. Only 8% of residents are able to receive public PHC services within the threshold range of 15 minutes walking distance, while 67% of residents can receive private PHC services.

We assumed that residents who have medical insurance covering private outpatient service would prefer service from private PHC institutes, while residents without medical insurance covering outpatient service, will take public PHC as their first choice in the process of seeking PHC. The Ai index was used here to express the availability of PHC services for Medicare and non-Medicare residents. For people without medical insurance seeking public PHC services, 58% are assigned an Ai value of less than 0.5; and 33% of them an Ai value greater than 0.5 and less than 1. This means that in a month, only 9% of residents who have no medical insurance can afford PHC service. For people with medical insurance seeking private PHC services, 0.2% of them have an Ai value less than 1, while 5.2% have an Ai value greater than 1 and less than 10, and 94% of them have an Ai value greater than 10. This means that in a month, almost all residents (99.8%) who have medical insurance can afford the PHC they need within the threshold range of 15 minutes walking distance.

In addition to medical insurance, the medical voucher scheme for the elderly also increases the affordability of medical products for the target population. Assuming that all Hong Kong seniors over the age of 65 are eligible for the medical voucher scheme and are willing to obtain private services to avoid crowding public services, then for people without medical voucher or insurance, 20% can afford PHC within the threshold range of 15 minutes walking distance. By contrast, for people with insurance or medical voucher, 95% can afford PHC.

In summary, with health insurance, 12% of residents in Hong Kong can afford PHC service within the threshold range of 15 minutes walking distance (Ai>1) in a month. The other 88% residents cannot afford it. With the additional support of medical vouchers, 32% residents in Hong Kong can afford PHC service within the threshold range of 15 minutes walking distance (Ai>1) in a month, while 68% residents cannot afford it. Thus, there remains a large gap between the availability and affordability of PHC in Hong Kong, even with help from medical insurance and medical vouchers for the elderly (Fig. 5).

With regards to the accessibility of PHC in Hong Kong, our results reveal that 95% of residents can access a PHC facility within 15 minutes, which is a good achievement from the perspective of geographic access time compared with other regions [27, 56, 69‐72]. It indicates that the medical resources of PHC in Hong Kong, are fairly evenly distributed among residents such that most residents can access one PHC institute conveniently. It is noteworthy that the private medical institutions make a major contribution to the accessibility of PHC for Hong Kong residents, as the number of private PHC facility in Hong Kong is much greater than that of public PHC facilities. As the results reveal, only 56% of residents can reach a public PHC institution within 15 minutes, while 95% of residents can reach a private PHC institution within 15 minutes.

If PHC is accessible, is the supply of services sufficient to meet residents’ demand? Considering the capacity to provide primary health services, we found that 8% of residents can receive a public PHC service within the threshold range of 15 minutes’ walking distance in a month, while 67% of residents can receive a private PHC service. Combining the capacity of public and private PHC, 83% residents in Hong Kong can receive a PHC service. Thus, it is evident that private institutions have great advantages over public institutions in terms of capacity to meet residents' primary health care needs.

If PHC is available, then can people afford it? The public medical system in Hong Kong, including PHC services, is free and accessible to all residents. Due to the high cost of private medical systems, private medical services have a threshold for most residents, especially those who do not possess private medical insurance. In Hong Kong, people's medical insurance mainly covers hospital products, while the coverage rate of doctor consultation is low [64]. If uninsured residents seek public services and insured residents seek private services, the availability of primary medical services for Hong Kong residents will become less affordable. As the results indicated, only 12% of residents in Hong Kong can afford PHC service; while with the support of Medical Voucher, 32% residents in Hong Kong can afford PHC service. Even with the intervention of both medical insurance and medical vouchers, 68% of Hong Kong residents still cannot afford PHC.

The poorer accessibility among high-income groups, reflects that the distance between residential address and PHC facilities is longer compared with other groups, which means they have poorer access to PHC. But this kind of inconvenience is because we set walking as the way for residents to seek health care providers. High-income people have access to more modes of travel, and walking is not the main way they seek medical treatment. The use of walking as the main mode of travel for PHC accessibility in this paper therefore underestimates the actual accessibility and availability of PHC services, especially for high-income residents. What’s more, it’s noticeable that among these subgroups, the highest income group has the largest differences in accessibility and accessibility from the other groups, while the differences between the other groups are not particularly large. One possible explanation is that the truly wealthy people tend to live in the suburbs, which may make their physical distance to health care even longer with lower accessibility and availability to PHC on foot.

During the outbreak of the COVID-19 pandemic, the PHC system has provided not only primary medical services, but also partial public health services. In Hong Kong, some of the primary medical institutions are entrusted by the government to provide vaccination services to the public and consultations and drugs to COVID-19 patients. Sound PHC systems can enhance regional public health response capacity, by making full use of its basic network of close ties with residents. Accessibility of PHC essentially reflects the spatial distribution of institutions, and the accessibility focuses on the spatial distribution of medical resources. Thus, a good accessibility and availability of PHC reflects that the network layout of existing PHC systems in Hong Kong is reasonable and can meet the needs of most residents.

The accessibility and availability of PHC in Hong Kong is good, but the affordability of PHC is not ideal. Low coverage of medical insurance in outpatient services has aggravated the crowding of public medical services and led to the underutilization of private PHC resources. It will be hard to realize the role of private institutions without addressing the issue of affordability of private services. Therefore, we suggest diverting patients to private institutions through medical insurance, medical vouchers, or other ways to relieve the pressure on the public system and make full use of the existing PHC network in Hong Kong. The government has been aware of the problem of crowded utilization of public medical resources and taken active measures to divert patients from public medical institutions to private institutions. In 2019, it launched a policy called the Voluntary Health Insurance Scheme (VHIS) [73]. The purpose of VHIS is to provide residents with an additional choice of using private healthcare services through hospital insurance and relieve the pressure on the public healthcare system. However, the scheme covers only hospital products. If primary health care products are included, it can also help alleviate crowding of primary health care in public institutions.

Our study possesses some limitations. First, we use walking as the mode of travel to seek PHC services, ignoring other possible options. This makes the results of accessibility in this paper pessimistic. However, the accessibility of primary health care services in Hong Kong is already good on foot, and if other transport factors are considered, the coverage of PHC will increase, and the time of access will be further reduced with additional costs. PHC is set for the whole residents, but the elderly are an important target group for primary health care. Moreover, the elderly experience restricted mobility. Thus, it is appropriate to take walking as the mode of travel to access primary health care. If precise geographic accessibility is required, other travel patterns may need to be added. Second, the number of visits accepted by public PHC institutions for a month was obtained from official reference data, but the number of visits accepted by private institutions was estimated. Assuming each private clinic has one registered doctor who works five days a week and treats 20 patients per day, then a private clinic can provide 400 visits per month. In calculating the availability, we set 400 visits as the monthly volume of services provided by each private clinic. A clinic may have more than one registered physician, so the availability of private primary health care services measured in this paper is underestimated. Even if the number of services provided by private clinics was underestimated, the accessibility of residents calculated in this paper was good, and the accessibility value would be even greater. Third, in calculating the affordability of PHC services, we assumed that people without medical insurance prefer public services, while people with medical insurance prefer private services. Moreover, all residents aged over 65 were assumed to be eligible for medical vouchers and to prefer private services in seeking PHC. Other factors affecting health behavior like the severity of the disease and economic status were ignored in this paper. Finally, in analyzing potential factors impacting the accessibility, availability, and affordability, the smallest geographic unit used in this paper is the LSBGs, and variables, such as age, education, income status, were all descripting characteristics of LSBGs. If we could get more detailed individual data in future, we can further analyze a more completed demographic factor analysis, including gender, marriage status, career and so on. What’s more, this paper aims to provide a general view of accessibility, availability, and affordability of residents of PHC to residents in Hong Kong. In future, in-depth analysis is needed to research on the primary care services for vulnerable groups, such as the differently abled and critically ill patients.