ROP in New York State

Abstract

We read with interest the article by Chiang et al1Chiang M.F. Arons R.R. Flynn J.T. Starren J.B. Incidence of retinopathy of prematurity from 1996 to 2000 Analysis of a comprehensive New York state patient database.Ophthalmology. 2004; 111: 1317-1325Abstract Full Text Full Text PDF PubMed Scopus (108) Google Scholar on the incidence of retinopathy of prematurity (ROP) in New York State as determined by International Classification of Diseases 9, Clinical Modification (ICD9CM), and Current Procedural Terminology 4 information extracted from the Statewide Planning and Research Cooperative System (SPARCS). The authors took pains to explain their data-harvesting methods, which included an internal consistency check to detect birth weight coding errors. They found that 1 of 625 babies with a birth weight of 2500 to 2999 g and 16 of 697 babies with a birth weight of 2000 to 2499 g were diagnosed with ROP, but none required treatment. They postulate that “some of these cases in larger infants may instead have been extremely rare diseases with very similar phenotypes to ROP.” Unfortunately, the authors overlooked the practical difficulties encountered in coding for ROP examinations. Technically, those infants who are not found to have active ROP should not be coded with ICD9CM diagnosis 362.21 (ROP). Options include V29.8 (observation and evaluation of newborns and infants for suspected condition not found; other specified suspected condition) or 362.89 (other retinal disorders). Reimbursement for V codes is miserable or nonexistent in most cases; thus, it is likely that some infants with a discharge code of 362.21 in the SPARCS never had ROP but were given a vague written diagnosis of ROP (e.g., “ROP exam,” “ROP stage zero”) so that the examiner could be reimbursed. Another possibility is that hospital discharge coders could not reliably distinguish between an infant who was examined for ROP and one who actually had ROP. A third possibility is that fundus examinations performed for other reasons (e.g., fungal vitritis) on an infant in the neonatal intensive care unit erroneously were considered to be related to ROP out of habit. Any of these is more likely than a child weighing >2500 g at birth actually having ROP or having a rare but phenotypically similar disease. Only in those cases where treatment was required can we be sure that a baby in the SPARCS with a diagnosis of 362.21 truly had ROP. If the necessary data could be extracted from the SPARCS, Chiang et al might investigate this problem by determining the total number of infants in the “ROP screening discretionary” weight categories who underwent fundus examinations. For example, if 200 of the 2948 infants who were 1500 to 1999 g at birth were examined and 115 of these had a diagnosis of ROP, either the neonatologists in New York State have an extraordinary ability to detect high-risk infants in this weight range or diagnosis coding errors are occurring routinely. In the latter case, regrettably, the data set used in this interesting study was corrupted, and the analysis and conclusions are unsupported. Incidence of retinopathy of prematurity from 1996 to 2000: Analysis of a comprehensive New York state patient databaseOphthalmologyVol. 111Issue 7PreviewTo determine the current incidence of retinopathy of prematurity (ROP) in New York state. Full-Text PDF Incidence of ROP in New York State: Author replyOphthalmologyVol. 112Issue 4PreviewWe are very aware of the practical difficulties associated with analyzing diagnostic codes in administrative databases. Brown and Farr have described several hypothetical scenarios in which overcoding for retinopathy of prematurity (ROP) (International Classification of Diseases 9, Clinical Modification, 362.21) could occur because of reimbursement concerns or human errors. We certainly agree that the accuracy of our results is dependent on the extent to which discharge diagnoses are properly coded, and our article addresses that topic in detail. Full-Text PDF