Explaining disparate findings across studies: Issues of recruitment and follow-up, selection bias, and exposure definition Conference Paper