OBJECTIVE: To investigate the diagnostic accuracy of screening serum P in diagnosis of ectopic pregnancy (EP) and to identify a cutoff value that provides the best compromise between test sensitivity and specificity. DESIGN: Retrospective analysis. SETTING: University hospital. INTERVENTIONS: Observation only. PATIENTS: First trimester pregnant women at risk for EP. MAIN OUTCOME MEASURE: Single P measurements were obtained from 3,674 pregnancies with outcomes defined as EP, viable intrauterine pregnancy (IUP), and spontaneous abortion (SAB). Diagnostic accuracy of the test was analyzed by generating receiver operating characteristic (ROC) curves, which quantify the ability of the test to distinguish EP and SAB from IUP. RESULTS: Diagnostic accuracy for EP versus IUP was 88.7% +/- 0.1% (mean +/- SEM); for SAB versus IUP, 93.8% +/- 0.4%; and for SAB + EP versus IUP, 92.8% +/- 0.4%. Diagnostic accuracy for SAB versus EP was only 39.4% +/- 0.2%. In the interval of 15.0 to 19.9 ng/mL (47.7 to 63.3 nmol/L), P missed 5.3% of the EPs and incorrectly included 84.3% of the viable IUPs; in the interval of 20.0 to 24.9 ng/mL (63.6 to 79.2 nmol/L), sensitivity improved in that only 3.5% of the EPs were missed but 88.8% of viable IUPs were included incorrectly. A cutoff value of > or = 17.5 ng/mL (55.7 nmol/L), the median point of the 15.0 to 19.9 ng/mL (47.7 to 63.3 nmol/L) interval, missed only 35 of 423 (8.3%) total EPs in the study. CONCLUSION: Analysis of ROC curves demonstrates that single serum P has high diagnostic accuracy for differentiating accidents of pregnancy (SAB and EP) from viable IUP, both individually (SAB versus IUP and EP versus IUP) and collectively (SAB + EP versus IUP); it cannot efficiently discriminate SAB versus EP. We conclude that for P > or = 17.5 ng/mL (55.7 nmol/L), patients thought to be at risk for EP may be followed reasonably without ultrasound or further invasive diagnostic studies.
OBJECTIVE: To investigate the diagnostic accuracy of screening serum P in diagnosis of ectopic pregnancy (EP) and to identify a cutoff value that provides the best compromise between test sensitivity and specificity. DESIGN: Retrospective analysis. SETTING: University hospital. INTERVENTIONS: Observation only. PATIENTS: First trimester pregnant women at risk for EP. MAIN OUTCOME MEASURE: Single P measurements were obtained from 3,674 pregnancies with outcomes defined as EP, viable intrauterine pregnancy (IUP), and spontaneous abortion (SAB). Diagnostic accuracy of the test was analyzed by generating receiver operating characteristic (ROC) curves, which quantify the ability of the test to distinguish EP and SAB from IUP. RESULTS: Diagnostic accuracy for EP versus IUP was 88.7% +/- 0.1% (mean +/- SEM); for SAB versus IUP, 93.8% +/- 0.4%; and for SAB + EP versus IUP, 92.8% +/- 0.4%. Diagnostic accuracy for SAB versus EP was only 39.4% +/- 0.2%. In the interval of 15.0 to 19.9 ng/mL (47.7 to 63.3 nmol/L), P missed 5.3% of the EPs and incorrectly included 84.3% of the viable IUPs; in the interval of 20.0 to 24.9 ng/mL (63.6 to 79.2 nmol/L), sensitivity improved in that only 3.5% of the EPs were missed but 88.8% of viable IUPs were included incorrectly. A cutoff value of > or = 17.5 ng/mL (55.7 nmol/L), the median point of the 15.0 to 19.9 ng/mL (47.7 to 63.3 nmol/L) interval, missed only 35 of 423 (8.3%) total EPs in the study. CONCLUSION: Analysis of ROC curves demonstrates that single serum P has high diagnostic accuracy for differentiating accidents of pregnancy (SAB and EP) from viable IUP, both individually (SAB versus IUP and EP versus IUP) and collectively (SAB + EP versus IUP); it cannot efficiently discriminate SAB versus EP. We conclude that for P > or = 17.5 ng/mL (55.7 nmol/L), patients thought to be at risk for EP may be followed reasonably without ultrasound or further invasive diagnostic studies.
Authors: Jorine Verhaegen; Ioannis D Gallos; Norah M van Mello; Mohamed Abdel-Aziz; Yemisi Takwoingi; Hoda Harb; Jonathan J Deeks; Ben W J Mol; Arri Coomarasamy Journal: BMJ Date: 2012-09-27