Objective: To review the literature on rates follow-up after abnormal medical tests with a particular focus on follow-up after abnormal mammograms and also propose a study using the Carolina Mammography Registry which will 1) determine if the CMR data can be used for this purpose, 2) establish rates of incomplete follow-up after abnormal mammograms in the patients in the CMR data base, 3) compare rates of incomplete follow-up among participating CMR mammography centers, and 4) identify patient and radiological demographics associated with lower rates of follow-up after an abnormal screening mammogram. Data Sources: Most of the articles in the review portion of the paper were found via PubMed searches and reports published by the Institute of Medicine. Design: The proposed study is a descriptive study using patients in the CMR who had an abnormal mammogram requiring follow-up between 1998 and 2003. The number of women without follow-up divided by the total number who had abnormal mammograms requiring follow-up will define the rate of incomplete followed-up. Setting: The CMR is a member of the National Cancer Institutes Breast Cancer Surveillance Consortium. The CMR sites include mainly community based mammography facilities representing hospital-based, private radiology and multi-specialty facilities in 39 counties in North Carolina. As of 2004, data on over 1,450,000 records on more that 460,000 NC women are included in the database. Patients: The study will include all women in the CMR aged 40 years old and older without personal histories of breast cancer who had abnormal screening mammograms by Breast Imaging and Reporting Data System (BI-RADS®) assessment in between January 1998 and December 2003. Those who received follow-up within three months and those who did not receive follow-up within three months will be compared. Demographic data to be extracted include: race, age, education level, family history of a first-degree relative with breast cancer, personal history of breast procedures, time interval since the last mammogram, BIRADS assessment code, recommended follow- up procedure, and radiology facility. Follow-up interval will be established by assessing the number of days in between the mammogram reading and first date pertaining to follow-up information. The follow-up date will be abstracted from pathology specimen laboratory information, date of return to a follow-up facility as ascertained by the mammography sites, the Lineberger Comprehensive Cancer Center's Rapid Case Ascertainment system, the North Carolina Cancer Registry or from other patient information supplied to a CMR site. Main outcome measure: Overall rates of abnormal screening mammograms that are not resolved within 3 months after the index abnormal MMG. This will be established for the CMR in the 1998-2003 interval and for individual radiology practices. Results: Rates of follow-up for the CMR will be calculated as well as rates of each particular mammography site. Bivariate analysis will be used to examine the relationship between lack of follow-up and patient and radiographic characteristics. Multivariate logistic regression will be performed to find the set of risk factors that best predict the probability of not returning for follow-up. Associations between demographic characteristics and the probability of not following up will be presented as adjusted odds ratios with 95% confidence intervals. Conclusions: The literature review supports the use of tracking and reminder systems by health care providers and care systems to improve upon the rates of follow-up after abnormal screening and other medical tests. We hope that findings will reveal that the CMR practices have higher rates of complete patient follow-up after abnormal mammograms than what is detailed in the literature. This would then suggest that the CMR systems are effective for tracking and resolving mammographic abnormalities, thus provide an element of safety for its participants. Information on characteristics associated with more concerning rates of follow-up will then be able to guide the direction of improvement efforts. The information may help in the design of better systems or offer comparative data to stakeholders as they make decisions regarding other computerized data systems for their patients and staff.