The J-EDIT Study Group: Principal Investigator: Hideki Ito M.D., Ph.D., Department of Diabetes, Metabolism and Endocrinology, Tokyo Metropolitan Geriatric Hospital and Institute of Gerontology, 35-2 Sakae-cho, Itabashi-ku, Tokyo 173-0015, Japan.
Lower physical activity is a strong predictor of cardiovascular events in elderly patients with type 2 diabetes mellitus beyond traditional risk factors: The Japanese elderly diabetes intervention trial
Article first published online: 21 MAR 2012
© 2012 Japan Geriatrics Society
Geriatrics & Gerontology International
Special Issue: The Japanese Elderly Diabetes Intervention Trial (J-EDIT). Guest Editor: Hideki Ito
Volume 12, Issue Supplement s1, pages 77–87, April 2012
How to Cite
Iijima, K., Iimuro, S., Shinozaki, T., Ohashi, Y., Sakurai, T., Umegaki, H., Araki, A., Ouchi, Y., Ito, H. and the Japanese Elderly Diabetes Intervention Trial Investigator Group (2012), Lower physical activity is a strong predictor of cardiovascular events in elderly patients with type 2 diabetes mellitus beyond traditional risk factors: The Japanese elderly diabetes intervention trial. Geriatrics & Gerontology International, 12: 77–87. doi: 10.1111/j.1447-0594.2011.00815.x
- Issue published online: 21 MAR 2012
- Article first published online: 21 MAR 2012
- Accepted for publication 7 November 2011.
- engagement in physical activity;
- Japanese Elderly Diabetes Intervention Trial study;
- physical activity;
- risk reduction;
- type 2 diabetes mellitus
Aim: It is well known that a decline in physical activity is associated with lifestyle-related diseases including cardiovascular (CV) events. However, little is known about the association between physical activity and CV events in elderly patients, because recent accumulating reports have mainly dealt with middle-aged populations. In this study, we investigated the correlation between physical activity and CV events in Japanese elderly patients with type 2 diabetes mellitus (T2DM).
Methods: A total of 938 Japanese elderly patients with T2DM (447 men and 491 women, mean age 71.9 years) enrolled (2000–2002) in the Japanese Elderly Diabetes Intervention Trial (J-EDIT) were used in this study. Physical activity consisting of three components, work, sports and leisure-time, of their lifestyle was evaluated using the Baecke questionnaire at baseline. Total activity score (TAS) as a sum of each activity score was divided into four quartiles (Q1 to Q4).
Results: During a follow-up period of 65.2 months, 165 events and 71 deaths in total occurred. Higher TAS grade was associated with reduced risk of all events (hazard ratios: 0.82, 0.77 and 0.54 in Q2, Q3 and Q4, respectively) with statistical significance. Even after multivariate adjustment for covariates, higher TAS grade was a strong predictor of all events, and the prediction by TAS of cerebrovascular events was more effective than that of cardiac events. In contrast, all-cause mortality gradually decreased according to TAS grade; however, no statistical significance was found. Among the four grades of TAS, no significant change in several parameters, such as profiles of lipid and glucose metabolism, blood pressure, physical measurements, cognitive function and depression scale, was found throughout the follow-up period, suggesting that the higher level of physical activity itself was associated with the risk reduction of primary events.
Conclusion: Lower physical activity is a strong and independent predictor of all CV events in the elderly with T2DM beyond traditional risk factors. In addition to strict management of each atherosclerotic risk factor, engagement with patients to augment and maintain the level of physical activity in their lifestyle is also essential in clinical practice. Geriatr Gerontol Int 2012; 12 (Suppl. 1): 77–87.
A decline in physical activity has been shown to lead to increased risk of several cardiovascular (CV) diseases, such as cerebrovascular disease (CVD) and coronary heart disease (CHD).1–6 In developed countries, 80% of all deaths from CV disease occur in people aged 65 years and older.7 Unfortunately, 60% or more of USA adults are not physically active in their lifestyle.8 Accordingly, in 1995 the Centers for Disease Control (CDC) and the American College of Sports Medicine (ACSM) recommended a moderate amount of physical activity on most days, and preferably all days, of the week.9 However, the precise mechanisms through which physical activity lowers the risk of CV disease are not well understood.
Little is known about the crucial correlation between physical activity and CV events in elderly patients, because recent accumulating reports have mainly dealt with middle-aged populations. In fact, the evidence as a whole has been derived from studies targeting the middle-aged and the elderly combined, including three previous studies in Japan.10–12 In addition, few studies have evaluated the association between physical activity and long-term outcomes in Japanese. The Framingham Heart Study showed an inverse association between physical activity and mortality risk as a result of CV disease, even in 285 elderly individuals.13 However, no statistical significance was reached, possibly as a result of the limited number of events. To our knowledge, how effective and beneficial encouragement of physical activity is in the elderly is still controversial. In general, although activity has been believed to be beneficial even in the elderly,14–16 some studies have emphasized that physical activity might be harmful to the elderly.13,17 Therefore, it is essential to investigate the precise association of physical activity with CV events and mortality in Japanese elderly patients.
In the present study, the correlation between grade of physical activity and events (all CV events and all-cause mortality) was investigated in Japanese elderly patients with type 2 diabetes mellitus (T2DM). In addition, analysis was also carried out to address which component was more effective as a good predictor.
The subjects were participants who were enrolled in the Japanese Elderly Diabetic Intervention Trial (J-EDIT), a randomized, double-blind, recently completed trial of intensive or standard treatment of diabetes for the prevention of CV disease in elderly patients with T2DM. J-EDIT involved 1173 diabetic subjects who were aged 65 years or older (mean age 71.8 ± 4.6 years) and whose serum glycated hemoglobin A1c (HbA1c) level was >7.4% from 39 institutions and hospitals (the University of Tokyo Hospital, Kobe University Hospital, Nagoya University Hospital and Tokyo Metropolitan Geriatric Hospital etc.) in Japan. Written informed consent was obtained from all patients.
From among these patients enrolled in the J-EDIT, we selected 938 patients in whom complete data regarding physical activity (Baecke physical activity questionnaire) were obtained at baseline. We excluded participants who had difficulty communicating, dementia or serious deterioration of activities of daily living (ADL).
Physical activity assessed by Baecke questionnaire
To evaluate physical activity at enrolment in this trial, Baecke physical activity questionnaire was carried out as previously reported.18,19 The reliability of this score has been confirmed by many previous reports. Therefore, it is suggested that it might be a valuable monitoring tool for assessing the association of multiple domains of physical activity with the metabolic syndrome (MetS) in elderly patients with T2DM, with acceptable reliability and validity. The activity score is classified into three domains: work activity, sports activity and non-sporting leisure activity. These three components consisted of items on the frequency, duration, average amount of time spent weekly on walking, hobbies and so on, and the average amount of time spent on odd jobs and sports monthly. The types of odd jobs, sports and hobbies (e.g. dancing, gardening or fishing) were also assessed.
In analyses, each component was also divided into several groups as follows: three groups in “work activity” score (WAS; None: =0, Low: ≥1 and <3.25, High: >3.25), two groups in “sports activity” score (SAS; Low: <4.0, High: ≥4.0), and four groups (quartiles) in “non-sporting leisure-time activity” score (LTAS; L-Q1: <2.0, L-Q2: ≥2.0 and <2.25, L-Q3: ≥2.25 and <2.75, L-Q4: ≥2.75). Scores from each component were summed to yield total physical activity score (TAS; maximum 15 points). TAS was divided into four quartiles (Q1 to Q4) as follows: Q1: <5.7, Q2: ≥5.7 and <7.7, Q3: ≥7.7 and <10.5, Q4: ≥10.5.
Physical measurement, cognitive function, ADL and mood status
At enrolment, body mass index (BMI) was calculated by physical measurement of body height and weight. Both waist circumference and hip circumference were also measured, and waist-to-hip ratio (W/H ratio) was calculated. Cognitive function was determined by Mini-Mental State Examination (MMSE). Each basic or instrumental ADL was determined by Barthel index or Tokyo Metropolitan Institute of Gerontology (TMIG) index,21 respectively. Mood status was checked using the Geriatric Depression Scale (GDS-15).
Blood pressure and laboratory measurements
Blood pressure (BP) was measured in the clinic. Laboratory data obtained from blood sample collection at enrolment included lipid profile (total cholesterol [TC]), high-density lipoprotein cholesterol (HDL-C), triglyceride (TG) and low-density lipoprotein cholesterol (LDL-C; calculated by Friedewald equation: TC-HDL-C-TG/5), profile of glucose metabolism (fasting blood glucose; FBS, HbA1c, fasting insulin concentration [FIRI]), and renal function (serum creatinine). All parameters were obtained every year. Changes in each parameter were calculated as the difference between baseline and end of follow up.
Patients in the present study were continuously monitored for the occurrence of all events and deaths. In this trial, the CV events according to our definition were specified clearly as cardiac events including coronary heart disease (CHD; angina pectoris and myocardial infarction), cerebrovascular disease (CVD) including transient ischemic attack (TIA), stroke and cerebral hemorrhage, peripheral artery disease (PAD), and heart failure (HF). Individual diagnoses were classified according to the 9th International Classification of Disease (ICD-9) codes. We also classified each event into diabetes-related events (CHD, CVD, coronary revascularization, heart failure, sudden death, renal death, diabetic foot) and diabetes-independent events.
Differences in baseline characteristics across the four quartiles of physical activity (Q1 to Q4) were evaluated using analysis of variance for normally distributed variables. P-values for sex and previous CHD/CVD were calculated based on the Cochran–Armitage trend test, and others were based on the linear contrast test. Event-free survival during the follow-up period was analyzed using Kaplan–Meier curves and log–rank test. Hazard ratios (HR) for all CV events and all-cause deaths were analyzed using a Cox proportional hazards model. HR, 95% confidence intervals (CI) and P-values were presented using the lowest quartile (Q1) as the reference category.
The baseline characteristics of all the eligible subjects (n = 938) are shown according to TAS category, which was divided into four quartiles (Q1 to Q4), in Table 1.
|All (n = 938)||TAS category||P-values for linear trend|
|Quartile 1 (n = 232)||Quartile 2 (n = 229)||Quartile 3 (n = 230)||Quartile 4 (n = 247)|
|Age at baseline (years)†||71.9(4.7)||72.7(4.8)||72.1(4.8)||72.1(4.7)||70.8(4.2)||<.0001|
|Waist circumference (cm)†||84(10.3)||86(10.7)||84.2(9.6)||83.4(10.6)||82.6(9.9)||0.0002|
|Hip circumference (cm)†||93.9(8.1)||94.9(8.5)||94.1(7.8)||93.2(7.9)||93.4(7.9)||0.022|
First, patients with higher TAS grade tended to be slightly younger. Scores of each of the three components (work, sports and non-sporting leisure time) were positively associated with TAS. No significant association between BP and TAS was observed. In the lipid profile, HDL-C and TG were positively correlated with TAS grade (P-value for trend: P = 0.021 and P = 0.028, respectively); however, other lipid parameters (TC and LDL-C) showed no statistical significance. In addition, serum creatinine also tended to be lower according to TAS grade; however, the difference was very slight. Each parameter of glucose metabolism showed no statistical significance between the four quartiles of TAS.
With regard to physical measurements, TAS was negatively associated with waist circumference. BMI and hip circumference also tended to be associated with TAS grade, but without statistical significance.
Cognitive function, as determined by MMSE score, was higher according to increasing TAS grade. TMIG index as instrumental ADL also showed a similar positive association; however, Barthel index as basic ADL did not (data not shown). In addition, GDS-15 score as mood status showed an inverse association with TAS grade, suggesting an association between lower activity and depressive mood.
Regarding previous CHD, there was no significant difference among the groups. In contrast, previous CVD was less frequent in the TAS Q4 group.
Incidence of events during follow-up period
During the follow-up period of approximately 6 years (average 65.2 months), 165 all-CV events and 71 deaths in total occurred. All CV events, defined as a first event, included 45 CHD events (10 fatal and 35 non-fatal), 52 CVD events (4 fatal and 48 non-fatal), 29 diabetes-related events (9 fatal and 20 non-fatal) other than CHD and CVD, and 39 fatal diabetes-independent events. All-deaths included 11 deaths from CHD, four deaths from CVD, 11 diabetes-related deaths and 45 diabetes-independent deaths.
According to TAS grade, the incidence of these events was evaluated in all participants. As shown in Figure 1a, few all-CV events were found in the higher TAS group with statistical significance (log–rank test; P = 0.0065). In contrast to all-CV events, all-cause deaths gradually decreased according to TAS grade; however, no statistical significance was found. In analysis without adjustment, HR of each TAS grade were 0.82, 0.77 and 0.54 in Q2, Q3 and Q4, respectively, compared with Q1 as reference (P-value for trend; P = 0.006; Fig. 1b). In comparison between the lowest group (Q1) and highest group (Q4), a significant difference in incidence of first events was found from the early phase after randomization (log–rank test [Q1 vs Q4]; P = 0.006).
To determine how several parameters or TAS contributed to the reduction in all events, we carried out additional analysis after adjustment for several variables as potential confounders (Table 2). Among some variables except TAS, statistical significance was found for HbA1c, age and sex (female). Strikingly, TAS showed strong predictive power for all CV events. HR of each TAS grade were 0.74, 0.77 and 0.62 in Q2, Q3 and Q4, respectively, compared with Q1 as reference, and Q4 group statistically showed a significance (P = 0.037). However, variables including TAS except age and sex (female) were not associated with all-cause deaths. After addition of each of previous CVD or CHD to these adjusting variables, HR was 1.52 in previous CVD (95% CI 1.06–2.41, P = 0.1006) and HR was 1.45 in previous CHD (95% CI 1.90–2.33, P = 0.1236), and HR of TAS Q4 was reduced HR 0.64 (95% CI 0.35–1.17, P = 0.1447). In addition, in case of adjustment using previous CVD or CHD, its presence of each previous vascular event showed statistical significance (HR 1.52, 95% CI 1.06–2.41, P = 0.1006). Unfortunately, HR of TAS Q4 was similarly reduced HR 0.65 (95% CI 0.35–1.18, P = 0.1549).
|Variants||All CV Events||All-Cause Deaths|
|HR (95% CI)||P-value||HR (95% CI)||P-VALUE|
|Q1||1.00 Reference||1.00 Reference|
Significant correlation of TAS with cerebrovascular events compared with cardiac events
The association of TAS with cerebrovascular events and cardiac events (AP, MI, coronary revascularization and heart failure) was evaluated. TAS was significantly associated with cerebrovascular disease including both fatal and non-fatal events, although there was no significant association between cardiac events and TAS (Fig. 2).
Next, we compared the predictive power of TAS according to sex and age (young-elderly aged 65–74 years and old-elderly patients aged ≥75 years; Table 3). First, with regard to sex, the predictive power of TAS for all CV events was stronger in women than in men. In addition, TAS in the young-elderly significantly predicted all CV events. In the old-elderly, a similar tendency was observed; however, the association did not reach statistical significance.
|HR (95% CI)||P-value||HR (95% CI)||P-value|
|All CV events|
|Q1||1 Reference||1 Reference|
|Q1||1 Reference||1 Reference|
|HR(95% CI)||HR(95% CI)|
|All CV events|
|Q1||1 Reference||1 Reference|
|Q1||1 Reference||1 Reference|
Comparison of predictive power of each component in TAS
As described in the Methods section, TAS consists of three components: work activity, sports activity and non-sporting leisure-time activity. Subanalysis clearly showed that the predictive power of “work activity” for all CV events was stronger than that of the other components (log–rank test; P = 0.0003) (Table 4). After adjustment, its power remained. The risk reduction of work activity was also significant, even in all-cause mortality (log–rank test; P = 0.004; data not shown). There was no statistical significance for sports activity. Regarding leisure-time activity, the risk reduction of it for all CV events in Q3 was strongest; however, statistical analysis did not show significance (log–rank test; P = 0.11).
|HR||95% CI||P-value||HR||95% CI||P-value|
|None||1 Reference||1 Reference|
|Low||1 Reference||1 Reference|
|Q1||1 Reference||1 Reference|
Changes in each parameter during follow-up period
To explore which parameter contributed to the risk reduction of all primary events, the changes in values (from baseline to the end of the follow-up period) of each parameter were calculated according to TAS grade (Table 5). Among the parameters, the differences in laboratory data, including lipid parameters and glucose metabolism, BP, physical measurements, cognitive function and depression scale, between TAS grades were not significant, suggesting that a higher level of physical activity itself was important in the risk reduction of events in elderly patients with T2DM.
|Variables||TAS category||P for trend|
|Quartile 1||Quartile 2||Quartile 3||Quartile 4|
|Waist circumference (cm)||–0.8(0.3)||–0.1(0.4)||0.6(0.5)||–0.1(0.3)||0.1608|
|Hip circumference (cm)||0.5(0.4)||0.0(0.4)||0.4(0.4)||–0.1(0.4)||0.6398|
Physical activity has been shown to reduce the risk of CV events; however, the biological mechanisms underlying this finding are still unclear. In the present study, the association of physical activity, as determined by TAS at baseline, with all CV events and all-cause mortality was evaluated in the J-EDIT study.
Higher TAS grade was significantly associated with a risk reduction in non-fatal all CV events; however, the association with all-cause mortality was not significant. In addition, among the three components of TAS, the predictive power of “work activity” was stronger than that of the other components – sports and leisure-time activity. There was no significant difference in the change in almost all parameters among TAS grades throughout the follow-up period in the present trial. Therefore, we emphasize the following conclusion: (i) our data provide evidence that lower physical activity is a strong and independent predictor of CV events in the elderly with T2DM beyond traditional risk factors; (ii) in addition to routine strict management of laboratory data in clinical practice, engagement with patients to enhance and/or maintain physical activity in their lifestyle is also important. In fact, the elderly have retired from their routine jobs. The results of the present study show that they should do at least slight work routinely in their daily life, such as cooking or gardening. It might be meaningful for them to carry out some activity and continue it by themselves.
Physical activity is a well-established approach to reducing the risk of many chronic diseases. Most studies have shown a significant relative reduction in the incidence of CV events in physically active participants; however, the range of benefit showed considerable variation. For example, Myers et al. reported a marked reduction in all-cause mortality of 72% between active and inactive male participants during 6 years of follow up,21 whereas Lee et al. found a risk reduction of just 13%.22 It is clear that the risk reduction might vary depending on adjustment for important covariables, such as BP and profiles of lipid and glucose metabolism. With regard to adjustment for several relevant risk factors, a meta-analysis handling a total of 33 studies with 883 372 participants (follow-up period from 4 years to over 20 years) clearly showed an important correlation of higher physical activity with a risk reduction in CV mortality of 35% (95% CI 30–40%).23 In addition, all-cause mortality was also reduced by 33% (95% CI 28–37%). This systematic review by meta-analysis emphasized that physical activity was associated with a marked decrease in CV and all-cause mortality in both sexes, even after adjusting for other relevant risk factors. In the present study, there was a good correlation between TAS and all CV events. However, after adjustment of previous atherosclerotic diseases, its presence of previous CVD or CHD showed a significant association with CV events during the follow-up period. Consequently, the predictive power of TAS against all CV events was slightly decreased. These observations might suggest a high risk of recurrence of CV events in T2DM patients beyond TAS grade at the baseline. Therefore, further subanalysis to simply evaluate the predictive power of TAS as primary prevention against CV events using elderly patients without both previous CVD and CHD is required. In addition, regarding all-cause mortality, TAS tended to show an association with it; however, no statistical significance was reached. One of the hypotheses to explain the relationship and discrepancy is that the sample size was relatively small and non-fatal CV events rather than fatal events might be frequently observed in all participants with T2DM at the baseline.
Next, we focused on cognitive function and depressive mood. The presence of geriatric syndrome including cognitive dysfunction has been shown to be a major factor in decline in physical activity level in the older elderly. Besides traditional risk factors, it has been clearly shown that “depressive mood” readily causes a decline in physical activity, leading to increased risk of CV disease.24 In addition, patients with depression had a worse prognosis than those without depression after a myocardial infarction.25,26 Prospective studies have shown that depressed people develop a more sedentary lifestyle and become less physically active.27,28 In fact, the GDS score was higher in the lower TAS group in the present study as well. Therefore, this evidence that the importance of physical activity in the risk reduction of CV events is also associated with depressive mood is consistent with previous reports. However, the average GDS score was not so high (range 3–5 points). In addition, depressive score did not decrease, even in the lowest TAS group (Q1), throughout the follow-up period. In subanalysis, the inverse correlation between GDS and TAS was more clearly found in young-elderly patients, compared with old-elderly patients. In view of these results, especially in young-elderly patients, detailed assessment of the patient's mentality, including depressive mood, should be considered more aggressively and routinely.
The predictive power of TAS was compared between CHD and CVD. The relative risk of CVD decreased with increasing TAS, with statistical significance; however, no significance was found for CHD. In fact, we found some evidence regarding this discrepancy. Although the incidence of CHD among physically active elderly men in the Honolulu Heart Program study was less than half that in more sedentary men,29 no clear association was observed in the Established Populations for Epidemiologic Studies of the Elderly study.30 Compared with CHD, the correlation of physical activity with stroke has not been extensively examined for any age group.31 However, a meta-analysis handling five epidemiological case–control studies has reported that all studies had consistent evidence showing a large advantage of higher physical activity in reducing the risk of stroke.32
As another interest in our data, “work activity” was the most potent predictor of first events among the three components. This suggests that, even if the patient's age is over 65 years, the fact that they are motivated to routinely try to do at least any slight work might enhance their total physical activity.
The present analysis was based on the physical activity score measured once at enrolment in the J-EDIT study. However, during the follow-up period of this trial, some patients showed a gradual decline in TAS. It is notable that new development or progression of diabetic complications, such as neuropathy and retinopathy, readily leads to a decline in physical activity. Therefore, further investigation to evaluate which factor mainly caused the decline in TAS throughout the follow-up period is necessary. This could provide supportive information on the cause–effect relationship of the associations found in this trial.
In addition, the current associations might not be extended to all populations, because the enrolled participants in the present trial were patients with T2DM. Whether the observed associations can be generalized to populations of much older ages and populations without T2DM is unknown.
This prospective follow-up study confirmed that lower physical activity is a strong independent predictor of onset of CV events, even in Japanese elderly with T2DM. The data in the present study suggest the potential of activity to enhance overall health and well-being with aging. Ultimately, the key is to aggressively translate these findings into public health efforts.
The majority of elderly patients still have a primarily sedentary life. Numerous studies have already addressed the importance of physical activity in health management; however, unfortunately, medical staff might not have been educated about how to promote and augment the level of physical activity in elderly patients. Therefore, as well as strict management of each atherosclerotic risk factor, we should aggressively assess physical activity (especially working) and encourage elderly patients to increase or maintain their level of physical activity.
We thank all patients, physicians, and staff who took part in the J-EDIT study.
The registration number for this clinical trial was UMIN000000890. This study was financially supported by Research Grants for Longevity Sciences from the Ministry of Health and Labour, and Welfare (H12-Choju-016, H15-Chojyu-016, H17-Choju-Ordinal-013) and the Japan Foundation for Aging and Health.
Conflict of interest
There is no conflict of interest. The J-EDIT Study Group has not cleared any potential conflicts.
- 7WHO Study Group on Epidemiology and Prevention of Cardiovascular Diseases in the Elderly. Epidemiology and Prevention of Cardiovascular Diseases in Elderly People: Report of A WHO Study Group. Geneva: World Health Organization, 1995; WHO Technical Report Series No. 853.
- 8US Department of Health and Human Services. Physical Activity and Health: A Report of the Surgeon General. Atlanta, GA: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, 1996.