Continuity of diabetes care among elderly Japanese patients: a medical record review study in a specialized diabetes clinic

Abstract

Aims

Continuity of diabetes care is relevant among elderly patients. The aim of this study is to investigate the impact of clinical characteristics on continuing outpatient visits to a specialized diabetes clinic in elderly Japanese patients with diabetes.

Methods

We included outpatients with type 2 diabetes aged ≥ 65 years who first visited our clinic from 2006 to 2009. The information of patients’ characteristics was obtained through medical record review from the CoDiC database. We have tracked whether the patients continued to visit the clinic until May 31, 2019. A Cox proportional hazards regression model identified variables related to withdrawal.

Results

Among 128 patients, 63 patients (49.2%) were withdrawn during the follow-up periods. The average visit duration of withdrawals was 4.6 (range 1, 10) years. The patients who discontinued to visit were older (72.6 vs. 69.5 years old, p = 0.005) compared with those who continued to visit. No significant differences in clinical conditions such as complication of diabetes, Charlson Comorbidity Index and polypharmacy between the first and last visit were observed in each group. Age (≥ 75 years) was significantly associated with withdrawal (hazard ratio 2.72 [95% confidence interval 1.59, 4.63], p < 0.001). Except for age, no significant differences were observed in all variables when adjusted for confounders.

Conclusions

Our findings indicated that continuous outpatient visits were difficult in elderly Japanese patients with diabetes. Older age (≥ 75 years) independently affected withdrawal. Future multicenter studies with adequate populations and social and geriatric factors are necessary to confirm our findings.

Introduction

Many countries currently address health-related issues in elderly people with diabetes [1, 2]. In Japan, one in five people over the age of 60 has diabetes [3]. The lifespan of diabetes has reached 72.6 years old [4]. Diabetes care predominantly depends on individual self-care. Self-care among aging patients may decline, while complex diabetes care is required for older patients with multiple advanced health conditions [1, 5]. We anticipate that lifelong diabetes care for elderly Japanese patients will become more complicated.

To support them, clinics play an important role because Japanese patients with diabetes are largely outpatients [6]. All patients are allowed free access to clinics on their own by universal health insurance [7]. National guidance recommends continuity of diabetes care in a clinic, such as patient education, annual check-ups of diabetic complications, and contact for regular visits, leading to reduce the severity of diabetes and health care burden [7, 8].

Although diabetes care across Japan relies on outpatient visits, little is discussed regarding the continuous visits to clinics in elders with diabetes. Since diabetic complications, comorbidity, and polypharmacy are known as barriers to access to diabetes care [8, 9], continuous access to diabetes care through outpatient visits is challenging for older patients who are susceptible to these clinical conditions.

Therefore, it is useful for continuity of diabetes care among elderly patients to clarify the clinical characteristics of those who quit visiting a clinic. This study aimed to investigate the characteristics of elderly Japanese outpatients regularly visiting a clinic and compare the impact of clinical variables between the group with continuous visits and the group withdrew.

Methods

Study design

We conducted a facility-based cohort study by reviewing medical records from the Computerized Diabetes Care (CoDiC) database [10].

Setting

The study setting was a specialized diabetes clinic in rural Japan, responsible for not only primary and secondary diabetes care, but also outpatient care of common diseases for patients with diabetes.

Participants

We included patients with type 2 diabetes aged ≥ 65 years who first visited our clinic from June 1, 2006 to May 31, 2009. There were no restrictions on diabetic conditions, comorbidities, or treatment regimens.

The participants were classified into two groups according to whether they continued the visits to the clinic on a monthly and bimonthly frequency or withdrew until May 31, 2019. We defined withdrawals as the cases of patients who had not visited our clinic more than 3 months after their last visit for any reason.

Variables

Anonymous data at the first and last visit dates of the continuous visits and withdrawal were retrieved from the CoDiC database by May 31, 2019. We extracted the following variables: age, sex, living alone or not, living area either inside or outside of the city where the clinic is located, duration of diabetes, previous medication, self-interruption of past medications, duration of visits to the clinic, HbA1c level, body mass index, prevalence of diabetic complications based on the national standard valid [5], comorbidities, and prescriptions.

Measurements

The severity of comorbidities was assessed using the Charlson Comorbidity Index (CCI) updated version validated in the Japanese national data [11]. The CCI score was cut off at 2 scores with reference to the median of Japanese people who received long-term care in a previous study [12]. The number of prescriptions was counted for the types of medications prescribed by the physicians in the clinic. Polypharmacy was classified as more than six types of prescriptions that lasted over the latest 4 weeks.

Statistical analysis

Clinical conditions between the groups were compared using the Chi-square test or Fisher’s exact test for categorical variables and the unpaired t test for continuous variables. When the continuous data were skewed, a non-parametric test was applied. Unadjusted overall time to withdrawal (i.e., years of outpatient visit) was derived by the Kaplan–Meier analysis and log-rank test. The Cox proportional hazards regression model identified variables at the first visit related to the incidence of withdrawal by adjusting for distributed visit durations and potential confounders. The proportional hazards assumption was confirmed using Schoenfield residuals. The statistical power was set at 0.8. Two-tailed p values of < 0.05 indicated statistical significance. All analyses were performed using Stata, version 16.0, software (StataCorp LLC, College Station, TX, USA). The study protocol was approved by the research ethics committee of the Faculty of Medicine, University of Tsukuba (approved on April 10, 2019; No. 1391). The participants were given the opportunity to opt out.

Results

Among the 978 new patients with diabetes who visited our clinic from June 1, 2006 to May 31, 2009, the data of 225 potential subjects were retrieved from the database (Fig. 1). We excluded 77 patients who left the clinic within 1 year. They were those who returned to their family doctor or referral hospital after temporary consultation. We also excluded 20 patients who revisited our clinic after hospital discharge. Finally, this study included 128 patients.

Fig. 1.

Extraction of study participants

Of the 128 participants, 63 (49.2%) patients withdrew from our clinic visits. The mean durations of visits were 10.7 (range 8, 12) and 4.6 (1, 10) years in those with continuous visits and withdrew, respectively (p < 0.001). The reasons for 63 withdrawals were as follows: referred to another medical institute, 30 (47.6%); died, 19 (30.2%); dropped out (i.e., no medical consultation for diabetes anywhere for more than 3 months), 5 (7.9%); moved, 2 (3.2%); unknown, 7 (11.1%).

Table 1 shows the characteristics of study participants at the first visit according to continuous outpatient visits or not. Patients who discontinued to visit were older (72.6 vs. 69.5 years old, p = 0.005) and less living alone (3.2 vs. 15.4%, p = 0.030) compared with those who continued to visit.

Table 1.

Characteristics of study participants at the first visit according to continuous outpatient visits

	Continued (n = 65)	Withdrew (n = 63)	p value
Age (years)a	69.5 (3.9)	72.6 (6.2)	0.005
65–69	37 (56.9%)	23 (36.5%)
70–79	27 (41.5%)	29 (46.0%)
80 ≤	1 (1.5%)	11 (17.5%)
Male	43 (66.2%)	37 (58.7%)	0.466
Living alone (yes)	10 (15.4%)	2 (3.2%)	0.030
Living area (outside of the city)	40 (61.5%)	41 (65.1%)	0.716
Duration of diabetes (years)a	10.1 (8.0)	11.9 (10.8)	0.558
Previous medication (yes)	49 (75.4%)	48 (76.2%)	1.000
Self-interruption of past medications (yes)	5 (7.7%)	4 (6.3%)	1.000

^aMean (SD)

Table 2 presents the differences of clinical conditions at the first and last visits according to continuous outpatient visits. The mean difference of the HbA1c level improved by -1.2% in continuous visits and by − 1.0% in withdrawals. The prevalence of retinopathy decreased at the last visit in both groups, while the prevalence of other complications increased. Although the CCI scores increased in both groups, no statistical significance was observed between the first and last visits. Diabetic complications (61.5% vs. 74.6%) and malignancies (18.5% vs. 20.6%) were frequently observed in the CCI (not shown in Table 2). Polypharmacy increased in each group. Over 90% of patients in both groups were prescribed medications other than antidiabetic agents. No significant differences in the above clinical conditions were observed between the groups.

Table 2.

Differences in clinical conditions at the first visit and the last visit according to continuous outpatient visits

	First visit			Last visit
	Continued (n = 65)	Withdrew (n = 63)	p value	Continued (n = 65)	Withdrew (n = 63)	p value
Duration of visits to the clinic (years)
Mean (SD)	0 (0)	0 (0)	NA	10.7 (1.0)	4.6 (2.6)	< 0.001
Median (range)	0 (0, 0)	0 (0, 0)		11 (8, 12)	4 (1, 10)
HbA1c (NGSP%)
Mean (SD)	8.5 (1.8)	8.4 (2.1)	0.469	7.3 (0.9)	7.4 (1.1)	0.610
Median (range)	8.2 (6.0, 15.1)	8.0 (5.8, 17.1)		7.2 (5.4, 11.0)	7.0 (5.8, 11,7)
Body mass index (kg/m2)
Mean (SD)	24.8 (3.5)	24.5 (4.0)	0.675	24.2 (3.4)	24.0 (4.1)	0.777
Median (range)	24.3 (16.8, 33.4)	24.3 (16.7, 34.9)		24.0 (17.0, 33.1)	23.9 (15.7, 34.2)
Diabetic complications n (%)
Retinopathy	27 (41.5%)	30 (47.6%)	0.594	25 (38.5%)	29 (46.0%)	0.474
Nephropathy	24 (36.9%)	28 (44.4%)	0.472	30 (46.2%)	33 (52.4%)	0.596
Neuropathy	15 (23.1%)	13 (20.6%)	0.832	17 (26.2%)	27 (42.9%)	0.063
Cardiovascular disease	1 (1.5%)	6 (9.5%)	0.060	8 (12.3%)	9 (14.3%)	0.799
Cerebrovascular disease	5 (7.7%)	4 (6.3%)	1.000	9 (13.8%)	10 (15.9%)	0.807
Charlson comorbidity index (scores)
Mean (SD)	0.9 (0.9)	1.1 (1.2)	0.194	1.1 (1.1)	1.7 (1.9)	0.116
Median (range)	1 (0, 3)	1 (0, 7)		1 (0, 5)	1 (0, 7)
Polypharmacy n (%)	5 (7.7%)	9 (14.3%)	0.268	27 (41.5%)	25 (39.7%)	0.859
Antidiabetic agent	54 (83.1%)	53 (84.1%)	1.000	58 (89.2%)	57 (90.5%)	1.000
Others	61 (93.8%)	59 (93.7%)	1.000	63 (96.9%)	62 (98.4%)	1.000

SD: standard deviation, NA: not applicable, Polypharmacy: ≥ 6 types of medications

Figure 2 showed the Kaplan–Meier curves for withdrawal according to each variable at the first visit. Those who were aged ≥ 75 years at the first visit presented a significantly increased rate of withdrawal compared to those aged < 75 years.

Fig. 2.

Kaplan–Meier analysis of unadjusted overall time to quit regular visiting in each clinical variable at the first visit

Table 3 shows the hazard ratios for withdrawal. Age (≥ 75 years) was significantly associated with withdrawal (hazard ratio 2.72 [95% confidence interval 1.59, 4.63], p < 0.001). Except for age, no significant differences were observed in sex, CCI (score ≥ 2) and polypharmacy.

Table 3.

Cox regression analysis to estimate hazard ratio for withdrew continuous outpatient visits

Variable	Univariate analysis		Multivariate analysis
	HR (95% CI)	p value	HR (95% CI)	p value
Age ≥ 75	2.76 (1.65, 4.61)	< 0.001	2.72 (1.59, 4.63)	< 0.001
Male	0.81 (0.49, 1.34)	0.410	0.75 (0.44, 1.26)	0.273
CCI score ≥ 2	1.23 (0.66, 2.31)	0.513	1.13 (0.58, 2.21)	0.717
Polypharmacy	1.51 (0.74, 3.05)	0.254	1.13 (0.54, 2.36)	0.752

HR: hazard ratio, 95% CI: 95% confidence interval

Cox multivariate regression model estimated HR of the clinical variables at the first visit on withdrew in a total of 128 elderly diabetic patients including the 65 continued visits cases and the 63 withdrawals

Discussion

Our study indicated that half of the elderly patients with diabetes withdrew from the regular visits to our clinic. Patients who withdrew were older at the first visit and the average duration of visits was only 4 years. Older age (≥ 75 years) significantly affected continuous outpatient visits.

The characteristics of clinical conditions between the first and last visits were revealed among elderly Japanese outpatients. The diabetic complications, severity of comorbidity, and polypharmacy were progressive regardless of whether older patients continued regular visits. Our findings would be useful for the continuity of diabetes care among elderly outpatients, even if they will attend other medical institutions.

Although polypharmacy seemed to increase in the continued visits group (Table 2), it would have affected withdrawal (Fig. 2, Table 3). Given the mixed results, the issue of polypharmacy needs further discussion, including patients’ preferences [8].

The severity of comorbidities measured by the CCI scores was mild in our study participants. Although the CCI may help predict health service use, such as long-term care [12], its sensitivity among elderly outpatients should be considered.

Our study had several limitations. First, this study aimed to assess the continuity of diabetes care in a local clinic. The study sample size was consequently small leading to insufficient statistical power and external validity. Thus, our findings should be confirmed in larger populations and in various healthcare settings. Second, this study could not enough discuss the patients’ sociodemographic information because of the restricted available medical records. Although the patients living alone were more likely to continue regular visits (Table 1), which may indicate individual mobility [9], we could not consider living arrangements due to lack of numbers and follow-up information. Besides, data on geriatric factors such as frailty, fall, and mild cognitive impairment were limited in our database.

In conclusion, continuous outpatient visits were difficult in elderly Japanese patients with diabetes. Older age (≥ 75 years) was independently associated with withdrawal. Future large multicenter studies are needed considering the social and geriatric factors.

Funding

This study was supported by JSPS KAKENHI (Grant No. JP20K18871).

Declarations

Conflict of interest

The authors declare that there are no conflicts of interest.

Ethical standards

All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation and with the Helsinki Declaration of 1964 and later versions.