Randomized Trial of Computerized Quantitative Pretest Probability in Low-Risk Chest Pain Patients: Effect on Safety and Resource Use

Abstract

We hypothesize that the presentation of a quantitative pretest probability of acute coronary syndrome would safely reduce unnecessary resource use in low-risk emergency department (ED) chest pain patients. Randomized controlled trial of adult patients with chest pain paired with their clinicians. Patients had neither obvious evidence of acute coronary syndrome nor obvious other reason for admission. Clinicans provided their unstructured point estimate for pretest probability before randomization. Clinicans and patients in the intervention group received a printout of pretest probability of acute coronary syndrome result displayed numerically and graphically. Controls received no printout. Patients were followed for 45 days for predefined criteria of acute coronary syndrome and efficacy endpoints. Endpoints were compared between groups, with 95% confidence intervals (CIs) for differences. Four hundred were enrolled, and 31 were excluded for cocaine use or elopement from care. The mean pretest probability estimates of acute coronary syndrome were 4 (SD 5%) from clinicians and 4 (SD 6%) from the computer. Safety and efficacy endpoints for controls (n=185) versus intervention patients (n=184) were as follows: (1) delayed or missed diagnosis of acute coronary syndrome: 1 of 185 versus 0 of 184 (95% CI for difference -2.8% to 15.0%); (2) hospital admission with no significant cardiovascular diagnosis, 11% versus 5% (-0.2% to 11%); (3) thoracic imaging imparting greater than 5 mSv radiation with a negative result, 20% versus 9% (95% CI for difference = 3.8% to 18.0%); (4) median length of stay, 11.4 hours versus 9.2 hours (95% CI for difference = -2.9 to 7.6 hours); (5) reported feeling "very satisfied" with clinician explanation of problem on follow-up survey, 38% versus 49% (95% CI for difference = 0.9% to 21.0%); (6) readmitted within 7 days, 11% versus 4% (95% CI for difference = 2.5% to 13.2%). Presentation of a quantitative estimate of the pretest probability of acute coronary syndrome to clinicians and low-risk ED chest pain patients was associated with reduced resource use, without evidence of increased rate of premature discharge of patients with acute coronary syndrome.