American Journal of Epidemiology Vol. 127, No. 4: 831-842
Copyright © 1988 by The Johns Hopkins University School of Hygiene and Public Health
research-article |
EVALUATING OBSTETRIC RISK SCORES BY RECEIVER OPERATING CHARACTERISTIC CURVES
1Institute for Biostatistics (Tranavaal) of the South African Medical Research Council Johannesburg, South Africa
2Department of Community Health, University of Witwatersrand, Medical School Johannesburg, South Africa
5Department of Statistics, University of Pretoria Pretoria, South Africa
Address reprint requests to Dr. Allen A. B. Herman
The purpose of obstetric risk scoring systems is to identify pregnancies that will end in poor perinatal outcomes such as perinatal death. The proportion of such pregnancies identified by a risk scoring system can be increased by choosing a more lenient threshold criterion for the score beyond which the pregnancy is regarded as being at risk. This increase in sensitivity is gained at the expense of a decrease in specificity. Thresholds differ between published risk scoring systems, making it difficult to assess their relative merits using the measures of sensitivity and specificity. However, risk scores can be compared by constructing receiver operating characteristic curves, i.e., plotting the true positive rate against the false positive rate for multiple thresholds and calculating the area under the receiver operating characteristic curve (Ø). In published systems used in this paper, values of Ø, used to rank published obstetric risk scoring systems, ranged from 0.490.95. All scores used prepregnancy and early antenatal information. The risk scores with the highest predictive accuracy tended to include factors representing past reproductive experience and those from the late antenatal and intrapartum periods. Studies with statistical weighting of risk factors tended to use fewer late prenatal and intrapartum risk factors.
epidemiologic methods; perinatology; pregnancy complications; prenatal care; prenatal diagnosis
3Present address: Gertrude H. Sergievsky Center, Columbia University, 630 West 168th Street, New York, NY 10032.
4Present address: School of Public Health and Tropical Medicine, University of Sydney, Sydney, Australia.
![]()
CiteULike
Connotea
Del.icio.us What's this?
This article has been cited by other articles:
![]() |
M. Isohanni, H. Oja, I. Moilanen, P. Rantakallio, and M. Koiranen The relation between teenage smoking and drinking, with special reference to non-standard family background Scand J Public Health, March 1, 1993; 21(1): 24 - 30. [Abstract] [PDF] |
||||
![]() |
J. Hilden The Area under the ROC Curve and Its Competitors Med Decis Making, June 1, 1991; 11(2): 95 - 101. [Abstract] [PDF] |
||||

