Effect of recruitment strategy on types of subjects entered into a primary prevention clinical trial