Table 1 Be Fit, Be Well recruitment strategies and results

Identify potentially eligible participants through medical record review

• Difficult to maintain large pool of potential participants for contact

1. Provider referral: providers could submit participants via faxed or emailed referral form; drop boxes for referrals placed in

1. Received less than 20 provider referrals. Most were people already identified through medical record review;

2. Patient self-referral: flyers were placed in waiting and patient rooms with brief study description and contact information;

2. Received no inquiries from patient self-referral

3. Refer-a-friend: enrolled participants were asked to tell their friends about the study;

3. Received less than 10 suggestions from enrolled participants. Most were ineligible.

4. Newspaper ads: Two ads were run in a publication that was distributed to people riding public transportation; one in paper serving the African-American community; Another in a Spanish language paper serving the Hispanic/Latino community

4. Received less than 20 inquiries from newspaper ads. Most were not patients at one of the three health centers and were ineligible. One person was ultimately enrolled.

>95% of enrolled participants were identified via medical record review

Participant names submitted to their primary medical provider for approval to contact for study enrollment

• Long delays in receipt of provider approval

1. Implemented passive provider approval system. We divided patients into ‘Needs Confirmation’ and ‘Confirmed’ groups. Based on medical record review, patients in the ‘Confirmed’ group met all eligibility criteria and were free of diabetes, CVD and peripheral vascular disease. Providers had the option to exclude patients from this list, but we initiated patient contact after 10 business days if providers had not responded. The ‘Needs Confirmation’ group included people with at least one of the aforementioned health conditions. This group still required explicit approval from providers before contact.

1. Prior to passive approval, we had submitted 431 names to providers and had received a response on 212 (49.1%) of them. With passive approval in place, there were just 29 still awaiting provider approval at the end of recruitment. Providers also expressed appreciation for the passive process that, given their demanding schedules, reduced their study related workload considerably.

Passive approval decreased time from patient identification to initial contact significantly.

Approved participants sent introductory letter and called by research assistants to confirm eligibility using contact information in medical records

• Incorrect contact information;

1. Collaborated with health center administrative staff to obtain regular updates of patient contact information;

1. Health center staff generally did not have more up to date contact information than what they had originally provided us;

Large number of calls (> 4000) required to garner each scheduled baseline visit

2. Purchased subscription to online people search website to find new addresses and phone numbers for potential participants;

2. Website provide correct contact information for some potential participants;

3. Mailed potentially eligible participants a self-addressed, postage-paid card with our contact phone number which requested the three best times to call, best phone number to use and phone number type (home, work, cell), any alternate phone numbers. Card completion qualified them for a $75 Target gift card raffle.

3. Few cards were returned among participants with incorrect phone numbers. Many were returned to sender as incorrect phone numbers were highly correlated with incorrect mailing addresses.

4. Hired an off-site call center to make intake calls. Trained survey assistants at the University of Massachusetts Amherst Survey Research Center made all English calls. Research assistants continued to make calls in Spanish.

4. Call center improved the weekly call volume and allowed the research assistants to focus on other recruitment tasks.

454 (17%) potentially eligible participants had never been reached at the end of recruitment.

Two research assistants responsible for completion of all baseline and follow-up assessments

• Baseline and follow-up visits occurring simultaneously

1. Hired and trained short-term part-time staff to help with completion of baseline and follow-up visits

1. Seven short-term staff were largely students hired through institutional internship programs. Generally there were only two working during any given time period, but there was one 3-month period with four. Each short-term staff member underwent training and certification in measuring height, weight and blood pressure, shadowed the full-time RAs for 2–3 visits, and led their first three visits with a full-time RA present to ensure data quality.

They greatly facilitated completion of visits, particularly during a point when baseline, 6-month and 12-month visits were all ongoing.