In the analysis of all-cause and ACSC-related hospitalization, RVF = 1 was again used as the reference group (Figs. 3 and 4). Kaplan -- Meier plots for all-cause and ACSC-related hospitalizations were shown in Figure S6 & S7. Participants with an RVF of 0 had the highest HR for hospitalization (HR: 2.08, 95% CI: 2.04-2.12, p < 0.0001). In patients with RVF of > 1, higher RVFs were associated with an increasing HR for all-cause hospitalization, although these were all lower than the HR for an RVF of 0 (Fig. 3). For ACSC-related hospitalizations, participants with an RVF of 0 also had the highest HR (HR: 2.08, 95% CI: 2.00-2.16, p < 0.01). Compared to participants with an RVF of 1, those with RVFs of 2 and 3 had slightly lower HRs (HR: 0.96, 95% CI: 0.94-0.98, p < 0.0001 for RVFs of 2, HR: 0.97, 95% CI: 0.95-0.99, p = 0.0054 for RVFs of 3), while those with RVFs of > 5 had a higher HR for ACSC-related hospitalizations (HR: 1.13, 95% CI: 1.06-1.22, p < 0.01) (Fig. 4). Outpatient medical costs were also compared using RVF = 1 as the reference. An upward trend was observed with increasing RVF levels. Participants with RVFs > 5 had outpatient costs 3.21 times higher than those with an RVF of 1 (95% CI: 3.17-3.26, p < 0.0001) (Fig. 5).
In this large-scale study, polydoctoring was associated with a lower all-cause mortality rate. However, polydoctoring was also associated with an increased hospitalization rate and higher outpatient medical costs. Thus, while polydoctoring may contribute to improved survival, it comes at the cost of increased healthcare utilization and expenses, raising concerns about sustainability from a healthcare system perspective. A U-shaped relationship was observed with ACSC-related hospitalizations, indicating that RVFs of 2-3 may strike an optimal balance between minimizing care fragmentation and maximizing benefits, while RVFs of ≥ 5 warrant caution due to increased risks. Our findings also suggest that regular medical care is crucial for older adults with chronic conditions, as those with no RVFs exhibited the highest rates of mortality, hospitalizations, and ACSC-related admissions, underscoring the importance of usual care for this population. While previous studies have described the concept of polydoctoring or assessed care fragmentation using provider-level continuity indices, this study is novel in evaluating polydoctoring based on a facility-level measure (RVF) and linking it to multiple patient outcomes -- including mortality, hospitalization, and outpatient costs -- using large-scale claims data from an elderly population with multimorbidity in Japan. By using a standardized metric applicable across diverse clinical settings, our findings offer empirical evidence for understanding the system-level implications of care fragmentation in Japan's specialist-driven healthcare model.
The observed trend of lower mortality with increased RVFs could be attributed to the benefits of specialized care. Previous studies have demonstrated that specialist care can lead to better outcomes in patients with certain conditions, such as diabetes and heart failure, compared to generalist care. The improved survival rate in our study may be attributed to the expertise and targeted treatment offered by specialists managing individual chronic conditions. However, not all studies have demonstrated the superiority of specialist care for every condition, with some reporting equivalent outcomes between specialist and generalist care. This suggests that the benefits of polydoctoring may vary depending on the combination of conditions a patient has. Therefore, further research is needed to investigate how different disease patterns influence the impact of polydoctoring. Additionally, it may be necessary to examine the combination of conditions and optimal number of healthcare providers, considering patient and physician capacity when determining the appropriate RVF. Machine learning approaches may be useful in identifying combinations of conditions that are associated with better outcomes depending on the level of RVF.
Regarding ACSC-related hospitalizations, we observed a slight decrease in admissions for patients with RVFs of 2 or 3 but an increase in those with RVFs of ≥ 5. This suggests that RVFs of 2 or 3 might lead to better care coordination and a reduction in ACSC-related hospitalizations. This could be because care coordination is more manageable at this level, whereas having many providers (RVFs ≥ 5) may result in care fragmentation and increased hospitalization. Previous studies have demonstrated that maintaining higher continuity of care leads to fewer ACSC hospitalizations and emergency visits. Our findings support this, suggesting that involving too many healthcare providers may interrupt this continuity, reducing its overall benefits.
Though this study focuses on Japan, the findings have broader implications for countries facing aging populations, especially given Japan's status as one of the countries with the highest life expectancy. Japan's healthcare system, with its free access to specialists and the lack of a GP gatekeeper model, provides a unique environment for studying the effects of polydoctoring. In contrast to countries such as the UK, where GPs act as gatekeepers, Japan's system allows for more frequent and direct access to specialists, leading to higher rates of polydoctoring. Gatekeeping may help reduce healthcare costs by restricting access to specialists, but it also carries the risk of worsening patient outcomes, and evidence on its overall effectiveness remains inconsistent. This specialist-driven primary care model in Japan may have contributed to the reduction in mortality observed herein, particularly among older adults.
However, an important finding from this study is the strong association between higher RVFs and increased outpatient medical costs. This is particularly relevant in countries like Japan, where a rapidly aging population is expected to place significant strain on healthcare resources. As multimorbidity becomes increasingly common among older adults in high-income countries, financial sustainability of healthcare systems is becoming a pressing concern. Previous research has shown that when specialists are the usual source of care, continuity tends to decrease, whereas continuity remains higher and healthcare costs lower when generalists provide primary care. Recognizing these challenges, Japan has begun to focus on developing GPs capable of delivering comprehensive care, starting from undergraduate medical education. Historically, Japan's primary care system has been largely provided by specialists. Therefore, many conditions associated with polydoctoring, such as osteoarthritis, osteoporosis, benign prostatic hyperplasia, and allergic conjunctivitis, fall within the scope of family medicine in other countries. Furthermore, the development of GPs as specialists in Japan has only recently gained momentum. Investigating whether an increased presence of GPs can help reduce healthcare costs without compromising mortality rates is essential. Additionally, the optimal management approach for multimorbidity in older adults may vary significantly depending on the prioritized outcomes. As highlighted by the Quadruple Aim framework proposed in the United States, assessing cost, patient experience, equity, and the well-being of care teams is crucial. A comprehensive, multidimensional analysis, including patient-reported outcomes, is necessary to ensure that healthcare systems effectively meet the diverse needs of both patients and healthcare providers. Furthermore, future studies should investigate regional disparities in polydoctoring and their potential impact on patient outcomes.
This study has some limitations. First, the RVF measure used in this study only captures the number of healthcare facilities visited without assessing the quality or coordination of care among providers. Care coordination, which is difficult to measure, may have influenced outcomes among patients with polydoctoring. Second, our findings were not adjusted for patients' socioeconomic backgrounds, such as education level and income, which could impact healthcare access and utilization and cofound our results. Future research should adjust for these variables to clarify the relationship between polydoctoring and patient outcomes. Third, as this study only examined the outpatient medical costs associated with polydoctoring, a more detailed cost-effectiveness analysis, including lifetime healthcare costs, is necessary to evaluate the sustainability of the healthcare system. Fourth, the study cohort was limited to individuals aged 75 to 89 years with multimorbidity in specific geographic regions in Japan, which may limit the generalizability of our findings to other populations or settings. Fifth, our definition of polydoctoring was based on the number of RVFs. However, due to data limitations, we could not determine whether the same physician or care team provided care at multiple facilities. Moreover, we were unable to account for situations in which patients were treated by multiple physicians within the same facility. As a result, our measure may not fully capture the actual number of distinct providers or the complexity of care fragmentation.