Patient history and physician suspicion accurately exclude pregnancy

Abstract

Objectives

Reliance upon patient assessment in excluding pregnancy is questionable. Physicians are encouraged to obtain pregnancy tests in all women of childbearing age. We affirmed the accuracy of women and their physicians in predicting pregnancy.

Methods

This was a prospective, observational study performed in a suburban academic emergency department on adult women with an ordered pregnancy test. A standardized gynecologic history was obtained by trained research assistants. Subjects estimated their likelihood of pregnancy as impossible, possible, or definite. Emergency department physicians estimated the likelihood of pregnancy as high, moderate, or low. All women had either a serum or urine β-human chorionic antigen. The diagnostic characteristics of patient and physician predictions of pregnancy were calculated with 95% confidence intervals (CIs).

Results

We enrolled 377 subjects. Median age was 29 (interquartile range, 22-37) years. Twelve percent of the women were pregnant. Women's estimates of pregnancy were as follows: impossible, 64.7%; possible, 22.5%; and definite, 12.7%. The pregnancy rates among women with estimates of impossible, possible, and definite were 0% (95% CI, 0%-1.5%), 4.7% (95% CI, 1.9%-11.5%), and 89.6% (95% CI, 77.8%-95.5%) (P < .001). Physicians' suspicions of pregnancy were high (13.7%), moderate (11.3%), and low (75.1%). The rate of pregnancy among low, moderate, and high physician suspicion groups were 0% (95% CI, 0%-1.4%), 9.5% (95% CI, 3.8%-22%), and 84.3% (95% CI, 72%-92%) (P < .001).

Conclusions

There were no pregnancies among women who estimated pregnancy as impossible or whose physicians thought that the likelihood of pregnancy was low. Routine pregnancy testing before radiological imaging and medication administration may not be required in adult women of childbearing age.

Introduction

Women of childbearing age account for nearly 30 million emergency department (ED) visits [1]. Knowledge of a patients' pregnancy status is an important component of diagnostic reasoning and treatment. In 2005, nearly 5 million pregnancy tests were ordered in the ED [1]. In many cases, ordering of a urine or serum β-human chorionic antigen (HCG) pregnancy test increases the cost and length of stay for these patients particularly if no blood or urine tests would have been ordered if pregnancy were not a concern. There are conflicting results on the reliability of patient history and assessment in excluding pregnancy in the ED set ting. Ramoska et al [2] documented an 11.5% rate of pregnancy in patient reporting “no chance” of pregnancy. Strote and Chen [3] found a 0.3% rate of pregnancy in patients reporting “no chance” of pregnancy. With such discrepancy reported in the literature, our goal was to confirm the reliability of patient assessment in excluding pregnancy. We hypothesized that the likelihood of pregnancy would be very low (<1%) in women (and/or their physicians) who felt that pregnancy was impossible.