An investigation of report bias in a case-control study of pregnancy outcome.