Validation of an algorithm for identifying type 1 diabetes in adults based on electronic health record data

Heterogeneity in the effect of type 2 diabetes on dementia incidence in a diverse cohort of Asian American and non-Latino White older adults

Dementia incidence is lower among Asian Americans than among Whites, despite higher prevalence of type 2 diabetes, a well-known dementia risk factor. Determinants of dementia, including type 2 diabetes, have rarely been studied in Asian Americans. We followed 4846 Chinese, 4129 Filipino, 2784 Japanese, 820 South Asian, and 123 360 non-Latino White members of a California-based integrated health-care delivery system from 2002 to 2020. We estimated dementia incidence rates by race/ethnicity and type 2 diabetes status, and we fitted Cox proportional hazards and Aalen additive hazards models for the effect of type 2 diabetes (assessed 5 years before baseline) on age of dementia diagnosis, controlling for sex/gender, educational attainment, nativity, height, race/ethnicity, and a race/ethnicity × diabetes interaction. Type 2 diabetes was associated with higher dementia incidence in Whites (hazard ratio [HR] = 1.46; 95% CI, 1.40-1.52). Compared with Whites, the estimated effect of diabetes was larger in South Asians (HR = 2.26; 95% CI, 1.48-3.44), slightly smaller in Chinese (HR = 1.32; 95% CI, 1.08-1.62) and Filipino (HR = 1.31; 95% CI, 1.08-1.60) individuals, and similar in Japanese individuals (HR = 1.44; 95% CI, 1.15-1.81). Heterogeneity in this association across Asian subgroups may be related to type 2 diabetes severity. Understanding this heterogeneity may inform prevention strategies to prevent dementia for all racial and ethnic groups.

Antihypertensive Medication Use Trajectories After Bariatric Surgery: A Matched Cohort Study

BACKGROUND

Metabolic and bariatric surgery (MBS) is the most effective and durable treatment for obesity. We aimed to compare the trajectories of antihypertensive medication (AHM) use among obese individuals treated and not treated with MBS.

METHODS

Adults with a body mass index of ≥35 kg/m2 were identified in the Merative Database (US employer-based claims database). Individuals treated with versus without MBS were matched 1:1 using baseline demographic and clinical characteristics as well as AHM utilization. Monthly AHM use was examined in the 3 years after the index date using generalized estimating equations. Subanalyses investigated rates of AHM discontinuation, AHM initiation, and apparent treatment-resistant hypertension.

RESULTS

The primary cohort included 43 206 adults who underwent MBS matched with 43 206 who did not. Compared with no MBS, those treated with MBS had sustained, markedly lower rates of AHM use (31% versus 15% at 12 months; 32% versus 17% at 36 months). Among patients on AHM at baseline, 42% of patients treated with MBS versus 7% treated medically discontinued AHM use (P<0.01). The risk of apparent treatment-resistant hypertension was 3.41× higher (95% CI, 2.91-4.01; P<0.01) 2 years after the index date in patients who did not undergo MBS. Among those without hypertension treated with MBS versus no MBS, 7% versus 21% required AHM at 2 years.

CONCLUSIONS

MBS is associated with lower rates of AHM use, higher rates of AHM discontinuation, and lower rates of AHM initiation among patients not taking AHM. These findings suggest that MBS is both an effective treatment and a preventative measure for hypertension.

New-Onset Type 2 Diabetes after Bariatric Surgery: A Matched Cohort Study

INTRODUCTION

The objective of this study is to determine the difference in rates of new-onset type 2 diabetes (T2D) for individuals who have had metabolic and bariatric surgery (MBS) and similar individuals who did not have MBS, and to determine whether differences in new-onset T2D differ depending on whether the individual had prediabetes at baseline.

METHODS

This study used data from a large United States employer-based retrospective claims database from 2016 to 2021 (analysis completed in 2023). Individuals who did and did not have MBS were matched 1:1 on index year, sex, age, health plan type, region, body mass index, baseline healthcare costs, other obesity-related comorbidities, prediabetes diagnosis, and inpatient admissions in the year before the index date. New-onset T2D was examined at 1 (18,752 matches) and 3 (5,416 matches) years after the index date and stratified by baseline prediabetes.

RESULTS

Among the full cohort of individuals with and without prediabetes at baseline, 0.1% and 2.7% of individuals who had did and did not have MBS developed T2D within 1 year after the index date, respectively (difference=2.6, 95% CI 2.4-2.8), and 0.3% and 8.4% of individuals who did and did not have MBS developed T2D within 3 years after the index date, respectively (difference=8.1, 95% CI 7.3-8.8). The difference in new-onset T2D was greatest among individuals with prediabetes at baseline.

CONCLUSIONS

This study demonstrated patients with obesity and without T2D who undergo MBS are significantly less likely to develop new-onset T2D compared to matched non-MBS patients.

COVID-19 infection and incident diabetes in American Indian and Alaska Native people: a retrospective cohort study

Background

Evidence suggests an increased risk of new-onset diabetes following COVID-19 infection. American Indian/Alaska Native (AI/AN) people were disparately impacted by the COVID-19 pandemic and historically have had higher diabetes incidence than other racial/ethnic groups in the US. We measured the association between COVID-19 infection and incident diabetes in AI/AN people.

Methods

We conducted a retrospective cohort study using de-identified patient data from the Indian Health Service's (IHS) National Patient Information Reporting System. We estimated age-adjusted diabetes incidence rates, incidence rate ratios, and adjusted hazard ratios among three cohorts spanning pre-pandemic (1/1/2018-2/28/2020) and pandemic (3/1/2020-12/31/2021) timeframes: 1) pre-pandemic cohort (1,503,085 individuals); 2) no-COVID-19 pandemic cohort (1,344,339 individuals); and 3) COVID-19 cohort (176,483 individuals).

Findings

The COVID-19 cohort had an increased hazard of diabetes compared to the no-COVID-19 group (adjusted hazard ratio (aHR) = 1.56; 95% CI: 1.50-1.62) and the pre-pandemic group (aHR = 1.27; 95% CI: 1.22-1.32). The association between COVID-19 infection and new-onset diabetes was stronger in those with severe COVID-19 illness. A sensitivity analysis comparing the COVID-19 cohort to members of other cohorts that had acute upper respiratory infections showed an attenuated but higher risk of new-onset diabetes in those with COVID-19.

Interpretation

AI/AN people diagnosed with COVID-19 had an elevated risk of a new diabetes diagnosis when compared to the no-COVID-19 group and the pre-pandemic group. The increased diabetes risk in the COVID-19 group remained in a sensitivity analysis that limited the comparator groups to individuals with an AURI diagnosis.

Funding

US National Institute of Diabetes and Digestive and Kidney Diseases.

Safety and Efficacy of Insulins in Critically Ill Patients Receiving Continuous Enteral Nutrition

OBJECTIVE

There is a relative lack of consensus regarding the optimal management of hyperglycemia in patients receiving continuous enteral nutrition (EN), with or without a diagnosis of diabetes.

METHODS

This retrospective study examined 475 patients (303 with known diabetes) hospitalized in critical care setting units in 2019 in a single center who received continuous EN. Rates of hypoglycemia, hyperglycemia, and glucose levels within the target range (70-180 mg/dL) were compared between patients with and without diabetes, and among patients treated with intermediate-acting (IA) biphasic neutral protamine Hagedorn 70/30, long-acting (LA) insulin, or rapid-acting insulin only.

RESULTS

Among those with type 2 diabetes mellitus, IA and LA insulin regimens were associated with a significantly higher proportion of patient-days in the target glucose range and fewer hyperglycemic days. Level 1 (<70 mg/dL) and level 2 (<54 mg/dL) hypoglycemia occurred rarely, and there were no significant differences in level 2 hypoglycemia frequency across the different insulin regimens.

CONCLUSION

Administration of IA and LA insulin can be safe and effective for those receiving insulin doses for EN-related hyperglycemia.

Observational cohort investigating health outcomes and healthcare costs after metabolic and bariatric surgery: a study protocol

INTRODUCTION

As the rate of obesity increases, so does the incidence of obesity-related comorbidities. Metabolic and bariatric surgery (MBS) is the most effective treatment for obesity, yet this treatment is severely underused. MBS can improve, resolve, and prevent the development of obesity-related comorbidities; this improvement in health also results in lower healthcare costs. The studies that have examined these outcomes are often limited by small sample sizes, reliance on outdated data, inconsistent definitions of outcomes, and the use of simulated data. Using recent real-world data, we will identify characteristics of individuals who qualify for MBS but have not had MBS and address the gaps in knowledge around the impact of MBS on health outcomes and healthcare costs.

METHODS AND ANALYSIS

Using a large US employer-based retrospective claims database (Merative), we will identify all obese adults (21+) who have had a primary MBS from 2016 to 2021 and compare their characteristics and outcomes with obese adults who did not have an MBS from 2016 to 2021. Baseline demographics, health outcomes, and costs will be examined in the year before the index date, remission and new-onset comorbidities, and healthcare costs will be examined at 1 and 3 years after the index date.

ETHICS AND DISSEMINATION

As this was an observational study of deidentified patients in the Merative database, Institutional Review Board approval and consent were exempt (in accordance with the Health Insurance Portability and Accountability Act Privacy Rule). An IRB exemption was approved by the wcg IRB (#13931684). Knowledge dissemination will include presenting results at national and international conferences, sharing findings with specialty societies, and publishing results in peer-reviewed journals. All data management and analytic code will be made available publicly to enable others to leverage our methods to verify and extend our findings.

Patterns and Trends in Continuous Glucose Monitoring Utilization Among Commercially Insured Individuals With Type 1 Diabetes: 2010-2013 to 2016-2019

Prior studies suggest that only ∼30% of patients with type 1 diabetes use continuous glucose monitoring (CGM), but most studies to date focused on children and young adults seen by endocrinologists or in academic centers. This study examined national trends in CGM utilization among commercially insured children and adults with type 1 diabetes. Overall, CGM utilization was 20.12% in 2010-2013 and 49.78% in 2016-2019, reflecting a 2.5-fold increase in utilization within a period of <10 years. Identifying populations with low CGM use is a necessary first step in developing targeted interventions to increase CGM uptake.

Type 2 Diabetes Remission After Bariatric Surgery and Its Impact on Healthcare Costs

PURPOSE

Bariatric surgery is the most effective and durable treatment of obesity and can put type 2 diabetes (T2D) into remission. We aimed to examine remission rates after bariatric surgery and the impacts of post-surgical healthcare costs.

MATERIALS AND METHODS

Obese adults with T2D were identified in Merative™ (US employer-based retrospective claims database). Individuals who had bariatric surgery were matched 1:1 with those who did not with baseline demographic and health characteristics. Rates of remission and total healthcare costs were compared at 6-12 and 6-36 months after the index date.

RESULTS

Remission rates varied substantially by baseline T2D complexity; differences in rates at 1 year ranged from 41% for those with high-complexity T2D to 66% for those with low- to mid-complexity T2D. At 3 years, those who had bariatric surgery had 56% higher remission rates than those who did not have bariatric surgery, with differences of 73%, 59%, and 35% for those with low-, mid-, and high-complexity T2D at baseline. Healthcare costs were $3401 and $20,378 lower among those who had bariatric surgery in the 6 to 12 months and 6 to 36 months after the index date, respectively, than their matched controls. The biggest cost differences were seen among those with high-complexity T2D; those who had bariatric surgery had $26,879 lower healthcare costs in the 6 to 36 months after the index date than those who did not.

CONCLUSION

Individuals with T2D undergoing bariatric surgery have substantially higher rates of T2D remission and lower healthcare costs.

Prevalence of diagnosed depression, anxiety, and ADHD among youth with type 1 or type 2 diabetes mellitus

We examined the prevalence of diagnosed depression, anxiety, and ADHD among youth by diabetes type, insurance type, and race/ethnicity. These mental disorders were more prevalent among youth with diabetes, particularly those with type 2 diabetes, with non-Hispanic White youth with Medicaid and diabetes having a higher prevalence than other races/ethnicities.

Racial and Ethnic Differences in the Association Between Mean Glucose and Hemoglobin A1c

Background: Studies have reported significantly higher hemoglobin A1c (A1C) in African American patients than in White patients with the same mean glucose, but less is known about other racial/ethnic groups. We evaluated racial/ethnic differences in the association between mean glucose, based on continuous glucose monitor (CGM) data, and A1C. Methods: Retrospective study among 1788 patients with diabetes from Kaiser Permanente Northern California (KPNC) who used CGM devices during 2016 to 2021. In this study population, there were 5264 A1C results; mean glucose was calculated from 124,388,901 CGM readings captured during the 90 days before each A1C result. Hierarchical mixed models were specified to estimate racial/ethnic differences in the association between mean glucose and A1C. Results: Mean A1C was 0.33 (95% confidence interval: 0.23-0.44; P < 0.0001) percentage points higher among African American patients relative to White patients for a given mean glucose. A1C results for Asians, Latinos, and multiethnic patients were not significantly different from those of White patients. The slope of the association between mean glucose and A1C did not differ significantly across racial/ethnic groups. Variance for the association between mean glucose and A1C was substantially greater within groups than between racial/ethnic groups (65% vs. 9%, respectively). Conclusions: For African American patients, A1C results may overestimate glycemia and could lead to premature diabetes diagnoses, overtreatment, or invalid assessments of health disparities. However, most of the variability in the mean glucose-A1C association was within racial/ethnic groups. Treatment decisions driven by guideline-based A1C targets should be individualized and supported by direct measurement of glycemia.

Algorithms to define diabetes type using data from administrative databases: A systematic review of the evidence

AIMS

To find the best-performing algorithms to distinguish type 1 and type 2 diabetes in administrative data.

METHODS

Embase and MEDLINE databases were searched from January 2000 until January 2023. Papers evaluating the performance of algorithms to define type 1 and type 2 diabetes by reporting diagnostic metrics against a range of reference standards were selected. Study quality was evaluated using the Quality Assessment of Diagnostic Accuracy Studies.

RESULTS

Of the 24 studies meeting the eligibility criteria, 19 demonstrated a low risk of bias and low concerns about the applicability of the study population across all domains. Algorithms considering multiple diabetes diagnostic codes alone were sensitive and specific approaches to classify diabetes type (both metrics >92.1% for type 1 diabetes; >86.9% for type 2 diabetes). Among the top 10-performing algorithms to detect type 1 and type 2 diabetes, 70% and 100% featured multiple criteria, respectively. Information on insulin use was more sensitive and specific for detecting diabetes type than were criteria based on use of oral hypoglycaemic agents.

CONCLUSIONS

Algorithms based on multiple diabetes diagnostic codes and insulin use are the most accurate approaches to distinguish type 1 from type 2 diabetes using administrative data. Approaches with more than one criterion may also increase sensitivity in distinguishing diabetes type.

Data-driven classification of health status of older adults with diabetes: The diabetes and aging study

BACKGROUND

We set out to identify empirically-derived health status classes of older adults with diabetes based on clusters of comorbid conditions which are associated with future complications.

METHODS

We conducted a cohort study among 105,786 older (≥65 years of age) adults with type 2 diabetes enrolled in an integrated healthcare delivery system. We used latent class analysis of 19 baseline comorbidities to derive health status classes and then compared incident complication rates (events per 100 person-years) by health status class during 5 years of follow-up. Complications included infections, hyperglycemic events, hypoglycemic events, microvascular events, cardiovascular events, and all-cause mortality.

RESULTS

Three health status classes were identified: Class 1 (58% of the cohort) had the lowest prevalence of most baseline comorbidities, Class 2 (22%) had the highest prevalence of obesity, arthritis, and depression, and Class 3 (20%) had the highest prevalence of cardiovascular conditions. The risk for incident complications was highest for Class 3, intermediate for Class 2 and lowest for Class 1. For example, the age, sex and race-adjusted rates for cardiovascular events (per 100 person-years) for Class 3, Class 2 and Class 1 were 6.5, 2.3, and 1.6, respectively; 2.1, 1.2, 0.7 for hypoglycemia; and 8.0, 3.8, and 2.3 for mortality.

CONCLUSIONS

Three health status classes of older adults with diabetes were identified based on prevalent comorbidities and were associated with marked differences in risk of complications. These health status classes can inform population health management and guide the individualization of diabetes care.

Predictors of insulin pump initiation among people with type 2 diabetes from a US claims database using machine learning

OBJECTIVE

Insulin pump use is increasing among people with type 2 diabetes (T2D), albeit at a slower rate compared to people with type 1 diabetes (T1D). Factors associated with insulin pump initiation among people with T2D in the real-world are understudied.

METHODS

This retrospective, nested case-control study aimed to identify predictors of insulin pump initiation among people with T2D in the United States (US). Adults with T2D who were new to bolus insulin use were identified from the IBM MarketScan Commercial database (2015-2020). Candidate variables of pump initiation were entered into conditional logistic regression (CLR) and penalized CLR models.

RESULTS

Of the 32,104 eligible adults with T2D, 726 insulin pump initiators were identified and matched to 2,904 non-pump initiators using incidence density sampling. Consistent predictors of insulin pump initiation across the base case, sensitivity, and post hoc analyses included continuous glucose monitor (CGM) use, visiting an endocrinologist, acute metabolic complications, higher count of HbA1c tests, lower age, and fewer diabetes-related medication classes.

CONCLUSIONS

Many of these predictors could represent a clinical indication for treatment intensification, greater patient engagement in diabetes management, or proactive management by healthcare providers. Improved understanding of predictors for pump initiation may lead to more targeted efforts to improve access and acceptance of insulin pumps among persons with T2D.

Validation of Register-Based Diabetes Classifiers in Danish Data

Purpose

To validate two register-based algorithms classifying type 1 (T1D) and type 2 diabetes (T2D) in a general population using Danish register data.

Patients and Methods

After linking data on prescription drug usage, hospital diagnoses, laboratory results and diabetes-specific healthcare services from nationwide healthcare registers, diabetes type was defined for all individuals in Central Denmark Region age 18-74 years on 31 December 2018 according to two distinct register-based classifiers: 1) a novel register-based diabetes classifier incorporating diagnostic hemoglobin-A1C measurements, the Open-Source Diabetes Classifier (OSDC), and 2) an existing Danish diabetes classifier, the Register for Selected Chronic Diseases (RSCD). These classifications were validated against self-reported data from the Health in Central Denmark survey - overall and stratified by age at onset of diabetes. The source-code of both classifiers was made available in the open-source R package osdc.

Results

A total of 2633 (9.0%) of 29,391 respondents reported having any type of diabetes, divided across 410 (1.4%) self-reported cases of T1D and 2223 (7.6%) cases of T2D. Among all self-reported diabetes cases, 2421 (91.9%) were classified as diabetes cases by both classifiers. In T1D, sensitivity of OSDC-classification was 0.773 [95% CI 0.730-0.813] (RSCD: 0.700 [0.653-0.744]) and positive predictive value (PPV) 0.943 [0.913-0.966] (RSCD: 0.944 [0.912-0.967]). In T2D, sensitivity of OSDC-classification was 0.944 [0.933-0.953] (RSCD: 0.905 [0.892-0.917]) and PPV 0.875 [0.861-0.888] (RSCD: 0.898 [0.884-0.910]). In age at onset-stratified analyses of both classifiers, sensitivity and PPV were low in individuals with T1D onset after age 40 and T2D onset before age 40.

Conclusion

Both register-based classifiers identified valid populations of T1D and T2D in a general population, but sensitivity was substantially higher in OSDC compared to RSCD. Register-classified diabetes type in cases with atypical age at onset of diabetes should be interpreted with caution. The validated, open-source classifiers provide robust and transparent tools for researchers.

Polygenic Scores Help Reduce Racial Disparities in Predictive Accuracy of Automated Type 1 Diabetes Classification Algorithms

OBJECTIVE

Automated algorithms to identify individuals with type 1 diabetes using electronic health records are increasingly used in biomedical research. It is not known whether the accuracy of these algorithms differs by self-reported race. We investigated whether polygenic scores improve identification of individuals with type 1 diabetes.

RESEARCH DESIGN AND METHODS

We investigated two large hospital-based biobanks (Mass General Brigham [MGB] and BioMe) and identified individuals with type 1 diabetes using an established automated algorithm. We performed medical record reviews to validate the diagnosis of type 1 diabetes. We implemented two published polygenic scores for type 1 diabetes (developed in individuals of European or African ancestry). We assessed the classification algorithm before and after incorporating polygenic scores.

RESULTS

The automated algorithm was more likely to incorrectly assign a diagnosis of type 1 diabetes in self-reported non-White individuals than in self-reported White individuals (odds ratio 3.45; 95% CI 1.54-7.69; P = 0.0026). After incorporating polygenic scores into the MGB Biobank, the positive predictive value of the type 1 diabetes algorithm increased from 70 to 97% for self-reported White individuals (meaning that 97% of those predicted to have type 1 diabetes indeed had type 1 diabetes) and from 53 to 100% for self-reported non-White individuals. Similar results were found in BioMe.

CONCLUSIONS

Automated phenotyping algorithms may exacerbate health disparities because of an increased risk of misclassification of individuals from underrepresented populations. Polygenic scores may be used to improve the performance of phenotyping algorithms and potentially reduce this disparity.

Associations between alcohol brief intervention in primary care and drinking and health outcomes in adults with hypertension and type 2 diabetes: a population-based observational study

OBJECTIVES

To evaluate associations between alcohol brief intervention (BI) in primary care and 12-month drinking outcomes and 18-month health outcomes among adults with hypertension and type 2 diabetes (T2D).

DESIGN

A population-based observational study using electronic health records data.

SETTING

An integrated healthcare system that implemented system-wide alcohol screening, BI and referral to treatment in adult primary care.

PARTICIPANTS

Adult primary care patients with hypertension (N=72 979) or T2D (N=19 642) who screened positive for unhealthy alcohol use between 2014 and 2017.

MAIN OUTCOME MEASURES

We examined four drinking outcomes: changes in heavy drinking days/past 3 months, drinking days/week, drinks/drinking day and drinks/week from baseline to 12-month follow-up, based on results of alcohol screens conducted in routine care. Health outcome measures were changes in measured systolic and diastolic blood pressure (BP) and BP reduction ≥3 mm Hg at 18-month follow-up. For patients with T2D, we also examined change in glycohaemoglobin (HbA1c) level and 'controlled HbA1c' (HbA1c<8%) at 18-month follow-up.

RESULTS

For patients with hypertension, those who received BI had a modest but significant additional -0.06 reduction in drinks/drinking day (95% CI -0.11 to -0.01) and additional -0.30 reduction in drinks/week (95% CI -0.59 to -0.01) at 12 months, compared with those who did not. Patients with hypertension who received BI also had higher odds for having clinically meaningful reduction of diastolic BP at 18 months (OR 1.05, 95% CI 1.00 to 1.09). Among patients with T2D, no significant associations were found between BI and drinking or health outcomes examined.

CONCLUSIONS

Alcohol BI holds promise for reducing drinking and helping to improve health outcomes among patients with hypertension who screened positive for unhealthy drinking. However, similar associations were not observed among patients with T2D. More research is needed to understand the heterogeneity across diverse subpopulations and to study BI's long-term public health impact.

Identifying type 1 and 2 diabetes in research datasets where classification biomarkers are unavailable: assessing the accuracy of published approaches

OBJECTIVES

We aimed to compare the performance of approaches for classifying insulin-treated diabetes within research datasets without measured classification biomarkers, evaluated against two independent biological definitions of diabetes type.

STUDY DESIGN AND SETTING

We compared accuracy of ten reported approaches for classifying insulin-treated diabetes into type 1 (T1D) and type 2 (T2D) diabetes in two cohorts: UK Biobank (UKBB) n = 26,399 and Diabetes Alliance for Research in England (DARE) n = 1,296. The overall performance for classifying T1D and T2D was assessed using: a T1D genetic risk score and genetic stratification method (UKBB); C-peptide measured at >3 years diabetes duration (DARE).

RESULTS

Approaches' accuracy ranged from 71% to 88% (UKBB) and 68% to 88% (DARE). When classifying all participants, combining early insulin requirement with a T1D probability model (incorporating diagnosis age and body image issue [BMI]), and interview-reported diabetes type (UKBB available in only 15%) consistently achieved high accuracy (UKBB 87% and 87% and DARE 85% and 88%, respectively). For identifying T1D with minimal misclassification, models with high thresholds or young diagnosis age (<20 years) had highest performance. Findings were incorporated into an online tool identifying optimum approaches based on variable availability.

CONCLUSION

Models combining continuous features with early insulin requirement are the most accurate methods for classifying insulin-treated diabetes in research datasets without measured classification biomarkers.

Algorithmic identification of atypical diabetes in electronic health record (EHR) systems

AIMS

Understanding atypical forms of diabetes (AD) may advance precision medicine, but methods to identify such patients are needed. We propose an electronic health record (EHR)-based algorithmic approach to identify patients who may have AD, specifically those with insulin-sufficient, non-metabolic diabetes, in order to improve feasibility of identifying these patients through detailed chart review.

METHODS

Patients with likely T2D were selected using a validated machine-learning (ML) algorithm applied to EHR data. "Typical" T2D cases were removed by excluding individuals with obesity, evidence of dyslipidemia, antibody-positive diabetes, or cystic fibrosis. To filter out likely type 1 diabetes (T1D) cases, we applied six additional "branch algorithms," relying on various clinical characteristics, which resulted in six overlapping cohorts. Diabetes type was classified by manual chart review as atypical, not atypical, or indeterminate due to missing information.

RESULTS

Of 114,975 biobank participants, the algorithms collectively identified 119 (0.1%) potential AD cases, of which 16 (0.014%) were confirmed after expert review. The branch algorithm that excluded T1D based on outpatient insulin use had the highest percentage yield of AD (13 of 27; 48.2% yield). Together, the 16 AD cases had significantly lower BMI and higher HDL than either unselected T1D or T2D cases identified by ML algorithms (P<0.05). Compared to the ML T1D group, the AD group had a significantly higher T2D polygenic score (P<0.01) and lower hemoglobin A1c (P<0.01).

CONCLUSION

Our EHR-based algorithms followed by manual chart review identified collectively 16 individuals with AD, representing 0.22% of biobank enrollees with T2D. With a maximum yield of 48% cases after manual chart review, our algorithms have the potential to drastically improve efficiency of AD identification. Recognizing patients with AD may inform on the heterogeneity of T2D and facilitate enrollment in studies like the Rare and Atypical Diabetes Network (RADIANT).

Productivity Loss and Medical Costs Associated With Type 2 Diabetes Among Employees Aged 18-64 Years With Large Employer-Sponsored Insurance

OBJECTIVE

To estimate productivity losses and costs and medical costs due to type 2 diabetes (T2D) among employees aged 18-64 years.

RESEARCH DESIGN AND METHODS

Using 2018-2019 MarketScan databases, we identified employees with T2D or no diabetes among those with records on workplace absences, short-term disability (STD), and long-term disability (LTD). We estimated per capita mean annual time loss attributable to T2D and its associated costs, calculated by multiplying time loss by average hourly wage. We estimated direct medical costs of T2D in total and by service type (inpatient, outpatient, and prescription drugs). We used two-part models (productivity losses and costs and inpatient and drug costs) and generalized linear models (total and outpatient costs) for overall and subgroup analyses by age and sex. All costs were in 2019 U.S. dollars.

RESULTS

Employees with T2D had 4.2 excess days lost (20.8 vs. 20.3 absences, 6.4 vs. 3.3 STD days, and 1.0 vs. 0.4 LTD days) than those without diabetes. Productivity costs were 13.3% ($680) higher and medical costs were double (total $11,354 vs. $5,101; outpatient $4,558 vs. $2,687, inpatient $3,085 vs. $1,349, prescription drugs $4,182 vs. $1,189) for employees with T2D. Employees aged 18-34 years had higher STD days and outpatient costs. Women had more absences and STD days and higher outpatient costs than men.

CONCLUSIONS

T2D contributes nearly $7,000 higher annual per capita costs, mostly due to excess medical costs. Our estimates may assist employers to assess potential financial gains from efforts to help workers prevent or better manage T2D.

Association Between Real-Time Continuous Glucose Monitor Use and Diabetes-Related Medical Costs for Patients with Type 2 Diabetes

Little is known about the impact of real-time continuous glucose monitoring (rtCGM) on diabetes-related medical costs within the type 2 diabetes (T2D) population. A retrospective analysis of administrative claims data from the Optum Research Database was conducted. Changes in diabetes-related health care resource utilization costs were expressed as per-patient-per-month (PPPM) costs. A total of 571 T2D patients (90% insulin treated) met study inclusion criteria. Average PPPM for diabetes-related medical costs decreased by -$424 (95% confidence interval [CI] -$816 to -$31, P = 0.035) after initiating rtCGM. These reductions were driven, in part, by reductions in diabetes-related inpatient medical costs: -$358 (95% CI -$706 to -$10, P = 0.044). Inpatient hospital admissions were reduced on average -0.006 PPPM (P = 0.057) and total hospital days were reduced an average of -0.042 PPPM (P = 0.139). These findings provide real-world evidence that rtCGM use was associated with diabetes-related health care resource utilization cost reductions in patients with T2D.

Mortality trends in type 1 diabetes: a multicountry analysis of six population-based cohorts

AIMS/HYPOTHESIS

Mortality has declined in people with type 1 diabetes in recent decades. We examined how the pattern of decline differs by country, age and sex, and how mortality trends in type 1 diabetes relate to trends in general population mortality.

METHODS

We assembled aggregate data on all-cause mortality during the period 2000-2016 in people with type 1 diabetes aged 0-79 years from Australia, Denmark, Latvia, Scotland, Spain (Catalonia) and the USA (Kaiser Permanente Northwest). Data were obtained from administrative sources, health insurance records and registries. All-cause mortality rates in people with type 1 diabetes, and standardised mortality ratios (SMRs) comparing type 1 diabetes with the non-diabetic population, were modelled using Poisson regression, with age and calendar time as quantitative variables, describing the effects using restricted cubic splines with six knots for age and calendar time. Mortality rates were standardised to the age distribution of the aggregate population with type 1 diabetes.

RESULTS

All six data sources showed a decline in age- and sex-standardised all-cause mortality rates in people with type 1 diabetes from 2000 to 2016 (or a subset thereof), with annual changes in mortality rates ranging from -2.1% (95% CI -2.8%, -1.3%) to -5.8% (95% CI -6.5%, -5.1%). All-cause mortality was higher for male individuals and for older individuals, but the rate of decline in mortality was generally unaffected by sex or age. SMR was higher in female individuals than male individuals, and appeared to peak at ages 40-70 years. SMR declined over time in Denmark, Scotland and Spain, while remaining stable in the other three data sources.

CONCLUSIONS/INTERPRETATION

All-cause mortality in people with type 1 diabetes has declined in recent years in most included populations, but improvements in mortality relative to the non-diabetic population are less consistent.

Obesity and Chronic Kidney Disease in US Adults With Type 1 and Type 2 Diabetes Mellitus

OBJECTIVE

Obesity is a global public health challenge and strongly associated with type 2 diabetes (T2D), but its burden and effects are not well understood in people with type 1 diabetes (T1D). Particularly, the link between obesity and chronic kidney disease (CKD) in T1D is poorly characterized.

RESEARCH DESIGN AND METHODS

We included all T1D and, for comparison, T2D in the Geisinger Health System from 2004 to 2018. We evaluated trends in obesity (body mass index ≥ 30 kg/m2), low estimated glomerular filtration rate (eGFR) (≤60 mL/min/1.73m2), and albuminuria (urine albumin-to-creatinine ratio ≥ 30 mg/g). We used multivariable logistic regression to evaluate the independent association of obesity with CKD in 2018.

RESULTS

People with T1D were younger than T2D (median age 39 vs 62 years). Obesity increased in T1D over time (32.6% in 2004 to 36.8% in 2018), while obesity in T2D was stable at ~60%. The crude prevalence of low eGFR was higher in T2D than in T1D in all years (eg, 30.6% vs 16.1% in 2018), but after adjusting for age differences, prevalence was higher in T1D than T2D in all years (eg, 16.2% vs 9.3% in 2018). Obesity was associated with increased odds of low eGFR in T1D [adjusted odds ratio (AOR) = 1.52, 95% CI 1.12-2.08] and T2D (AOR = 1.29, 95% CI 1.23-1.35).

CONCLUSIONS

Obesity is increasing in people with T1D and is associated with increased risk of CKD. After accounting for age, the burden of CKD in T1D exceeded the burden in T2D, suggesting the need for increased vigilance and assessment of kidney-protective medications in T1D.

All in the Family: A Qualitative Study of the Early Experiences of Adults with Younger Onset Type 2 Diabetes

OBJECTIVE

Adults with type 2 diabetes diagnosed at a younger age are at increased risk for poor outcomes. We examined life stage-related facilitators and barriers to early self-management among younger adults with newly diagnosed type 2 diabetes.

RESEARCH DESIGN AND METHODS

We conducted 6 focus groups that each met twice between November 2017 and May 2018. Participants (n = 41) were aged 21 to 44 years and diagnosed with type 2 diabetes during the prior 2 years. Transcripts were coded using thematic analysis and themes were mapped to the Capability-Opportunity-Motivation-Behavior framework.

RESULTS

Participants were 38.4 (±5.8) years old; 10 self-identified as Latinx, 12 as Black, 12 as White, and 7 as multiple or other races. We identified 9 themes that fell into 2 categories: (1) the impact of having an adult family member with diabetes, and (2) the role of nonadult children. Family members with diabetes served as both positive and negative role models, and, for some, personal familiarity with the disease made adjusting to the diagnosis easier. Children facilitated their parents' self-management by supporting self-management activities and motivating their parents to remain healthy. However, the stress and time demands resulting from parental responsibilities and the tendency to prioritize children's needs were perceived as barriers to self-management.

CONCLUSIONS

Our results highlight how the life position of younger-onset individuals with type 2 diabetes influences their early experiences. Proactively addressing perceived barriers to and facilitators of self-management in the context of family history and parenthood may aid in efforts to support these high-risk, younger patients.

Comparison of cognitive function in older adults with type 1 diabetes, type 2 diabetes, and no diabetes: results from the Study of Longevity in Diabetes (SOLID)

INTRODUCTION

The incidence of both type 1 diabetes (T1D) and type 2 diabetes (T2D) is increasing. Life expectancy is improving in T1D, resulting in a growing population of elderly adults with diabetes. While it is well established that older adults with T2D are at increased risk of cognitive impairment, little is known regarding cognitive aging in T1D and how their cognitive profiles may differ from T2D.

RESEARCH DESIGN AND METHODS

We compared baseline cognitive function and low cognitive function by diabetes status (n=734 T1D, n=232 T2D, n=247 without diabetes) among individuals from the Study of Longevity in Diabetes (mean age=68). We used factor analysis to group cognition into five domains and a composite measure of total cognition. Using linear and logistic regression models, we examined the associations between diabetes type and cognitive function, adjusting for demographics, comorbidities, depression, and sleep quality.

RESULTS

T1D was associated with lower scores on total cognition, language, executive function/psychomotor processing speed, and verbal episodic memory, and greater odds of low executive function/psychomotor processing speed (OR=2.99, 95% CI 1.66 to 5.37) and verbal episodic memory (OR=1.92, 95% CI 1.07 to 3.46), compared with those without diabetes. T2D was associated with lower scores on visual episodic memory. Compared with T2D, T1D was associated with lower scores on verbal episodic memory and executive function/psychomotor processing speed and greater odds of low executive function/psychomotor processing speed (OR=1.74, 95% CI 1.03 to 2.92).

CONCLUSIONS

Older adults with T1D had significantly poorer cognition compared with those with T2D and those without diabetes even after accounting for a range of comorbidities. Future studies should delineate how to reduce risk in this vulnerable population who are newly surviving to old age.

Using a Constraint-Based Method to Identify Chronic Disease Patients Who Are Apt to Obtain Care Mostly Within a Given Health Care System: Retrospective Cohort Study

BACKGROUND

For several major chronic diseases including asthma, chronic obstructive pulmonary disease, chronic kidney disease, and diabetes, a state-of-the-art method to avert poor outcomes is to use predictive models to identify future high-cost patients for preemptive care management interventions. Frequently, an American patient obtains care from multiple health care systems, each managed by a distinct institution. As the patient's medical data are spread across these health care systems, none has complete medical data for the patient. The task of building models to predict an individual patient's cost is currently thought to be impractical with incomplete data, which limits the use of care management to improve outcomes. Recently, we developed a constraint-based method to identify patients who are apt to obtain care mostly within a given health care system. Our method was shown to work well for the cohort of all adult patients at the University of Washington Medicine for a 6-month follow-up period. It is unknown how well our method works for patients with various chronic diseases and over follow-up periods of different lengths, and subsequently, whether it is reasonable to perform this predictive modeling task on the subset of patients pinpointed by our method.

OBJECTIVE

To understand our method's potential to enable this predictive modeling task on incomplete medical data, this study assesses our method's performance at the University of Washington Medicine on 5 subgroups of adult patients with major chronic diseases and over follow-up periods of 2 different lengths.

METHODS

We used University of Washington Medicine data for all adult patients who obtained care at the University of Washington Medicine in 2018 and PreManage data containing usage information from all hospitals in Washington state in 2019. We evaluated our method's performance over the follow-up periods of 6 months and 12 months on 5 patient subgroups separately-asthma, chronic kidney disease, type 1 diabetes, type 2 diabetes, and chronic obstructive pulmonary disease.

RESULTS

Our method identified 21.81% (3194/14,644) of University of Washington Medicine adult patients with asthma. Around 66.75% (797/1194) and 67.13% (1997/2975) of their emergency department visits and inpatient stays took place within the University of Washington Medicine system in the subsequent 6 months and in the subsequent 12 months, respectively, approximately double the corresponding percentage for all University of Washington Medicine adult patients with asthma. The performance for adult patients with chronic kidney disease, adult patients with chronic obstructive pulmonary disease, adult patients with type 1 diabetes, and adult patients with type 2 diabetes was reasonably similar to that for adult patients with asthma.

CONCLUSIONS

For each of the 5 chronic diseases most relevant to care management, our method can pinpoint a reasonably large subset of patients who are apt to obtain care mostly within the University of Washington Medicine system. This opens the door to building models to predict an individual patient's cost on incomplete data, which was formerly deemed impractical.

INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID)

RR2-10.2196/13783.

Real-world treatment escalation from metformin monotherapy in youth-onset Type 2 diabetes mellitus: A retrospective cohort study

BACKGROUND

Due to high rates of comorbidities and rapid progression, youth with Type 2 diabetes may benefit from early and aggressive treatment. However, until 2019, the only approved medications for this population were metformin and insulin.

OBJECTIVE

To investigate patterns and predictors of treatment escalation within 5 years of metformin monotherapy initiation for youth with Type 2 diabetes in clinical practice.

SUBJECTS

Commercially-insured patients with incident youth-onset (10-18 years) Type 2 diabetes initially treated with metformin only.

METHODS

Retrospective cohort study using a patient-level medical claims database with data from 2000 to 2020. Frequency and order of treatment escalation to insulin and non-insulin antihyperglycemics were determined and categorized by age at diagnosis. Cox proportional hazards regression was used to evaluate potential predictors of treatment escalation, including age, sex, race/ethnicity, comorbidities, complications, and metformin adherence (medication possession ratio ≥ 0.8).

RESULTS

The cohort included 829 (66% female; median age at diagnosis 15 years; 19% Hispanic, 17% Black) patients, with median 2.9 year follow-up after metformin initiation. One-quarter underwent treatment escalation (n = 207; 88 to insulin, 164 to non-insulin antihyperglycemic). Younger patients were more likely to have insulin prescribed prior to other antihyperglycemics. Age at diagnosis (HR 1.14, 95% CI 1.07-1.21), medication adherence (HR 4.10, 95% CI 2.96-5.67), Hispanic ethnicity (HR 1.83, 95% CI 1.28-2.61), and diabetes-related complications (HR 1.78, 95% CI 1.15-2.74) were positively associated with treatment escalation.

CONCLUSIONS

In clinical practice, treatment escalation for pediatric Type 2 diabetes differs with age. Off-label use of non-insulin antihyperglycemics occurs, most commonly among older adolescents.

Intensive Care Unit Admission, Mechanical Ventilation, and Mortality Among Patients With Type 1 Diabetes Hospitalized for COVID-19 in the U.S

OBJECTIVE

To assess whether risk of severe outcomes among patients with type 1 diabetes mellitus (T1DM) hospitalized for coronavirus disease 2019 (COVID-19) differs from that of patients without diabetes or with type 2 diabetes mellitus (T2DM).

RESEARCH DESIGN AND METHODS

Using the Premier Healthcare Database Special COVID-19 Release records of patients discharged after COVID-19 hospitalization from U.S. hospitals from March to November 2020 (N = 269,674 after exclusion), we estimated risk differences (RD) and risk ratios (RR) of intensive care unit admission or invasive mechanical ventilation (ICU/MV) and of death among patients with T1DM compared with patients without diabetes or with T2DM. Logistic models were adjusted for age, sex, and race or ethnicity. Models adjusted for additional demographic and clinical characteristics were used to examine whether other factors account for the associations between T1DM and severe COVID-19 outcomes.

RESULTS

Compared with patients without diabetes, T1DM was associated with a 21% higher absolute risk of ICU/MV (RD 0.21, 95% CI 0.19-0.24; RR 1.49, 95% CI 1.43-1.56) and a 5% higher absolute risk of mortality (RD 0.05, 95% CI 0.03-0.07; RR 1.40, 95% CI 1.24-1.57), with adjustment for age, sex, and race or ethnicity. Compared with T2DM, T1DM was associated with a 9% higher absolute risk of ICU/MV (RD 0.09, 95% CI 0.07-0.12; RR 1.17, 95% CI 1.12-1.22), but no difference in mortality (RD 0.00, 95% CI -0.02 to 0.02; RR 1.00, 95% CI 0.89-1.13). After adjustment for diabetic ketoacidosis (DKA) occurring before or at COVID-19 diagnosis, patients with T1DM no longer had increased risk of ICU/MV (RD 0.01, 95% CI -0.01 to 0.03) and had lower mortality (RD -0.03, 95% CI -0.05 to -0.01) in comparisons with patients with T2DM.

CONCLUSIONS

Patients with T1DM hospitalized for COVID-19 are at higher risk for severe outcomes than those without diabetes. Higher risk of ICU/MV in patients with T1DM than in patients with T2DM was largely accounted for by the presence of DKA. These findings might further guide recommendations related to diabetes management and the prevention of COVID-19.

Incidence of diabetic ketoacidosis and its trends in patients with type 1 diabetes mellitus identified using a U.S. claims database, 2007-2019

Diabetic ketoacidosis (DKA) is a common complication of type 1 diabetes mellitus (T1DM). We found that the incidence of DKA was 55.5 per 1000 person-years in US commercially insured patients with T1DM; age-sex-standardized incidence decreased at an average annual rate of 6.1% in 2018-2019 after a steady increase since 2011.

Disclosure of New Type 2 Diabetes Diagnoses to Younger Adults: a Qualitative Study

BACKGROUND

Adults diagnosed with type 2 diabetes at a younger age are at increased risk for poor outcomes. Yet, little is known about the early experiences of these individuals, starting with communication of the diagnosis. Addressing this knowledge gap is important as this initial interaction may shape subsequent disease-related perceptions and self-management.

OBJECTIVE

We examined diagnosis disclosure experiences and initial reactions among younger adults with newly diagnosed type 2 diabetes.

PARTICIPANTS

Purposive sample of adult members of Kaiser Permanente Northern California, an integrated healthcare delivery system, diagnosed with type 2 diabetes before age 45 years.

APPROACH

We conducted six focus groups between November 2017 and May 2018. Transcribed audio recordings were coded by two coders using thematic analysis.

KEY RESULTS

Participants (n = 41) were 38.4 (± 5.8) years of age; 10 self-identified as Latinx, 12 as Black, 12 as White, and 7 as multiple or other races. We identified variation in diagnosis disclosure experiences, centered on four key domains: (1) participants' sense of preparedness for diagnosis (ranging from expectant to surprised); (2) disclosure setting (including in-person, via phone, via secure message, or via review of results online); (3) perceived provider tone (from nonchalant, to overly fear-centered, to supportive); and (4) participants' emotional reactions to receiving the diagnosis (including acceptance, denial, guilt, and/or fear, rooted in personal and family experience).

CONCLUSIONS

For younger adults, the experience of receiving a diabetes diagnosis varies greatly. Given the long-term consequences of inadequately managed diabetes and the need for early disease control, effective initial disclosure represents an opportunity to optimize initial care. Our results suggest several opportunities to improve the type 2 diabetes disclosure experience: (1) providing pre-test counseling, (2) identifying patient-preferred settings for receiving the news, and (3) developing initial care strategies that acknowledge and address the emotional distress triggered by this life-altering, chronic disease diagnosis.

Performance assessment of different machine learning approaches in predicting diabetic ketoacidosis in adults with type 1 diabetes using electronic health records data

Purpose

To assess the performance of different machine learning (ML) approaches in identifying risk factors for diabetic ketoacidosis (DKA) and predicting DKA.

Methods

This study applied flexible ML (XGBoost, distributed random forest [DRF] and feedforward network) and conventional ML approaches (logistic regression and least absolute shrinkage and selection operator [LASSO]) to 3400 DKA cases and 11 780 controls nested in adults with type 1 diabetes identified from Optum® de‐identified Electronic Health Record dataset (2007–2018). Area under the curve (AUC), accuracy, sensitivity and specificity were computed using fivefold cross validation, and their 95% confidence intervals (CI) were established using 1000 bootstrap samples. The importance of predictors was compared across these models.

Results

In the training set, XGBoost and feedforward network yielded higher AUC values (0.89 and 0.86, respectively) than logistic regression (0.83), LASSO (0.83) and DRF (0.81). However, the AUC values were similar (0.82) among these approaches in the test set (95% CI range, 0.80–0.84). While the accuracy values >0.8 and the specificity values >0.9 for all models, the sensitivity values were only 0.4. The differences in these metrics across these models were minimal in the test set. All approaches selected some known risk factors for DKA as the top 10 features. XGBoost and DRF included more laboratory measurements or vital signs compared with conventional ML approaches, while feedforward network included more social demographics.

Conclusions

In our empirical study, all ML approaches demonstrated similar performance, and identified overlapping, but different, top 10 predictors. The difference in selected top predictors needs further research.

Electronic Medical Record-Based Case Phenotyping for the Charlson Conditions: Scoping Review

Background

Electronic medical records (EMRs) contain large amounts of rich clinical information. Developing EMR-based case definitions, also known as EMR phenotyping, is an active area of research that has implications for epidemiology, clinical care, and health services research.

Objective

This review aims to describe and assess the present landscape of EMR-based case phenotyping for the Charlson conditions.

Methods

A scoping review of EMR-based algorithms for defining the Charlson comorbidity index conditions was completed. This study covered articles published between January 2000 and April 2020, both inclusive. Embase (Excerpta Medica database) and MEDLINE (Medical Literature Analysis and Retrieval System Online) were searched using keywords developed in the following 3 domains: terms related to EMR, terms related to case finding, and disease-specific terms. The manuscript follows the Preferred Reporting Items for Systematic reviews and Meta-analyses extension for Scoping Reviews (PRISMA) guidelines.

Results

A total of 274 articles representing 299 algorithms were assessed and summarized. Most studies were undertaken in the United States (181/299, 60.5%), followed by the United Kingdom (42/299, 14.0%) and Canada (15/299, 5.0%). These algorithms were mostly developed either in primary care (103/299, 34.4%) or inpatient (168/299, 56.2%) settings. Diabetes, congestive heart failure, myocardial infarction, and rheumatology had the highest number of developed algorithms. Data-driven and clinical rule–based approaches have been identified. EMR-based phenotype and algorithm development reflect the data access allowed by respective health systems, and algorithms vary in their performance.

Conclusions

Recognizing similarities and differences in health systems, data collection strategies, extraction, data release protocols, and existing clinical pathways is critical to algorithm development strategies. Several strategies to assist with phenotype-based case definitions have been proposed.

Development and evaluation of a patient-centered quality indicator for the appropriateness of type 2 diabetes management

INTRODUCTION

Current diabetes quality measures are agnostic to patient clinical complexity and type of treatment required to achieve it. Our objective was to introduce a patient-centered indicator of appropriate diabetes therapy indicator (ADTI), designed for patients with type 2 diabetes, which is based on hemoglobin A1c (HbA1c) but is also contextualized by patient complexity and treatment intensity.

RESEARCH DESIGN AND METHODS

A draft indicator was iteratively refined by a multidisciplinary Delphi panel using existing quality measures, guidelines, and published literature. ADTI performance was then assessed using OptumLabs Data Warehouse data for 2015. Included adults (n=206 279) with type 2 diabetes were categorized as clinically complex based on comorbidities, then categorized as treated appropriately, overtreated, or undertreated based on a matrix of clinical complexity, HbA1c level, and medications used. Associations between ADTI and emergency department/hospital visits for hypoglycemia and hyperglycemia were assessed by calculating event rates for each treatment intensity subset.

RESULTS

Overall, 7.4% of patients with type 2 diabetes were overtreated and 21.1% were undertreated. Patients with high complexity were more likely to be overtreated (OR 5.60, 95% CI 5.37 to 5.83) and less likely to be undertreated (OR 0.65, 95% CI 0.62 to 0.68) than patients with low complexity. Overtreated patients had higher rates of hypoglycemia than appropriately treated patients (22.0 vs 6.2 per 1000 people/year), whereas undertreated patients had higher rates of hyperglycemia (8.4 vs 1.9 per 1000 people/year).

CONCLUSIONS

The ADTI may facilitate timely, patient-centered treatment intensification/deintensification with the goal of achieving safer evidence-based care.

Incidence and seasonality of type 1 diabetes: a population-based 3-year cohort study using the National Database in Japan

INTRODUCTION

To investigate the incidence of type 1 diabetes by age group (0-19, 20-39, 40-59, ≥60 years) in Japan and whether there is seasonality in this incidence.

RESEARCH DESIGN AND METHODS

The incidence of type 1 diabetes from September 2014 to August 2017 was estimated using 2013-2018 data from the National Database of Health Insurance Claims and Specific Health Check-ups of Japan. The incidence rate was analyzed using Tango's Index and the self-controlled case series method.

RESULTS

Overall, 10 400 of the 79 175 553 included individuals were diagnosed with type 1 diabetes. The incidence of type 1 diabetes from September 2014 to August 2017 was 4.42/100 000 person-years. The incidence rates for men aged 0-19, 20-39, 40-59, and ≥60 years were 3.94, 5.57, 5.70, and 3.48, respectively. Among women, the incidence rates for the same age ranges were 5.22, 4.83, 4.99, and 3.31, respectively. Tango's index showed that the incidence rate of type 1 diabetes was significantly associated with seasons among those aged 0-19 years. Further, the self-controlled case series method showed a significant interaction between age and season, with the incidence of type 1 diabetes being higher in spring for patients younger than 20 years of age.

CONCLUSIONS

In Japan, men aged 40-59 years and women aged 0-19 years are the groups with the highest incidence of type 1 diabetes. Further, the incidence of younger-onset diabetes in Japan was higher in spring (from March to May).

Burden of Cardiovascular Disease in Adult Patients with Type 1 Diabetes in the US

BACKGROUND AND OBJECTIVES

The burden imposed by cardiovascular disease (CVD) on patients with type 1 diabetes (T1D) in the US has not been thoroughly addressed. In a retrospective observational analysis of the Optum^® Clinformatics™ Data Mart database, the prevalence of CVD and cardiovascular risk factors (CVRF) as well as health economic outcomes were evaluated in adults with T1D.

METHODS

Patients with at least one T1D medical claim between January 1, 2016, and December 31, 2016, were divided into cohorts based on the presence of CVD and/or CVRF. Descriptive and multivariate analyses enabled comparisons of healthcare resource utilization and costs between the cohorts.

RESULTS

The analysis included 12,687 patients: CVD, 2871; CVRF, 5371; and no CVD/CVRF, 4445. The period prevalence of CVD and CVRF in the combined baseline and follow-up periods was 27% and 44%, respectively. Fewer patients in the no-CVD/CVRF cohort had a claim of a diabetes-related inpatient admission compared with the CVD cohort (8% vs. 26%, respectively; P < 0.001, standardized mean difference [SMD] > 0.1). Likewise, fewer patients with no CVD/CVRF visited the emergency department vs. those with CVRF or CVD (diabetes-related: 4% vs. 7% and 18%, respectively; P < 0.001, SMD > 0.1). Higher overall costs were observed for the CVD and CVRF vs. the no-CVD/CVRF cohort ($30,241 and $16,220, respectively, vs. $11,761; P < 0.05 and SMD ≥ 0.1 for both).

CONCLUSIONS

Cardiovascular comorbidities are common among US adults with T1D. Considering their significant economic burden, optimal management is of the utmost importance to improve patient outcomes and reduce healthcare costs.

Predictors of Hyperkalemia and Hypokalemia in Individuals with Diabetes: a Classification and Regression Tree Analysis

BACKGROUND

Both hyperkalemia and hypokalemia can lead to cardiac arrhythmias and are associated with increased mortality. Information on the predictors of potassium in individuals with diabetes in routine clinical practice is lacking.

OBJECTIVE

To identify predictors of hyperkalemia and hypokalemia in adults with diabetes.

DESIGN

Retrospective cohort study, with classification and regression tree (CART) analysis.

PARTICIPANTS

321,856 individuals with diabetes enrolled in four large integrated health care systems from 2012 to 2013.

MAIN MEASURES

We used a single serum potassium result collected in 2012 or 2013. Hyperkalemia was defined as a serum potassium ≥ 5.5 mEq/L and hypokalemia as < 3.5 mEq/L. Predictors included demographic factors, laboratory measurements, comorbidities, medication use, and health care utilization.

KEY RESULTS

There were 2556 hypokalemia events (0.8%) and 1517 hyperkalemia events (0.5%). In univariate analyses, we identified concordant predictors (associated with increased probability of both hyperkalemia and hypokalemia), discordant predictors, and predictors of only hyperkalemia or hypokalemia. In CART models, the hyperkalemia "tree" had 5 nodes and a c-statistic of 0.76. The nodes were defined by prior potassium results and eGFRs, and the 5 terminal "leaves" had hyperkalemia probabilities of 0.2 to 7.2%. The hypokalemia tree had 4 nodes and a c-statistic of 0.76. The hypokalemia tree included nodes defined by prior potassium results, and the 4 terminal leaves had hypokalemia probabilities of 0.3 to 17.6%. Individuals with a recent potassium between 4.0 and 5.0 mEq/L, eGFR ≥ 45 mL/min/1.73m², and no hypokalemia in the previous year had a < 1% rate of either hypokalemia or hyperkalemia.

CONCLUSIONS

The yield of routine serum potassium testing may be low in individuals with a recent serum potassium between 4.0 and 5.0 mEq/L, eGFR ≥ 45 mL/min/1.73m², and no recent history of hypokalemia. We did not examine the effect of recent changes in clinical condition or medications on acute potassium changes.

Validation of a type 1 diabetes algorithm using electronic medical records and administrative healthcare data to study the population incidence and prevalence of type 1 diabetes in Ontario, Canada

INTRODUCTION

We aimed to develop algorithms distinguishing type 1 diabetes (T1D) from type 2 diabetes in adults ≥18 years old using primary care electronic medical record (EMRPC) and administrative healthcare data from Ontario, Canada, and to estimate T1D prevalence and incidence.

RESEARCH DESIGN AND METHODS

The reference population was a random sample of patients with diabetes in EMRPC whose charts were manually abstracted (n=5402). Algorithms were developed using classification trees, random forests, and rule-based methods, using electronic medical record (EMR) data, administrative data, or both. Algorithm performance was assessed in EMRPC. Administrative data algorithms were additionally evaluated using a diabetes clinic registry with endocrinologist-assigned diabetes type (n=29 371). Three algorithms were applied to the Ontario population to evaluate the minimum, moderate and maximum estimates of T1D prevalence and incidence rates between 2010 and 2017, and trends were analyzed using negative binomial regressions.

RESULTS

Of 5402 individuals with diabetes in EMRPC, 195 had T1D. Sensitivity, specificity, positive predictive value and negative predictive value for the best performing algorithms were 80.6% (75.9-87.2), 99.8% (99.7-100), 94.9% (92.3-98.7), and 99.3% (99.1-99.5) for EMR, 51.3% (44.0-58.5), 99.5% (99.3-99.7), 79.4% (71.2-86.1), and 98.2% (97.8-98.5) for administrative data, and 87.2% (81.7-91.5), 99.9% (99.7-100), 96.6% (92.7-98.7) and 99.5% (99.3-99.7) for combined EMR and administrative data. Administrative data algorithms had similar sensitivity and specificity in the diabetes clinic registry. Of 11 499 711 adults in Ontario in 2017, there were 24 789 (0.22%, minimum estimate) to 102 140 (0.89%, maximum estimate) with T1D. Between 2010 and 2017, the age-standardized and sex-standardized prevalence rates per 1000 person-years increased (minimum estimate 1.7 to 2.56, maximum estimate 7.48 to 9.86, p<0.0001). In contrast, incidence rates decreased (minimum estimate 0.1 to 0.04, maximum estimate 0.47 to 0.09, p<0.0001).

CONCLUSIONS

Primary care EMR and administrative data algorithms performed well in identifying T1D and demonstrated increasing T1D prevalence in Ontario. These algorithms may permit the development of large, population-based cohort studies of T1D.

Initial Glycemic Control and Care Among Younger Adults Diagnosed With Type 2 Diabetes

OBJECTIVE

The prevalence of type 2 diabetes is increasing among adults under age 45. Onset of type 2 diabetes at a younger age increases an individual's risk for diabetes-related complications. Given the lasting benefits conferred by early glycemic control, we compared glycemic control and initial care between adults with younger onset (21-44 years) and mid-age onset (45-64 years) of type 2 diabetes.

RESEARCH DESIGN AND METHODS

Using data from a large, integrated health care system, we identified 32,137 adults (aged 21-64 years) with incident diabetes (first HbA_1c ≥6.5% [≥48 mmol/mol]). We excluded anyone with evidence of prior type 2 diabetes, gestational diabetes mellitus, or type 1 diabetes. We used generalized linear mixed models, adjusting for demographic and clinical variables, to examine differences in glycemic control and care at 1 year.

RESULTS

Of identified individuals, 26.4% had younger-onset and 73.6% had mid-age-onset type 2 diabetes. Adults with younger onset had higher initial mean HbA_1c values (8.9% [74 mmol/mol]) than adults with onset in mid-age (8.4% [68 mmol/mol]) (P < 0.0001) and lower odds of achieving an HbA_1c <7% (<53 mmol/mol) 1 year after the diagnosis (adjusted odds ratio [aOR] 0.70 [95% CI 0.66-0.74]), even after accounting for HbA_1c at diagnosis. Adults with younger onset had lower odds of in-person primary care contact (aOR 0.82 [95% CI 0.76-0.89]) than those with onset during mid-age, but they did not differ in telephone contact (1.05 [0.99-1.10]). Adults with younger onset had higher odds of starting metformin (aOR 1.20 [95% CI 1.12-1.29]) but lower odds of adhering to that medication (0.74 [0.69-0.80]).

CONCLUSIONS

Adults with onset of type 2 diabetes at a younger age were less likely to achieve glycemic control at 1 year following diagnosis, suggesting the need for tailored care approaches to improve outcomes for this high-risk patient population.

Association Between Rotavirus Vaccination and Type 1 Diabetes in Children

IMPORTANCE

Because rotavirus infection is a hypothesized risk factor for type 1 diabetes, live attenuated rotavirus vaccination could increase or decrease the risk of type 1 diabetes in children.

OBJECTIVE

To examine whether there is an association between rotavirus vaccination and incidence of type 1 diabetes in children aged 8 months to 11 years.

DESIGN, SETTING, AND PARTICIPANTS

A retrospective cohort study of 386 937 children born between January 1, 2006, and December 31, 2014, was conducted in 7 US health care organizations of the Vaccine Safety Datalink. Eligible children were followed up until a diagnosis of type 1 diabetes, disenrollment, or December 31, 2017.

EXPOSURES

Rotavirus vaccination for children aged 2 to 8 months. Three exposure groups were created. The first group included children who received all recommended doses of rotavirus vaccine by 8 months of age (fully exposed to rotavirus vaccination). The second group had received some, but not all, recommended rotavirus vaccines (partially exposed to rotavirus vaccination). The third group did not receive any doses of rotavirus vaccines (unexposed to rotavirus vaccination).

MAIN OUTCOMES AND MEASURES

Incidence of type 1 diabetes among children aged 8 months to 11 years. Type 1 diabetes was identified by International Classification of Diseases codes: 250.x1, 250.x3, or E10.xx in the outpatient setting. Cox proportional hazards regression models were used to analyze time to type 1 diabetes incidence from 8 months to 11 years. Hazard ratios and 95% CIs were calculated. Models were adjusted for sex, race/ethnicity, birth year, mother's age, birth weight, gestational age, number of well-child visits, and Vaccine Safety Datalink site.

RESULTS

In a cohort of 386 937 children (51.1% boys and 41.9% non-Hispanic white), 360 169 (93.1%) were fully exposed to rotavirus vaccination, 15 765 (4.1%) were partially exposed to rotavirus vaccination, and 11 003 (2.8%) were unexposed to rotavirus vaccination. Children were followed up a median of 5.4 years (interquartile range, 3.8-7.8 years). The total person-time follow-up in the cohort was 2 253 879 years. There were 464 cases of type 1 diabetes in the cohort, with an incidence rate of 20.6 cases per 100 000 person-years. Compared with children unexposed to rotavirus vaccination, the adjusted hazard ratio was 1.03 (95% CI, 0.62-1.72) for children fully exposed to rotavirus vaccination and 1.50 (95% CI, 0.81-2.77) for children partially exposed to rotavirus vaccination.

CONCLUSIONS AND RELEVANCE

The findings of this study suggest that rotavirus vaccination does not appear to be associated with type 1 diabetes in children.

Racial and Ethnic Disparities in Diabetes Care Quality among Women of Reproductive Age in an Integrated Delivery System

BACKGROUND

Diabetes is increasingly prevalent among women of reproductive age, yet little is known about quality of diabetes care for this population at increased risk of diabetes complications and poor maternal and infant health outcomes. Previous studies have identified racial/ethnic disparities in diabetes care, but patterns among women of reproductive age have not been examined.

METHODS

This retrospective cohort study analyzed 2016 data from Kaiser Permanente Northern California, a large integrated delivery system. Outcomes were quality of diabetes care measures-glycemic testing, glycemic control, and medication adherence-among women ages 18 to 44 with type 1 or type 2 diabetes (N = 9,923). Poisson regression was used to estimate the association between patient race/ethnicity and each outcome, adjusting for other patient characteristics and health care use.

RESULTS

In this cohort, 83% of participants had type 2 diabetes; 31% and 36% of women with type 2 and type 1 diabetes, respectively, had poor glycemic control (hemoglobin A1c of ≥9%), and approximately one-third of women with type 2 diabetes exhibited nonadherence to diabetes medications. Compared with non-Hispanic White women with type 2 diabetes, non-Hispanic Black women (adjusted risk ratio, 1.2; 95% confidence interval, 1.1-1.3) and Hispanic women (adjusted risk ratio, 1.2; 95% confidence interval, 1.1-1.3) were more likely to have poor control. Findings among women with type 1 diabetes were similar.

CONCLUSIONS

Our findings indicate opportunities to decrease disparities and improve quality of diabetes care for reproductive-aged women. Elucidating the contributing factors to poor glycemic control and medication adherence in this population, particularly among Black, Hispanic, and Asian women, should be a high research and practice priority.

Prevalence and incidence of type 1 diabetes in Ireland: a retrospective cross-sectional study using a national pharmacy claims data from 2016

OBJECTIVES

The aim of this study is to estimate the prevalence and incidence of type 1 diabetes in the Irish population using a national pharmacy claims database in the absence of a national diabetes register.

DESIGN

National, population-based, retrospective, cross-sectional study.

SETTING

Community care with data available through the Health Service Executive Pharmacy Claims Reimbursement Scheme from 2011 to 2016.

PARTICIPANTS

Individuals with type 1 diabetes were identified by coprescription of insulin and glucometer test strips without any prolonged course (>12 months) of oral hypoglycaemic agents prior to commencing insulin. Those claiming prescriptions for long-acting insulin only, without any prandial insulin, were excluded from the analysis. Incidence was estimated based on the first claim for insulin in 2016, with no insulin use in the preceding 12 months.

MAIN OUTCOME MEASURES

Prevalence of type 1 diabetes in children (<18 years) and adults (≥18 years); incidence of type 1 diabetes in children (≤14 years) and adolescents and adults (>14 years).

RESULTS

There were 20 081 prevalent cases of type 1 diabetes in 2016. The crude prevalence was 0.42% (95% CI 0.42% to 0.43%). Most prevalent cases (n=17 053, 85%) were in adults with a prevalence of 0.48% (95% CI 0.47% to 0.48%). There were 1527 new cases of type 1 diabetes in 2016, giving an incidence rate of 32 per 100 000 population/year (95% CI 30.5 to 33.7). There was a significant positive linear trend for age, for prevalence (p<0.0001) and incidence (p=0.014). The prevalence and incidence were 1.2-fold and 1.3-fold higher in men than women, respectively. Significant variations in prevalence (p<0.0001) and incidence (p<0.001) between the different geographical regions were observed.

CONCLUSIONS

This study provides epidemiological estimates of type 1 diabetes across age groups in Ireland, with the majority of prevalent cases in adults. Establishing a national diabetes register is essential to enable updated epidemiological estimates of diabetes and for planning of services in Ireland.

Estimation of Annual Health Care Costs for Adults with Type 1 Diabetes in the United States

BACKGROUND

Diabetes health care resource utilization (HCRU) studies tend to focus on patients with type 2 diabetes (T2D) or pool patients with T2D and type 1 diabetes (T1D). There is a paucity of recent data on the cost of treating patients with T1D in the United States.

OBJECTIVES

To (a) estimate the per-patient per-year (PPPY) HCRU and costs, from a payer perspective, associated with treating U.S. adults with T1D and (b) compare these with the HCRU and costs for patients with T2D.

METHODS

This retrospective cohort study used claims data from the Optum Clinformatics database between January 2015 and December 2017. Adults (aged ≥ 18 years) with a diagnosis of T1D were propensity score-matched to adults with T2D. Overall and nondiabetes-related HCRU and costs were assessed for T1D and T2D and compared between the 2 groups.

RESULTS

Propensity scores were used to match 10,103 patient pairs from T1D and T2D cohorts (mean ages 54.4 and 56.9 years, respectively). In the T1D cohort, inpatient, emergency department (ED), outpatient, and prescription claims occurred in 14.0%, 17.3%, 85.5%, and 100% of patients, respectively, resulting in a mean total cost of U.S. $18,817 PPPY (diabetes-related = $11,002; nondiabetes-related = $7,816). The T1D cohort had significantly higher mean total costs than the T2D cohort ($18,817 vs. $14,148 PPPY; P < 0.001). When extrapolating these findings to a commercial health plan with 1 million covered lives, the estimated total direct medical costs of T1D would be $103.4 million.

CONCLUSIONS

This study showed that the total annual cost of managing an adult with T1D is significantly higher than that of an adult with T2D. Nondiabetes costs accounted for 40% of the total per-patient cost, similar to patients with T2D, confirming that as patients with T1D live longer lives, they may also be at greater risk for cardiometabolic complications.

DISCLOSURES

This study was funded by Sanofi U.S. and Lexicon Pharmaceuticals as part of a business partnership in a diabetes program at the time this study was conducted. Joish and Davies are employees and stockholders of Lexicon Pharmaceuticals. Zhou, Preblick, and Paranjape are employees and stockholders of Sanofi. Lin was a postdoctoral fellow at Sanofi through Rutgers University during this project. Deshpande provided consulting services through Communication Symmetry. Verma is an employee of Evidera, which was contracted by Sanofi for work on this study. Pettus is a consultant for Diasome, Insulet, Lexicon, Lilly, Mannkind, Novo Nordisk, Sanofi, and Senseonics.

Development and validation of algorithms to classify type 1 and 2 diabetes according to age at diagnosis using electronic health records

BACKGROUND

Validated algorithms to classify type 1 and 2 diabetes (T1D, T2D) are mostly limited to white pediatric populations. We conducted a large study in Hong Kong among children and adults with diabetes to develop and validate algorithms using electronic health records (EHRs) to classify diabetes type against clinical assessment as the reference standard, and to evaluate performance by age at diagnosis.

METHODS

We included all people with diabetes (age at diagnosis 1.5-100 years during 2002-15) in the Hong Kong Diabetes Register and randomized them to derivation and validation cohorts. We developed candidate algorithms to identify diabetes types using encounter codes, prescriptions, and combinations of these criteria ("combination algorithms"). We identified 3 algorithms with the highest sensitivity, positive predictive value (PPV), and kappa coefficient, and evaluated performance by age at diagnosis in the validation cohort.

RESULTS

There were 10,196 (T1D n = 60, T2D n = 10,136) and 5101 (T1D n = 43, T2D n = 5058) people in the derivation and validation cohorts (mean age at diagnosis 22.7, 55.9 years; 53.3, 43.9% female; for T1D and T2D respectively). Algorithms using codes or prescriptions classified T1D well for age at diagnosis < 20 years, but sensitivity and PPV dropped for older ages at diagnosis. Combination algorithms maximized sensitivity or PPV, but not both. The "high sensitivity for type 1" algorithm (ratio of type 1 to type 2 codes ≥ 4, or at least 1 insulin prescription within 90 days) had a sensitivity of 95.3% (95% confidence interval 84.2-99.4%; PPV 12.8%, 9.3-16.9%), while the "high PPV for type 1" algorithm (ratio of type 1 to type 2 codes ≥ 4, and multiple daily injections with no other glucose-lowering medication prescription) had a PPV of 100.0% (79.4-100.0%; sensitivity 37.2%, 23.0-53.3%), and the "optimized" algorithm (ratio of type 1 to type 2 codes ≥ 4, and at least 1 insulin prescription within 90 days) had a sensitivity of 65.1% (49.1-79.0%) and PPV of 75.7% (58.8-88.2%) across all ages. Accuracy of T2D classification was high for all algorithms.

CONCLUSIONS

Our validated set of algorithms accurately classifies T1D and T2D using EHRs for Hong Kong residents enrolled in a diabetes register. The choice of algorithm should be tailored to the unique requirements of each study question.

Paradox of glycemic management: multimorbidity, glycemic control, and high-risk medication use among adults with diabetes

INTRODUCTION

Glycemic targets and glucose-lowering regimens should be individualized based on multiple factors, including the presence of comorbidities. We examined contemporary patterns of glycemic control and use of medications known to cause hypoglycemia among adults with diabetes across age and multimorbidity.

RESEARCH DESIGN AND METHODS

We retrospectively examined glycosylated hemoglobin (HbA_1c) levels and rates of insulin/sulfonylurea use as a function of age and multimorbidity using administrative claims and laboratory data for adults with type 2 diabetes included in OptumLabs Data Warehouse, 1 January 2014 to 31 December 2016. Comorbidity burden was assessed by counts of any of 16 comorbidities specified by guidelines as warranting relaxation of HbA_1c targets, classified as being diabetes concordant (diabetes complications or risk factors), discordant (unrelated to diabetes), or advanced (life limiting).

RESULTS

Among 194 157 patients with type 2 diabetes included in the study, 45.2% had only concordant comorbidities, 30.6% concordant and discordant, 2.7% only discordant, and 13.0% had ≥1 advanced comorbidity. Mean HbA_1c was 7.7% among 18-44 year-olds versus 6.9% among ≥75 year-olds, and was higher among patients with comorbidities: 7.3% with concordant only, 7.1% with discordant only, 7.1% with concordant and discordant, and 7.0% with advanced comorbidities compared with 7.4% among patients without comorbidities. The odds of insulin use decreased with age (OR 0.51 (95% CI 0.48 to 0.54) for age ≥75 vs 18-44 years) but increased with accumulation of concordant (OR 5.50 (95% CI 5.22 to 5.79) for ≥3 vs none), discordant (OR 1.72 (95% CI 1.60 to 1.86) for ≥3 vs none), and advanced (OR 1.45 (95% CI 1.25 to 1.68) for ≥2 vs none) comorbidities. Conversely, sulfonylurea use increased with age (OR 1.36 (95% CI 1.29 to 1.44) for age ≥75 vs 18-44 years) but decreased with accumulation of concordant (OR 0.76 (95% CI 0.73 to 0.79) for ≥3 vs none), discordant (OR 0.70 (95% CI 0.64 to 0.76) for ≥3 vs none), but not advanced (OR 0.86 (95% CI 0.74 to 1.01) for ≥2 vs none) comorbidities.

CONCLUSIONS

The proportion of patients achieving low HbA_1c levels was highest among older and multimorbid patients. Older patients and patients with higher comorbidity burden were more likely to be treated with insulin to achieve these HbA_1c levels despite potential for hypoglycemia and uncertain long-term benefit.

Association of Cumulative Multimorbidity, Glycemic Control, and Medication Use With Hypoglycemia-Related Emergency Department Visits and Hospitalizations Among Adults With Diabetes

IMPORTANCE

Severe hypoglycemia is a serious and potentially preventable complication of diabetes, with some of the most severe episodes requiring emergency department (ED) care or hospitalization. A variety of health conditions increase the risk of hypoglycemia. People with diabetes often have multiple comorbidities, and the association of such multimorbidity with hypoglycemia risk in the context of other risk factors is uncertain.

OBJECTIVE

To examine the associations of age, cumulative multimorbidity, glycated hemoglobin (HbA1c) level, and use of glucose level-lowering medication with hypoglycemia-related ED visits and hospitalizations.

DESIGN, SETTING, AND PARTICIPANTS

Cohort study of claims and laboratory data from OptumLabs Data Warehouse, an administrative claims database of commercially insured and Medicare Advantage beneficiaries in the United States. Participants were adults (aged ≥18 years) with diabetes who had an available HbA1c level result in 2015. Data from January 1, 2014, to December 31, 2016, were analyzed. Final analyses were conducted from December 2017 to September 2018.

MAIN OUTCOMES AND MEASURES

This study calculated rates of hypoglycemia-related ED visits and hospitalizations during the year after the index HbA1c level was obtained, stratified by patient demographic characteristics, diabetes type, comorbidities (from 16 guideline-specified high-risk conditions), index HbA1c level, and glucose level-lowering medication use. The association of each variable with hypoglycemia-related ED and hospital care was examined using multivariable Poisson regression analysis overall and by diabetes type.

RESULTS

The study cohort was composed of 201 705 adults with diabetes (mean [SD] age, 65.8 [12.1] years; 102 668 [50.9%] women; 118 804 [58.9%] white; mean [SD] index HbA1c level, 7.2% [1.5%]). Overall, there were 9.06 (95% CI, 8.64-9.47) hypoglycemia-related ED visits and hospitalizations per 1000 persons per year. The risk of hypoglycemia-related ED visits and hospitalizations was increased by age 75 years or older (incidence rate ratio [IRR], 1.56 [95% CI, 1.23-2.02] vs 18-44 years), black race/ethnicity (IRR, 1.30 [95% CI, 1.16-1.46] vs white race/ethnicity), lower annual household income (IRR, 0.63 [95% CI, 0.53-0.74] for ≥$100 000 vs <$40 000), number of comorbidities (increasing from IRR of 1.66 [95% CI, 1.42-1.95] in the presence of 2 comorbidities to IRR of 4.12 [95% CI, 3.07-5.51] with ≥8 comorbidities compared with ≤1), prior hypoglycemia-related ED visit or hospitalization (IRR, 6.60 [95% CI, 5.77-7.56]), and glucose level-lowering treatment regimen (IRR, 6.73 [95% CI, 4.93-9.22] for sulfonylurea; 12.53 [95% CI, 8.90-17.64] for basal insulin; and 27.65 [95% CI, 20.32-37.63] for basal plus bolus insulin compared with other medications). Independent of these factors, having type 1 diabetes was associated with a 34% increase in the risk of hypoglycemia-related ED visits or hospitalizations (IRR, 1.34 [95% CI, 1.15-1.55]). The index HbA1c level was associated with hypoglycemia-related ED visits and hospitalizations when both low (IRR, 1.45 [95% CI, 1.12-1.87] for HbA1c level ≤5.6% vs 6.5%-6.9%) and high (IRR, 1.24 [95% CI, 1.02-1.50] for HbA1c level ≥10%).

CONCLUSIONS AND RELEVANCE

In this cohort study of adults with diabetes, the risk of an ED visit or hospitalization for hypoglycemia appeared to be highest among patients with type 1 diabetes, multiple comorbidities, prior severe hypoglycemia, and sulfonylurea and/or insulin use. At-risk patients may benefit from individualized treatment regimens to decrease their risk of hypoglycemia.

Use of Sodium-Glucose Cotransporter 2 Inhibitors in Patients With Type 1 Diabetes and Rates of Diabetic Ketoacidosis

OBJECTIVE

To estimate real-world off-label use of sodium-glucose cotransporter 2 (SGLT2) inhibitors in patients with type 1 diabetes, estimate rates of diabetic ketoacidosis (DKA), and compare them with DKA rates observed in sotagliflozin clinical trials.

RESEARCH DESIGN AND METHODS

We identified initiators of SGLT2 inhibitors in the Sentinel System from March 2013 to June 2018, determined the prevalence of type 1 diabetes using a narrow and a broad definition, and measured rates of DKA using administrative claims data. Standardized incidence ratios (SIRs) were calculated using age- and sex-specific follow-up time in Sentinel and age- and sex-specific DKA rates from sotagliflozin trials 309, 310, and 312.

RESULTS

Among 475,527 initiators of SGLT2 inhibitors, 0.50% and 0.92% met narrow and broad criteria for type 1 diabetes, respectively. Rates of DKA in the narrow and broad groups were 7.3/100 person-years and 4.5/100 person-years, respectively. Among patients who met narrow criteria for type 1 diabetes, rates of DKA were highest for patients aged 25-44 years, especially females aged 25-44 years (19.7/100 person-years). More DKA events were observed during off-label use of SGLT2 inhibitors in Sentinel than would be expected based on sotagliflozin clinical trials (SIR = 1.83; 95% CI 1.45-2.28).

CONCLUSIONS

Real-world off-label use of SGLT2 inhibitors among patients with type 1 diabetes accounted for a small proportion of overall SGLT2 inhibitor use. However, the risk for DKA during off-label use was notable, especially among young, female patients. Although real-word rates of DKA exceeded the expectation based on clinical trials, results should be interpreted with caution due to differences in study methods, patient samples, and study drugs.

Adoption of New Glucose-Lowering Medications in the U.S.-The Case of SGLT2 Inhibitors: Nationwide Cohort Study

Background: High-quality diabetes care is evidence-based, timely, and equitable. Sodium-glucose cotransporter-2 inhibitors (SGLT2i) are the most recently approved class of glucose-lowering medications with additional cardio- and renal-protective benefits and low risk of hypoglycemia. Cardiovascular and kidney disease are among the most common chronic diabetes complications, whereas hypoglycemia is the most prevalent adverse effect of glucose-lowering therapy. We examine the sociodemographic and clinical factors associated with early SGLT2i initiation and appropriateness of use based on contemporaneous scientific evidence. Materials and Methods: Retrospective analysis of medical and pharmacy claims data from OptumLabs® Data Warehouse for commercially insured and Medicare Advantage adult beneficiaries with diabetes types 1 and 2, who filled any glucose-lowering medication between January 1, 2013 and December 31, 2016. Demographic (age, sex, race, income), clinical (comorbidities), and insurance-related factors affecting first prescription for a SGLT2i were examined using multivariable logistic regression. Results: Among 1,054,727 adults with pharmacologically treated diabetes, 7.2% (n = 75,500) initiated a SGLT2i. Patients with prior myocardial infarction (MI) (odds ratio [OR]: 0.94, 95% confidence interval [CI]: 0.91-0.96), heart failure (HF) (OR: 0.93, 95% CI: 0.91-0.94), kidney disease (OR: 0.80, 95% CI: 0.78-0.81), and severe hypoglycemia (OR: 0.96, 95% CI: 0.94-0.98) were all less likely to start a SGLT2i; P < 0.001 for all. SGLT2i were also less likely to be started by patients ≥75 years (OR: 0.57, 95% CI: 0.55-0.59, vs. 18-44 years), Black patients (OR: 0.93, 95% CI: 0.91-0.95, vs. White), and those with Medicare Advantage insurance (OR: 0.63, 95% CI: 0.62-0.64, vs. commercial). Conclusions: Younger, healthier, non-Black patients with commercial health insurance were most likely to start taking SGLT2i. Patients with MI, HF, kidney disease, and prior hypoglycemia were less likely to use SGLT2i, despite evidence supporting their preferential use in these patients. Efforts to address this treatment-risk paradox may help improve health outcomes among patients with type 2 diabetes.

Validity of ICD-10-CM codes for determination of diabetes type for persons with youth-onset type 1 and type 2 diabetes

OBJECTIVE

Diagnosis codes might be used for diabetes surveillance if they accurately distinguish diabetes type. We assessed the validity of International Classification of Disease, 10th Revision, Clinical Modification (ICD-10-CM) codes to discriminate between type 1 diabetes mellitus (T1DM) and type 2 diabetes mellitus (T2DM) among health plan members with youth-onset (diagnosis age <20 years) diabetes.

RESEARCH DESIGN AND METHODS

Diabetes case identification and abstraction of diabetes type was done as part of the SEARCH for Diabetes in Youth Study. The gold standard for diabetes type is the physician-assigned diabetes type documented in patients' medical records. Using all healthcare encounters with ICD-10-CM codes for diabetes, we summarized codes within each encounter and determined diabetes type using percent of encounters classified as T2DM. We chose 50% as the threshold from a receiver operating characteristic curve because this threshold yielded the largest Youden's index. Persons with ≥50% T2DM-coded encounters were classified as having T2DM. Otherwise, persons were classified as having T1DM. We calculated sensitivity, specificity, positive and negative predictive values, and accuracy overall and by demographic characteristics.

RESULTS

According to the gold standard, 1911 persons had T1DM and 652 persons had T2DM (mean age (SD): 19.1 (6.5) years). We obtained 90.6% (95% CI 88.4% to 92.9%) sensitivity, 96.3% (95% CI 95.4% to 97.1%) specificity, 89.3% (95% CI 86.9% to 91.6%) positive predictive value, 96.8% (95% CI 96.0% to 97.6%) negative predictive value, and 94.8% (95% CI 94.0% to 95.7%) accuracy for discriminating T2DM from T1DM.

CONCLUSIONS

ICD-10-CM codes can accurately classify diabetes type for persons with youth-onset diabetes, showing promise for rapid, cost-efficient diabetes surveillance.