Quality of cancer incidence data in North Carolina: an evaluation of the Central Cancer Registry 1990 database

Г п ^STUDIES A SPECIAL REPORT SERIES BY THE N.C. DEPARTMENT OF ENVIRONMENT, HEALTH, AND NATURAL RESOURCES STATE CENTER FOR HEALTH AND ENVIRONMENTAL STATISTICS P.O. BOX 29536, RALEIGH, N.C. 276264)538 No. 67 June 1992 THE QUALITY OF CANCER INCIDENCE DATA IN NORTH CAROLINA An Evaluation of the Central Cancer Registry 1990 Database by Carol R. Dickinson and Tim E. Aldrich ABSTRACT The North Carolina Central Cancer Registry (CCR) has completed its first year of statewide data collection. During 1990, 127 of 132 facilities submitted cancer incidence data; 29 of those were tumor registries. For the facilities withouttumorregistries, the CCR conducted a complete review of all incidence reports for quality assurance and completeness, and provided training to all new staff members. This quality control process led to two-thirds of these facilities having an accuracy rate in excess of 96 percent. For facilities with new tumor registrars, CCR offered training sessions and a 100 percent evaluation of their abstracts for at least two months. The CCR conducted a study to assess the reliability of the data submitted. From the 127 reporting facilities, 87 persons participated in an abstracting exercise of four sample charts. Tumor registrars were consistently more accurate than the data collectors from hospitals without tumor registrars. Items presenting the most problems for all persons were the determination of the exact anatomic primary site and summary staging. Medical records personnel who abstract cancer data in hospital s without tumor registri es also need to improve on the accuracy of the reporting of ethnicity, date of diagnosis, date first seen at their facility, and the sequence coding for multiple primary tumors. In addition to the above quality control activities, the CCR performed two evaluations of the completeness of case ascertainment. First, a standard measure of completeness of reporting, the incidence/ mortality ratio, was calculated for all cancers by race-sex group. This measure consistently agreed with national standards and those from the state of Georgia. Also, a casefinding audit was performed for a sample of hospitals. The results highlighted areas that need strengthening in the future. This report is presented as documentation of the quality control activities for the CCR 1990 incidence database and the results of those activities.