Abstract
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 X 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.
| Original language | English (US) |
|---|---|
| Pages (from-to) | 463-469 |
| Number of pages | 7 |
| Journal | Journal of Ultrasound in Medicine |
| Volume | 30 |
| Issue number | 4 |
| State | Published - Apr 1 2011 |
| Externally published | Yes |
Fingerprint
Keywords
- Down syndrome
- Genetic sonogram
- Likelihood ratios
- Logistic regression coefficients
ASJC Scopus subject areas
- Radiology Nuclear Medicine and imaging
- Radiological and Ultrasound Technology
Cite this
The genetic sonogram : Comparing the use of likelihood ratios versus logistic regression coefficients for down syndrome screening. / Zhong, Yan; Longman, Ryan E; Bradshaw, Rachael; Odibo, Anthony O.
In: Journal of Ultrasound in Medicine, Vol. 30, No. 4, 01.04.2011, p. 463-469.Research output: Contribution to journal › Article
}
TY - JOUR
T1 - The genetic sonogram
T2 - Comparing the use of likelihood ratios versus logistic regression coefficients for down syndrome screening
AU - Zhong, Yan
AU - Longman, Ryan E
AU - Bradshaw, Rachael
AU - Odibo, Anthony O.
PY - 2011/4/1
Y1 - 2011/4/1
N2 - 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 X 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.
AB - 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 X 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.
KW - Down syndrome
KW - Genetic sonogram
KW - Likelihood ratios
KW - Logistic regression coefficients
UR - http://www.scopus.com/inward/record.url?scp=79953309689&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=79953309689&partnerID=8YFLogxK
M3 - Article
C2 - 21460145
AN - SCOPUS:79953309689
VL - 30
SP - 463
EP - 469
JO - Journal of Ultrasound in Medicine
JF - Journal of Ultrasound in Medicine
SN - 0278-4297
IS - 4
ER -