This prospective observational study was performed on a convenience sample of 186 female patients who presented to 1 of 2 EDs because of pelvic or abdominal symptoms or both and who had a bimanual pelvic examination as part of the routine ED evaluation. The study was performed from August 1996 through August 1997 at a private university hospital and a public county hospital that are staffed by attending emergency medicine physicians who share a second-through fourth-year emergency medicine residency program. The hospitals have a combined annual ED census of more than 90,000 patients, and residents, who do a 2-week obstetrics rotation during their first residency year, perform about 200 bimanual examinations in the ED by the beginning of the third year of residency training, as 1 of us (P L D) has estimated.
Research assistants approached senior (3rd- and 4th- year) residents and attending emergency physicians who were preparing to do a pelvic examination as part of the routine evaluation of an ED patient and asked them to participate in the study. If the physician agreed, the patient was asked to participate. Eligible patients included any woman with a presenting complaint thought by the clinician to require a pelvic examination in the normal course of evaluation. Exclusion criteria included age younger than 18 years or having had a hysterectomy.
After the treating physician filled out a standardized data form regarding the bimanual examination, a second emergency physician with similar experience (3rd- or 4th- year resident or attending physician) was recruited to perform a second pelvic examination and to then complete a standardized data form. The 2 physicians were counseled not to communicate with each other regarding the examination before completing the data forms.
The patient's chief complaint and history of present illness were available to both examiners, and each physician could ask the patient any questions he or she thought would aid in the evaluation. No specific instruction was given to examiners about how to perform a bimanual examination, and definitions of abnormal (positive) and normal (“negative”) findings were not provided. Rectovaginal examination was neither suggested nor recorded. Examiners could perform a speculum examination as desired, although this was not a measured aspect of the study. The time at which each physician examination occurred was recorded.
The data form included the evaluation of cervical motion tenderness, uterine tenderness, right adnexal tenderness, right adnexal mass, left adnexal tenderness, and left adnexal mass. For the purposes of this study, responses for each variable were limited to “clinically significant,” “not clinically significant,” and “unsure” because of examination limitations. Research assistants were instructed to clarify, if asked by the clinician, that “clinically significant” implied a finding about which the physician would then take some (diagnostic or therapeutic) action. Subjects were also asked to evaluate uterine size. For this variable, the physician could give a number in centimeters or answer “unsure” because of examination limitations. When a physician answered “unsure,” he or she was asked to note the factors that limited the examination.
The percentages of agreement and of positive agreement were calculated for each variable. The percentage of agreement is defined as the number of observations in which the examiners agree, divided by the total number of observations. The percentage of positive agreement evaluates the subset of patients for whom 1 or both examiners reported an abnormality. It is defined as the number of observations in which the examiners agreed on an abnormal finding, divided by the total number of observations in which 1 or both examiners reported an abnormality. For adnexal mass and adnexal tenderness, the results for the right and left sides were combined.
The study population was subdivided according to whether the examiners categorized the examination as limited. In addition, to evaluate whether the results of the examinations were found to be unreliable because of actual differences in the patient's examination over time, we divided subjects into 2 groups based on the interval between examinations. We grouped a response of “unsure” with responses of “clinically significant” because physicians are likely to continue the evaluation (ie, order an ultrasonogram, expedite follow-up) if uncertain about physical examination findings.
Uterine sizes determined by either examiner to be less than 12 cm were evaluated. Examiner's estimates that differed by more than 2 cm defined discordance.
The human subjects review board at both institutions approved the study with a waiver of written consent. Verbal informed consent was required for patient participation.