Search

OBJECTIVE Although diabetic retinopathy is a leading cause of blindness worldwide, diabetes-related blindness can be prevented through effective screening, detection, and treatment of disease. The study goal was to develop risk stratification algorithms for the onset of retinal complications of diabetes, including proliferative diabetic retinopathy, referable retinopathy, and macular edema. RESEARCH DESIGN AND METHODS Retrospective cohort analysis of patients from the Kaiser Permanente Northern California Diabetes Registry who had no evidence of diabetic retinopathy at a baseline diabetic retinopathy screening during 2008–2020 was performed. Machine learning and logistic regression prediction models for onset of proliferative diabetic retinopathy, diabetic macular edema, and referable retinopathy detected through routine screening were trained and internally validated. Model performance was assessed using area under the curve (AUC) metrics. RESULTS The study cohort (N = 276,794) was 51.9% male and 42.1% White. Mean (±SD) age at baseline was 60.0 (±13.1) years. A machine learning XGBoost algorithm was effective in identifying patients who developed proliferative diabetic retinopathy (AUC 0.86; 95% CI, 0.86–0.87), diabetic macular edema (AUC 0.76; 95% CI, 0.75–0.77), and referable retinopathy (AUC 0.78; 95% CI, 0.78–0.79). Similar results were found using a simpler nine-covariate logistic regression model: proliferative diabetic retinopathy (AUC 0.82; 95% CI, 0.80–0.83), diabetic macular edema (AUC 0.73; 95% CI, 0.72–0.74), and referable retinopathy (AUC 0.75; 95% CI, 0.75–0.76). CONCLUSIONS Relatively simple logistic regression models using nine readily available clinical variables can be used to rank order patients for onset of diabetic eye disease and thereby more efficiently prioritize and target screening for at risk patients.

ImportanceThe levels of glycemic control associated with the lowest risk of dementia in people with type 2 diabetes are unknown. This knowledge is critical to inform patient-centered glycemic target setting.ObjectiveTo examine the associations between cumulative exposure to various ranges of glycated hemoglobin (HbA1c) concentrations with dementia risk across sex and racial and ethnic groups and the association of current therapeutic glycemic targets with dementia risk.Design, Setting, and ParticipantsThis cohort study included members of the Kaiser Permanente Northern California integrated health care system with type 2 diabetes who were aged 50 years or older during the study period from January 1, 1996, to September 30, 2015. Individuals with fewer than 2 HbA1c measurements during the study period, prevalent dementia at baseline, or less than 3 years of follow-up were excluded. Data were analyzed from February 2020 to January 2023.ExposuresTime-updated cumulative exposure to HbA1c thresholds. At each HbA1c measurement, participants were categorized based on the percentage of their HbA1c measurements that fell into the following categories: less than 6%, 6% to less than 7%, 7% to less than 8%, 8% to less than 9%, 9% to less than 10%, and 10% or more of total hemoglobin (to convert percentage of total hemoglobin to proportion of total hemoglobin, multiply by 0.01).Main Outcomes and MeasuresDementia diagnosis was identified using International Classification of Diseases, Ninth Revision codes from inpatient and outpatient encounters. Cox proportional hazards regression models estimated the association of time-varying cumulative glycemic exposure with dementia, adjusting for age, race and ethnicity, baseline health conditions, and number of HbA1c measurements.ResultsA total of 253 211 participants were included. The mean (SD) age of participants was 61.5 (9.4) years, and 53.1% were men. The mean (SD) duration of follow-up was 5.9 (4.5) years. Participants with more than 50% of HbA1c measurements at 9% to less than 10% or 10% or more had greater risk of dementia compared with those who had 50% or less of measurements in those categories (HbA1c 9% to 1c≥10%: aHR, 1.74 [95% CI, 1.62-1.86]). By contrast, participants with more than 50% of HbA1c concentrations less than 6%, 6% to less than 7%, or 7% to less than 8% had lower risk of dementia (HbA1c1c 6% to 1c 7% to Conclusions and RelevanceIn this study dementia risk was greatest among adults with cumulative HbA1c concentrations of 9% or more. These results support currently recommended relaxed glycemic targets for older people with type 2 diabetes.

Continuous glucose monitoring (CGM) is indicated in poorly controlled insulin-treated patients with type 2 diabetes (T2D) to improve glycemic control and reduce the risk of hypoglycemia, but the benefits of CGM for lower risk patients have not been well studied. Among 17,422 insulin-treated patients with T2D with hemoglobin A1c (HbA1c)

Individuals with anemias are often excluded from diabetes research when the population or outcome of interest is defined by HbA1c, as abnormal red blood cell turnover rates raise concern about the accuracy of measured HbA1c values. While racial/ethnic differences in the prevalence of certain anemias are well-known, the potential bias introduced by excluding individuals with anemias from research samples is less studied. We examined the prevalence of diagnosed anemias by race/ethnicity in a population of adults diagnosed with diabetes. Using data from a large healthcare system, we identified adults (≥21 years) diagnosed with type 2 diabetes from 2010-2016. We used ICD-9/codes to identify individuals diagnosed with anemia in the year prior to diabetes diagnosis and inherited hemoglobinopathies through follow-up. We used chi-square tests to compare the prevalence of these diagnoses by race/ethnicity. Among 78,2individuals, 5,434 (6.9%) had an anemia diagnosis in the 12 months prior to their diabetes diagnosis, with the most common being unspecified anemia (3.3%) and iron deficiency anemia (2.3%) . Black adults had the highest prevalence of any anemia (11% vs. Other race/ethnicity 8.8% vs. White 7% vs. Latino 6.1% vs. Asian 5.9%, p≤.0001) , unspecified anemia (5.4% vs. Other race/ethnicity 3.8% vs. White 3.5% vs. Latino 2.8% vs. Asian 2.4%, p≤.0001) , and iron deficiency anemia (3.6% vs. Other race/ethnicity 2.9% vs. White 2.2% vs. Latino 2.4% vs. Asian 1.9%, p≤.0001) . Black and Asian adults had the highest prevalence of inherited hemoglobinopathies (Asian 2.3% vs. Black 1.9% vs. Other race/ethnicity 1.3% vs. White 0.3% vs. Latino 0.3%, p≤.0001) . In conclusion, among adults diagnosed with diabetes, Black adults had the highest prevalence of diagnosed anemias (>1 in Black adults had a diagnosis) . Excluding individuals from diabetes research based on diagnosis of anemia may disproportionately impact the relevance of the findings to Black adults. Disclosure C.Board: None. A.Gopalan: None. A.J.Karter: Research Support; Dexcom, Inc. R.W.Grant: None. O.Sofrygin: None. Funding National Institutes of Health (1R21DK130018-01)

Purpose: The purpose of this study is to develop a diabetes outcomes model for the US (DOMUS) using a multi-ethnic, real-world-data cohort of newly diagnosed Type II diabetics that can predict outcomes for a U.S. multi-ethnic type 2 diabetes population. Methods: The Kaiser Permanente Northern California (KPNC) Diabetes Registry, which is a well-described epidemiologic cohort with up to 13 years of follow-up from EMR and claims, were used to identify over 150,000 newly diagnosed diabetes patients between 2005-2016 with up to 13-year follow-up. The DOMUS model integrates separate, but interdependent risk equations to predict events for each of the micro and macro-vascular events, hypoglycemia, dementia, depression, and death, and predictive models for eight biomarker levels. Model accounted for static demographic factors (e.g., race) , neighborhood deprivation, and dynamic factors, such as age, duration of diabetes, fifteen-possible glucose -lowering treatment combinations, biomarker levels, and history of diabetes-related events. Moreover, the models explicitly allow for a legacy effect (average A1c in the first year after diagnosis) for all outcomes. Results: Data were randomly split into 75,000, 18,750, and 18,750 patients to perform estimation, out-of-sample calibration, and validation respectively. Model predictions in the validation sample closely aligned with the observed longitudinal trajectory of biomarkers and outcomes. Moreover, we examine the model performance within by age, race/ethnicity, and sex and found excellent predictive performance within subgroups. Conclusion: The DOMUS Model is able to simulate event histories and biomarker trajectories that closely match observed outcomes in a real-world, multi-ethnic population of newly diagnosed diabetics. DOMUS has the potential to carry out many complex analyses that examine the effectiveness and cost-effectiveness of new medications, value of diabetes prevention, and serve as decision support tools in health care settings. Disclosure A.N.Winn: Consultant; Takeda Pharmaceutical Company Limited. E.Huang: Advisory Panel; Twin Health, Stock/Shareholder; AbbVie Inc. P.Clarke: None. A.J.Karter: Research Support; Dexcom, Inc. W.Wan: n/a. D.R.Miller: None. M.Franco: None. A.Deckard: None. N.Laiteerapong: None.

This cohort study uses data from continuous glucose monitoring to validate a hypoglycemia risk stratification tool.

The substantial expansion of secure messaging (SM) via the patient portal in the last decade suggests that it is becoming a standard of care, but few have examined SM use longitudinally. We examined SM patterns among a diverse cohort of patients with diabetes (N = 19 921) and the providers they exchanged messages with within a large, integrated health system over 10 years (2006-2015), linking patient demographics to SM use. We found a 10-fold increase in messaging volume. There were dramatic increases overall and for patient subgroups, with a majority of patients (including patients with lower income or with self-reported limited health literacy) messaging by 2015. Although more physicians than nurses and other providers messaged throughout the study, the distribution of health professions using SM changed over time. Given this rapid increase in SM, deeper understanding of optimizing the value of patient and provider engagement, while managing workflow and training challenges, is crucial.

BACKGROUND: Mail order pharmacy (MOP) use has been linked to improved medication adherence and health outcomes among patients with diabetes. However, no large-scale intervention studies have assessed the effect of encouraging MOP use on medication adherence. OBJECTIVE: To assess an intervention to encourage MOP services to increase its use and medication adherence. DESIGN: Randomized encouragement trial. PATIENTS: 63,012 diabetes patients from three health care systems: Kaiser Permanente Northern California (KPNC), Kaiser Permanente Hawaii (KPHI), and Harvard Pilgrim Health Care (HPHC) who were poorly adherent to at least one class of cardiometabolic medications and had not used MOP in the prior 12 months. INTERVENTION: Patients were randomized to receive either usual care (control arm) or outreach encouraging MOP use consisting of a mailed letter, secure email message, and automated telephone call outlining the potential benefits of MOP use (intervention arm). HPHC intervention patients received the letter only. MEASUREMENTS: We compared the percentages of patients that began using MOP and that became adherent to cardiometabolic medication classes during a 12-month follow-up period. We also conducted a race/ethnicity-stratified analysis. RESULTS: During follow-up, 10.6% of intervention patients began using MOP vs. 9.3% of controls (p

Objective To develop novel, scalable, and valid literacy profiles for identifying limited health literacy patients by harnessing natural language processing. Data Source With respect to the linguistic content, we analyzed 283 216 secure messages sent by 6941 diabetes patients to physicians within an integrated system's electronic portal. Sociodemographic, clinical, and utilization data were obtained via questionnaire and electronic health records. Study Design Retrospective study used natural language processing and machine learning to generate five unique “Literacy Profiles” by employing various sets of linguistic indices: Flesch‐Kincaid (LP_FK); basic indices of writing complexity, including lexical diversity (LP_LD) and writing quality (LP_WQ); and advanced indices related to syntactic complexity, lexical sophistication, and diversity, modeled from self‐reported (LP_SR), and expert‐rated (LP_Exp) health literacy. We first determined the performance of each literacy profile relative to self‐reported and expert‐rated health literacy to discriminate between high and low health literacy and then assessed Literacy Profiles’ relationships with known correlates of health literacy, such as patient sociodemographics and a range of health‐related outcomes, including ratings of physician communication, medication adherence, diabetes control, comorbidities, and utilization. Principal Findings LP_SR and LP_Exp performed best in discriminating between high and low self‐reported (C‐statistics: 0.86 and 0.58, respectively) and expert‐rated health literacy (C‐statistics: 0.71 and 0.87, respectively) and were significantly associated with educational attainment, race/ethnicity, Consumer Assessment of Provider and Systems (CAHPS) scores, adherence, glycemia, comorbidities, and emergency department visits. Conclusions Since health literacy is a potentially remediable explanatory factor in health care disparities, the development of automated health literacy indicators represents a significant accomplishment with broad clinical and population health applications. Health systems could apply literacy profiles to efficiently determine whether quality of care and outcomes vary by patient health literacy; identify at‐risk populations for targeting tailored health communications and self‐management support interventions; and inform clinicians to promote improvements in individual‐level care.

OBJECTIVE In children with type 1 diabetes (T1D), severe hypoglycemia (SH) is associated with poorer cognition, but the association of SH with cognitive function in late life is unknown. Given the increasing life expectancy in people with T1D, understanding the role of SH in brain health is crucial. RESEARCH DESIGN AND METHODS We examined the association between SH and cognitive function in 718 older adults with T1D from the Study of Longevity in Diabetes (SOLID). Subjects self-reported recent SH (previous 12 months) and lifetime history of SH resulting in inpatient/emergency department utilization. Global and domain-specific cognition (language, executive function, episodic memory, and simple attention) were assessed. The associations of SH with cognitive function and impaired cognition were evaluated via linear and logistic regression models, respectively. RESULTS Thirty-two percent of participants (mean age 67.2 years) reported recent SH and 50% reported lifetime SH. Compared with those with no SH, subjects with a recent SH history had significantly lower global cognition scores. Domain-specific analyses revealed significantly lower scores on language, executive function, and episodic memory with recent SH exposure and significantly lower executive function with lifetime SH exposure. Recent SH was associated with impaired global cognition (odds ratio [OR] 3.22, 95% CI 1.30, 7.94) and cognitive impairment on the language domain (OR 3.15, 95% CI 1.19, 8.29). CONCLUSIONS Among older adults with T1D, recent SH and lifetime SH were associated with worse cognition. Recent SH was associated with impaired global cognition. These findings suggest a deleterious role of SH on the brain health of older patients with T1D and highlight the importance of SH prevention.

The relationship between achieved low-density lipoprotein cholesterol (LDL-C) levels and risk of incident atherosclerotic cardiovascular disease events among patients with diabetes and metabolic dyslipidemia has not been well described.We conducted an observational cohort study of statin-treated adults (ages 21-90 years) with type 2 diabetes without established atherosclerotic cardiovascular disease (as of January 1, 2006) who had metabolic dyslipidemia (elevated triglycerides ≥150 mg/dL and low high-density lipoprotein cholesterol,50 mg/dL [women] and40 mg/dL [men]). All subjects were members of Kaiser Permanente Northern California, an integrated health care delivery system. Adjusted multivariable Cox models were specified to estimate hazard ratios (HRs) for incident atherosclerotic cardiovascular disease events by achieved LDL-C levels (50, 50-70, 70-100, and ≥100 mg/dL). Incident atherosclerotic cardiovascular disease events were defined as a composite of nonfatal myocardial infarction, ischemic stroke, or coronary heart disease death through December 31, 2013.A total of 19,095 individuals met the selection criteria. Mean age was 63.4 years, 53.5% were women, and the mean follow-up was 5.9 years. Unadjusted rates of atherosclerotic cardiovascular disease events were not significantly different across specified LDL-C categories. In models adjusted for demographics and clinical characteristics, the risk was significantly lower with decreasing achieved LDL-C levels (P0.0001 for trend). Relative to achieved LDL-C ≥100 mg/dL, LDL-C50 mg/dL had an hazard ratio of 0.66 (95% confidence interval [CI] 0.52-0.82).In a large, contemporary cohort of statin-treated patients with type 2 diabetes and metabolic dyslipidemia without established atherosclerotic cardiovascular disease, lower achieved LDL-C levels were associated with a monotonically lower risk of incident atherosclerotic cardiovascular disease events. The benefits of achieving very-low LDL-C (50 mg/dL) in this population requires further evaluation in prospective interventional studies.

IntroductionThe 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 methodsWe 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.ResultsT1D 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).ConclusionsOlder 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.

BACKGROUND: Low literacy skills impact important aspects of communication, including health-related information exchanges. Unsuccessful communication on the part of physician or patient contributes to lower quality of care, is associated with poorer chronic disease control, jeopardizes patient safety and can lead to unfavorable healthcare utilization patterns. To date, very little research has focused on digital communication between physicians and patients, such as secure messages sent via electronic patient portals. METHOD: The purpose of the current study is to develop an automated readability formula to better understand what elements of physicians’ digital messages make them more or less difficult to understand. The formula is developed using advanced natural language processing (NLP) to predict human ratings of physician text difficulty. RESULTS: The results indicate that NLP indices that capture a diverse set of linguistic features predict the difficulty of physician messages better than classic readability tools such as Flesch Kincaid Grade Level. Our results also provide information about the textual features that best explain text readability. CONCLUSION: Implications for how the readability formula could provide feedback to physicians to improve digital health communication by promoting linguistic concordance between physician and patient are discussed.

ObjectiveTo determine whether severe hypoglycemic and hyperglycemic events are associated with longitudinal dementia risk in older adults with type 1 diabetes.MethodsA longitudinal cohort study followed up 2,821 members of an integrated health care delivery system with type 1 diabetes from 1997 to 2015. Hypoglycemic and hyperglycemic events requiring emergency room or hospitalization were abstracted from medical records beginning January 1, 1996, through cohort entry. Participants were followed up for dementia diagnosis through September 30, 2015. Dementia risk was examined with Cox proportional hazard models adjusted for age (as time scale), sex, race/ethnicity, hemoglobin A1c, depression, stroke, and nephropathy.ResultsAmong 2,821 older adults (mean age 56 years) with type 1 diabetes, 398 (14%) had a history of severe hypoglycemia, 335 (12%) had severe hyperglycemia, and 87 (3%) had both. Over a mean 6.9 years of follow-up, 153 individuals (5.4%) developed dementia. In fully adjusted models, individuals with hypoglycemic events had 66% greater risk of dementia than those without a hypoglycemic event (hazard ratio [HR] 1.66, 95% confidence interval [CI] 1.09, 2.53), while those with hyperglycemic events had >2 times the risk (HR 2.11, 95% CI 1.24, 3.59) than those without a hyperglycemic event. There was a 6-fold greater risk of dementia in individuals with both severe hypoglycemia and hyperglycemia vs those with neither (HR 6.20, 95% CI 3.02, 12.70).ConclusionsFor older individuals with type 1 diabetes, severe hypoglycemic and hyperglycemic events are associated with increased future risk of dementia.

Study objective The objective was to examine the association between sleep quality and global and domain-specific cognitive function among older individuals with type 1 diabetes (T1D). Methods We evaluated 695 individuals with T1D aged 60 years or above who participated in the baseline assessment of the Study of Longevity in Diabetes (SOLID), which captured subjective sleep quality (Pittsburgh Sleep Quality Index) and global and domain-specific (language, executive function, episodic memory, and simple attention) cognitive function. Multivariable linear regressions estimated the associations between sleep quality quartiles and overall and domain-specific cognitive function adjusting for age, sex, race/ethnicity, education, depressive symptoms, and severe hypoglycemic episodes. Sensitivity analyses examined the associations between aspects of sleep quality and global cognitive function. Results The worst sleep quality quartile was associated with lower global cognition (β=-0.08; 95% confidence interval: -0.17, -0.01) and lower executive function (β=-0.17, 95% confidence interval: -0.30, -0.03) compared with the best quartile of sleep quality adjusting for demographics and comorbidities. Sleep quality was not associated with language, episodic memory, or simple attention. Sleep medications and daytime dysfunction were most strongly associated with global cognition. Conclusion Our results suggest that sleep quality may be a modifiable risk factor for global cognitive function and executive function among elderly individuals with T1D.

BACKGROUND: Little is known about patients who have caregiver proxies communicate with healthcare providers via portal secure messaging (SM). Since proxy portal use is often informal (e.g., sharing patient accounts), novel methods are needed to estimate the prevalence of proxy-authored SMs. OBJECTIVE: (1) Develop an algorithm to identify proxy-authored SMs, (2) apply this algorithm to estimate predicted proxy SM (PPSM) prevalence among patients with diabetes, and (3) explore patient characteristics associated with having PPSMs. DESIGN: Retrospective cohort study. PARTICIPANTS: We examined 9856 patients from Diabetes Study of Northern California (DISTANCE) who sent ≥ 1 English-language SM to their primary care physician between July 1, 2006, and Dec. 31, 2015. MAIN MEASURES: Using computational linguistics, we developed ProxyID, an algorithm that identifies phrases frequently found in registered proxy SMs. ProxyID was validated against blinded expert categorization of proxy status among an SM sample, then applied to identify PPSM prevalence across patients. We examined patients’ sociodemographic and clinical characteristics according to PPSM penetrance, “none” (0%), “low” (≥ 0–50%), and “high” (≥ 50–100%). KEY RESULTS: Only 2.3% of patients had ≥ 1 registered proxy-authored SM. ProxyID demonstrated moderate agreement with expert classification (Κ = 0.58); 45.7% of patients had PPSMs (40.2% low and 5.5% high). Patients with high percent PPSMs were older than those with low percent and no PPSMs (66.5 vs 57.4 vs 56.2 years, p

Limited health literacy (HL) partially mediates health disparities. Measurement constraints, including lack of validity assessment across racial/ethnic groups and administration challenges, have undermined the field and impeded scaling of HL interventions. We employed computational linguistics to develop an automated and novel HL measure, analyzing >300,000 messages sent by >9,000 diabetes patients via a patient portal to create a Literacy Profiles. We carried out stratified analyses among White/non-Hispanics, Black/non-Hispanics, Hispanics, and Asian/Pacific Islanders to determine if the Literacy Profile has comparable criterion and predictive validities. We discovered that criterion validity was consistently high across all groups (c-statistics 0.82–0.89). We observed consistent relationships across racial/ethnic groups between HL and outcomes, including communication, adherence, hypoglycemia, diabetes control, and ED utilization. While concerns have arisen regarding bias in AI, the automated Literacy Profile appears sufficiently valid across race/ethnicity, enabling HL measurement at a scale that could improve clinical care and population health among diverse populations.

Objective In the National Library of Medicine funded ECLIPPSE Project (Employing Computational Linguistics to Improve Patient-Provider Secure Emails exchange), we attempted to create novel, valid, and scalable measures of both patients’ health literacy (HL) and physicians’ linguistic complexity by employing natural language processing (NLP) techniques and machine learning (ML). We applied these techniques to > 400,000 patients’ and physicians’ secure messages (SMs) exchanged via an electronic patient portal, developing and validating an automated patient literacy profile (LP) and physician complexity profile (CP). Herein, we describe the challenges faced and the solutions implemented during this innovative endeavor. Materials and methods To describe challenges and solutions, we used two data sources: study documents and interviews with study investigators. Over the five years of the project, the team tracked their research process using a combination of Google Docs tools and an online team organization, tracking, and management tool (Asana). In year 5, the team convened a number of times to discuss, categorize, and code primary challenges and solutions. Results We identified 23 challenges and associated approaches that emerged from three overarching process domains: (1) Data Mining related to the SM corpus; (2) Analyses using NLP indices on the SM corpus; and (3) Interdisciplinary Collaboration. With respect to Data Mining, problems included cleaning SMs to enable analyses, removing hidden caregiver proxies (e.g., other family members) and Spanish language SMs, and culling SMs to ensure that only patients’ primary care physicians were included. With respect to Analyses, critical decisions needed to be made as to which computational linguistic indices and ML approaches should be selected; how to enable the NLP-based linguistic indices tools to run smoothly and to extract meaningful data from a large corpus of medical text; and how to best assess content and predictive validities of both the LP and the CP. With respect to the Interdisciplinary Collaboration, because the research required engagement between clinicians, health services researchers, biomedical informaticians, linguists, and cognitive scientists, continual effort was needed to identify and reconcile differences in scientific terminologies and resolve confusion; arrive at common understanding of tasks that needed to be completed and priorities therein; reach compromises regarding what represents “meaningful findings” in health services vs. cognitive science research; and address constraints regarding potential transportability of the final LP and CP to different health care settings. Discussion Our study represents a process evaluation of an innovative research initiative to harness “big linguistic data” to estimate patient HL and physician linguistic complexity. Any of the challenges we identified, if left unaddressed, would have either rendered impossible the effort to generate LPs and CPs, or invalidated analytic results related to the LPs and CPs. Investigators undertaking similar research in HL or using computational linguistic methods to assess patient-clinician exchange will face similar challenges and may find our solutions helpful when designing and executing their health communications research.

Introduction: Studies evaluating the relationship of severe hypoglycemia (SH) with the risk of atherosclerotic cardiovascular disease (ASCVD) events are limited by size, confounding and inconsistent outcome definitions. The aim of this study was to assess the association of SH, as defined by hypoglycemia-related hospital utilization, with risk of ASCVD. Methods: We conducted an observational cohort study of adults with diabetes with or without established ASCVD (as of 1/1/2014). All subjects were members of Kaiser Permanente Northern California, an integrated health care delivery system. Baseline was determined by the date of the first hypoglycemia-related utilization (primary diagnosis of hypoglycemia in the emergency department or principal diagnosis in the hospital) during 1/1/2013 - 12/31/2013. Baseline for those without SH, was a randomly assigned date in 2013. Multivariate Cox proportional hazard models were specified to estimate hazard ratios (HRs) for SH in adjusted models of time to incident ASCVD events. This outcome was defined as a composite of nonfatal myocardial infarction, ischemic stroke, or coronary heart disease death through 12/31/2017. Results: Among 233,696 eligible individuals, mean age was 63.6 years, 47.6% were women, and mean follow-up was 3.8 years. There were 2,179 episodes of SH. Age-sex-race adjusted risk of ASCVD among those with SH versus those without was HR 2.59 (95% CI 2.3-2.9). When additionally adjusted for prior ASCVD, the HR was still more than double, HR 2.32 (95% CI 2.1-2.6). In a fully saturated model with additional adjustments for diabetes type, diabetes duration, HbA1c, GFR, Charlson score, insulin use, sulfonylureas, and quartiles of neighborhood deprivation, SH was associated with a 30% increased risk (HR 1.31, 95% CI 1.16-1.5). The substantial attenuation in risk after more comprehensive adjustment suggests that patients at higher risk of SH are likely at higher risk of ASCVD. Conclusions: In a large, diverse and contemporary cohort of patients with diabetes, SH as defined by hypoglycemia-related hospital utilization is associated with greater risk of ASCVD. Increased awareness in recognizing this risk and vigilance in care for this higher risk population is warranted.

Meta-analyses and reviews of observational cohorts have reported a relationship between severe hypoglycemia (SH) and atherosclerotic cardiovascular disease (ASCVD) events, but estimates of effect vary widely and many studies are limited by size and diversity, ability to adjust for confounders, and inconsistent definitions of exposures and outcomes (1–4). We studied the association of SH events and risk of ASCVD in a large, diverse cohort of patients. This study was approved by the Kaiser Permanente Northern California (KPNC) Institutional Review Board. This observational cohort study included adults (age ≥18 years) with diabetes who were members of KPNC, an integrated health care delivery system, with continuous membership during the 2 years prior to baseline. The outcomes of interest were ASCVD events, defined as a composite of nonfatal myocardial infarction, fatal or nonfatal ischemic stroke, or death due to coronary heart disease. The exposure of interest …

Objective: We examined how diabetes medication initiation after first diagnosis varied by race/ethnicity. Method: We conducted a retrospective cohort study of 66,939 patients with newly diagnosed diabetes using data from a large integrated healthcare system. We classified patients into 12 race/ethnicity groups (see Table) and assessed their associations with diabetes medication initiation within 1, 12 and 18 months after diagnosis. We used a modified Poisson regression to estimate risk ratios adjusted for pre-diagnosis demographics and clinical characteristics. Results: Medications were started in 30%, 43% and 48% of patients within 1, 12 and 18 months, respectively. Within 1 month of diagnosis, Chinese, Filipino, and Native Hawaiian/Pacific islander were less likely to initiate a medication compared to whites. Results were consistent at 12 and 18 months, except that Japanese and blacks/African Americans were less likely to initiate compared to whites and that there was no difference between Native Hawaiian/Pacific islanders and whites at 18 months. Conclusions: We observed persistent racial differences between patients receiving any medication following initial diabetes diagnosis, with the largest differences for several Asian populations. Disclosure A.N. Winn: None. A.J. Karter: Research Support; Self; Dexcom, Inc. E.M. Staab: None. J. Liu: None. W. Wan: None. R. Skandari: None. A.C. Knitter: None. H.H. Moffet: None. D.R. Miller: None. M.E. Peek: Research Support; Self; Merck Foundation. E. Huang: None. N. Laiteerapong: None.

IntroductionDiabetic ketoacidosis (DKA) is a serious complication of diabetes. DKA is associated with poorer cognition in children with type 1 diabetes (T1D), but whether this is the case in older adults with T1D is unknown. Given the increasing life expectancy in T1D, understanding the role of DKA on brain health in older adults is crucial.Research design and methodsWe examined the association of DKA with cognitive function in 714 older adults with T1D from the Study of Longevity in Diabetes. Participants self-reported lifetime exposure to DKA resulting in hospitalization; DKA was categorized into 0 hospitalization, 1 hospitalization or ≥2 hospitalizations (recurrent DKA). Global and domain-specific cognition (language, executive function/psychomotor speed, episodic memory and simple attention) were assessed. The association of DKA with cognitive function was evaluated via linear and logistic regression models.ResultsTwenty-eight percent of participants (mean age=67 years; mean age at diagnosis=28 years; average duration of diabetes=39 years) reported a lifetime history of DKA resulting in hospitalization (18.5% single DKA; 9.7% recurrent DKA). In fully adjusted models, those with recurrent DKA had lower global cognitive function (β=−0.13; 95% CI −0.22 to 0.02) and lower scores on the executive function/psychomotor speed domain (β=−0.34; 95% CI −0.51 to 0.17). Individuals with recurrent DKA were also more likely to have the lowest level of cognitive function on the executive function/psychomotor speed domain (defined as 1.5 SD below the population mean; OR=3.26, 95% CI 1.43 to 7.42).ConclusionsAmong 714 older adults with T1D, recurrent DKA was associated with lower global cognitive function, lower scores on the executive function/psychomotor speed domain and 3.3 times greater risk of having the lowest level of cognitive function in our sample on the executive function/psychomotor speed domain. These findings suggest that recurrent DKA may negatively impact the brain health of older patients with T1D and highlight the importance of DKA prevention.

Severe hypoglycemia in patients with diabetes is an adverse drug event that is associated with poor outcomes, such as cardiovascular events, falls, high rehospitalization rates, and increased mortality (1,2). Several strategies, including adjustment of glucose-lowering therapy and glycemic targets, have been recommended in patients with severe hypoglycemia (1,3). However, it is not clear how changes in diabetes management are maintained after severe hypoglycemia among patients with type 2 diabetes. We conducted a retrospective analysis of data from the OptumLabs Data Warehouse, a de-identified administrative claims database of over 200 million individuals enrolled in commercial and Medicare Advantage plans across the U.S. (4). Data were accessed in adherence with the Health Insurance Portability and Accountability Act of 1996, and the Yale School of Medicine Human Investigations Committee deemed the study exempt from institutional review board review. We included commercially insured or Medicare Advantage beneficiaries who were aged 18 years or older, had pharmacologically treated type 2 diabetes, and had a severe hypoglycemic event. Severe hypoglycemia was identified as the first event during an observation window of 1 July 2013 through 30 June 2014, based on primary/principal diagnosis codes for hypoglycemia from an emergency department visit, observation stay, or hospital …

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 HbA1c ≥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 HbA1c 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 HbA1c 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.

Introduction: Increasing incidence of type 1 diabetes (T1D) coupled with increasing life expectancy have resulted in an unprecedented number of older adults living with T1D. However, little is known about the burden of aging and diabetes-related complications in this unique group. We hypothesized that older adults with T1D would have greater diabetes and aging-related burden compared to an age, sex, race/ethnicity, and education-matched group of older adults with type 2 diabetes (T2D). Methods: We compared the following characteristics by diabetes type among older adults (aged ≥60) with T1D (n=805) and T2D (n=249) from the Study of Longevity in Diabetes (SOLID) using chi-squared tests: diabetes history (age of onset, diabetes duration); diabetes-related complications (retinopathy, neuropathy, nephropathy, severe hypo- and hyperglycemia resulting in hospitalization/emergency department utilization), cardiovascular disease (stroke, MI, coronary bypass), and geriatric syndromes (depression, incontinence, memory problems and functional impairment). Results: Average age at diagnosis and duration of diabetes, respectively, were 28 years old and 40 years duration for T1D and 56 years old and 13 years duration for T2D (Table 1). Compared to T2D, participants with T1D were more likely to report history of retinopathy, neuropathy, nephropathy, lifetime hypo- and hyperglycemic events resulting in hospitalization/emergency department utilization, and history of a coronary bypass. By contrast, those with T2D were more likely to be incontinent and have functional impairment. Conclusions: Our results show that diabetes-related complications are more prevalent in those with T1D than in comparable adults with T2D, while certain geriatric syndromes were more prevalent in those with T2D. Older adults with T1D are a growing population with unique diabetes-specific and aging-related considerations. Additional research is needed to understand the interplay of aging and diabetes in this group to inform patient care .

Introduction: Studies have shown poorer cognitive function in children and adolescents with type 1 diabetes (T1D) as compared to non-diabetic peers. However, little is known about cognitive function in older adults with T1D. Hypothesis: We hypothesized that older adults with T1D and type 2 diabetes (T2D) would have greater cognitive impairment than age, sex, race/ethnicity, and education-matched controls without diabetes. Methods: We compared baseline cognitive impairment among older adults (aged ≥60) from the Study of Longevity in Diabetes (SOLID) with T1D (n=771), T2D (=234) and no diabetes (n=253). Cognitive tests assessed three cognitive domains identified via factor analysis (language, executive function, episodic memory). All cognitive test scores were standardized and cognitive impairment was defined as 1.5 SD below the mean. In logistic regression models adjusted for age, sex, education, and race/ethnicity, we examined the association between diabetes status (T1D, T2D or no diabetes) and cognition on each cognitive domain and on global cognition (average of scores on the 3 domains). Results: In adjusted regression models, compared to older adults without diabetes, those with T1D were more likely to have impaired cognitive function on the language (OR=2.13, 95% CI: 1.08, 4.17) and executive function domains (OR=2.66, 95% CI: 1.36, 5.22). No significant differences in global cognitive impairment or impairment on the episodic memory domain were observed for T1D and no significant differences on any domain were observed for T2D. Conclusions: Our findings suggest that older adults with T1D have greater cognitive impairment than their peers without diabetes; findings were specific to the language and executive function domains, with episodic memory being unaffected. No increase in cognitive impairment was observed for older adults with T2D. Additional research is needed to understand the causes and potentially modifiable factors associated with impaired cognition among older adults with T1D.

OBJECTIVE Individuals with type 1 diabetes have experienced an increase in life expectancy, yet it is unknown what level of glycemic control is ideal for maintaining late-life brain health. We investigated the association of long-term glycemic control with dementia in older individuals with type 1 diabetes. RESEARCH DESIGN AND METHODS We followed 3,433 members of a health care system with type 1 diabetes, aged ≥50 years, from 1996 to 2015. Repeated measurements of hemoglobin A1c (HbA1c), dementia diagnoses, and comorbidities were ascertained from health records. Cox proportional hazards models were fit to evaluate the association of time-varying glycemic exposure with dementia, with adjustment for age, sex, race/ethnicity, baseline health conditions, and frequency of HbA1c measurement. RESULTS Over a mean follow-up of 6.3 years, 155 individuals (4.5%) were diagnosed with dementia. Patients with ≥50% of HbA1c measurements at 8–8.9% (64–74 mmol/mol) and ≥9% (≥75 mmol/mol) had 65% and 79% higher risk of dementia, respectively, compared with those with CONCLUSIONS Among older patients with type 1 diabetes, those with majority exposure to HbA1c 8–8.9% and ≥9% had increased dementia risk, while those with majority exposure to HbA1c 6–6.9% and 7–7.9% had reduced risk. Currently recommended glycemic targets for older patients with type 1 diabetes are consistent with healthy brain aging.

OBJECTIVE Both food insecurity (limited food access owing to cost) and living in areas with low physical access to nutritious foods are public health concerns, but their relative contribution to diabetes management is poorly understood. RESEARCH DESIGN AND METHODS This was a prospective cohort study. A random sample of patients with diabetes in a primary care network completed food insecurity assessment in 2013. Low physical food access at the census tract level was defined as no supermarket within 1 mile in urban areas and 10 miles in rural areas. HbA1c measurements were obtained from electronic health records through November 2016. The relationship among food insecurity, low physical food access, and glycemic control (as defined by HbA1c) was analyzed using hierarchical linear mixed models. RESULTS Three hundred and ninety-one participants were followed for a mean of 37 months. Twenty percent of respondents reported food insecurity, and 31% resided in an area of low physical food access. In adjusted models, food insecurity was associated with higher HbA1c (difference of 0.6% [6.6 mmol/mol], 95% CI 0.4–0.8 [4.4–8.7], P < 0.0001), which did not improve over time (P = 0.50). Living in an area with low physical food access was not associated with a difference in HbA1c (difference 0.2% [2.2 mmol/mol], 95% CI −0.2 to 0.5 [−2.2 to 5.6], P = 0.33) or with change over time (P = 0.07). CONCLUSIONS Food insecurity is associated with higher HbA1c, but living in an area with low physical food access is not. Food insecurity screening and interventions may help improve glycemic control for vulnerable patients.

OBJECTIVE Severe hyperglycemia and hypoglycemia (“severe dysglycemia”) are serious complications of type 1 diabetes (T1D). Depression has been associated with severe dysglycemia in type 2 diabetes but has not been thoroughly examined specifically in T1D. We evaluated bidirectional associations between depression and severe dysglycemia among older people with T1D. RESEARCH DESIGN AND METHODS We abstracted depression and severe dysglycemia requiring emergency room visit or hospitalization from medical health records in 3,742 patients with T1D during the study period (1996–2015). Cox proportional hazards models estimated the associations between depression and severe dysglycemia in both directions, adjusting for demographics, micro- and macrovascular complications, and HbA1c. RESULTS During the study period, 41% had depression and 376 (11%) and 641 (20%) had hyperglycemia and hypoglycemia, respectively. Depression was strongly associated with a 2.5-fold increased risk of severe hyperglycemic events (hazard ratio [HR] 2.47 [95% CI 2.00, 3.05]) and 89% increased risk of severe hypoglycemic events (HR 1.89 [95% CI 1.61, 2.22]). The association was strongest within the first 6 months (HRhyperglycemia 7.14 [95% CI 5.29, 9.63]; HRhypoglycemia 5.58 [95% CI 4.46, 6.99]) to 1 year (HRhyperglycemia 5.16 [95% CI 3.88, 6.88]; HRhypoglycemia 4.05 [95% CI 3.26, 5.04]) after depression diagnosis. In models specifying severe dysglycemia as the exposure, hyperglycemic and hypoglycemic events were associated with 143% (HR 2.43 [95% CI 2.03, 2.91]) and 74% (HR 1.75 [95% CI 1.49, 2.05]) increased risk of depression, respectively. CONCLUSIONS Depression and severe dysglycemia are associated bidirectionally among patients with T1D. Depression greatly increases the risk of severe hypoglycemic and hyperglycemic events, particularly in the first 6 months to 1 year after diagnosis, and depression risk increases after severe dysglycemia episodes.

Effective management of diabetes during the coronavirus disease 2019 (COVID-19) pandemic is critical given that hyperglycemia is both a risk factor for infection with the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and has been associated with poorer outcomes for those infected (1). Unfortunately, inadequate diabetes self-management due to medication nonadherence is prevalent among patients with diabetes and is a major contributor to poor diabetes outcomes (2,3). Research suggests that mail order pharmacy (MOP) use can improve medication adherence (3). However, there is little data on how the COVID-19 pandemic is related to MOP use. Regrettably, racial/ethnic minority patients with diabetes have poorer medication adherence than White patients, and research shows marked disparities among racial/ethnic minorities in SARS-CoV-2 infection rates and outcomes, but little is known about racial/ethnic differences in MOP use during the COVID-19 pandemic (3,4). The objectives of our study were to 1 ) assess the association between the COVID-19 pandemic shelter-in-place order and MOP use and 2 ) examine these associations by patient race/ethnicity. We conducted a retrospective cohort study using a cohort from a randomized encouragement trial conducted in 2017–2018 at Kaiser Permanente Northern California (KPNC), the Encouraging Mail Order Pharmacy Use to Improve Outcomes and Reduce Disparities …

Importance Continuous glucose monitoring (CGM) is recommended for patients with type 1 diabetes; observational evidence for CGM in patients with insulin-treated type 2 diabetes is lacking. Objective To estimate clinical outcomes of real-time CGM initiation. Design, Setting, and Participants Exploratory retrospective cohort study of changes in outcomes associated with real-time CGM initiation, estimated using a difference-in-differences analysis. A total of 41 753 participants with insulin-treated diabetes (5673 type 1; 36 080 type 2) receiving care from a Northern California integrated health care delivery system (2014-2019), being treated with insulin, self-monitoring their blood glucose levels, and having no prior CGM use were included. Exposures Initiation vs noninitiation of real-time CGM (reference group). Main Outcomes and Measures Ten end points measured during the 12 months before and 12 months after baseline: hemoglobin A1c(HbA1c); hypoglycemia (emergency department or hospital utilization); hyperglycemia (emergency department or hospital utilization); HbA1clevels lower than 7%, lower than 8%, and higher than 9%; 1 emergency department encounter or more for any reason; 1 hospitalization or more for any reason; and number of outpatient visits and telephone visits. Results The real-time CGM initiators included 3806 patients (mean age, 42.4 years [SD, 19.9 years]; 51% female; 91% type 1, 9% type 2); the noninitiators included 37 947 patients (mean age, 63.4 years [SD, 13.4 years]; 49% female; 6% type 1, 94% type 2). The prebaseline mean HbA1cwas lower among real-time CGM initiators than among noninitiators, but real-time CGM initiators had higher prebaseline rates of hypoglycemia and hyperglycemia. Mean HbA1cdeclined among real-time CGM initiators from 8.17% to 7.76% and from 8.28% to 8.19% among noninitiators (adjusted difference-in-differences estimate, −0.40%; 95% CI, −0.48% to −0.32%;P Conclusions and Relevance In this retrospective cohort study, insulin-treated patients with diabetes selected by physicians for real-time continuous glucose monitoring compared with noninitiators had significant improvements in hemoglobin A1cand reductions in emergency department visits and hospitalizations for hypoglycemia, but no significant change in emergency department visits or hospitalizations for hyperglycemia or for any reason. Because of the observational study design, findings may have been susceptible to selection bias.

Objective To estimate the effect of out-of-pocket (OOP) cost on nonadherence to classes of cardiometabolic medications among patients with diabetes. Data Sources/Setting Electronic health records from a large, health care delivery system for 223,730 patients with diabetes prescribed 842,899 new cardiometabolic medications during 2006–2012. Study Design Observational, new prescription cohort study of the effect of OOP cost on medication initiation and adherence. Data Collection Adherence and OOP costs were based on pharmacy dispensing records and benefits. Principal Findings Primary nonadherence (never dispensed) increased monotonically with OOP cost after adjusting for demographics, neighborhood socioeconomic status, Medicare, medical financial assistance, OOP maximum, deductibles, mail order pharmacy incentive and use, drug type, generic or brand, day's supply, and comorbidity index; 7 percent were never dispensed the new medication when OOP cost ≥$11, 5 percent with OOP cost of $1–$10, and 3 percent when the medication was free of charge (p

Seven-year surveillance study (2005-2011) to evaluate race/ethnic differences in the trends in rates of severe hypoglycemia (SH) in a population of insured, at-risk adults with diabetes.SH events were identified via any primary or principal diagnosis from emergency department or inpatient encounters among African American, Asian, Latino and White adult diabetes patients treated with insulin or secretagogues (Sulfonylureas or Meglitinides), receiving care from integrated healthcare delivery systems across the United States. We calculated age- and sex-standardized annual SH rates and average annual percent change (AAPC) in SH rates.Annual SH rates ranged from 1.8% to 2.1% during this 7-year observation period (2,200,471 person-years). African Americans had consistently higher SH rates compared with Whites, while Latinos and Asians had consistently lower rates compared with Whites in each of the 7 years (all p0.01). The trend increased significantly only among African Americans (AAPC = +4.3%; 95% CI: +2.1, +6.5%); in the other groups, the AAPC was not significantly different from zero.Surveillance efforts should monitor the racial/ethnic-specific rates. The factors underlying substantially higher rates of hypoglycemia in African Americans should be evaluated. Clinically and culturally-appropriate strategies to reduce the risk of SH need to be developed and tested.

The housing foreclosure crisis was harmful to the financial well-being of many households. In the present study, we investigated the health effects of the housing foreclosure crisis on glycemic control within a population of patients with diabetes. We hypothesized that an increase in the neighborhood foreclosure rate could worsen glycemic control by activating stressors such as higher neighborhood crime, lower housing prices, and erosion of neighborhood social cohesion. To test this, we linked public foreclosure records at the census-block level with clinical records from 2006 to 2009 of patients with diabetes. We specified individual fixed-effects models and controlled for individual time-invariant confounders and area-level time-varying confounders, including housing prices and unemployment rate, to estimate the effect of the foreclosure rate per census-block group on glycated hemoglobin. We found no statistically significant relationship between changes in the neighborhood foreclosure rate per block group in the prior year and changes in glycated hemoglobin. There is no evidence that increased foreclosure rates worsened glycemic control in this continuously insured population with diabetes. More research is needed to inform our knowledge of the role of insurance and health-care delivery systems in protecting the health of diabetic patients during times of economic stress.

Aims To classify trajectories of long term HbA1c values in patients after diagnosis of type 2 diabetes and examine each trajectory's associations with subsequent microvascular and macrovascular events and mortality. Methods A longitudinal follow-up of 28,016 patients newly diagnosed with type 2 diabetes was conducted. Latent growth mixture modeling was used to identify ten-year HbA1c trajectories. Cox proportional hazards models were used to assess how HbA1c trajectories were associated with events (microvascular and macrovascular) and mortality. Results We identified 5 HbA1c trajectories: “low stable” (82.5%), “moderate increasing late” (5.1%), “high decreasing early” (4.9%), “moderate peaking late” (4.1%) and “moderate peaking early” (3.3%). After adjusting for average HbA1c, compared to the low stable trajectory, all non-stable trajectories were associated with higher incidences of microvascular events (hazard ratio (HR) range, 1.28 (95% CI, 1.08–1.53) (high decreasing early) to 1.45 (95% CI, 1.20–1.75) (moderate peaking early)). The high decreasing early trajectory was associated with an increased mortality risk (HR, 1.27 (95% CI, 1.03–1.58)). Trajectories were not associated with macrovascular events. Conclusions Non-stable HbA1c trajectories were associated with greater risk of microvascular events and mortality. These findings suggest a potential benefit of early diabetes detection, prioritizing good glycemic control, and maintaining control over time.

Medication non-adherence is a major contributor to poor diabetes outcomes. Observational studies show positive associations between mail order pharmacy (MOP) use and adherence, and suggest that mail order is cost-saving to patients. However, no prospective trials have examined the effect of encouraging MOP use on medication adherence. We conducted a randomized encouragement trial in 2017 in three health plans: Kaiser Permanente (KP) Northern California, KP Hawaii, and Harvard Pilgrim Health Care (HPHC). Patients were eligible if they had diabetes, poor adherence (Proportion of Days Covered Disclosure J. Schmittdiel: None. C.M. Trinacty: None. J. Wharam: None. O. Duru: Advisory Panel; Self; Healthwise. R. Neugebauer: None. S.D. Brown: None. D. Ross-Degnan: None. A.J. Karter: None. W.T. Dyer: None. B. Ramachandran: None. Funding National Institute of Diabetes and Digestive and Kidney Diseases