Potential conflict of interest: Dr Hernandez-Diaz reports having served on a drug safety monitoring board for Novartis and a Study Advisory Board for Serono. Dr Ray reports having received grant support from Pfizer Incorporated.
Positive predictive value of computerized records for major congenital malformations†
Article first published online: 17 DEC 2007
Copyright © 2007 John Wiley & Sons, Ltd.
Pharmacoepidemiology and Drug Safety
Volume 17, Issue 5, pages 455–460, May 2008
How to Cite
Cooper, W. O., Hernandez-Diaz, S., Gideon, P., Dyer, S. M., Hall, K., Dudley, J., Cevasco, M., Thompson, A. B. and Ray, W. A. (2008), Positive predictive value of computerized records for major congenital malformations. Pharmacoepidem. Drug Safe., 17: 455–460. doi: 10.1002/pds.1534
- Issue published online: 25 APR 2008
- Article first published online: 17 DEC 2007
- Manuscript Accepted: 5 NOV 2007
- Manuscript Revised: 16 OCT 2007
- Manuscript Received: 13 JUL 2007
- Food and Drug Administration. Grant Number: FDA 223-02-3003
- the Agency for Healthcare Research and Quality, Centers for Education and Research on Therapeutics. Grant Number: HS1-0384
- S. H. D.. Grant Number: R01 HD046595
- congenital malformations;
- vital records;
- positive predictive value
To assess the positive predictive value of computerized records in a linked database of vital records and infant claims, with medical record confirmation to detect congenital malformations in a Medicaid population.
Study subjects were selected from cases identified for three studies of congenital malformations in the Tennessee Medicaid (TennCare) population including 173 827 (studies 1 and 2) and 519 465 (study 3) mother/infant pairs. Possible malformations were identified from computerized databases of birth certificates linked with maternal and infant claims. Medical records were reviewed for all possible congenital malformations and positive predictive values were calculated for each data source and for each malformation.
Among 1430 potential congenital malformations identified from either birth certificates or inpatient claims, 67.7% were confirmed by medical record review. The positive predictive value varied considerably depending on the data source and the organ system. For example, cardiac defects had a very low positive predictive value when identified from birth certificates, and somewhat higher positive predictive value when identified from inpatient claims. Orofacial defects had 90.9% positive predictive value from birth certificates and inpatient claims. Requiring evidence of a diagnostic or therapeutic procedure increased the positive predictive value to >90% for specific defects, but substantially reduced the number of included cases.
Depending on the defect, computerized claims data linked to vital records offer opportunities for identifying birth defects in populations of vulnerable persons. However, for many defects, medical record confirmation is likely to be required to provide valid identification of malformation occurrence. Copyright © 2007 John Wiley & Sons, Ltd.