Abstract 12359: Measuring Continuous Changes in Individual Cardiovascular Risk for People With Diabetes and Pre-Diabetes

Abstract

Objective: For people attempting to lower cardiovascular disease (CV) risk through self-care, feedback on progress can be a strong motivator. Many CV risk scores, however, are based on binary inputs (e.g., whether the subject is diagnosed with diabetes or hypertension). In such cases, incremental improvements in health metrics (e.g., blood glucose, BG; blood pressure, BP), though known to reduce CV risk, will not necessarily reduce the risk score. We modified a well-known risk score to accept continuous inputs and used the modified score to estimate CV risk variability over time for a large sample of people with diabetes. Method: One Drop has collected over 25 billion data points from members in all 195 countries, including BP, BG, weight, medication, food, physical activity, heart rate, sleep, comorbidities and demographics. The INTERHEART risk score (IRS) uses CV risk factor information to compute a score ranging from 0 to 40, each point representing a 12% increase in risk. We transformed the risk ratios from published studies of CV risk factors (i.e., BG, BP, BMI, BP medications) to align with the IRS scale, interpolating among reported values by fitting functions via constrained optimization. We applied the modified risk assessment retrospectively to data collected between 2016-2021. Results: Among the 2,150 One Drop members included, the majority did not report gender (43%; 32% male) and had type 1 diabetes (78%; 14% type 2; 1% prediabetes; 7% other) with BMI of 28.8±6.9. CV risk scores varied from 2 to 23 IRS points, with six-month standard deviations ranging from 0.008 to 6.379 IRS points. Variations over time for individuals were driven by changes in BG (r = 0.89) more than BP (r = 0.70). Conclusion: Established CV risk scores can be modified to replace binary inputs with continuous inputs. This provides a more granular measure of risk to enable more specific intervention personalization and detect smaller changes in risk over time when evaluating interventions. Moreover, people working to lower their CV risk could benefit from seeing that even small changes in their BG, BP, weight and use of medications lead to a reduction of CV risk.