Diet and Food and Nutrition Insecurity and Cardiometabolic Disease.

Poor nutrition is the leading cause of poor health, health care spending, and lost productivity in the United States and globally, which acts through cardiometabolic diseases as precursors to cardiovascular disease, cancer, and other conditions. There is great interest in how the social determinants of health (the conditions in which people are born, live, work, develop, and age) impact cardiometabolic disease. Food insecurity is an example of a powerful social determinant of health that impacts health outcomes. Nutrition insecurity, a distinct but related concept to food insecurity, is a direct determinant of health. In this article, we provide an overview of how diet in early life relates to cardiometabolic disease and then continue to focus on the concepts of food insecurity and nutrition insecurity. In the discussions herein we make important distinctions between the concepts of food insecurity and nutrition insecurity and provide a review of their concepts, histories, measurement and assessment devices, trends and prevalence, and links to health and health disparities. The discussions here set the stage for future research and practice to directly address the negative consequences of food and nutrition insecurity.

Food Is Medicine: A Presidential Advisory From the American Heart Association.

Unhealthy diets are a major impediment to achieving a healthier population in the United States. Although there is a relatively clear sense of what constitutes a healthy diet, most of the US population does not eat healthy food at rates consistent with the recommended clinical guidelines. An abundance of barriers, including food and nutrition insecurity, how food is marketed and advertised, access to and affordability of healthy foods, and behavioral challenges such as a focus on immediate versus delayed gratification, stand in the way of healthier dietary patterns for many Americans. Food Is Medicine may be defined as the provision of healthy food resources to prevent, manage, or treat specific clinical conditions in coordination with the health care sector. Although the field has promise, relatively few studies have been conducted with designs that provide strong evidence of associations between Food Is Medicine interventions and health outcomes or health costs. Much work needs to be done to create a stronger body of evidence that convincingly demonstrates the effectiveness and cost-effectiveness of different types of Food Is Medicine interventions. An estimated 90% of the $4.3 trillion annual cost of health care in the United States is spent on medical care for chronic disease. For many of these diseases, diet is a major risk factor, so even modest improvements in diet could have a significant impact. This presidential advisory offers an overview of the state of the field of Food Is Medicine and a road map for a new research initiative that strategically approaches the outstanding questions in the field while prioritizing a human-centered design approach to achieve high rates of patient engagement and sustained behavior change. This will ideally happen in the context of broader efforts to use a health equity-centered approach to enhance the ways in which our food system and related policies support improvements in health.

Financial Hardship Among Traditional Medicare and Medicare Advantage Enrollees With and Without Food Insecurity.

BACKGROUND: Compared to traditional Medicare (TM), Medicare Advantage (MA) plans impose out-of-pocket cost limits and offer extra benefits, potentially providing financial relief for MA enrollees, especially for those with food insecurity. OBJECTIVE: To examine whether the prevalence of food insecurity differs between TM and MA enrollees at baseline and then examine whether MA enrollment in a baseline year is associated with less financial hardships in the following year, relative to TM enrollment, especially for those experiencing food insecurity. DESIGN: We conducted a retrospective longitudinal cohort study. PARTICIPANTS: Our analysis included 2807 Medicare beneficiaries (weighted sample size, 23,963,947) who maintained continuous enrollment in either TM or MA in both 2020 and 2021 from the Medical Expenditure Panel Survey. MAIN MEASURES: We assessed outcomes related to financial hardships in health care and non-health care domains (measured in 2021). Our primary independent variables were food insecurity and MA enrollment (measured in 2020). RESULTS: The point estimate of food insecurity prevalence was greater among MA enrollees than TM enrollees, but the difference was not statistically significant (1.1 percentage points [95% CI, - 1.0, 3.4]). Furthermore, there is evidence that compared to TM enrollment, MA enrollment did not mitigate the risk of financial hardship, particularly for food-insecure enrollees. Rather, food-secure MA enrollees faced greater financial hardship in the following year than food-secure TM enrollees (11.2% [8.9-13.6] and 7.6% [6.9-8.3] for problems paying medical bills and 5.5% [4.6-6.4] and 2.8% [2.1-3.6] for paying medical bills over time). Moreover, the point estimate of financial hardship was higher among food-insecure MA enrollees than food-insecure TM enrollees (21.5% [5.4-37.5] and 11.2% [4.1-18.4] and 23.7% [9.6-37.9] and 6.9% [0.5-13.3]) despite the lack of statistical significance. CONCLUSION: These findings suggest that the promise of financial protection offered by MA plans has not been fully realized, particularly for those with food insecurity.

Association of Patient-Reported Social Needs with Emergency Department Visits and Hospitalizations Among Federally Qualified Health Center Patients.

BACKGROUND: Health care systems are increasingly screening for unmet social needs. The association between patient-reported social needs and health care utilization is not well understood. OBJECTIVE: To investigate the association between patient-reported social needs, measured by the Protocol for Responding to and Assessing Patients' Assets, Risks, and Experiences (PRAPARE), and inpatient and emergency department (ED) utilization. DESIGN: This cohort study analyzed merged 2017-2019 electronic health record (EHR) data across multiple health systems. PARTICIPANTS: Adult patients from a federally qualified health center (FQHC) in central North Carolina who completed PRAPARE as part of a primary care visit with behavioral health services. MAIN MEASURES: The count of up to 12 unmet social needs, aggregated as 0, 1, 2, or 3 + . Outcomes include the probability of an ED visit and hospitalization 12 months after PRAPARE assessment, modeled by logistic regressions controlling for age, sex, race, ethnicity, comorbidity burden, being uninsured, and prior utilization in the past 12 months. KEY RESULTS: The study population consisted of 1924 adults (38.7% male, 50.1% Black, 36.3% Hispanic, 55.9% unemployed, 68.2% of patients reported 1 + needs). Those with more needs were younger, more likely to be unemployed, and experienced greater comorbidity burden. 35.3% of patients had ED visit(s) and 36.3% had hospitalization(s) 1 year after PRAPARE assessment. In adjusted analysis, having 3 + needs was associated with a percentage point increase in the predicted probability of hospitalization (average marginal effect 0.06, SE 0.03, p < 0.05) compared with having 0 needs. Similarly, having 2 needs (0.07, SE 0.03, p < 0.05) or 3 + needs (0.06, SE 0.03, p < 0.05) was associated with increased probability of ED visits compared to 0 needs. CONCLUSIONS: Patient-reported social needs were common and associated with health care utilization patterns. Future research should identify interventions to address unmet social needs to improve health and avoid potentially preventable escalating medical intervention.

Multisector Collaboration vs. Social Democracy for Addressing Social Determinants of Health.

Policy Points Multisector collaboration, the dominant approach for responding to health harms created by adverse social conditions, involves collaboration among health care insurers, health care systems, and social services organizations. Social democracy, an underused alternative, seeks to use government policy to shape the civil (e.g., civil rights), political (e.g., voting rights), and economic (e.g., labor market institutions, property rights, and the tax-and-transfer system) institutions that produce health. Multisector collaboration may not achieve its goals, both because the collaborations are difficult to accomplish and because it does not seek to transform social conditions, only to mitigate their harms. Social democracy requires political contestation but has greater potential to improve population health and health equity.

"We don't separate out these things. Everything is related": Partnerships with Indigenous Communities to Design, Implement, and Evaluate Multilevel Interventions to Reduce Health Disparities.

Multilevel interventions (MLIs) are appropriate to reduce health disparities among Indigenous peoples because of their ability to address these communities' diverse histories, dynamics, cultures, politics, and environments. Intervention science has highlighted the importance of context-sensitive MLIs in Indigenous communities that can prioritize Indigenous and local knowledge systems and emphasize the collective versus the individual. This paradigm shift away from individual-level focus interventions to community-level focus interventions underscores the need for community engagement and diverse partnerships in MLI design, implementation, and evaluation. In this paper, we discuss three case studies addressing how Indigenous partners collaborated with researchers in each stage of the design, implementation, and evaluation of MLIs to reduce health disparities impacting their communities. We highlight the following: (1) collaborations with multiple, diverse tribal partners to carry out MLIs which require iterative, consistent conversations over time; (2) inclusion of qualitative and Indigenous research methods in MLIs as a way to honor Indigenous and local knowledge systems as well as a way to understand a health disparity phenomenon in a community; and (3) relationship building, maintenance, and mutual respect among MLI partners to reconcile past research abuses, prevent extractive research practices, decolonize research processes, and generate co-created knowledge between Indigenous and academic communities.

Benefits, Challenges, and Opportunities in Addressing Medicaid Beneficiaries' Non-Medical Risks to Health: A Qualitative Analysis.

CONTEXT: North Carolina's Healthy Opportunities Pilots (HOP) is a Medicaid 1115 Waiver program that seeks to address nonmedical risks to health for Medicaid beneficiaries through multisector collaboration. Among other stakeholders, HOP involves collaboration between human services organizations that deliver interventions, network leads, which establish and oversee the human services organizations within a region of the state. OBJECTIVE: To understand how employees at human services organizations and network leads prepared to deliver HOP services. DESIGN: Qualitative analysis of semistructured interviews. Interviews were conducted between April and June 2022. Interviews were recorded, transcribed verbatim, coded thematically, and analyzed using a conceptual model derived from the consolidated framework for implementation research. SETTING: Organizations within North Carolina counties participating in HOP. PARTICIPANTS: Employees of human services and network lead organizations across all 3 HOP regions of North Carolina. RESULTS: The researchers interviewed 37 participants. Overall, organizations experienced benefits from HOP participation, including capacity-building resources, flexibility in allocating resources, and creating community-wide enthusiasm for addressing nonmedical risks to health. There were also key challenges. These included the time needed to build capacity, adjustments to the work processes and regulations inherent to multisector collaboration, geographic variation in availability of services to offer, and the difficulty of addressing different needs. Finally, participants recognized substantial opportunities that HOP presented, including membership in a more extensive network, exposure to a learning community, and a more sustainable funding source. CONCLUSIONS: The perspectives of individuals preparing to deliver HOP services offer important lessons for those developing and implementing large-scale programs that can address nonmedical threats to health.

Eliminating Food Insecurity in the USA: a Target Trial Emulation Using Observational Data to Estimate Effects on Health-Related Quality of Life.

BACKGROUND: Food insecurity is associated with many aspects of poor health. However, trials of food insecurity interventions typically focus on outcomes of interest to funders, such as healthcare use, cost, or clinical performance metrics, rather than quality of life outcomes that may be prioritized by individuals who experience food insecurity. OBJECTIVE: To emulate a trial of a food insecurity elimination intervention, and quantify its estimated effects on health utility, health-related quality of life, and mental health. DESIGN: Target trial emulation using longitudinal, nationally representative data, from the USA, 2016-2017. PARTICIPANTS: A total of 2013 adults in the Medical Expenditure Panel Survey screened positive for food insecurity, representing 32 million individuals. MAIN MEASURES: Food insecurity was assessed using the Adult Food Security Survey Module. The primary outcome was the SF-6D (Short-Form Six Dimension) measure of health utility. Secondary outcomes were mental component score (MCS) and physical component score (PCS) of the Veterans RAND 12-Item Health Survey (a measure of health-related quality of life), Kessler 6 (K6) psychological distress, and Patient Health Questionnaire 2-item (PHQ2) depressive symptoms. KEY RESULTS: We estimated that food insecurity elimination would improve health utility by 80 QALYs per 100,000 person-years, or 0.008 QALYs per person per year (95% CI 0.002 to 0.014, p = 0.005), relative to the status quo. We also estimated that food insecurity elimination would improve mental health (difference in MCS [95% CI]: 0.55 [0.14 to 0.96]), physical health (difference in PCS: 0.44 [0.06 to 0.82]), psychological distress (difference in K6: -0.30 [-0.51 to -0.09]), and depressive symptoms (difference in PHQ-2: -0.13 [-0.20 to -0.07]). CONCLUSIONS: Food insecurity elimination may improve important, but understudied, aspects of health. Evaluations of food insecurity interventions should holistically investigate their potential to improve many different aspects of health.

Changes in Food Insecurity and Changes in Patient-Reported Outcomes: a Nationally Representative Cohort Study.

BACKGROUND: Cross-sectional studies have found that health-related quality of life and mental health are worse among food-insecure compared with food-secure individuals. However, how these outcomes change as food insecurity changes is unclear. OBJECTIVE: To evaluate how common patient-reported health-related quality of life and mental health scales change in response to changes in food security. DESIGN: Retrospective cohort study using data representative of the civilian, adult, non-institutionalized population of the USA. PARTICIPANTS: Food insecure adults who completed the 2016-2017 Medical Expenditure Panel Survey. MAIN MEASURES: Mental health, as measured by the mental component score of the Veterans Rand 12-Item Health Survey (VR-12) (primary outcome), along physical health (physical component score of the VR-12), self-rated health status, psychological distress (Kessler 6), depressive symptoms (PHQ2), and the SF-6D measure of health utility. We fit linear regression models adjusted for baseline outcome level, age, gender, race/ethnicity, education, health insurance, and family size followed by predictive margins to estimate the change in outcome associated with a 1-point improvement in food security. KEY RESULTS: A total of 1,390 food-insecure adults were included. A 1-point improvement in food security was associated with a 0.38 (95%CI 0.62 to 0.14)-point improvement in mental health, a 0.15 (95%CI 0.02 to 0.27)-point improvement in psychological distress, a 0.05 (95%CI 0.01 to 0.09)-point improvement in depressive symptoms, and a 0.003 (95%CI 0.000 to 0.007)-point improvement in health utility. Point estimates for physical health and self-rated health were in the direction of improvement, but were not statistically significant. CONCLUSIONS: Improvement in food insecurity was associated with improvement in several patient-reported outcomes. Further work should investigate whether similar changes are seen in food insecurity interventions, and the most useful scales for assessing changes in health-related quality of life and mental health in food insecurity interventions.

The Logic of Policies to Address Income-Related Health Inequity: A Problem-Oriented Approach.

Policy Points  Income is a fundamental cause of health across the life course. To address income-related health inequities, we need a set of overlapping and complementary policy approaches rather than focusing on a single policy.  During their lives, individuals inhabit different roles with regard to their ability to earn wages, and at any given time, only about 50% of the US population are expected to earn wages, while the rest (e.g., children, older adults, those who are disabled, unemployed, students, and/or caregivers) are not.  Three key "branch points" for designing policy approaches to address income-related health inequity are (1) should the needed good or service be obtained on the market? (2) do policy beneficiaries currently earn income? and (3) have policy beneficiaries earned income previously? The responses to these questions suggest one of four policy approaches: social services, social enfranchisement, social insurance, or social assistance.

Supplemental Nutrition Assistance Program Participation and Health Care Use in Older Adults : A Cohort Study.

BACKGROUND: Older adults dually eligible for Medicare and Medicaid have particularly high food insecurity prevalence and health care use. OBJECTIVE: To determine whether participation in the Supplemental Nutrition Assistance Program (SNAP), which reduces food insecurity, is associated with lower health care use and cost for older adults dually eligible for Medicare and Medicaid. DESIGN: An incident user retrospective cohort study design was used. The association between participation in SNAP and health care use and cost using outcome regression was assessed and supplemented by entropy balancing, matching, and instrumental variable analyses. SETTING: North Carolina, September 2016 through July 2020. PARTICIPANTS: Older adults (aged ≥65 years) dually enrolled in Medicare and Medicaid but not initially enrolled in SNAP. MEASUREMENTS: Inpatient admissions (primary outcome), emergency department visits, long-term care admissions, and Medicaid expenditures. RESULTS: Of 115 868 persons included, 5093 (4.4%) enrolled in SNAP. Mean follow-up was approximately 22 months. In outcome regression analyses, SNAP enrollment was associated with fewer inpatient hospitalizations (-24.6 [95% CI, -40.6 to -8.7]), emergency department visits (-192.7 [CI, -231.1 to -154.4]), and long-term care admissions (-65.2 [CI, -77.5 to -52.9]) per 1000 person-years as well as fewer dollars in Medicaid payments per person per year (-$2360 [CI, -$2649 to -$2071]). Results were similar in entropy balancing, matching, and instrumental variable analyses. LIMITATION: Single state, no Medicare claims data available, and possible residual confounding. CONCLUSION: Participation in SNAP was associated with fewer inpatient admissions and lower health care costs for older adults dually eligible for Medicare and Medicaid. PRIMARY FUNDING SOURCE: National Institutes of Health.

Estimated Effect on Life Expectancy of Alleviating Primary Care Shortages in the United States.

BACKGROUND: Prior studies have reported that greater numbers of primary care physicians (PCPs) per population are associated with reduced population mortality, but the effect of increasing PCP density in areas of low density is poorly understood. OBJECTIVE: To estimate how alleviating PCP shortages might change life expectancy and mortality. DESIGN: Generalized additive models, mixed-effects models, and generalized estimating equations. SETTING: 3104 U.S. counties from 2010 to 2017. PARTICIPANTS: Children and adults. MEASUREMENTS: Age-adjusted life expectancy; all-cause mortality; and mortality due to cardiovascular disease, cancer, infectious disease, respiratory disease, and substance use or injury. RESULTS: Persons living in counties with less than 1 physician per 3500 persons in 2017 had a mean life expectancy that was 310.9 days shorter than for persons living in counties above that threshold. In the low-density counties (n = 1218), increasing the density of PCPs above the 1:3500 threshold would be expected to increase mean life expectancy by 22.4 days (median, 19.4 days [95% CI, 0.9 to 45.6 days]), and all such counties would require 17 651 more physicians, or about 14.5 more physicians per shortage county. If counties with less than 1 physician per 1500 persons (n = 2636) were to reach the 1:1500 threshold, life expectancy would be expected to increase by 56.3 days (median, 55.6 days [CI, 4.2 to 105.6 days]), and all such counties would require 95 754 more physicians, or about 36.3 more physicians per shortage county. LIMITATION: Some projections are based on extrapolations of the actual data. CONCLUSION: In counties with fewer PCPs per population, increases in PCP density would be expected to substantially improve life expectancy. PRIMARY FUNDING SOURCE: None.

Examining Changes in Food Security, Perceived Stress, and Dietary Intake in a Cohort of Low-Wage Workers Experiencing an Increase in Hourly Wage.

OBJECTIVE: To determine whether an increase in hourly wages was associated with changes in food security and perceived stress among low-wage workers. We also determined whether changes in food security and stress were associated with changes in diet. SETTING: Wages is a prospective cohort study following 974 low-wage workers in Minneapolis, MN, where an ordinance is incrementally increasing minimum wage to US$15/hr from 2018 to 2022, and a comparison community with no minimum wage ordinance (Raleigh, NC). Interaction models were estimated using generalized estimating equations. PARTICIPANTS: Analyses used two waves of data (2018 [baseline], 2019) and included 219 and 321 low-wage workers in Minneapolis and Raleigh (respectively). RESULTS: Average hourly wages increased from US$9.77 (SD US$1.69) to US$11.67 (SD US$4.02). Changes in wages were not associated with changes in food security (odds ratio = 1.05, 95% confidence interval [CI] [0.89, 1.23], p = .57) or stress (ß = -0.01, 95% CI [-0.04, 0.03], p = .70) after 1 year of policy implementation. Changes in food security were not associated with changes in diet. However, we found significant changes in the frequency of fruit and vegetable intake across time by levels of stress, with decreased intake from Wave 1 to 2 at low levels of stress, and increased intake at high levels of stress (incidence rate ratio = 1.17, 95% CI [1.05, 1.31], p = .01). CONCLUSIONS: Changes in wages were not associated with changes in food security or stress in a sample of low-wage workers. Future research should examine whether full implementation of a minimum wage increase is associated with changes in these outcomes.

North Carolina's Healthy Opportunities Pilots Focus Attention on Housing as a Health Care Intervention.

The social conditions in which people live, including their housing, can have a profound impact on their health. The North Carolina Healthy Opportunities Pilots aim to address housing problems and other social needs to improve health. The results from the Pilots' evaluation will have important implications for health policy and practice.

Association between Food Insecurity and Health-Related Quality of Life: a Nationally Representative Survey.

BACKGROUND: Food insecurity, limited or uncertain access to enough food for an active, healthy life, affected over 37 million Americans in 2018. Food insecurity is likely to be associated with worse health-related quality of life (HRQoL), but this association has not been measured with validated instruments in nationally representative samples. Given growing interest understanding food insecurity's role in health outcomes, it would be useful to learn what HRQoL measures best capture the experience of those with food insecurity. OBJECTIVE: To determine the association between food insecurity and several validated HRQoL instruments in US adults. DESIGN: Cross-sectional. PARTICIPANTS: US adults (age ≥ 18), weighted to be nationally representative. MAIN MEASURES: Food insecurity was assessed with three items derived from the USDA Household Food Security Survey Module. HRQoL was assessed using PROMIS-Preference (PROPr), which contains 7 PROMIS domains, self-rated health (SRH), Euroqol-5D-5L (EQ-5D), Health Utilities Index (HUI), and Short Form-6D (SF-6D). KEY RESULTS: In December 2017, 4142 individuals completed at least part of the survey (31% response rate), of whom 4060 (98.0%) reported food security information. Of survey respondents, 51.7% were women, 12.5% self-identified as black, 15.8% were Hispanic, and 11.0% did not have a high school diploma. 14.1% of respondents reported food insecurity. In adjusted analyses, food insecurity was associated with worse HRQoL across all instruments and PROMIS domains (p < .0001 for all). The magnitude of the difference between food-insecure and food-secure participants was largest with the SF-6D, EQ-5D, and PROPr; among individual PROMIS domain scores, the largest difference was for ability to participate in social roles. CONCLUSIONS: Food insecurity is strongly associated with worse HRQoL, with differences between food-secure and food-insecure individuals best measured using the SF-6D, EQ-5D, and PROPr. Future work should develop a specific instrument to measure changes in HRQoL in food insecurity interventions.

Transporting experimental results with entropy balancing.

We show how entropy balancing can be used for transporting experimental treatment effects from a trial population onto a target population. This method is doubly robust in the sense that if either the outcome model or the probability of trial participation is correctly specified, then the estimate of the target population average treatment effect is consistent. Furthermore, we only require the sample moments of the effect modifiers drawn from the target population to consistently estimate the target population average treatment effect. We compared the finite-sample performance of entropy balancing with several alternative methods for transporting treatment effects between populations. Entropy balancing techniques are efficient and robust to violations of model misspecification. We also examine the results of our proposed method in an applied analysis of the Action to Control Cardiovascular Risk in Diabetes Blood Pressure trial transported to a sample of US adults with diabetes taken from the National Health and Nutrition Examination Survey cohort.

Association between hourly wages and dietary intake after the first phase of implementation of the Minneapolis minimum wage ordinance.

OBJECTIVE: In 2018, Minneapolis began phased implementation of an ordinance to increase the local minimum wage to $15/h. We sought to determine whether the first phase of implementation was associated with changes in frequency of consumption of fruits and vegetables (F&V), whole-grain-rich foods, and foods high in added sugars among low-wage workers. DESIGN: Natural experiment. SETTING: The Wages Study is a prospective cohort study of 974 low-wage workers followed throughout the phased implementation of the ordinance (2018-2022). We used difference-in-difference analysis to compare outcomes among workers in Minneapolis, Minnesota, to those in a comparison city (Raleigh, North Carolina). We assessed wages using participants' pay stubs and dietary intake using the National Cancer Institute Dietary Screener Questionnaire. PARTICIPANTS: Analyses use the first two waves of Wages data (2018 (baseline), 2019) and includes 267 and 336 low-wage workers in Minneapolis and Raleigh, respectively. RESULTS: After the first phase of implementation, wages increased in both cities, but the increase was $0·84 greater in Minneapolis (P = 0·02). However, the first phase of the policy's implementation was not associated with changes in daily frequency of consumption of F&V (IRR = 1·03, 95 % CI: 0·86, 1·24, P = 0·73), whole-grain-rich foods (IRR = 1·23, 95 % CI: 0·89, 1·70, P = 0·20), or foods high in added sugars (IRR = 1·13, 95 % CI: 0·86, 1·47, P = 0·38) among workers in Minneapolis compared to Raleigh. CONCLUSIONS: The first phase of implementation of the Minneapolis minimum wage policy was associated with increased wages, but not with changes in dietary intake. Future research should examine whether full implementation is associated dietary changes.

Association of Glycemic Control Trajectory with Short-Term Mortality in Diabetes Patients with High Cardiovascular Risk: a Joint Latent Class Modeling Study.

BACKGROUND: The relationship between risk factor or biomarker trajectories and contemporaneous short-term clinical outcomes is poorly understood. In diabetes patients, it is unknown whether hemoglobin A1c (HbA1c) trajectories are associated with clinical outcomes and can inform care in scenarios in which a single HbA1c is uninformative, for example, after a diagnosis of coronary artery disease (CAD). OBJECTIVE: To compare associations of HbA1c trajectories and single HbA1c values with short-term mortality in diabetes patients evaluated for CAD DESIGN: Retrospective observational cohort study PARTICIPANTS: Diabetes patients (n = 7780) with and without angiographically defined CAD MAIN MEASURES: We used joint latent class mixed models to simultaneously fit HbA1c trajectories and estimate association with 2-year mortality after cardiac catheterization, adjusting for clinical and demographic covariates. KEY RESULTS: Three HBA1c trajectory classes were identified: individuals with stable glycemia (class A; n = 6934 [89%]; mean baseline HbA1c 6.9%), with declining HbA1c (class B; n = 364 [4.7%]; mean baseline HbA1c 11.6%), and with increasing HbA1c (class C; n = 482 [6.2%]; mean baseline HbA1c 8.5%). HbA1c trajectory class was associated with adjusted 2-year mortality (3.0% [95% CI 2.8, 3.2] for class A, 3.1% [2.1, 4.2] for class B, and 4.2% [3.4, 4.9] for class C; global P = 0.047, P = 0.03 comparing classes A and C, P > 0.05 for other pairwise comparisons). Baseline HbA1c was not associated with 2-year mortality (P = 0.85; hazard ratios 1.01 [0.96, 1.06] and 1.02 [0.95, 1.10] for HbA1c 7-9% and ≥ 9%, respectively, relative to HbA1c < 7%). The association between HbA1c trajectories and mortality did not differ between those with and without CAD (interaction P = 0.1). CONCLUSIONS: In clinical settings where single HbA1c measurements provide limited information, HbA1c trajectories may help stratify risk of complications in diabetes patients. Joint latent class modeling provides a generalizable approach to examining relationships between biomarker trajectories and clinical outcomes in the era of near-universal adoption of electronic health records.

Use of Machine Learning Approaches in Clinical Epidemiological Research of Diabetes.

PURPOSE OF REVIEW: Machine learning approaches-which seek to predict outcomes or classify patient features by recognizing patterns in large datasets-are increasingly applied to clinical epidemiology research on diabetes. Given its novelty and emergence in fields outside of biomedical research, machine learning terminology, techniques, and research findings may be unfamiliar to diabetes researchers. Our aim was to present the use of machine learning approaches in an approachable way, drawing from clinical epidemiological research in diabetes published from 1 Jan 2017 to 1 June 2020. RECENT FINDINGS: Machine learning approaches using tree-based learners-which produce decision trees to help guide clinical interventions-frequently have higher sensitivity and specificity than traditional regression models for risk prediction. Machine learning approaches using neural networking and "deep learning" can be applied to medical image data, particularly for the identification and staging of diabetic retinopathy and skin ulcers. Among the machine learning approaches reviewed, researchers identified new strategies to develop standard datasets for rigorous comparisons across older and newer approaches, methods to illustrate how a machine learner was treating underlying data, and approaches to improve the transparency of the machine learning process. Machine learning approaches have the potential to improve risk stratification and outcome prediction for clinical epidemiology applications. Achieving this potential would be facilitated by use of universal open-source datasets for fair comparisons. More work remains in the application of strategies to communicate how the machine learners are generating their predictions.