New Edits Promote Enhanced Quality

The NAACCR Edits Work Group has developed a new group of edits for central registry use, identified as QC_Chk (quality control) edits. They are included in special edit sets rather than in the traditional standard setter edit sets. The traditional edits routinely assess and promote data quality. Most traditional edits will only fail if one or more of the fields included are coded incorrectly. To resolve the edit one or more of the values must be changed. QC_Chk edits differ as they force a review of data without an assured resolution for a case that fails. The edits may be used to support communication between the central registry and reporting facilities on quality issues. However, they should not be used directly in metafiles distributed to reporting facilities as they could disrupt routine data submissions. Some of the QC_Chk edits compare what was directly coded with what would be expected treatment based on standardized treatment protocols. The QC_Chk edits require follow-up to satisfy their conditions. Others carry known constraints in their application to data without the provision of overrides. These edits promote the collection of complete treatment information, and correct information for specific circumstances in a majority of cases.

These edits have been developed from reviews of reported case data, and also from inspection of codes and coding instructions. NAACCR will provide a guide to the edit descriptions and logic for those central registries that choose to construct database queries to identify relevant cases for review, outside of the edit framework.

The edits are listed in the accompanying table in three categories: Treatment Check Edits, Checks with Warnings, and Checks with Failures. Checks with Failures are edits that will identify incorrect values or combinations of values within a record. Checks with Warnings are edits where the edit logic will identify values within a record that may be problematic and should be reviewed. Treatment Checks are edits that produce warnings that indicate a review is needed due to incomplete treatment data.

Treatment Check Edits

The larger group of edits under Treatment Checks identifies patient profiles through selection criteria, and then edits for anticipated treatment for those profiles. The second smaller group of edits in this category identifies records where treatment is coded as recommended unknown if given. Correction of any warnings from these edits requires follow-up to obtain complete treatment information, which may legitimately remain unknown. N7134, Breast, Surgery, Systemic Treatment, is a good example of the treatment checks. Selection criteria include Schema ID, Sex, Behavior, Age at Diagnosis, Summary Stage 2018, Tumor Size, Surgery of Primary Site, and whether positive or negative Estrogen Receptor Summary and Progesterone Summary. Data items checked are sequence between surgery and chemotherapy, and whether chemotherapy, immunotherapy, or hormone therapy is given based on age and receptors.

Checks with Warnings

The second category, Checks with Warnings, lists inter-field edits that will legitimately not meet the stated logic criteria in all cases reviewed. The first two edits on name in the smaller group here are longstanding edits (N6370 is an updated version of a previous edit on maiden name) and have been in single-field edit sets in the past. N6370 may flag an “unknown” requested by a registry. N2038 does have an over-ride, but its link between name and sex may be tenuous and cause a lot of failures requiring over-ride. N6434 may identify conflicts between records coded at different times that are difficult to resolve.

The larger group in this second category edit for coding patterns that will be correct most, but not necessarily all of the time. The edited code for testis primary site is allowed in certain situations and edited urothelial histologies are allowed in rare circumstances. Registries may collect non-reportable histologies by agreement. The tumor size ranges have been tested by SEER and reviewed with medical personnel. SEER uses an over-ride with its implementation of these edits, but an over-ride on tumor size is not currently available in the NAACCR metafile. The first two radiation edits, N6804 and N6452, flag unlikely, but not impossible similarities of codes over multiple phases of radiation. The second two radiation edits, N6891 and N6816, reference an imperfect table of paired sites to identify radiation treatment volumes.

Checks with Failures

The third category, Checks without Warnings, edit for coding patterns that will correctly fail, but the edits may be of questionable utility or hard to evaluate in a testing situation. N7101 edits for RX Hosp treatment codes for Class of Case where no treatment occurs at the reporting facility. N7099 and N7098 edit for unknown pathologic grade where bladder surgery does not support pathologic information.

Conclusion

The treatment edits, N7129 through N7139, are included in the NPCR Treatment Checks edit set in the NAACCR v25 metafile. The other edits are all included in the Quality Control Edits-Central Registry edit set in the metafile. These edit sets can be modified, or any of the edits pulled into other sets, similarly to any other edits sets or edits within the metafile. Edits that issue warnings rather than failures will be listed separately on GenEdits Plus reports, but they will continue to issue warnings whenever they are run until the logic criteria are met. The QC_Chk edits expand the utility of the metafile. They join the _SYS edits as a special category, intended for a particular use apart from promoting the routine quality of cancer abstracts across the range of coded data items. Like the _SYS edits, some of these edits may eventually pass into standard use, given experience with their application and refinement of edit logic. They offer a flexible approach to the benefits of data editing – helping to complete case information, correcting case information where data outliers may exist or edit logic is not truly comprehensive, and finally offering options on questioned edits.

TABLE 1. QC_CHECK EDITS – 2025