Search

We developed a new Physical Score (PS) consisting of comprehensive physical fitness indicators and elucidated the association between the resultant PS and metabolic diseases, i.e., diabetes, hypertension, dyslipidemia, fatty liver, and metabolic syndrome (MetS), among Japanese. Analyzed were 49,850 persons (30,039 men) aged 30 to 69 y who underwent physical fitness tests. Principal component analysis was performed on the correlation matrix of the physical fitness test results (relative grip strength, single-leg balance with eyes closed, and forward bending) according to sex and age. We defined the PS as the first principal component score. A formula was developed for various age groups comprised of men and women from 30 to 69 years of age from which the PS for each age and sex was calculated. The PS for both men and women was normally distributed with a value of 0 ± 1.15-1.16. Multivariate logistic regression analysis showed that the risk of metabolic diseases increased approximately 1.1-1.6 times per each 1-point reduction in the PS. The association between PS and MetS was particularly strong in that a 1-point reduction in the PS increased the risk of MetS by 1.54 times (95% confidence interval 1.46 to 1.62) in men and by 1.21 times (1.15 to 1.28) in women. The association between a lower PS and disease risk was stronger in younger men for fatty liver and in older men for MetS. Conversely, in women, the association between a lower PS and disease risk was stronger in older women for fatty liver and in younger women for MetS. For diabetes, hypertension, and dyslipidemia, the change in the impact of PS reductions across age groups was small. The PS is a useful and simple non-invasive tool for screening Japanese people for metabolic diseases.

Background: Although the impact of a history of coronary artery disease (CAD), cerebrovascular disease (CVD) or both and diabetes mellites (DM) on cardiovascular disease is essential for cardiovascular risk management in clinical practice, little is known about such an impact. Therefore, we investigated the association of prior CAD, prior CVD, DM and their combinations with the risk of subsequent cardiovascular disease. Methods: This historical cohort study included 342,033 participants aged 18-72 years followed for ≥5 years between 2008-2016. Eight groups were created according to whether they were with or without prior CAD, prior CVD and/or DM. DM was defined by fasting plasma glucose, HbA1c and antidiabetic drug prescription. Prior and subsequent CAD and CVD were identified by claims using ICD-10 codes, medical procedures and questionnaires. Cox regression models evaluated the risk of cardiovascular events. Results: Median follow-up period was 6.4 years. Incidence of composite cardiovascular events due to CAD and/or CVD in CAD-/CVD-, CAD+/CVD-, CAD-/CVD+ and CAD+/CVD+ groups were 1.92, 6.94, 25.14 and 31.98 per 1,000 person-years in non-DM and 8.66, 18.04, 39.98 and 60.72 in DM, respectively. Hazard ratio of cardiovascular events compared to CAD-/CVD-/non-DM were 1.66 (95% CI 1.55-1.78) in CAD-/CVD-/DM and 1.84 (1.56-2.18) in CAD+/CVD-/non-DM. CVD+ increased the risk of cardiovascular events approximately 4- to 7- fold regardless of CAD+ or DM. Conclusions:DM increase the risk of cardiovascular disease as high as a history of CAD, while prior CVD alone increased the risk of future cardiovascular disease without additional effects by DM.

Aims Little is known about the relationship between medication adherence for oral hypoglycemic agents (OHAs) and glycemic control after adjusting healthy adherer effect in large scale study. Thus, adjusting for health-related behaviors, we investigated the clinical variables associated with medication adherence and the relationship between medication adherence and glycemic control using a large claims database. Methods Analyzed were 8805 patients with diabetes whose medication records for OHA were available for at least 1 year. Medication adherence was evaluated by the proportion of days covered (PDC). Multivariate logistic regression model was used to identify clinical variables significantly associated with non-adherence. Multiple regression analysis evaluated the relationship between PDC and HbA1c after adjusting for health-related behaviors. Results Mean PDC was 80.1% and 32.8% of patients were non-adherence. Logistic analysis indicated that older age and taking concomitant medications were significantly associated with adherence while skipping breakfast (odds ratio 0.66 [95% CI 0.57–0.76]), late-night eating (0.86 [0.75–0.98]), and current smoking (0.89 [0.80–0.99]) were significantly associated with non-adherence. Conclusions Skipping breakfast, late-night eating and current smoking were significantly associated with medication adherence, suggesting that clinicians pay attention to those health-related behaviors to achieve good medication adherence.

Background Although both a history of cerebrovascular disease (CVD) and glucose abnormality are risk factors for CVD, few large studies have examined their association with subsequent CVD in the same cohort. Thus, we compared the impact of prior CVD, glucose status, and their combinations on subsequent CVD using real-world data. Methods This is a retrospective cohort study including 363,627 men aged 18–72 years followed for ≥ 3 years between 2008 and 2016. Participants were classified as normoglycemia, borderline glycemia, or diabetes defined by fasting plasma glucose, HbA1c, and antidiabetic drug prescription. Prior and subsequent CVD (i.e. ischemic stroke, transient ischemic attack, and non-traumatic intracerebral hemorrhage) were identified according to claims using ICD-10 codes, medical procedures, and questionnaires. Results Participants’ mean age was 46.1 ± 9.3, and median follow up was 5.2 (4.2, 6.7) years. Cox regression analysis showed that prior CVD + conferred excess risk for CVD regardless of glucose status (normoglycemia: hazard ratio (HR), 8.77; 95% CI 6.96–11.05; borderline glycemia: HR, 7.40, 95% CI 5.97–9.17; diabetes: HR, 5.73, 95% CI 4.52–7.25). Compared with normoglycemia, borderline glycemia did not influence risk of CVD, whereas diabetes affected subsequent CVD in those with CVD- (HR, 1.50, 95% CI 1.34–1.68). In CVD-/diabetes, age, current smoking, systolic blood pressure, high-density lipoprotein cholesterol, and HbA1c were associated with risk of CVD, but only systolic blood pressure was related to CVD risk in CVD + /diabetes. Conclusions Prior CVD had a greater impact on the risk of CVD than glucose tolerance and glycemic control. In participants with diabetes and prior CVD, systolic blood pressure was a stronger risk factor than HbA1c. Individualized treatment strategies should consider glucose tolerance status and prior CVD.

Our purpose in the research was to clarify the impact of medication adherence to oral hypoglycemic agents during a 1-year period and subsequent glycemic control on the risk of micro- and macrovascular diseases.Examined was a nationwide claims database on 13,256 individuals with diabetic eye disease without requiring prior treatment, 7,862 without prior initiation of dialysis, 15,556 without prior coronary artery disease, 16,243 without prior cerebrovascular disease, and 19,386 without prior heart failure from 2008 to 2016 in Japan. Medication adherence was evaluated by the proportion of days covered. Patients were considered to have poor adherence if the proportion of days covered was80%. Multivariate Cox regression model identified risks of micro- and macrovascular diseases.In each group, mean age was 53 to 54 years, HbA1c was 7.1% to 7.2%, and median follow-up period was 4.6 to 5.1 years, and the percentage of poor adherence was approximately 30%. During the study period, 532 treatment-requiring diabetic eye disease, 75 dialysis, 389 coronary artery disease, 316 cerebrovascular disease, and 144 heart failure events occurred. Multivariate Cox regression model revealed that the hazard ratio (95% confidence interval) of dialysis in the poor adherence group was 2.04 (1.27-3.30) compared with the good adherence group. The hazard ratios in the poor adherence/poor glycemic control group were 3.34 (2.63-4.24) for treatment-requiring diabetic eye disease, 4.23 (2.17-8.26) for dialysis, 1.69 (1.23-2.31) for coronary artery disease, and 2.08 (1.25-3.48) for heart failure compared with the good adherence/good glycemic control group.Poor medication adherence was an independent risk factor for the initiation of dialysis, suggesting that clinicians must pay close attention to these patients.

Background Machine learning (ML) algorithms have been widely introduced to diabetes research including those for the identification of hypoglycemia. Objective The objective of this meta-analysis is to assess the current ability of ML algorithms to detect hypoglycemia (ie, alert to hypoglycemia coinciding with its symptoms) or predict hypoglycemia (ie, alert to hypoglycemia before its symptoms have occurred). Methods Electronic literature searches (from January 1, 1950, to September 14, 2020) were conducted using the Dialog platform that covers 96 databases of peer-reviewed literature. Included studies had to train the ML algorithm in order to build a model to detect or predict hypoglycemia and test its performance. The set of 2 × 2 data (ie, number of true positives, false positives, true negatives, and false negatives) was pooled with a hierarchical summary receiver operating characteristic model. Results A total of 33 studies (14 studies for detecting hypoglycemia and 19 studies for predicting hypoglycemia) were eligible. For detection of hypoglycemia, pooled estimates (95% CI) of sensitivity, specificity, positive likelihood ratio (PLR), and negative likelihood ratio (NLR) were 0.79 (0.75-0.83), 0.80 (0.64-0.91), 8.05 (4.79-13.51), and 0.18 (0.12-0.27), respectively. For prediction of hypoglycemia, pooled estimates (95% CI) were 0.80 (0.72-0.86) for sensitivity, 0.92 (0.87-0.96) for specificity, 10.42 (5.82-18.65) for PLR, and 0.22 (0.15-0.31) for NLR. Conclusions Current ML algorithms have insufficient ability to detect ongoing hypoglycemia and considerate ability to predict impeding hypoglycemia in patients with diabetes mellitus using hypoglycemic drugs with regard to diagnostic tests in accordance with the Users’ Guide to Medical Literature (PLR should be ≥5 and NLR should be ≤0.2 for moderate reliability). However, it should be emphasized that the clinical applicability of these ML algorithms should be evaluated according to patients’ risk profiles such as for hypoglycemia and its associated complications (eg, arrhythmia, neuroglycopenia) as well as the average ability of the ML algorithms. Continued research is required to develop more accurate ML algorithms than those that currently exist and to enhance the feasibility of applying ML in clinical settings. Trial Registration PROSPERO International Prospective Register of Systematic Reviews CRD42020163682; http://www.crd.york.ac.uk/PROSPERO/display_record.php?ID=CRD42020163682

To determine associations between severity of hypertension and risk of starting dialysis in the presence or absence of diabetes mellitus (DM). A nationwide database with claims data on 258 874 people with and without DM aged 19-72 years in Japan was used to elucidate the impact of severity of hypertension on starting dialysis. Initiation of dialysis was determined from claims using International Classification of Diseases-10 codes and medical procedures. Using multivariate Cox modeling, we investigated the severity of hypertension to predict the initiation of dialysis with and without DM. Hypertension was significantly associated with the initiation of dialysis regardless of DM. The incidence of starting dialysis in those with systolic blood pressure (SBP) ≤119 mm Hg and DM (DM+) was almost the same as in those with SBP ≥150 mm Hg and absence of DM (DM−). In comparison with SBP ≤119 mm Hg, SBP ≥150 mm Hg significantly increased the risk of the initiation of dialysis about 2.5 times regardless of DM+ or DM−. Compared with DM− and SBP ≤119 mm Hg, the HR for DM+ and SBP ≥150 mm Hg was 6.88 (95% CI 3.66 to 12.9). Although the risks of hypertension differed only slightly regardless of the presence or absence of DM, risks for starting dialysis with DM+ and SBP ≤119 mm Hg were equivalent to DM− and SBP ≥150 mm Hg, indicating more strict blood pressure interventions in DM+ are needed to avoid dialysis. Future studies are required to clarify the cut-off SBP level to avoid initiation of dialysis considering the risks of strict control of blood pressure.

OBJECTIVE To determine associations of systolic blood pressure (SBP) and diastolic blood pressure (DBP) with new-onset coronary artery disease (CAD) or cerebrovascular disease (CVD) according to glucose status. RESEARCH DESIGN AND METHODS Examined was a nationwide claims database from 2008 to 2016 on 593,196 individuals. A Cox proportional hazards model identified risks of CAD and CVD events among five levels of SBP and DBP. RESULTS During the study period 2,240 CAD and 3,207 CVD events occurred. Compared with SBP ≤119 mmHg, which was the lowest quintile of SBP, hazard ratios (95% CI) for CAD/CVD in the 4 higher quintiles (120–129, 130–139, 140–149, ≥150 mmHg) gradually increased from 2.10 (1.73–2.56)/1.46 (1.27–1.68) in quintile 2 to 3.21 (2.37–4.34)/4.76 (3.94–5.75) in quintile 5 for normoglycemia, from 1.39 (1.14–1.69)/1.70 (1.44–2.01) in quintile 2 to 2.52 (1.95–3.26)/4.12 (3.38–5.02) in quintile 5 for borderline glycemia, and from 1.50 (1.19–1.90)/1.72 (1.31–2.26) in quintile 2 to 2.52 (1.95–3.26)/3.54 (2.66–4.70) in quintile 5 for diabetes. A similar trend was observed for DBP across 4 quintiles (75–79, 80–84, 85–89, and ≥90 mmHg) compared with ≥74 mmHg, which was the lowest quintile. CONCLUSIONS Results indicated that cardiovascular risks gradually increased with increases in SBP and DBP regardless of the presence of and degree of a glucose abnormality. Further interventional trials are required to apply findings from this cohort study to clinical practice.

BACKGROUND Machine learning (ML) algorithms have been widely introduced to diabetes research including those for the identification of hypoglycemia. OBJECTIVE The objective of this meta-analysis is to assess the current ability of ML algorithms to detect hypoglycemia (ie, alert to hypoglycemia coinciding with its symptoms) or predict hypoglycemia (ie, alert to hypoglycemia before its symptoms have occurred). METHODS Electronic literature searches (from January 1, 1950, to September 14, 2020) were conducted using the Dialog platform that covers 96 databases of peer-reviewed literature. Included studies had to train the ML algorithm in order to build a model to detect or predict hypoglycemia and test its performance. The set of 2 × 2 data (ie, number of true positives, false positives, true negatives, and false negatives) was pooled with a hierarchical summary receiver operating characteristic model. RESULTS A total of 33 studies (14 studies for detecting hypoglycemia and 19 studies for predicting hypoglycemia) were eligible. For detection of hypoglycemia, pooled estimates (95% CI) of sensitivity, specificity, positive likelihood ratio (PLR), and negative likelihood ratio (NLR) were 0.79 (0.75-0.83), 0.80 (0.64-0.91), 8.05 (4.79-13.51), and 0.18 (0.12-0.27), respectively. For prediction of hypoglycemia, pooled estimates (95% CI) were 0.80 (0.72-0.86) for sensitivity, 0.92 (0.87-0.96) for specificity, 10.42 (5.82-18.65) for PLR, and 0.22 (0.15-0.31) for NLR. CONCLUSIONS Current ML algorithms have insufficient ability to detect ongoing hypoglycemia and considerate ability to predict impeding hypoglycemia in patients with diabetes mellitus using hypoglycemic drugs with regard to diagnostic tests in accordance with the Users’ Guide to Medical Literature (PLR should be ≥5 and NLR should be ≤0.2 for moderate reliability). However, it should be emphasized that the clinical applicability of these ML algorithms should be evaluated according to patients’ risk profiles such as for hypoglycemia and its associated complications (eg, arrhythmia, neuroglycopenia) as well as the average ability of the ML algorithms. Continued research is required to develop more accurate ML algorithms than those that currently exist and to enhance the feasibility of applying ML in clinical settings. CLINICALTRIAL PROSPERO International Prospective Register of Systematic Reviews CRD42020163682; http://www.crd.york.ac.uk/PROSPERO/display_record.php?ID=CRD42020163682

BACKGROUND Applications of machine learning for the early detection of diseases for which a clear-cut diagnostic gold standard exists have been evaluated. However, little is known about the usefulness of machine learning approaches in the decision-making process for decisions such as insulin initiation by diabetes specialists for which no absolute standards exist in clinical settings. OBJECTIVE The objectives of this study were to examine the ability of machine learning models to predict insulin initiation by specialists and whether the machine learning approach could support decision making by general physicians for insulin initiation in patients with type 2 diabetes. METHODS Data from patients prescribed hypoglycemic agents from December 2009 to March 2015 were extracted from diabetes specialists’ registries, resulting in a sample size of 4860 patients who had received initial monotherapy with either insulin (n=293) or noninsulin (n=4567). Neural network output was insulin initiation ranging from 0 to 1 with a cutoff of >0.5 for the dichotomous classification. Accuracy, recall, and area under the receiver operating characteristic curve (AUC) were calculated to compare the ability of machine learning models to make decisions regarding insulin initiation to the decision-making ability of logistic regression and general physicians. By comparing the decision-making ability of machine learning and logistic regression to that of general physicians, 7 cases were chosen based on patient information as the gold standard based on the agreement of 8 of the 9 specialists. RESULTS The AUCs, accuracy, and recall of logistic regression were higher than those of machine learning (AUCs of 0.89-0.90 for logistic regression versus 0.67-0.74 for machine learning). When the examination was limited to cases receiving insulin, discrimination by machine learning was similar to that of logistic regression analysis (recall of 0.05-0.68 for logistic regression versus 0.11-0.52 for machine learning). Accuracies of logistic regression, a machine learning model (downsampling ratio of 1:8), and general physicians were 0.80, 0.70, and 0.66, respectively, for 43 randomly selected cases. For the 7 gold standard cases, the accuracies of logistic regression and the machine learning model were 1.00 and 0.86, respectively, with a downsampling ratio of 1:8, which were higher than the accuracy of general physicians (ie, 0.43). CONCLUSIONS Although we found no superior performance of machine learning over logistic regression, machine learning had higher accuracy in prediction of insulin initiation than general physicians, defined by diabetes specialists’ choice of the gold standard. Further study is needed before the use of machine learning–based decision support systems for insulin initiation can be incorporated into clinical practice.

Objective Although glucose abnormality status (GAS), prior coronary artery disease (CAD), and other traditional risk factors affect the incidence of subsequent CAD, their impact in the same cohort has been scantly studied. Research design and methods We analyzed data from a nationwide claims database in Japan that was accumulated during 2008–2016 involving 138,162 men aged 18–72 years. Participants were classified as having normoglycemia, borderline glycemia, or diabetes mellitus (DM) with prior CAD (CAD+) or without prior CAD (CAD−). Cox regression model identified variables related to the incidence of CAD. Results Among CAD−, management of traditional risks differed from those with and without subsequent CAD events. On the other hand, such differences were weaker in borderline glycemia and DM CAD+, and the influence of traditional risk factors on subsequent CAD was not observed. Cox regression model showed that borderline glycemia and DM confer approximately 1.2- and 2.8-fold excess risks of CAD, respectively, compared with CAD− with normoglycemia. CAD+ confers approximately a 5- to 8-fold increased risk. The impacts of DM and prior CAD additively reached a hazard ratio (HR) of 15.74 (95% confidence interval [CI]: 11.82–21.00). However, the HR in those with borderline glycemia and CAD+ was 7.20 (95% CI: 5.01–10.34), which was not different from those with normoglycemia and CAD+. Conclusion Control status of traditional risk factors and impact on subsequent CAD differ among categories of glycemic status with and without prior CAD. Individualizing treatment strategies is needed in consideration of risk factors, such as GAS and CAD+.

Background Applications of machine learning for the early detection of diseases for which a clear-cut diagnostic gold standard exists have been evaluated. However, little is known about the usefulness of machine learning approaches in the decision-making process for decisions such as insulin initiation by diabetes specialists for which no absolute standards exist in clinical settings. Objective The objectives of this study were to examine the ability of machine learning models to predict insulin initiation by specialists and whether the machine learning approach could support decision making by general physicians for insulin initiation in patients with type 2 diabetes. Methods Data from patients prescribed hypoglycemic agents from December 2009 to March 2015 were extracted from diabetes specialists’ registries, resulting in a sample size of 4860 patients who had received initial monotherapy with either insulin (n=293) or noninsulin (n=4567). Neural network output was insulin initiation ranging from 0 to 1 with a cutoff of >0.5 for the dichotomous classification. Accuracy, recall, and area under the receiver operating characteristic curve (AUC) were calculated to compare the ability of machine learning models to make decisions regarding insulin initiation to the decision-making ability of logistic regression and general physicians. By comparing the decision-making ability of machine learning and logistic regression to that of general physicians, 7 cases were chosen based on patient information as the gold standard based on the agreement of 8 of the 9 specialists. Results The AUCs, accuracy, and recall of logistic regression were higher than those of machine learning (AUCs of 0.89-0.90 for logistic regression versus 0.67-0.74 for machine learning). When the examination was limited to cases receiving insulin, discrimination by machine learning was similar to that of logistic regression analysis (recall of 0.05-0.68 for logistic regression versus 0.11-0.52 for machine learning). Accuracies of logistic regression, a machine learning model (downsampling ratio of 1:8), and general physicians were 0.80, 0.70, and 0.66, respectively, for 43 randomly selected cases. For the 7 gold standard cases, the accuracies of logistic regression and the machine learning model were 1.00 and 0.86, respectively, with a downsampling ratio of 1:8, which were higher than the accuracy of general physicians (ie, 0.43). Conclusions Although we found no superior performance of machine learning over logistic regression, machine learning had higher accuracy in prediction of insulin initiation than general physicians, defined by diabetes specialists’ choice of the gold standard. Further study is needed before the use of machine learning–based decision support systems for insulin initiation can be incorporated into clinical practice.

ObjectiveDeclining healthy life expectancy due to functional disability is relevant and urgent because of its association with decreased quality of life and also for its enormous socioeconomic impact. The aim of this study is to examine the impact of diabetes, hypertension, dyslipidemia and physical activity habits on functional disability among community-dwelling Japanese adults.Research design and methodsThis is a population-based retrospective cohort study including 9673 people aged 39–98 years in Japan (4420, men). Functional disability was defined as a condition meeting Japan’s new long-term care insurance certification requirements for the need of assistance in the activities of daily living whether by caregivers or assistive devices. Cox proportional-hazards regression model identified variables related to functional disability.ResultsMedian follow-up was 3.7 years. During the study period, 165 disabilities occurred in the overall study population. Multivariate analysis showed that diabetes (HR 1.74 (95% CI 1.12 to 2.68)) and no physical activity habit (HR 1.83 (1.27 to 2.65)) presented increased risks for disability. HR for disability increased with the number of risk factors (HR of individuals with four conditions, 3.96 (1.59 to 9.99) vs individuals with none of those conditions as a reference). HR for disability among patients with diabetes with and without a physical activity habit was 1.68 (0.70 to 4.04) and 3.19 (1.79 to 5.70), respectively, compared with individuals without diabetes with a physical activity habit.ConclusionsThe combination of diabetes and lack of habitual physical activity is predictive of functional disability in Japanese. Habitual physical activity attenuates the risk of functional disability in patients with diabetes.