# Sociodemographic characteristics of patients throughout the recruitment process into a randomized, controlled behavioral trial

## TL;DR

This study analyzed how different groups of people responded to recruitment for a physical activity trial targeting those with diabetes or prediabetes.

## Contribution

The study provides insights into recruitment disparities based on sociodemographic factors in a behavioral trial using electronic health records.

## Key findings

- Patients with diabetes, males, Hispanics, and Spanish speakers were less likely to participate in the trial.
- Those contacted by phone were more likely to be female and less likely to be Hispanic or Spanish speakers.
- Neighborhood deprivation did not differ between those randomized and those not randomized.

## Abstract

Recruiting participants into trials is important to ensure trial success. We report the results of a recruitment process using electronic health records (EHR) to identify potentially eligible patients with telephone follow-up to enroll participants into a randomized, controlled trial of a physical activity intervention.

Patients aged 18 to 77 years with prediabetes or diabetes not using insulin and reporting low physical activity were potentially eligible. An EHR program code was created to assess eligibility. Patients received an email or mailed letter explaining the trial and information that they will be contacted in the coming weeks. Recruiters made up to 5 contact attempts. Patients were categorized as never reached, contacted by phone, consented, and randomized. Information was collected from the EHR to identify sociodemographic characteristics through the recruitment stages. Pearson’s chi-squared test and Fisher’s exact test were used to test for differences in categorical variables; Wilcoxon rank sum test was used for continuous variables. Analyses were completed using R version 4.3.0.

Recruitment continued from July 2020 through September 2023 with a goal of 482 participants. A total of 11,152 patients received either an email (89.5%) or letter (10.5%) informing them of potential eligibility (pre-diabetes: 66%; female: 57%; Hispanic: 65%; Spanish-language preference: 25%; neighborhoods with a high deprivation index: 24%). Recruiters contacted 4033 patients by phone; those contacted were more likely to have prediabetes than diabetes (77%), to be female (61%), were less likely to be Hispanic (60%), or to have a Spanish-language preference (17%) compared to those not contacted by phone (all p < 0.001). Among those contacted by phone, patients who consented (N = 721) differed from those who did not: they were more likely to be pre-diabetic (87%), female (75%), and less likely to prefer Spanish language (10%) (all p < 0.001). Four hundred fifty-one were randomized; 23% lived in neighborhoods with a high deprivation index, which did not differ from those not randomized.

Following a recruitment process of patients identified in the EHR, people with diabetes, males, Hispanics, and Spanish speakers were less likely to participate in a physical activity behavior trial for patients with diabetes or prediabetes, with no apparent differences in neighborhood deprivation.

Randomized Trial of Exercise Promotion in Primary Care NCT0445168. Registered on June 24, 2020.

The online version contains supplementary material available at 10.1186/s13063-025-09225-7.