PEDIATRICS Vol. 119 No. 6 June 2007, pp. 1246-1247 (doi:10.1542/peds.2007-0176)
LETTER TO THE EDITOR |
Score for Neonatal Acute Physiology (SNAP) or Vermont Oxford Risk-Adjustment Model for Very Low Birth Weight Infants?
Luigi Gagliardi, MDDivision of Neonatology and Pediatrics,
Ospedale Versilia,
I-55043 Lido di Camaiore, Italy
Italian Neonatal Network
Roberto Bellù, MD
Division of Neonatology and Neonatal Intensive Care Unit
Ospedale "A Manzoni"
23900 Lecco, Italy
Italian Neonatal Network
To the Editor.—
We read with great interest the article by Zupancic et al,1 which compared the revised Score for Neonatal Acute Physiology (SNAP-II) and revised Score for Neonatal Acute Physiology Perinatal Extension (SNAPPE-II) scores with the Vermont Oxford Network risk-adjustment algorithm (VON-RA) in a large cohort of term and preterm infants from North America. In very low birth weight (VLBW) infants, both scores performed equally well, as judged by their receiver operating characteristic (ROC) analysis results, the area being 0.86 for the SNAPPE-II and 0.85 for the VON-RA.
Given that treatment policies of small preterm infants are different in different countries, it is of interest to know how these scores perform in other settings.
We previously published an article2 in which we obtained, in VLBW infants, an ROC area of 0.84 for the SNAPPE-II, which is very similar to that found in the Zupancic et al study. In infants without congenital anomalies, the ROC area was 0.86. Here we report the performance of the VON-RA model in this data set: using the original logistic coefficients (kindly provided by J. Horbar, MD, and J. Carpenter, MS), the VON-RA model had an ROC area of 0.906 for infants without congenital anomalies. This value was confirmed in the complete cohort of 2070 VLBW infants3 admitted to 14 NICUs in Lombardy, northern Italy, where ROC areas of 0.908 and 0.907 for infants with and without congenital anomalies, respectively, were found.
We conclude that in our setting, the SNAPPE-II had a performance similar to that found in the Zupancic et al study (and in previous articles), whereas the VON-RA model had better discrimination. Interestingly, different centers showed greater variability in SNAPPE scores (mean score: 18–36 [P < .0001] by analysis of variance) than in VON-RA scores (P = .22).
We agree with the authors1 that a measure of severity of illness that takes into account individual characteristics of the infant (eg, urine output, core temperature, and pH, as in the SNAPPE-II) should be better, theoretically, than a risk score based only on fixed covariates (eg, gender, multiple pregnancy, outborn status, etc). Unfortunately, as the authors admit, no severity scoring system can be used to judge an individual infant's risk of mortality because of the wide confidence limits of the estimate. In fact, an ROC area represents the probability that the score of a randomly selected infant who died will be greater than the score in of a randomly selected normal (surviving) infant.4 Thus, even an ROC area of 0.86 would not look so impressive, meaning a 14% overlapping of scores between deaths and survivals, which is a figure that is too high to guide therapeutic decisions.
On the other hand, we found that for risk adjustment in groups of subjects (eg, for comparing hospital performances), the VON-RA worked very well and can be calculated from data collected routinely, as opposed to the SNAPPE, which requires careful and time-consuming collection of data for this purpose.
Finally, congenital anomalies add an important risk of death, even without increasing physiologic instability (ie, even without increasing the SNAPPE-II), and up to now no clinically useful way to quantify this risk has been found. We think that an empirical classification of congenital anomalies, calculated from the very large database of the VON, is a major improvement, and we hope that this classification will be made publicly available soon.
FOOTNOTES
Statements appearing here are those of the writers and do not represent the official position of the American Academy of Pediatrics or its Committees. Comments on any topic, including the contents of PEDIATRICS, are invited from all members of the profession; those accepted for publication will not be subject to major editorial revision but generally must be no more than 400 words in length. The editors reserve the right to publish replies and may solicit responses from authors and others.
Please see www.pediatrics.org for instructions on submitting letters.
REFERENCES
- Zupancic JAF, Richardson DK, Horbar JD, Carpenter JH, Lee SK, Escobar GJ. Revalidation of the Score for Neonatal Acute Physiology in the Vermont Oxford Network. Pediatrics. 2007;119(1) . Available at: www.pediatrics.org/cgi/content/full/119/1/e156
- Gagliardi L, Cavazza A, Brunelli A, et al. Assessing mortality risk in very low birthweight infants: a comparison of CRIB, CRIB-II, and SNAPPE-II.
Arch Dis Child Fetal Neonatal Ed. 2004;89
:F419
–F422
[Abstract/Free Full Text] - Gagliardi L, Agosti M, Barera G, et al. Survival to discharge of a cohort of very low birth weight infants born in Lombardy between 1999–2002. Ital J Pediatr. 2006;32 :169 –176
- Altman DG, Bland JM. Diagnostic tests 3: ROC plots.
BMJ. 1994;309
:188
[Free Full Text]
PEDIATRICS (ISSN 1098-4275). ©2007 by the American Academy of Pediatrics
CiteULike 
Connotea 
Del.icio.us 
Digg 
Facebook 
Reddit 
Technorati 
Twitter What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
