Registry data quality improvement by identifying discrepancies between assigned codes and text descriptions of birth...

Birth defects surveillance programs support efforts to prevent and address population health. For 30 years, the New York State (NYS) Department of Health (DOH) Congenital Malformations Registry (CMR) has been receiving reports of children with birth defects diagnosed from birth to age 2 years as required by NYS DOH regulation. Our objective in this effort was to improve the accuracy of British Pediatric Association (BPA) codes assigned to case reports in the NYS CMR.

From 1998 forward, the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code and a text description for reported birth defects have been entered into the CMR database. Upon receipt, CMR staff use reported descriptions and ICD-9-CM codes to recode all birth defects into BPA codes to improve surveillance specificity. To improve CMR data completeness and accuracy for birth years 1998-2010, computer programming was used to identify possible mismatches between birth defect descriptions and BPA codes. First, descriptions were reviewed for text strings related to specific BPA codes, and common spellings, misspellings and abbreviations were identified. Next, 1998-2010 case reports were flagged if they had either the identified text strings or the related BPA codes but not both. All flagged records were reviewed manually by registry staff and necessary data corrections were made. This quality-control process was evaluated, overall and for 78 targeted birth defect groups, using frequency and percent. The number of records flagged for possible discrepancy between BPA codes and text strings, percent reviewed and percent corrected are presented. These findings helped us understand which BPA codes or birth defects required the most attention.

A total of 164,726 records were scanned for discrepancies between BPA code and text strings related to 78 birth defects. Of the records scanned, approximately 10 percent were found to have discrepancies and were flagged for manual review and approximately 8 percent required correction. Among case records for the targeted birth defects, 24.8 percent were reviewed and 19 percent were corrected. Of the records reviewed, 76.8 percent were corrected.

This approach to data quality improvement was effective in identifying and resolving inaccuracies in the NYS CMR. High quality data is valuable for monitoring birth defects trends, allowing interstate comparisons of specific defects, and supporting birth defects research. As a result, prevention efforts and policy decisions aimed at improving public health will be better informed.