The structure of this chapter is as follows. The first section contains a summary of major findings and their interpretations. The second section discusses policy-related implications and recommendations. The third section provides study limitations and suggestions for future studies.
The final section presents an overall conclusion.
I. Principal Findings
a. Efficiency disparities exist between the North and the South
At first glance, northern cities averaged in the high 80% to 90% on their overall results.
This finding is consistent with Chen (2019), which analyzed congestion and utilization rate of Taiwan’s pharmacies from 2011-2016 using prescription claims data. Most of these areas rest around the Taipei Basin which is bounded by the narrow valleys of Keelung, the northernmost city (Taipei City Government, 2017). While occupying about 20% of Taiwan’s landmass, these regions hold almost half of Taiwan’s total populations (Department of Household Registration, 2020). As far as utilization is concerned, northern regions have several factors that played into their favor.
Compared to southern urban centers, northern cities are more tightly connected by railways, highways, and bus lines. Their compact infrastructures make easy trips to hospitals and pharmacies.
Socio-economic status and education attainment level could have also played a role. As Taiwan follows an unusual system of prescribing and dispensing, utilizing community pharmacy as a refilling site for CIRP’s will require the patients to have some knowledge about CIRP’s rules and methods. Past research has indicated that the proportion of population with the correct relevant knowledge is only 64.1% (Chuang et al., 2013). This would likely give northern cities an edge
36
where residents are more informed of the ways to utilize their local pharmacies. Hence, it can be argued that health policies should continue to aim at promoting CIRP’s in the South.
b. Northern efficiency declines without hospital-less districts
With most of the rural districts removed, southern suburban townships would fare better than the districts in Taipei (Table 14). Many top ranking townships share the semblance of small settlements outside large urban areas. Perhaps attributing to their decent transportation system, these suburban townships are characterized by minimal hospital inputs and a considerable amount of outputs in the pharmacy division. Based on this result, it can be inferred that even the most efficient city in the North may still have rooms for improvement, especially in the pharmacy division. For instance, the highly dense infrastructure and transportation system in Taipei sometimes make hospital trips “too easy”, which could lead hospitals and community pharmacies into a competition for CIRP’s. For many northern residents, it is just as easy to check in a hospital as it is to visit a nearby pharmacy. As each dispensed prescription is allotted reimbursement points by the NHI, both hospitals and pharmacies are financially interested to discourage patients to refill at the other end. In this regard, the oversaturation of hospitals in the North could be seen as having a thwarting effect on the number released CIRP’s. To address these impediments, urban areas should pay more attention to coordinated development between the two divisions.
c. Pharmacy division is relatively less efficient across the nation
Divisional difference is present in both city and district scale. The projection tables would suggest that cities and counties to increase on the number of released CIRP’s and community pharmacies (Table 6 and Table 12). Although it is arguable that improvements have been made over the 6-year period, the model does consider each studied year to be more efficient than the
37
preceding year, and 2019, the last studied year, to be the benchmark, an inference that efficiency has yet reached a true maximum.
The pharmacy division experienced an efficiency drop from 86.5% to 39.2%. This decrease is an expected outcome considering that rural populations tend to rely on community pharmacies more where hospitals are absent (Henkel & Marvanova, 2018). The hospital-less districts and townships can thus be accounted for a large portion of released prescriptions, and their exclusion favors the hospital division.
II. Policy-related Implications and Recommendations
The results obtained from the city/county analysis reveal a North-South imbalance that can be explained by the government’s long-term over-investment in northern Taiwan. While there is uncertainty into whether the human resource associated with input variables can be lowered without jeopardizing healthcare quality, medical access in southern regions can certainly be improved. In Taiwan, hospital establishment is a challenging task that must consider target patient base and strict protocols. On the other hand, community pharmacies are owned by a private individual or organization, rather than by the state or a public body; their establishment process can be much simpler than hospitals. For a pharmacist aspiring to establish his or her own community pharmacy, site selection will be influenced by many factors including the presence of hospitals and clinics within proximity. Therefore, in extremely inefficient DMU’s such as the rural districts in Hualien or Taitung, it may be imperative for health policymakers to incentivize community pharmacy entrepreneurship and transportation building based on local needs and conditions.
38
An in-depth look at the results obtained from the district/township estimation indicates that southern suburban districts are not recommended to be used as benchmarks in practice. As the model uses the number of community pharmacists as one of the only two input variables for the pharmacy division, DMU’s with fewer pharmacists are considered relatively efficient. However, descriptive statistics indicate that districts in Kaohsiung, Taitung, and Pingtung have the ratio of pharmacists and pharmacist approaching almost 1:1. In other words, these pharmacists in the South are very likely to be operating a community pharmacy alone. Although efficient on paper, this scenario is not a powerful example of pharmaceutical care quality.
III. Strengths and Limitations
The current study contributes to the empirical literature on healthcare utilization management and health equity. Performance measurement approaches by Dynamic Network DEA was selected to examine change in efficiency of medical and pharmaceutical resource utilization between 2014 and 2019. The main strength of this study lies in its use of a two-stage model to incorporate medical and pharmaceutical care as subunits of the overall healthcare service in Taiwan, with the hope of improving efficiency at a local level. Consequently, this paper could be considered to be the first study that conceived and incorporated community pharmacy utilization as a main contributor to health equity.
However, there are some limitations associated with the availability of data. First, the compilation of panel data as required by the DNDEA model restricts the study period to be 2014- 2019, because older data uploaded by the Department of Health before 2013 were found to be incomplete and limited the study period. The second limitation lies in the district/township dataset, where the number of community pharmacists is an estimation of the community pharmacist workforce. As the exact numbers of practicing community pharmacists for each district are not
39
obtainable, the district/township analysis uses an estimation which subtracts the number of hospital and clinic pharmacists from the total number of registered pharmacists. As the estimation may contain pharmacists who registered as reserve or industrial pharmacists, it is expected to be slightly larger than the actual number of practicing community pharmacists. Nevertheless, the estimated number gives us a general representation of the distribution of pharmaceutical resources for each district, which is in line with the original intention of this paper.
Finally, clinics data is omitted with reluctance. While there are functional differences between hospitals and clinics, clinics are capable of issuing CIRP’s and do account for about 20%
of the total CIRP’s issued each year. Inclusion of clinics data would allow this study to generate a more complete representation of healthcare utilization efficiency. However, incorporating clinics data into either division would be inappropriate because: 1. Clinics, lacking an inpatient and emergency department, do not share the same output and carry-over variables with hospitals. 2.
Many clinics in Taiwan feature a physician-owned pharmacy where pharmacists play a reduced role in pharmaceutical care. In light of this omission, the study results do provide an indication of what efficient and inefficient cities are, as well as the factors that assist in their identification.
IV. Suggestions for Future Research
There are some empirical issues that may be investigated for future research and examination. To examine whether inefficiency is related to resource distribution and health policy, it is feasible to group administrative regions into the Medical Care Networks as intended by the MOHW in 1985. A meta-frontier model can then be employed to evaluate the comparable technical efficiencies for DMU’s operating under different networks (See et al, 2021). Based on the result of district/township analysis, it can be speculated that northern administrative region has reached a state where inputs are overly invested, especially on the number of healthcare
40
professionals. A congestion analysis may explain the drastic decline in efficiency for northern region.
Regional development can be analyzed in various context, includingenvironmental, social and economic factors. To explain inter-DMU efficiency differences, a Tobit regression of the external factors associated with inefficiency may shed light on how to increase healthcare utilization in practice. The estimated result would allow researchers to assess factors such as the influence of population density and the degree of urbanization on efficiency.
Lastly, hospitals currently have a complete data record for use as inputs and outputs because they report to the Ministry of Health and Welfare directly. Community pharmacy data, on the other hand, are collected by individual districts and townships, and some of these data are reported on a voluntary basis. These obstacles make many insightful pharmacy indices, such as the number of non-chronic illness prescriptions released, less likely to be valid in research setting.
However, as health information are increasingly digital today, more indices for the pharmacy division are expected to emerge in the future and provide new directions for research.
V. Concluding Remarks
This study is one of the first attempts at analyzing efficiency of community pharmacy utilization in Taiwan using Dynamic Network DEA methodology. The study illustrates that a large majority of Taiwan’s central and southern cities are operating inefficiently. This result supports the conventional beliefs that there is an uneven distribution of resources with regard to healthcare between the North and the South. Compared to the North, there are more inputs and carry-overs not utilized in the production of hospital and pharmacy services in the South. With this information, policymakers will be able to make educated decision according to the results.
41
The pharmacy division is found running less efficiently than the hospitals. Hospitals in Taiwan are instrumental for care coordination and integration, and they complement many other parts of the health system. It can thus be argued that community pharmacies are dependent on hospitals in many aspects, including store site selection and the amount of CIRP’s they receive.
Since hospital managers generally don’t have much control over their service productions, it may come down to policymakers to promote rural pharmacies and the releasing of CIRP’s. What pharmacy managers can do is to devote more attention to the quality of pharmaceutical care, with a focus on improving pharmacy workforce issues.
This paper opens up a new way of evaluating the multi-dimensional nature of healthcare utilization in Taiwan. Although some of the variables developed in this study are specific to the assessment of Taiwan’s unique prescribing and dispensing system, they could be generalized to measure different regional efficiency by selecting suitable inputs and outputs. The use of more specific inputs and outputs is also worth considering in the process of measuring pharmacy utilization efficiency. It is hoped that this study will encourage future research on DEA application when new pharmacy-related indices become available, as well as on strategies to improve regional healthcare utilization efficiency.
42