Introduction and background

The global COVID-19 pandemic has significantly impacted healthcare, including cancer care.1 To prevent infection spread and protect patients, several oncology societies such as the European Society for Medical Oncology or the American Society of Clinical Oncology have issued guidelines for cancer patients to reduce hospital visits or consider delaying treatment during the pandemic.2 3 Consequently, the number of cancer surgeries and patient visits significantly dropped temporarily in many countries, including Japan.4–7 Most decreases in patient visits were temporary, returning to prepandemic levels within months. However, lockdowns and outing restrictions may have permanently altered attitudes toward hospital visits.8 9

In Japan, regional healthcare is structured around a medical plan that segments the country into primary, secondary and tertiary service areas. In 2012, 348 secondary healthcare service areas (SHSAs) were established, providing healthcare from initial to advanced levels.10 The Cancer Control Act, promoting cancer prevention, was enacted in 2006 by the Ministry of Health, Labour and Welfare (MHLW) for cancer care.11 To achieve this, Designated Regional Cancer Hospitals (DRCHs), which specialise in cancer treatment, have been established across the country.12 Patients have the option to seek treatment at distant DRCHs or hospitals outside their SHSAs if they prefer. Reports indicate that about 20% of cancer patients in Japan choose hospitals far from their residences.13

This study aimed to investigate changes in long-distance visits outside of SHSAs among Japanese cancer patients during the COVID-19 pandemic. This will offer a chance to contemplate how hospitals should operate in the future pandemic.

Materials and methodsData collection

This retrospective observational study analysed inpatient data from the Japanese Diagnosis Procedure Combination (DPC) database. The DPC system standardises electronic claims and ensures hospital performance transparency. The MHLW uses DPC data to monitor healthcare trends in acute care hospitals, disclosing aggregated summaries online. Both patients and healthcare providers can access this information to choose hospitals based on clinical data and to enhance clinical practices.14 15 The database includes discharge and administrative claims data for all inpatients from over 1000 participating hospitals, covering 92% of all the tertiary care hospitals in Japan.16 17 Patient data from January 2018 to December 2021 were extracted from the database to compare the prepandemic and pandemic states. Patients diagnosed with cancer, as per the International Classification of Diseases 10th Revision (ICD-10) codes, were included: ICD codes C00 to C97 were selected.

Data analysis

We categorised the year 2018/2019 as prepandemic and 2020/2021 as postpandemic, following the identification of the first COVID-19 case in Japan in January 2020.18 We investigated whether the patient’s hospital was located within or outside their residential SHSAs. Admission rates outside the SHSAs were compared before and after the pandemic, focusing on age, sex, referrals from other hospitals and cancer sites based on the major diagnostic category. The following characteristics of patient residences were also analysed: COVID-19 infection rate, population density and the number of DRCHs. The infection level was categorised into three groups (high, medium and low) by dividing the 47 prefectures based on the total number of COVID-19 cases as of December 2021. The population density was also categorised into three groups (high, medium and low) of prefectures. The actual infection count and population density data were obtained from the MHLW, Japan. The three categories of high, medium and low were defined as patient populations in the upper, middle and lower 33rd percentile of the values, respectively. Prefectures were divided into two categories based on the number of prefectural DRCHs and regional advanced DRCHs, as per the MHLW January 2023 data:

‘high’ for prefectures with two or more hospitals, and ‘low’ for those with one hospital.

In the analysis of each cancer site’s treatment procedures, chemotherapy or surgery was categorised. The analysis for surgery was further refined by categorising only those surgeries performed for malignant tumours based on the medical practice code (K-code) defined by the MHLW. Lung cancer surgery was categorised as high technical surgery (including pneumonectomy with tracheoplasty, surgery with resection of adjacent organs, pleuropulmonary resection, lung segmentectomy and surgery for pyothorax) and other surgery, as per the classification by the Japanese Board of General Thoracic Surgery.

The correlations between categorical variables were analysed using χ2 tests. All tests were two-tailed and p values<0.05 were considered to indicate statistical significance. Univariate and multivariate logistic regression analyses were conducted to identify factors associated with the hospital admissions outside of SHSA.

The study protocol was approved by the Institutional Review Board of Keio University School of Medicine (20211151). All statistical data were analysed using R V.4.1.2 (2021; R Foundation for Statistical Computing, Vienna, Austria), and graphs were created with GraphPad Prism 9 (GraphPad Software, Boston, MA).

Patient and public involvement

Patients were not involved in setting the research question or formation of the study protocol.

ResultsChanges in hospital visit behaviour by patient characteristics

The data of a total of 2 394 759 patients were analysed. Table 1 presents a comparison of the actual number of patients admitted to hospitals outside of SHSAs. Overall, there was a significant decrease in admission rates outside the SHSA postpandemic compared with prepandemic years (26.67% and 27.58%, respectively; p<0.001). Figure 1A illustrates the percentage change in each age group before and after the pandemic. Regardless of the COVID-19 pandemic, younger patients were more likely to be admitted outside of SHSAs. Significant reductions were observed in all age groups over 40 years, with the highest decrease of −1.05% in those over 80 years old, followed by −0.75% in the 60 to 79 years age group. The percentage for the age group 19 years and younger increased, but not significantly. No significant gender differences were observed, with both genders showing substantial reductions (male: −1.13%, female: −0.58%) (figure 1B). In patients not referred from other hospitals, the rate significantly decreased (−2.60%, p<0.001), but not in referred patients (−0.11%, p=0.15) (figure 1C).

Figure 1

Effect of patient and residence characteristics on hospital visit behaviour. (A) Admission rates to hospitals outside of SHSA in each age category. (B, C) Rate changes in hospital admissions outside of SHSA based on gender (B) and referral status (C). Changes in hospital admission rates outside of SHSA based on COVID-19 infection severity (D), population density (E), and the number of DRCHs (F) in the patient’s residential area. Asterisks and black bars indicate p<0.05. DRCH, designated regional cancer hospital; SHSA, secondary healthcare service area.

Table 1

Comparison of hospital admissions outside of SHSA before and after COVID-19 pandemic

Changes in hospital visit behaviour by patient residences

Patient residence characteristics were deemed significant in influencing hospital visit behaviour. Figure 1D–F, respectively, illustrate the percentage changes based on COVID-19 infection grade, population density and the number of DRCHs at patient residences. Significant decreases of −1.55% and −0.82% were noted in areas with high and low infection rates, respectively (both p<0.001). Significant decreases of −0.83% and −1.66% were observed in areas of high and low population density, respectively (both p<0.001). Conversely, areas with medium infection rates and population density saw an increase of 0.40% and 0.74%, respectively (both p<0.001). The rate significantly decreased in patients living in areas with a high number of DRCHs (−1.19%, p<0.01), but not significantly in those in areas with a low number of DRCHs (−0.10%, p=0.40).

Changes in hospital visit behaviour by cancer sites

We then concentrated on the trend differences by cancer sites (online supplemental table 1). Regarding ‘Eye’, ‘Ear, nose, mouth and throat’, ‘Musculoskeletal system’, and ‘Endocrine’, where admissions outside of SHSAs exceeded 35% prepandemic, some showed a decrease of postpandemic, but none of the differences were significant. Significant decreases were observed in the ‘Respiratory system’ (−3.77%, p<0.001), ‘Nervous system’ (−1.80%, p<0.001), ‘Breast’ (−1.21%, p<0.001), ‘Kidney and urinary tract, male reproductive system’ (−0.82%, p<0.001), ‘Female reproductive system’ (−0.57%, p=0.025) and ‘Digestive system’ (−0.24%, p=0.006). Figure 2A shows the waterfall plot for the percentage changes.

Figure 2

Changes in hospital admission rates outside of SHSA, categorised by cancer sites. (A) A waterfall chart illustrating the changes. The black bars represent p<0.05. (B) A heatmap illustrating the rate changes in hospital admissions outside of SHSA, categorised by cancer sites and treatment procedures. SHSA, secondary healthcare service areas

Various factors were considered to influence the changes before and after the pandemic. A multivariate analysis, adjusting for those factors, revealed that the pandemic significantly decreased hospital admissions outside of SHSA (table 2, online supplemental table 2).

Table 2

Univariate and multivariate analysis of the effect of the pandemic on the hospital admissions outside of SHSA

Changes in hospital visit behaviour by treatment procedures

We then aimed to determine whether decreases in some cancer sites were due to any specific treatment procedure. Online supplemental table 3 compares the rate changes prepandemic and postpandemic in patients who underwent surgery, categorised by cancer sites. The ‘Eye’, ‘Respiratory system’, ‘Breast’ and ‘Digestive system’ showed significant decreases of −7.43%, −4.38%, −1.31% and −1.00%, respectively (all p<0.001). Online supplemental table 4 shows a similar comparison in patients who underwent chemotherapy. The ‘Endocrine’, ‘Respiratory system’ and ‘Breast’ showed significant decreases of −7.16% (p<0.001), –4.15% (p<0.001) and −0.99% (p=0.013), respectively. Conversely, ‘Skin, subcutaneous tissue’, ‘Musculoskeletal system’ and ‘Blood, blood-forming organs’ exhibited a significant increase of 5.13% (p=0.007), 3.06% (p=0.001) and 0.51 (p=0.048), respectively. Figure 2B presents a heatmap illustrating the percentage changes in major treatment procedures (overall, surgery and chemotherapy) for cancer sites that exhibited significant decreases. The decrease percentage was notably significant in the ‘Respiratory system’, regardless of treatments. We hypothesised that certain treatments may be responsible for this finding. Table 3 presents a comparison of the number of patients who underwent surgery for respiratory cancers, categorised by the type of surgery. No significant change was observed in high technical surgery (p=0.9). However, other surgeries showed a significant decrease (p<0.001), suggesting a possible association between the specialty and difficulty of treatments and changes in patient behaviour.

Table 3

Comparison of number of patients who received surgery for respiratory malignant diseases according to the surgery classification

Discussion

This study indicates that hospital admissions outside of SHSAs for cancer patients decreased following the COVID-19 pandemic, suggesting a change in their hospital visitation behaviour due to the pandemic. Significant reductions were observed in nearly all age groups over 40 years old. Conversely, trends varied among different cancer sites. A significant decrease was found in certain cancer sites like the respiratory system, where rates were already low before the pandemic, but not in others. These differences were suggested to be due to the complexity and specificity of treatment procedures.

Earlier studies on hospital visitation habits of cancer patients during the COVID-19 pandemic revealed a roughly 30% decrease in both visits and admissions.19 This is attributed to hospitals limiting medical care to treat COVID-19 patients and prevent infection spread, or a decrease in cancer diagnoses due to reduced cancer screenings.20 21 This study investigates the hospital visit behaviour of cancer patients, focusing on their hospital selection—an aspect not adequately explored previously. The findings revealed a significant decrease in remote hospital admissions outside the SHSAs following the pandemic. In Japan, a state of emergency was declared, restricting people’s outings. This may have altered their behaviour, discouraging them from visiting hospitals far from their residences. Although the difference was statistically significant, it accounted for only 1% of the total patient population, so careful interpretation is warranted. However, it represents a reduction of nearly 7000 individuals, which suggests that this difference should not be dismissed.

Previous reports indicate that certain cancers, such as bone sarcoma and prostate cancer, have higher survival rates in patients with longer travel times to the hospital.21 22 This indicates that patients treated at high-volume facilities far away have a better prognosis than those treated at nearby low-volume facilities. While different patients have varying priorities, there is a certain percentage of patients visiting distant hospitals for superior treatment. The current study demonstrated a significant decrease in nearly all age groups over 40 years old, with no significant change observed in the paediatric age group. The limited availability of hospitals for highly specialised paediatric oncology may have been an influencing factor.23

The patient’s place of residence is also an important factor in hospital selection. As previously stated, the presence of nearby DRCH is a significant influencing factor.13 The current study revealed a significant decrease for patients residing in areas with a high number of DRCHs, while showing minimal change for those in areas with a low number of DRCHs. Patients with better access to DRCHs did not visit distant hospitals during the pandemic. Our study also considered the impact of COVID-19 infection severity and population density on patient behaviour. Significant decreases were observed in both high and low areas for these factors. High population density areas may have more hospital options, leading to lower rates of distant hospital visits. Conversely, low population density areas have historically had higher rates of distant hospital visits, even prepandemic, hence the larger impact in the decrease. Conversely, a notable rise was observed in medium-density areas. The cause of this outcome needs further clarification. It is presumed that patients had to seek treatment at distant hospitals when local ones exceeded capacity due to treating COVID-19 patients, due to limited hospital options in the area. The same reason could be considered regarding the infection grade, as it correlates with population density in many prefectures.24

This study revealed significant differences according to the cancer site. Cancer sites that had a distant admission rate exceeding 35% prior to the COVID-19 pandemic were less impacted by the pandemic. Conversely, the rate of distant admission tended to decrease in those under 35%. This may be influenced by the trend of centralising patients in certain cancer treatments. Recent reports suggest a variance in long-term prognosis based on hospital volumes. Consequently, centralisation of patients is anticipated in some high-risk cancer surgeries.25–27 The current study revealed significant changes in the respiratory system due to both surgical and chemotherapeutic treatments. Subgroup analysis of respiratory surgery revealed a significant reduction in distant admission exclusively for non-high-difficulty surgeries. Since high-difficulty surgeries require advanced skills and experiences, as well as manpower for postoperative management, it is assumed that high-risk patients were still treated at distant high-volume hospitals even during the COVID-19 pandemic. Given the fact that patients with better access to DRCHs refrained from visiting distant hospitals, it may be necessary to consider optimal hospital distribution, which will serve as a measure for responding to future pandemics. More importantly, careful monitoring is required to ascertain the long-term prognostic effect of this change.

This study has limitations. First, patients’ residences were plotted in terms of SHSAs, disregarding the geographical distance between the actual residences and hospitals. Another limitation is that the severity or stage of each cancer, as well as the detailed treatments required, were not considered. Although this study’s results suggest that advanced and highly specialised treatments likely had minimal impact due to the COVID-19 pandemic, it may be influenced by such factors. Lastly, the patients seen at DPC medical facilities may not represent all patients seen in Japan. However, considering the characteristics of DPC medical facilities, the data are considered to be appropriate for the purpose of this study because they are likely to be representative of patients who require urgent and highly specialised medical care.

In conclusion, the COVID-19 pandemic has altered the hospital visit patterns of cancer patients. Admission to distant hospitals among cancer patients decreased following the pandemic, particularly for diseases not requiring advanced treatment, crucial for the hospital centralisation strategy. The appropriate allocation of specialised hospitals, considering patient behaviour patterns, will serve as a measure for the next pandemic.