|Grant Number:||5R03CA166719-02 Interpret this number|
|Primary Investigator:||Partin, Melissa|
|Organization:||University Of Minnesota|
|Project Title:||Measurement of Inappropriate Screening Tests (MIST)|
DESCRIPTION (provided by applicant): Although routine screening can reduce colorectal cancer (CRC) incidence and mortality, some screened patients may not be appropriate for screening. We define inappropriate screening as screening at intervals that are shorter than recommended by guidelines, screening high-risk patients with a method other than colonoscopy, and screening when the patient is unlikely to benefit because they are under age 50 over age 85 or of have a life expectancy of less than 5 years. Inappropriate CRC screening strains scarce gastroenterology resources, increases costs of healthcare, and exposes patients to inconvenience, stress and medical risks, including death. Little is known about the prevalence of inappropriate screening beyond single site estimates. Nor do we know about inter-facility variation or its causes. Both computerized clinical reminders (CCRs) which are used to prompt providers when a patient is due for screening, and CRC screening performance measure scores have been identified as important interventions to promote CRC screening. However, CCRs may contribute to over-screening if they do not facilitate critical evaluation of appropriateness, and performance measures may contribute to overuse by encouraging indiscriminate screening practices. Aims. The objective of this research is to improve the quality of CRC screening programs by minimizing the number of patients who receive inappropriate screening tests while maintaining high rates of appropriate CRC screening. This grant will quantify the extent of inappropriate screening in a large integrated health care system. It will also test hypotheses regarding the association of inappropriate CRC screening rates with facility CRC screening performance measure scores and CCR characteristics (i.e., whether the CCR contains input fields to document a rationale for not recommending screening). Approach. This one-year observational study will begin by identifying patients who had a CRC screening test at any Veterans Health Administration facility (n=139) in fiscal year 2010. An algorithm, developed for this study, will be applied to the administrative data to classify each screening procedure as inappropriate or appropriate based on patient demographics, screening history, and health status. To reduce error associated with automated classification, a manual chart review on a subsample of 10 patients from each of 32 facilities will be conducted. Predictive modeling and multiple imputation techniques will then be used to identify predictors of misclassification and to reclassify cases from the full sample that are identified as likely to have been initially misclassified. Facility estimates of inappropriate screening obtained from these analyses will then be entered as dependent variables into hierarchical logistic regression models to test proposed hypotheses regarding the relationships between facility CCR characteristics, Healthcare Effectiveness Data and Information Set (HEDIS)-based performance measure scores, and inappropriate CRC screening.