The most successful recruitment strategy was the targeted mail-out with 70% of all participants recruited by this method. The mail out from the VEC was so successful in the local Barwon/Geelong region that the team decided to expand the mail out to the Mornington Peninsula and increase the target number of participants from 1,500 to 2,300. This allowed the study to be completed in the original time frame compensating for an initial slow rate of recruitment. The mail-outs from the government agencies (VEC, databases from the local public hospital and local government home-help agency) each resulted in a similar recruitment rate of five to seven percent (% of participants recruited out of the total number of letters sent). The targeted mail-out from a large rural medical centre had a comparable response rate (7.1%).
A majority of participants resided in two distinct regions (Barwon/Geelong 69.4% and Mornington Peninsula 22.4%). The response rate from the mail-out was no higher in the local region (Barwon/Geelong) where the study had extensive exposure through other recruiting strategies. This was compared to the targeted mail-out at the Mornington Peninsula where no presentations took place. However the letter describing the study sent to all general practitioners in the Mornington region may have improved their response rate. Since participants were not asked if they had discussed the study with their GP we are unable to determine the influence of their local doctor on recruitment rates. General practitioners in the Barwon/Geelong region were informed of the study through advertisements in the regional GP Association's newsletter and the education forum offered to them as part of their continuing medical education requirement.
Engaging the services of a small public relations company enhanced the success of the mail-outs. The company revised the study brochure/information letter. Although the changes appeared minor, the positive response rate to the letter increased from 2 to 7%.
The 'bring a friend' morning tea offered to participants in the early stages of recruitment was also a worthwhile activity. The morning tea resulted in 30 new participants from the 200 invited participants.
The strategies that were labour-intensive and did not result in successful recruitment include the activities directed towards the GP medical centres. The team spent considerable time and effort (1) developing the accredited CME activity for GPs, (2) developing and installing the electronic version of the screening algorithm that was compatible with GP medical practice software and (3) giving approximately 25 presentations to medical staff. Although referral of potential participants from doctors and other health professionals comprised the second most successful recruitment strategy, the method was labour-intensive and only contributed 7.4% of all participants. The 'hands-on' approach of public and targeted presentations was similarly labour-intensive and while generating a lot of interest about osteoporosis and vitamin D, this approach averaged only one eligible participant per presentation.
A majority of the participants with a recruitment method classified as 'other or not recorded' are likely to have been contacted through the targeted mail-outs from VEC. This was a very busy period of recruitment and staff had little time to follow-up on the unanswered response relating to the recruitment method. Staffing levels during this recruitment period were 0.8 equivalent full time (EFT) research scientist, 0.3 EFT research nurse and 0.3 EFT administrative officer. At this time staff were also responsible for the monthly ascertainment of all falls and fractures of participants already enrolled onto the study.
Other recruitment reports of clinical trials have identified similar recruitment rates from direct mail-outs, 'word of mouth' and posters/brochures. Robinson and colleagues [12
] enrolled 70 families into an obesity prevention trial for young children and report 66% were obtained through direct mail-outs; 7% from other sources such as 'word of mouth' and 3% from posters and brochures. However, in contrast to our findings, Robinson's team acquired 24% through newspaper advertisements compared with 1% of our participants. Earlier research suggests that newspaper advertisements and media can promote widespread awareness and a positive image of the study [13
]. While casual feedback from participants suggests this is true, our similar recruitment rate for the direct mail-outs in the Mornington region would suggest our advertising strategy in the local region had little additional impact on the results from the direct mail-out. The discrepancy in advertising success between the Vital D study and Robinson's trial may be explained by the different age of participants. Older people are more likely to have a close relationship with their local doctor while their vision and age may impede their ability to be influenced by advertisements in the local newspaper.
Our rate of recruitment is consistent with that of an intervention trial on vitamin E in adults aged 50 to 80 years. The VECAT study recruited 1,204 participants in 60 weeks [14
]. Similarly the Vital D team recruited 2,317 older women over 104 weeks, although 74% were recruited in the last 20 weeks. The lessons learnt from the initial poor response rate of Vital D recruitment include changing strategies earlier when the initial efforts are unsuccessful. As the study's secondary aim was to demonstrate that this was a practical primary health care intervention the research team were reluctant to minimise the GP's involvement in recruitment. Garrett's study on vitamin E also reported a lack of recruiting success with direct approaches to community groups and via general practitioners [14
]. Although these researchers suggest their low response rate from GPs may be due to weak collaborative links and no follow-up with the doctors by study personnel, these reasons are not applicable in the Vital D study. The team had (1) strong links with a majority of the 200 GPs in the local region, (2) a Chief Investigator that was also Head of the department of General Practice at The University of Melbourne and (3) an Associate Investigator who was president of the local GP Association.
Garrett and colleagues also reported that recruiting via direct mail-out using the electoral roll was efficient and cost-effective. Their recruitment rate of 5% from this method was in agreement with our rate of 7%. Although Garrett's research team observed that newspaper advertising was equally cost-effective as the electoral roll mail-out, their high initial response could not be duplicated in subsequent advertisements [14
The study has a high retention rate with few participants choosing to withdraw. The total person-years lost was only 5.3% of the maximum possible person-years (7,309 person-years) with 3.9% (n = 87/2256) of participants dying during the course of the study. Casual feedback from participants suggests that retention rates were improved by (1) regular phone contact by the same team members, (2) regular newsletters that included 'fun' staff profiles and (3) Christmas cards.
The study was designed to investigate the effect of an annual dose of vitamin D in older women at high risk of falls and fractures. The eligibility criteria of 5 points scored on the screening algorithm aimed to identify high-risk women. Successful recruiting involves both reaching the target number of participants and selecting individuals with the appropriate risk profile that satisfies the original power calculations. While it can be tempting to accept eager potential participants who are in the correct age range, the 'healthy volunteer' bias is well-documented [15
]. Based on our eligibility algorithm 22% of potential participants screened were rejected thus hopefully eliminating the 'healthy volunteer' bias.