Predictive value of history and physical examination in patients with suspected ectopic pregnancy

Study objective: To identify historical and physical examination findings t hat are predictive of ectopic pregnancy (EP) in pregnant patients with abdo minal pain or bleeding. Methods: This study was conducted in an urban academic emergency department as a prospective observational study of consecutive patients from August 1 , 1991, to August 31, 1992, who had abdominal pain or vaginal bleeding and a positive beta-human chorionic gonadotropin level. Patients were excluded if they had a diagnostic ultrasound during a previous visit, or if the uter ine size was larger than 12 weeks by pelvic examination. Data were analyzed using chi(2) With a P value less than .05 identified as significant. Odds ratios were determined for significant variables. A classification and regr ession tree analysis was then performed using the predictive variables to d erive a decision tree. Results: Four hundred forty-one patients were enrolled, 57 of whom (13%) ha d an EP. Factors by history that increased the risk of EP included pain tha t was described as moderate to severe, lateral, or sharp. Pain located in t he midline decreased the risk of EP A history of previous intrauterine devi ce use, infertility, prior pelvic surgery, or tubal ligation were each foun d to be predictive. On physical examination, the presence of peritoneal sig ns, cervical motion tenderness, or lateral or bilateral abdominal or pelvic tenderness increased the risk of EP. A uterine size larger than 8 weeks by pelvic examination decreased the risk of EP Combinations of predictive var iables identified subsets of patients with either an increased or decreased frequency of EP but in no case was a combination identified that would con firm or exclude this diagnosis with a high degree of certainty. Conclusion: History and physical examination findings predictive of EP were identified. However no constellation of findings could confirm or exclude this diagnosis with a high degree of reliability.