Fresh Food Farmacy: A Randomized Controlled Trial

Glycated hemoglobin (HbA1c) is a measure of the average share of glucose that has been in the bloodstream over the 2-3 months preceding measurement. HbA1c is measured by performing assays on samples of the participants' blood.

Units are percentages; for differences between HbA1c values, units are percentage points.

An HbA1c value of 6.5% is the lower threshold for type 2 diabetes. HbA1c values are rarely above 15-20%. In the context of this study (patients with type 2 diabetes), higher scores are considered a worse outcome.

Glycated hemoglobin. Units are percentages (for differences, percentage points) reflecting the average share of glucose in the bloodstream over the previous 2-3 months.

An HbA1c value of 6.5% is the lower threshold for type 2 diabetes. HbA1c values are rarely above 15-20%.

In the context of this study (patients with type 2 diabetes), higher scores are considered a worse outcome.

Fasting blood glucose is a measure of blood sugar control over a short time period (the blood glucose content after an overnight fast).

Units are mg/dL. To convert to mmol/L (SI), multiply by 0.0113. Lower fasting blood glucose levels indicate better health within the context of this study.

Weight in pounds. In the context of this study, higher values are considered a worse outcome.

SI conversion factor: To convert pounds to kilograms, multiply by 0.4536.

Body Mass Index (BMI) in kg/m^2. BMI is computed using weight (lb) and height (in). The CDC's healthy reference range for BMI among adults aged 20 and older is 18.5 to 24.9. BMI values greater than or equal to 30.0 indicate obesity. In the context of this study, higher BMI is considered a worse outcome.

Total cholesterol in mg/dL. The CDC recommends total cholesterol levels around 150 mg/dL and classifies levels of 200 mg/dL and above as unhealthy (hyperlipidemia). In the context of this study, higher total cholesterol indicates a worse outcome.

SI conversion factor: To convert from mg/dL to mmol/L, multiply by 0.0259.

Low-density lipoprotein (LDL) cholesterol, measured in mg/dL. The CDC recommends LDL levels of 100 mg/dL or below. In the context of this study, higher levels of LDL indicate a worse outcome.

SI conversion factor: To convert from mg/dL to mmol/L, multiply by 0.0259.

High-density lipoprotein (HDL) cholesterol, measured in mg/dL. The CDC recommends HDL levels of at least 40 mg/dL in men and 50 mg/dL in women. In the context of this study, higher levels of HDL indicate a better outcome.

SI conversion factor: To convert mg/dL to mmol/L, multiply by 0.0259.

Triglycerides, measured in mg/dL. The CDC recommends levels of triglycerides below 150 mg/dL. In the context of this study, higher levels of triglycerides indicate worse outcomes.

SI conversion factor: To convert mg/dL to mmol/L, multiply by 0.0113.

Systolic blood pressure, measured in mm Hg. The CDC identifies normal healthy systolic blood pressure as below 120 mm Hg. The American College of Cardiology identifies 120 to 129 mm Hg as elevated blood pressure and 130 mm Hg and higher as hypertensive. Higher systolic blood pressure levels indicate worse outcomes.

Diastolic blood pressure, measured in mm Hg. The American College of Cardiology identifies normal diastolic blood pressure as below 80 mm Hg and elevated/hypertensive diastolic blood pressure as 80 mm Hg or above. Higher levels of diastolic blood pressure indicate worse outcomes.

Patient-reported measure of diet improvement: Number of times in the last 7 days when the patient ate fruit.

Patient-reported measure of diet improvement: Number of times in the last 7 days when the patient ate dark green vegetables.

Patient-reported measure of diet improvement: Number of times in the last 7 days when the patient drank sweetened beverages.

Patient-reported measure of diet improvement: Number of times in the last 7 days when the patient ate fast food, takeout, or at a restaurant.

Patient-reported measure of diet improvement: Survey question: "Do you believe your diet has improved compared to one year ago?".

The proportion of questions testing patients' knowledge of how different factors contribute to blood sugar answered correctly.

Survey-based index score for patient-reported positive engagement outcomes.

Scores range from 0 to 1. Scores are standardized so that the control group always has a mean score of 0 and standard deviation of 1. Higher scores reflect better outcomes---here, better engagement with the program. Scores composite results from multiple survey questions while adjusting for covariance between questions.

Healthcare utilization: Number of outpatient claims among continuously enrolled/insured subjects during the relevant time frame after trial enrollment.

Healthcare utilization: Number of inpatient and emergency department claims among continuously enrolled/insured subjects during the relevant time frame after enrollment.

The number of subjects who were continuously enrolled/insured who had any outpatient claims during the relevant time frame after trial enrollment. For statistical analyses, coefficients and errors represent proportions of participants, relative to the size of the group/treatment arm.

The number of subjects who were continuously enrolled/insured who had any inpatient or emergency department claims during the relevant time frame after trial enrollment. For statistical analyses, coefficients and errors represent proportions of participants, relative to the size of the group/treatment arm.

Healthcare utilization: Total value (USD) of inpatient and emergency department payments among continuously enrolled/insured subjects during the relevant time frame after enrollment.

Healthcare utilization: Value (USD) of outpatient payments among continuously enrolled/insured subjects during the relevant time frame after enrollment.

The number of times an individual participant completed a medical appointment with a dietician in the relevant time frame, based on electronic health records.

The number of times an individual participant completed a Diabetes Self-Management Training (DSMT) in the relevant time frame, based on electronic health records.

The number of times an individual participant completed a medical appointment with an endocrinologist in the relevant time frame, based on electronic health records.

The number of times an individual participant completed a medical appointment regarding weight loss or weight management in the relevant time frame, based on electronic health records.

Healthy behaviors observed in electronic health record, including preventive care and appointments kept

Healthy behaviors observed in paid claims using Healthcare Effectiveness Data and Information Set (HEDIS) measures. These measures were not collected for this study.

HbA1c available in EHR records

Weight in pounds as available in EHR records

BMI in kg/m^2 in pounds as available in EHR records

Cholesterol as available in EHR records

Blood pressure as available in EHR records

Triglycerides as available in EHR records

Fasting glucose as available in EHR records

Inpatient utilization for household members as available in EHR records

Outpatient utilization for household members as available in EHR records

Inpatient utilization for household members as available in paid claims

Outpatient utilization for household members as available in paid claims

Separate results by tercile of baseline HbA1c

Heterogeneity analysis by site of the clinic

Heterogeneity analysis by causal trees using the baseline characteristics to explore heterogeneity, sample size permitting.