Publication Abstract

Authors: Goodrich ME, Greene MA, Eliassen MS, Poplack SP, Wells WS, Carney PA

Title: A population-based mammography registry: how complete is its capture?

Journal: J Registry Manage 33(1):24-8

Date: 2006 Spring

Abstract: Background: Complete and accurate data from a statewide registry are essential for research, especially when services are frequent and complex. Objective: To identify and describe gaps in a mammography registry's data sources and assess the implications of these gaps on studies evaluating the accuracy of and adherence to mammography. Methods: All women in a statewide mammography registry with a screening or diagnostic mammogram between March 1, 2000 and March 1, 2001 with incomplete follow-up for an abnormal assessment were identified. We sought to obtain this information by contacting women's referring primary care providers (PCPs). Results: Of 91 380 women receiving mammography during the study time period, complete New Hampshire Mammography Network (NHMN) data capture was obtained on 89 969 (98.5%), leaving 1411 (1.5%) with no resolution of an abnormal mammogram. Of the 510 PCP study respondents, 156 (31%) offered complete follow-up information on 38% (530/1411) of the women. Eighty-nine percent (450/506) of the follow-up imaging was performed at the same or another participating NHMN site. A majority of services received were ultrasounds at 301 (57.3%), 167 (31.9%) were additional views, 30 (5.7%) were routine screenings, and 25 (4.7%) were short-term follow-up examinations. We received 66 follow-up pathology cases of which 2 (3%) were from NHMN participating pathology labs and 64 (97%) were unknown or not NHMN pathology labs. Missing data had less then 1% influence on calculations of sensitivity and specificity, and a small effect on adherence to screening ratios. Conclusion: Small gaps currently exist in our population-based mammography registry. Data collection processes may be interrupted when additional imaging (ultrasound and additional views) is needed to resolve an abnormal mammogram. The influence on studies of mammographic accuracy is low, but careful monitoring is essential to ensure completeness of registry outcomes data.