Serum progesterone testing to predict ectopic pregnancy in symptomatic first-trimester patients

Study objective: This study was conducted to prospectively measure the accuracy of serum progesterone levels to detect ectopic pregnancy. Methods: Seven hundred sixteen symptomatic first-trimester emergency department patients with abdominal pain or vaginal bleeding at a tertiary care military teaching hospital had progesterone levels measured by radioimmunoassay with results unavailable to the treating physician. All patients were monitored until a criterion standard diagnosis of jntrauterine pregnancy or ectopic pregnancy was confirmed. Results: A 14-month derivation phase (n=399) used receiver operating characteristic curve testing to select a cutoff value of progesterone less than 22 ng/mL. A 12-month validation phase (n=317) then retested this cutoff value. Combining both phases, there were 434 (61%) viable intrauterine pregnancies, 229 (32%) nonviable intrauterine pregnancies, and 52 (7.3%) ectopic pregnancies, of which 17 were ruptured. Sensitivity, specificity, positive predictive values, and negative predictive values (95% confidence intervals) for progesterone levels less than 22 ng/mL to detect ectopic pregnancy were 100% (94% to 100%), 27% (23% to 30%), 10% (7% to 12%), and 100% (98% to 100%), respectively. Conclusion: Given similar disease prevalence, roughly one fourth (178/716) of symptomatic patients can be classified as low risk (0%, 95% confidence interval 0 to 2%) for having an ectopic pregnancy using a progesterone cutoff of 22 ng/ml. Whether implementation of rapid progesterone testing can safely expedite care and reduce the need for urgent diagnostic evaluation or admission remains to be determined.