Meeting the needs of a population is one of the most important considerations when allocating healthcare resources in Canada, and worldwide . Previous research has advocated the allocation of resources according to the needs of the population as assessed by an estimation method  that considers demographic-based indicators (e.g., age, education, and smoking) [36, 37]. However, examining traditional estimation methods for service needs, Kephart and Asada  noted substantial differences between estimated and real needs in some regions. A possible explanation for the biased estimation is that the needs estimation method is simply a linear combination of all of the considered factors, without considering how these factors interact with one another. Therefore, an in-depth understanding of the potential interactions among geodemographic factors (with certain factors moderating the effects of others on healthcare service utilization) can shed light on the design of better estimation methods for healthcare service needs. Further, as LHINs are sub-provincial administrative units responsible for planning and funding healthcare services for their corresponding geographic areas in Ontario , our study not only uncovers interesting relationships between LHINs’ geodemographic factors and healthcare service utilization, but also provides valuable knowledge for LHIN administrators to consider in their planning and/or managing of healthcare service resources.
In this paper, we have demonstrated that service accessibility has a significant moderating effect on the population size-service utilization relationship, and that educational profile exerts significant moderating effects on both the population size-service utilization relationship and the age profile-service utilization relationship, which are novel findings that have not been reported previously. The results of our analysis confirm our prediction that service accessibility is negatively associated with service utilization, and that it weakens the effect of population size on service utilization. The results suggest that the more healthcare services are accessible in an area, the fewer cardiac surgery patient arrivals any one particular hospital in that area will have. Take the Hamilton Niagara Haldimand Brant LHIN (LHIN 4) and its neighbor, the Mississauga Halton LHIN (LHIN 6) as examples. In 2007, the proportions of patients receiving cardiac surgery services in their resident LHINs (referred to as the inside-LHIN proportion) were 82% and 72%, respectively , whereas the service accessibilities for the two LHINs were approximately 51.54% and 88.20%, respectively, as shown in Table 2. Because both LHIN 4 and LHIN 6 have only one hospital in their own areas, the higher accessibility of LHIN 6 (compared to LHIN 4) suggests that there are more accessible hospitals in the LHINs surrounding LHIN 6 than in those surrounding LHIN 4. As a result, patients dwelling in LHIN 4 are less likely to visit hospitals in other LHINs, compared to those dwelling in LHIN 6, and thus the inside-LHIN proportion for LHIN 4 is higher than that for LHIN 6. Accordingly, we can expect that for LHINs with better accessibility to cardiac surgery services (e.g., LHINs 3, 6, 7, and 11 as shown in Table 2), the pressure of population growth in each of these LHINs on the hospital(s) within the LHIN may decrease.
In contrast, the negative but insignificant moderating effect of service accessibility on the relationship between age profile and arrival may be related to the fact that older people are more willing to visit a familiar hospital or a hospital with familiar physicians . Consequently, service accessibility in an LHIN, which reflects patients’ options in healthcare services, may have little effect on the senior population’s decisions in choosing cardiac surgery services.
The negative moderating effects of educational profile suggest that the effect of population size and age profile on service utilization is less pronounced in a well-educated population than it is in a less well-educated population. A possible explanation is that well-educated individuals, including those in old age, may have healthier lifestyles . These individuals also incline to receive routine physical examinations and engage in self-care behavior . Consequently, they are less likely to develop severe cardiovascular disease that requires cardiac surgery services . As illustrated in Table 4, the educational profiles of LHINs varied from 61.25% to 74.16%, with a mean value of 67.38% and standard deviation of 4.24% in 2006. This suggests that the effects of population growth and aging on patient arrivals in each LHIN may vary depending on the educational profiles of the LHIN. Specifically, as shown in Table 2, LHINs 6, 7, 8, and 11, which have more educated populations (indicated by higher-than-average educational profiles), may bear a lighter burden of patient arrivals due to population growth and aging, compared to other LHINs.
In addition, previous research has identified population growth and aging as two important factors driving the need for healthcare services in Ontario , and thus affecting patient arrivals. Likewise, our findings reveal a significant relationship between population size and service utilization, and between age profile and service utilization. This finding suggests that, for healthcare administrators, monitoring the trends in population growth and aging is an effective precautionary approach to providing sustainable healthcare services.
Finally, prior literature has noted the significant positive effect service utilization exerts on hospital wait time, an important performance indicator [40, 41]. Our findings suggest that geodemographic factors, such as population size, age profile, service accessibility, and educational profile, may indirectly affect wait time for cardiac surgery services via their influence on patient arrivals. Therefore, healthcare administrators should consider the role that geodemographic factors may play in their effort to improve wait time for healthcare services.
This study concentrates on investigating the relationships between geodemographic profiles and the utilization of healthcare services at an LHIN level. However, since cities/towns or communities contained in each LHIN may have distinct geodemographic profiles such as age and education, further research may zoom into the LHINs to investigate the effects of geodemographic profiles at a city/town/community level to gain more insights into their impacts. In addition to the geodemographic profiles examined in this study, it would also be desirable to investigate the effects of other factors on healthcare service utilization. For instance, the spatial distribution of risk factors for cardiovascular disease (e.g., co-morbidity of diabetes ) may dictate cardiac surgery service utilization.
Moreover, it would be interesting to explore whether our findings still hold for other cardiac care services such as regular checkups, diagnostic cardiac catheterization, and non-invasive percutaneous coronary interventions. This is because cardiac surgery is invasive and commands relatively scarce resources, and thus the geographic accessibility to cardiac surgery is quite limited when compared to other cardiac care services. For example, in 2005, 18 hospitals in Ontario provided angiography tests, while only 11 hospitals performed cardiac surgery services . Further, the purposes and risks for receiving cardiac surgery and for receiving other cardiac care services are also different, thus the service utilization patterns of cardiac surgery may differ from those of other cardiac care services.
Future research may also be carried out to extend our SEM testing method. In this study, the moderating effects of educational profile and service accessibility on the population size-service utilization relationship and the age profile-service utilization relationship are tested separately due to the collinearity between the two moderators. In the future, the moderating effects of the two factors may be tested simultaneously in a comprehensive model by using primary data. Moreover, in this study, each latent variable has only one indicator. Future research may examine additional indicators for each of the latent variables so as to capture more dimensions of these variables.