Quality assurance is required for all relevant instruments and procedures in epidemiological studies just like for clinical trials. The structure and complexity of the monitoring was developed based on the monitoring in clinical trials and applied to an epidemiological cohort study on early detection of cervical cancer (MARZY Study). Analyses of the on-site monitoring in participating gynaecological practices during the baseline investigation of the MARZY cohort were presented. The baseline investigation of the MARZY study was conducted between 2005 and 2007 in the city of Mainz, the rural district of Mainz-Bingen and surrounding areas. Women, who were randomly selected via population registries, were invited to attend cervical cancer screening at a gynaecologist's office of their choice. All study participants received a study swab in addition to their routine Pap smear. The on-site monitoring included equipment and support of all participating gynaecological practices during study recruitment. Each participant and physician signed an informed consent form. In addition, the participant completed an epidemiological questionnaire. The gynaecologist took the study swab and completed the study documentation form. Prior to recruitment, standardised processes and documentation forms were developed for the monitoring process. The monitoring visits were carried out every six to eight weeks. During the baseline investigation, participants were included in the study among 121 gynaecological practices. In total, 2,892 monitoring documentation forms from 390 on-site monitoring visits in 96 gynaecological practices from the study region and surrounding areas were analysed. On-site monitoring visits were more frequently conducted during the first year of the study. The average time needed for an on-site visit was 107 minutes (minimum 73 minutes, maximum 200 minutes). Problems such as incomplete study documentation forms or erroneous inclusion into the study occurred among 975 study participants (33.7%). 664 study participants (68.1%) did not fully complete the study forms, and 89 (9.1%) were included in the study despite the fact that they met the exclusion criteria such as hysterectomy or pregnancy. Most of these problems could be sufficiently corrected during the on-site monitoring. Monitoring in epidemiological studies performed at physicians' offices should be carried out in accordance with the monitoring in clinical trials. On-site monitoring helped to avoid missing data and to ensure adherence to exclusion criteria. On-site monitoring considerably contributed to the correct and complete study inclusion of all eligible participants and a high quality of study data.
