Impact of the Coronavirus Disease 2019 Pandemic on Outpatient Visits for Diabetic Retinopathy in Japan: A Retrospective Cohort Study

Abstract

Purpose: Long-term ramifications of the coronavirus disease 2019 pandemic on various care-seeking characteristics of patients with diabetic retinopathy remain unclear. This study aimed to identify risk factors for dropout from regular fundus examinations (RFEs) in patients with diabetic retinopathy in Japan.

Methods: We extracted demographic and health checkup data (April 2018 to March 2021) from the JMDC database. Patients with diabetes identified using diagnosis-related and medication codes were included. The dropout and continuation groups included patients who discontinued and continued to undergo RFEs during the coronavirus disease 2019 pandemic, respectively.

Results: The number of RFEs was significantly lower during the mild lockdown period (April and May 2020) than during the prepandemic period. Of the 14,845 patients with diabetes, 2333 (15.7%) dropped out of RFEs during the pandemic, whereas before the pandemic, of the 11,536 patients with diabetes, 1666 (14.4%) dropped out of RFEs (P = 0.004). Factors associated with dropout in the multivariate logistic regression analysis included younger age, male sex, high triglyceride levels, high γ-glutamyl transpeptidase levels, smoking habit, alcohol consumption, weight gain of more than 10 kg since the age of 20 years, and certain stages of lifestyle improvement. Factors associated with continuation included low body mass index and high glycosylated hemoglobin levels.

Conclusions: Our findings can assist in identifying patients with diabetes at risk of dropout.

Translational Relevance: These results have implications for public health and identifying patients with diabetes at risk of dropout. Education and tailored monitoring regimens could be pivotal role in fostering adherence.

Introduction

The coronavirus disease 2019 (COVID-19) pandemic has impacted the global health care system substantially.¹ The pandemic posed a hurdle for routine eye care owing to the need for close contact between patients and providers during examinations, which increased the risk of infection.² During the pandemic, providers were advised strongly to wear protective equipment and limit patient contact by triaging patients who required in-person care. 

Diabetic retinopathy (DR) is a spectrum of retinal vascular dysfunction attributed to diabetes.³ It remains the leading cause of vision loss worldwide, and its incidence is expected to increase with increasing diabetes prevalence.³ DR progression can significantly deteriorate patients' visual quality. Because regular fundus examinations (RFEs) for early detection of the signs of DR may be highly cost effective, several expert societies currently recommend RFEs for patients with diabetes.^4,5 However, a significant proportion of patients with diabetes receive fundus examinations infrequently owing to barriers to health care, including the lack of awareness, financial cost, and uncomfortable examination process.^6,7 

The latest data show that DR examination dropouts increased significantly during the initial 9 weeks of the COVID-19 pandemic.⁸ However, the long-term ramifications of the pandemic on various care-seeking characteristics of patients with DR have not been studied. The universal and government-provided health insurance in Japan decreases out-of-pocket medical costs for health care for all citizens. Thus, access to medical care in Japan is less affected by various pandemic- or cost-related barriers.⁹ As the new postpandemic normal lifestyle changes are being adopted by society, assessing the changes in care-seeking patterns while considering access and cost limitations may guide the approach to treat patients with DR within the scope of public health care systems. 

In this study, we evaluated the changes in care-seeking patterns, such as dropouts, among patients with DR during the COVID-19 pandemic using the Japanese national insurance database. Additionally, we determined the characteristics and risk factors associated with the dropout of regular follow-ups in patients with DR during the pandemic. 

Methods

Study Design

For this retrospective cohort study, we used data from the insurance claims database provided by JMDC Inc. (Tokyo, Japan.)¹⁰ The study spanned from April 2017 to March 2021. The mild lockdown period in Japan was defined as April 2020 to May 2020, during which a nationwide state of emergency was declared by the Ministry of Health, Labour and Welfare of Japan. 

This study was approved by the independent Ethics Committee of the Juntendo University Faculty of Medicine (approval number: E21-0338). The requirement for informed consent was waived, because personal information in the JMDC claims database used in this study was completely anonymized. The principles outlined in the Declaration of Helsinki were followed. 

Data Source

JMDC Inc., is a private company that provides health insurance receipt data for medical research. The JMDC claims database is an epidemiological receipt database that accumulates receipts (inpatient, outpatient, and dispensing) and medical examination data received from multiple health insurance associations. The database contains encrypted personal identifiers, age, sex, diagnoses, procedure codes, and prescribed drugs for approximately 14 million insured people, mainly employees of Japanese companies and their family members, all aged ≤74 years.¹⁰ In addition to the original Japanese codes, diagnoses are recoded according to the International Classification of Diseases, Tenth Edition,¹¹ and drugs are recoded according to the European Association for Medicines Market Research's Anatomical Therapeutic Chemical Classification.¹² 

Definitions

Patients With Diabetes

We identified all individuals meeting the following selection criteria as patients with diabetes: (1) diagnosis-related codes for diabetes (Supplementary Table S1), (2) medication codes for diabetes (Supplementary Table S2), and (3) entered into the database before the study period. Criterion (3) was adopted to exclude individuals newly enrolled during the study period and to adhere to the cohort study design. 

Outpatient Visits for DR Fundus Examination

Outpatient visits of interest included a precision fundus examination (bilateral or unilateral) or a panvitreoretinal examination (bilateral or unilateral). If the same patient with diabetes had more than one record of procedures in the same month, duplicates were removed, and the number of outpatient visits was counted as one. 

Dropout and Continuation Groups

Fundus examination for DR is recommended at least once a year in Japan.¹³ Therefore, we classified patients with diabetes who had more than one examination in both fiscal years 2018 and 2019 into the regular outpatient visit group, and only those with available health checkup data were included in the analysis. Among the patients analyzed, those who had no visits in fiscal year 2020 were included in the dropout group, whereas those with at least one visit were included in the continuation group based on Japan's standard practice for fundus examination for DR (at least once a year). We also assessed individuals in fiscal year 2017 as the prepandemic period to compare the dropout rates between the pandemic and prepandemic periods. 

Universal health insurance in Japan is determined by the insured individual's employment status. Therefore, if an individual's occupation changes, they can withdraw from the health insurance system and join another health insurance system. Furthermore, in the event of death, they drop out of the universal health insurance in Japan. To prevent patients who withdraw from health insurance owing to job change or death being included in the dropout groups, we considered only patients with continuous enrolment records in the JMDC database during the study period, excluding the data of patients who withdrew from health insurance or died. 

Data Collection

We extracted the following data: age; sex; health checkup data, including blood data, urine dipstick test results (protein and glucose), and body mass index (BMI); lifestyle factors (smoking history, alcohol intake frequency, and restfulness from sleep); drug history (antihypertensive and antidyslipidemic drugs); medical history (cerebrovascular disease, cardiovascular disease, kidney failure, or hemodialysis); and questionnaires (weight change >10 kg since the age of 20 years, willingness to receive health instruction from public health nurses, and lifestyle modification stage). We categorized the patients into three groups based on glycosylated hemoglobin (HbA1c) levels—HbA1c of less than 7.0% (control group); HbA1c of 7.0% to 9.0%; and HbA1c of 9.0 or higher.^14,15 Furthermore, we classified other blood data as normal or abnormal based on the adult reference values in Japan.¹⁶ The urine dipstick protein and glucose test results were categorized as positive or negative (≥30 mg/dL and ≥100 mg/dL, respectively). Based on the classification of the Japan Society for the Study of Obesity, we classified the patients into the following five groups according to BMI: low body weight (<18.5), normal weight (18.5–25.0), obesity degree 1 (25.0–30.0), obesity degree 2 (30.0–35.0), and obesity degrees 3 and 4 (≥35).¹⁷ Motivation for lifestyle improvement was classified into five groups (maintenance, working on improving for >6 months; action, working on improving for <6 months; preparation, intending to improve in the near future; contemplation, having the willingness to improve; precontemplation, no willingness to improve) according to the stage of change theory.¹⁸ 

Statistical Analysis

Continuous variables are presented as mean (standard deviation), and categorical variables are presented as percentages. Mann–Whitney U tests were performed for continuous variables that were non-normally distributed in the Shapiro–Wilk test. We performed χ² tests for categorical variables to compare the number of outpatient visits during the mild lockdown period and the same period 2 years earlier, and the dropout rate during the pandemic and prepandemic periods. We identified factors associated with dropout using univariate and multivariate logistic regression analyses. Results are presented as odds ratios with 95% confidence intervals. All data were analyzed using Stata, version 15 (StataCorp, College Station, TX). Statistical significance was set at a P value of less than 0.05. 

Results

Study Samples

Figure 1 shows the flowchart for the study sample selection during the pandemic period. From the database, 85,910 patients who met the defined criteria for patients with diabetes were identified and included in the analyses of trends in the number of outpatient visits for fundus examination. Supplementary Table S3 shows population characteristics of patients with diabetes during the pandemic period. Among them, we excluded 57,798 patients who did not undergo regular checkups and included 28,122 patients in the regular checkup group. Furthermore, we excluded 6849 patients owing to the absence of health checkup data. From the remaining 21,273 patients, 6428 were further excluded because of missing values in their health checkup data. Finally, we included 14,845 patients in the analyses of the dropout rate and risk factors for dropout during the pandemic. 

Figure 1.

 

Flowchart of study sample selection during the pandemic.

Supplementary Figure S1 shows the flowchart for study sample selection during the prepandemic period. From the database, 75,358 patients who met the defined criteria for patients with diabetes were identified. Among them, we excluded 50,847 patients who did not undergo regular checkups and included 24,511 patients in the regular checkup group. Furthermore, we excluded 6161 patients owing to the absence of health checkup data. From the remaining 18,350 patients, 6814 were further excluded because of missing values in their health checkup data. Finally, we included 11,536 patients in the analyses of the dropout rate and risk factors for dropout during the prepandemic period. 

Trends in the Number of Outpatient Visits for Fundus Examination Among Patients With DR

Figure 2 shows the trends in the outpatient visits for fundus examination among patients with DR. A total of 314,809 outpatient visits were observed during the study period, and there were 103,153, 105,786, and 105,870 outpatient visits in the fiscal year 2018, 2019, and 2020, respectively. During the mild lockdown period (Fig. 2, red arrow), there was a significant decrease in the number of outpatient visits in both April and May 2020 compared with the same period 2 years earlier. The number of outpatient visits in April 2020 (7509 [8.7%]) was significantly lower than that in April 2018 (8586 [10.0%]; P < 0.001). Similarly, the number of outpatient visits in May 2020 (7609 [8.9%]) was significantly lower than that in May 2018 (8476 [9.9%], P < 0.001). 

Figure 2.

 

Trends in outpatient visits for fundus examination among patients with DR. The red double arrows indicate the mild lockdown period in Japan.

Fundus Examination Dropout Rates and Population Characteristics

Population characteristics of the dropout and continuation groups during the pandemic period are presented in Table 1. Of the 14,845 patients with diabetes, 2333 (15.7%) dropped out of regular outpatient visits during the pandemic period. The mean age of the dropout group was 55.9 years, and 79.4% were male. The dropout group showed higher BMI, lower HbA1c levels, higher triglyceride, low-density lipoprotein-cholesterol, and γ-glutamyl transpeptidase levels, and higher frequencies of smoking, regular alcohol consumption, and weight change of more than 10 kg since the age of 20 years than the continuation group. Moreover, 41.9% of patients who dropped out of regular outpatient visits were in the preparation and contemplation stages of lifestyle modifications. 

Table 1.

 

Characteristics of the Dropout and Continuation Groups During the Pandemic

 

During the prepandemic period, of the 11,536 patients with diabetes, 1666 (14.4%) dropped out of regular outpatient visits. The dropout rate during the pandemic was significantly higher than that during the prepandemic period (χ² tests; P = 0.004). 

Risk Factors for Dropout

Supplementary Table S4 shows the risk factors for dropout during the pandemic period in the univariate logistic regression analysis, and Table 2 shows the factors associated with dropout during the pandemic period in the multivariate logistic regression analysis. Factors associated with an increased dropout rate were younger age, male sex, higher triglyceride and γ-glutamyl transpeptidase levels, smoking habit, high alcohol intake frequency, weight change of more than 10 kg since the age of 20 years, and preparation and contemplation stages of lifestyle modifications. Factors associated with a low dropout rate were a low BMI and increased HbA1c levels. Supplementary Table S5 shows the factors associated with dropout during the prepandemic period in the multivariate logistic regression analysis. Factors associated with an increased dropout rate were younger age, male sex, higher γ-glutamyl transpeptidase levels, smoking habit, weight change of more than 10 kg since the age of 20 years, and contemplation stages of lifestyle modifications. 

Table 2.

 

Risk Factors for Dropout During the Pandemic

 

Discussion

The COVID-19 pandemic prompted significant adaptation in the global health care system, reducing outpatient visits and affecting monitoring frequency, including those for DR, which requires RFEs for timely intervention. The behavioral pattern shift caused by the recent pandemic may be crucial for determining the appropriate steps with regard to future decisions regarding public health. Using the Japanese national insurance claims database and outpatient data, we observed a significantly higher dropout rate during the pandemic compared with that during the prepandemic period. Additionally, we identified several risk factors for dropout in DR monitoring. Our present findings may have implications for offering more effective disease management regimens for patients with DR. 

A 15.7% dropout rate for regular visits was observed in patients with diabetes during the pandemic period in our study. This result was significantly higher than the dropout rate during the prepandemic period (14.4%) identified in our study. To date, several reports on regimen adherence by patients with DR have been published. In a study by Bresnick et al.¹⁹ among patients with DR who visited an ophthalmologist based on a primary-care screening recommendation, the dropout rates were 8% to 10% and 17% to 44% for those who required and did not need intervention, respectively. Moreover, Obeid et al.²⁰ reported a 25.4% dropout rate among patients with DR who received interventions, such as panretinal photocoagulation or intravitreal anti-vascular endothelial growth factor injections. Additionally, Angermann et al.²¹ reported an 18.9% dropout rate among patients with DR who received intravitreal anti-vascular endothelial growth factor injections covered by national insurance similar to Japan's system. Notably, a comparable trend during the COVID-19 pandemic was noted for a different retinal pathology; a 14.6% decrease was observed in regular outpatient monitoring for exudative age-related macular degeneration.²² Prajapati et al.²³ suggested that delayed monitoring during the pandemic resulted in disease progression and vision loss in patients with DR. These findings underscore the need for increased awareness of pandemic-related dropouts from RFEs for DR, which may result in further decline in the patient's vision globally. 

The decrease in outpatient visits for DR during the mild lockdown period likely reflects the forced physical limitation to visit medical facilities for nonurgent reasons to minimize disease spread through social distancing. An increased concern for COVID-19 spread through aerosolized viral particles to the conjunctiva possibly cultivated a fear against receiving ophthalmological care.²⁴ The observed decrease in outpatient visits might be attributed to both the physical and psychological components that arose from social distancing measures²⁵ and the fear of infection.²⁶ Indeed, an eye clinic, which was part of a large tertiary care academic institution, reported a significant increase in outpatient appointment cancellations when lockdown measures were implemented, many of them owing to COVID-19–related reasons.²⁷ Additionally, Ikesu et al.⁸ observed a 14.1% decrease in the total number of DR fundus examinations in Japan during the pandemic; our results during the mild lockdown phase align with this trend. However, a long-term observation indicated a recovery in the number of outpatient visits for DR after the pandemic, suggesting the mitigation of several pandemic-related psychological and physical barriers to health care.²⁸ The psychological barrier to outpatient visits for DR is likely best epitomized by the fear of contracting COVID-19 during in-person examinations. However, concerning outpatient visits, prevention of irreversible vision loss owing to DR progression may outweigh the risk of possible COVID-19 exposure.²⁹ This opinion was shared by the patient population; 57% indicated a concern for ocular disease progression owing to the decreased frequency of follow-up visits associated with the COVID-19 pandemic.³⁰ However, the widespread adoption of standard precautions for COVID-19 prevention² may have contributed to decreasing the psychological barrier to health care. Moreover, the physical limitation to visiting medical facilities was decreased markedly at the end of the nationwide state of emergency,³¹ which may have promoted the resumption of regular checkups by patients. Whereas DR-related outpatient visits were significantly decreased during the COVID-19 pandemic in the short term, the long-term visit counts showed a trend for recovery after the removal of pandemic-related barriers. 

We identified several factors that increased the dropout rate among patients with DR during the pandemic. Previous studies have reported younger age, male sex, smoking habit, high alcohol consumption rates, and severe uncontrolled diabetes (HbA1c ≥ 9.0%) as risk factors for nonadherence to DR screening before the pandemic.^32,33 Similarly, a recent Japanese study reported younger age and male sex as risk factors for nonadherence to DR screening.³⁴ In this study, younger age, male sex, higher γ-glutamyl transpeptidase levels, smoking habit, and contemplation stages of lifestyle modifications were identified as risk factors for drop out in prepandemic period. Younger age, male sex, and smoking and alcohol consumption habits were overlapping risk factors identified both before and after the pandemic. Notably, a comparatively higher adherence to monitoring during the pandemic was observed among patients with uncontrolled diabetes, as determined by their higher HbA1c levels (HbA1c ≥ 7.0%), which is contradictory to the findings identified in our study during the prepandemic period and the findings of prepandemic studies.^32,33 This observation is likely attributed to the selection process of this cohort study, reflecting the fact that it targeted patients with DR who were receiving RFEs for DR before the pandemic. Among those who regularly receive eye care, patients with uncontrolled diabetes may have an increased fear of DR progression,²⁹ which could motivate frequent fundus examinations. However, well-controlled diabetes (HbA1c <7.0%) was found to be a risk factor for dropout from RFE. Although counterintuitive, this finding suggests that care providers should be aware of the possibility that patients with well-controlled diabetes are more likely to be lost to follow-up for DR monitoring than those with poorly controlled diabetes. Consistently, several studies support the importance of RFEs for DR onset and progression monitoring, particularly in patients with well-controlled HbA1c levels and minimal end-organ damage.^35,36 Additionally, DR is multifactorial, and its progression is associated closely with various systemic pathologies; many of these pathologies are commonly found in patients with diabetes, including hypertension and dyslipidemia,³⁷ but they are not associated directly with HbA1c levels. Hence, near-normal HbA1c levels do not equate to a minimal risk of DR. Thus, increasing awareness among patients with diabetes is crucial, as diabetes increases the risk of developing DR regardless of HbA1c levels; moreover, health care providers should emphasize the risk of dropping out of regular DR examinations for disease monitoring. 

This study had some limitations. First, the JMDC database does not comprehensively represent the entire health care system in Japan, possibly limiting the generalizability of the study findings. Moreover, we included only patients with complete health checkup data in the analysis to identify risk factors for dropout (Supplementary Tables S6–S8). Second, the JMDC database includes only insurance recipients aged more than 75 years. Older patients have adjusted targets for glycemic control,¹⁵ which may have affected the HbA1c level classification of the selected cohort. Third, our diabetes diagnosis relied on the presence of prescribed diabetes medications, excluding patients managing diabetes through dietary and lifestyle modifications in the selected cohort. Fourth, the JMDC database does not include the social factors. Therefore, the known risk factors related to patients with DR's compliance with outpatient visits,^32,33 such as educational level, whether patients are physically disabled, accessibility to health care, income and marital status, could not be considered in this study. Fifth, DR severity was not assessed in this study because JMDC database does not contain the information for DR grading. Because the frequency of RFEs is recommended according to the severity of DR,³⁸ the severity of DR may have an impact on RFEs in this study. Finally, the dropout from RFEs for DR in this study could be attributed to pandemic-related effects or reduced adherence unrelated to the pandemic. 

In this study, we examined the changes in the RFE patterns for DR owing to the pandemic and elucidated the risk factors for dropouts during the pandemic using the Japanese universal health insurance database. Although outpatient visits for DR decreased during the mild lockdown period, a trend toward recovery to prepandemic levels was noted. These results have implications for promoting public health and accurately identifying patients with diabetes who may be at risk of dropout. Education and tailored monitoring regimens could play a pivotal role in fostering adherence. 

Acknowledgments

The authors thank all members of the Department of Ophthalmology, Department of Digital Medicine, Department of Hospital Administration, and Department of Telemedicine and Mobile Health, Juntendo University Graduate School of Medicine for providing critical comments on this manuscript. 

Supported by JSPS KAKENHI Grant Numbers 20K09810 (TI), 20KK0207 (TI), 21K17311 (AMI), 21K16884 (KF), 22K16959 (YO), 22K16983 (AE), and Alcon Japan Research Grants (SN). 

The sponsors had no role in the design or execution of the study; data collection and management; the analysis and interpretation of the data; the preparation, review, or approval of the manuscript; or the decision to submit the manuscript for publication. 

Disclosure: K. Hirosawa, None; T. Inomata, Lion Corporation and Sony Network Communications Inc. (S), Johnson & Johnson Vision Care Inc. (F), Yuimedi Inc. (F), ROHTO Pharmaceutical Co. Ltd. (F), Kobayashi Pharmaceutical Co. Ltd. (F), Kandenko Co. Ltd. (F), Fukoku Co. Ltd. (F), Santen Pharmaceutical Co. Ltd. (F), InnoJin Inc. (F), Ono Pharmaceutical Co., Ltd. not related to the submitted work (F); Y. Akasaki, None; J. Sung, None; A. Yee, None; M. Iwagami, None; K. Nagino, InnoJin Inc. outside the submitted work (F); Y. Okumura, InnoJin Inc. outside the submitted work (F); K. Fujimoto, None; A. Midorikawa-Inomata, InnoJin Inc. outside the submitted work (F); A. Eguchi, None; H. Shokirova, None; K. Fujio, None; T. Huang, None; Y. Morooka, None; H. Kobayashi, None; A. Murakami, None; S. Nakao, Kowa Company. Ltd. (F), Mitsubishi Tanabe Pharma Corporation, Alcon Japan, Ltd. (F), Santen Pharmaceutical Co. Ltd. (F), Machida Endoscope Co. Ltd. (F), Wakamoto Pharmaceutical Co. Ltd. (F), Bayer Yakuhin Ltd. (F), Senju Pharmaceutical Co. Ltd. (F), Nippon Boehringer Ingelheim Co. Ltd. (F), Chugai Pharmaceutical Co. Ltd. (F), Hoya Corporation (F), Novartis Pharma (F)