Clinical Factors Affecting the Accuracy of Ultrasonography in Symptomatic First-Trimester Pregnancy

OBJECTIVE: To evaluate factors associated with accuracy of initial ultrasonography in patients with symptomatic first-trimester pregnancy. METHODS: Ultrasound diagnosis in the emergency department was compared with an ultimate clinical diagnosis in women in need of a gynecologic consult. The sensitivity, specificity, positive predictive values, and accuracy of the initial ultrasound impression were calculated and stratified by criteria of diagnosis, serum human chorionic gonadotropin (hCG) levels, pain, and bleeding. RESULTS: Eighteen hundred eighty women were evaluated. Overall accuracy of initial ultrasound diagnosis was 78%. A probable ultrasound diagnosis of ectopic pregnancy (adnexal mass without the presence of a yolk sac or embryo) resulted in a higher sensitivity (42.1% compared with 13.2%: P<.001) but a lower positive predictive value (82.7% compared with 98%: P<.01) compared with a definite diagnosis. A probable ultrasound diagnosis of intrauterine pregnancy (double decidual sign without yolk sac or embryo) resulted in a higher sensitivity (36.0% compared with 4.0%; P<.001) and lower positive predictive value (58.8% compared with 87.0%; P>.001) compared with a definite diagnosis. The sensitivity (34.3% compared with 75.9%; P<.01) and positive predictive value (80.4% compared with 91.5%; P=.02) were lower for diagnosis of ectopic pregnancy when serum hCG level was less than 2,000 milli-international units/mL. Ultrasonography was less accurate when bleeding was the chief complaint (72.7% compared with 84.8% P<.006) but not substantially altered by pain as a chief complaint (78.0 compared with 77.8% P>.99). CONCLUSION: A substantial number of misdiagnoses can occur when initial ultrasound diagnosis of intrauterine pregnancy or ectopic pregnancy is made without evidence of a yolk sac or embryo, when hCG values are low, or when a patient has significant bleeding. (Obstet Gynecol 2011;117:299-306) DOI: 10.1097/AOG.0b013e3182050ed0