Telemedicine Low-Carbohydrate Nutrition Linked to Fewer Heart Events in Claims Study

Adults with type 2 diabetes or obesity who received an individualized, carbohydrate-reduced nutrition program through telemedicine experienced significantly fewer cardiovascular events than matched controls receiving usual care, according to a retrospective claims-based cohort study published in Cardiology and Therapy. The analysis, which drew on medical and prescription claims data from more than 300 million individuals in the Komodo Healthcare Map, compared 4877 program participants with 4877 propensity score-matched controls and found lower incidence of major adverse cardiovascular events, new-onset cardiovascular disease, and new-onset hypertension across a follow-up window ranging from six months to twelve years.

The study, led by researchers including Caroline G. P. Roberts and published on October 2, 2026, enrolled adults between January 1, 2016 and June 1, 2025 who had type 2 diabetes or obesity but no known cardiovascular disease at baseline. Participants in the intervention group received remote care including individualized nutrition coaching, daily biometric feedback, and peer support. The program guided patients to reduce carbohydrates to 30 grams per day or less, below 10 percent of total calories, achieving nutritional ketosis defined by beta-hydroxybutyrate levels above 0.3 millimoles per liter, with a target range of 0.5 to 3.0. Protein targets were set at 1.5 grams per kilogram of reference bodyweight, dietary fat was incorporated without limits on saturated fat, and patients were advised to eat three to five servings of non-starchy vegetables daily. The program offered no specific coaching on physical activity, smoking cessation, alcohol intake, or sleep practices.

The primary outcomes were fourfold: a three-point composite of major adverse cardiovascular events comprising nonfatal myocardial infarction, nonfatal stroke, or death from any cause; a six-point composite adding percutaneous coronary intervention, hospitalization for heart failure, or unstable angina; all new-onset cardiovascular disease; and new-onset hypertension. In the intervention group, the incidence of the three-point composite was 4.1 events per 1000 person-years compared with 9.3 in controls, yielding a hazard ratio of 0.44 with a 95 percent confidence interval of 0.29 to 0.65 and a P value below 0.001. For the six-point composite, incidence rates were 5.7 versus 10.8 per 1000 person-years, with a hazard ratio of 0.52 and a confidence interval of 0.37 to 0.73. New-onset cardiovascular disease occurred at 27.7 versus 36.9 events per 1000 person-years, corresponding to a hazard ratio of 0.70 with a confidence interval of 0.59 to 0.82, and new-onset hypertension showed a hazard ratio of 0.81 with a confidence interval of 0.70 to 0.93.

In relative terms, the program group had a 56 percent lower risk of the three-point composite, a 48 percent lower risk of the six-point composite, a 30 percent lower risk of developing cardiovascular disease, and a 19 percent lower risk of developing hypertension. All-cause mortality was significantly lower across the full follow-up period, with a hazard ratio of 0.36 and a confidence interval of 0.16 to 0.84, though the association did not reach statistical significance when mortality was assessed within the closed-claims period, where the hazard ratio was 0.47 with a confidence interval of 0.19 to 1.14. The mean age of participants was 51 years, 60.3 percent were female, two-thirds had obesity, one-third had type 2 diabetes, 54 percent were white, and 95 percent were commercially insured. Seventy-three percent of intervention participants achieved biomarker levels consistent with adherence to the dietary recommendations.

Matching on the propensity score incorporated exact matching on five-year age groups, sex, race, baseline type 2 diabetes and obesity status, and fifteen categories of medications including statins, insulin, metformin, GLP-1 receptor agonists, and SGLT2 inhibitors. The score also included hypertension, hyperlipidemia, chronic kidney disease stages 1 through 3, smoking, U.S. region, and payer type. After matching, all variables achieved standardized mean differences below 0.1, indicating adequate balance between groups. Four sensitivity analyses tested the robustness of the findings: inverse probability of treatment weighting, adjustment for follow-up medication coverage, restriction of controls to overlapping index years, and inclusion of intervention participants with fewer than six months of enrollment. The weighted model produced slightly smaller but directionally consistent results, and the model adjusting for medication use during follow-up yielded slightly stronger associations, reflecting a reduced medication burden in the intervention group.

The sensitivity analysis that included participants with shorter program enrollment attenuated the inverse associations with the composite endpoints and all new-onset cardiovascular events, and the confidence interval for incident hypertension included the null. Notably, the hazard of all arrhythmias was higher in the intervention group in that analysis, a finding not observed in the primary analysis or other sensitivity analyses. The authors caution that this result should be interpreted cautiously given the potential for informative early discontinuation, differences in outcome surveillance, and multiple comparisons. In the primary analysis, safety outcomes showed no statistically significant excess electrophysiologic risk: hazard ratios were numerically below 1 for new-onset atrial fibrillation at 0.65 with a confidence interval of 0.42 to 1.00, and arrhythmia showed a hazard ratio of 1.00 with a confidence interval of 0.83 to 1.21. Nine long QTc events occurred overall, with a hazard ratio of 0.46 and a confidence interval of 0.09 to 2.33.

An exploratory analysis within the intervention group examined whether the degree of ketosis, measured by beta-hydroxybutyrate levels categorized as below 0.3, 0.3 to 0.5, 0.5 to 1.0, and at or above 1.0 millimoles per liter, was associated with cardiovascular events. Modeled as ordinal categories, higher ketosis levels were inversely associated with new-onset cardiovascular disease, with a hazard ratio of 0.86 and a confidence interval of 0.77 to 0.96, indicating a 14 percent lower hazard per category increase. Ketosis level was not significantly associated with the composite endpoints or hypertension, possibly reflecting limited statistical power for these less frequent outcomes. A secondary exploratory analysis of 613 intervention participants with pre-existing cardiovascular disease and 29,969 weighted controls found no statistically significant effects on primary, secondary, or safety endpoints.

The authors place these findings in the context of a broader and often inconsistent literature on carbohydrate intake and cardiovascular health. Prospective cohort studies have reported both higher and lower all-cause mortality associated with lower carbohydrate intake depending on the population and analytic approach, and meta-analyses have likewise produced divergent results. The authors argue that these studies are limited by a narrow exposure contrast: in most cohorts, the lowest carbohydrate intake categories represent roughly 35 to 50 percent of total energy intake, well above the level of carbohydrate reduction used in clinical interventions designed to induce nutritional ketosis. They also note that the intervention differs substantially from the dietary patterns linked in case reports to sudden cardiac death or long QT syndrome, which involved children receiving ketogenic diets for seizure control or adults in semi-starvation programs, whereas the telemedicine program emphasized ad libitum, medically supervised, individualized dietary modification without caloric deprivation.

Several limitations constrain the interpretation of these associations. As an observational study using claims data, the analysis cannot eliminate bias from unmeasured confounders such as socioeconomic factors or health behaviors, and the magnitude of residual confounding cannot be estimated from the available data. The requirement of at least six months of program participation may have selected for individuals with greater engagement or motivation, potentially limiting generalizability to those with shorter participation, though a sensitivity analysis including all eligible participants regardless of duration showed modestly attenuated but still significant associations. Laboratory lipid values including baseline and longitudinal LDL cholesterol were unavailable, so the study cannot evaluate the contribution of LDL changes to the observed associations. Claims data may under-capture diagnoses, the study cannot disentangle the independent effects of nutrition from other components of the care model such as care team contact and peer support, and reasons for termination of claims coverage were unavailable, so informative censoring cannot be excluded. The follow-up period was relatively short, leaving longer-term associations uncertain.

The cohort comprised adults with obesity and type 2 diabetes, groups at elevated cardiovascular risk, so the findings cannot be generalized to metabolically healthy populations. Women were overrepresented in the sample, and nearly 90 percent of participants had commercial insurance, raising questions about whether the results extend to other insured populations. The authors emphasize that the study cannot compare the cardiovascular impact of the intervention’s dietary pattern with other dietary patterns such as Mediterranean or DASH, because claims data do not capture dietary intake and the control group is presumed to consume heterogeneous diets. They conclude that the consistent association between the telemedicine carbohydrate-reduced nutrition program and lower cardiovascular event incidence, alongside the absence of harm signals in the primary analysis, supports the rationale for prospective randomized clinical trials to evaluate the cardioprotective potential of this nutritional approach in high-risk populations.

Subject of Research: Cardiovascular outcomes associated with a telemedicine carbohydrate-reduced nutrition intervention in adults with type 2 diabetes or obesity, assessed in a retrospective claims-based cohort study

Article Title: Cardiovascular Outcomes Among Adults with Type 2 Diabetes or Obesity Receiving Carbohydrate-Reduced Nutrition via Telemedicine: A Retrospective Claims-Based Cohort Study

Article References: Roberts, C. G. P., Shanmugam, P. V., Athinarayanan, S. J., Wolfberg, A. J., Raskin, S., & Lopez-Jimenez, F. (2026). Cardiovascular Outcomes Among Adults with Type 2 Diabetes or Obesity Receiving Carbohydrate-Reduced Nutrition via Telemedicine: A Retrospective Claims-Based Cohort Study. Cardiology and Therapy. https://doi.org/10.1007/s40119-026-00478-0

Image Credits: AI Generated

DOI: 10.1007/s40119-026-00478-0

Keywords: type 2 diabetes, obesity, cardiovascular disease, carbohydrate-reduced nutrition, telemedicine, nutritional ketosis, major adverse cardiovascular events, hypertension, propensity score matching, claims data, observational study, beta-hydroxybutyrate