Premium
The Genetic Sonogram
Author(s) -
Zhong Yan,
Longman Ryan,
Bradshaw Rachael,
Odibo Anthony O.
Publication year - 2011
Publication title -
journal of ultrasound in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.574
H-Index - 91
eISSN - 1550-9613
pISSN - 0278-4297
DOI - 10.7863/jum.2011.30.4.463
Subject(s) - medicine
Objectives The purpose of this study was to compare the screening efficiency for Down syndrome using likelihood ratios versus logistic regression coefficients. Methods We conducted a retrospective study of women at increased risk for Down syndrome referred for a second‐trimester genetic sonogram. Likelihood ratios were calculated by multiplying the risk ratio from maternal serum screening by the likelihood ratios of sonographic markers. Logistic regression coefficients were calculated using a formula derived from β coefficients generated from a multivariable logistic regression model. The screening efficiency of both methods was tested in an independent population of patients. The McNemar test was used to compare the predictive ability of the two methods. Results In the validation population, the use of likelihood ratios had an area under the receiver operator characteristic curve of 0.90 for Down syndrome detection, whereas the use of logistic regression coefficients had an area under the curve of 0.86. Adopting a risk cutoff point of 1/270, the sensitivity of likelihood ratios was 77.4% (95% confidence interval [CI], 58.9%–90.4%) with a false‐positive rate of 17.9% (95% CI, 15.0%–21.1%), whereas the sensitivity of logistic regression coefficients was 93.5% (95% CI, 78.6%–99.2%) with a false‐positive rate of 34.6% (95% CI, 30.9%–38.4%). There was significant difference in screening efficiency for Down syndrome detection between the two methods (exact McNemar χ 2 , P < .001 ). Conclusions With a slight reduction in the Down syndrome detection rate, the use of the likelihood ratio approach was associated with a significantly lower false‐positive rate compared with the logistic regression approach.