Unintended Consequences of Expanding Electronic Pathology Reporting: The Inverse Relationship Between Data...

In 2016, the New Jersey State Cancer Registry (NJSCR) began expanding electronic laboratory reporting. As a result, the number of electronic pathology reports (EPRs) submitted to NJSCR increased markedly from 2015 to 2017. EPRs are more likely to contain incomplete or missing race than North American Association of Central Cancer Registry (NAACCR) abstracts from hospitals and physician offices. NJSCR staff conduct follow-back for additional information for laboratory-only cases, but response rates are poor, the process is lengthy, and laboratory reports often do not include physician information.

To assess the impact of increased EPR on the quality of race data.

NJSCR data sets created 24 months after the end of the diagnosis year-with data that were more than 98% complete-were used to calculate the percent of EPR-only cases by primary site and the percent of cases with unknown race. We calculated the relative risk of unknown race by site, compared to all sites, and used Spearman's ρ to assess the correlation between EPR-only cases and unknown race.

While the percent of cases with unknown race was within the standards for NAACCR Gold Certification (3%), it varied by cancer site. Sites less likely to be reported by hospitals had higher rates of unknown race in the 24-month data set: prostate, leukemia, melanoma, bladder. After follow-back and death clearance activities, ≥36 months after the diagnosis year, the percent of cases with unknown race is reduced, although the impact varies by cancer site.

Race-specific incidence rates for certain cancer types may be artificially depressed in the 24-month data set due to the unavailability of race for the increasing number of laboratory-only cases. While follow-back activities help to improve the collection of race data over time, these new values are not available until a revised data set is released. The higher proportion of unknown or other race in the 24-month data set impacts the accuracy of reporting the burden and trends of cancer by race. In addition, cases with unknown race may be ineligible for inclusion in cancer surveillance research studies.