Optimal blood ordering for emergency department patients.

A lack of documented criteria for determining which emergency patients should have blood crossmatched for possible transfusion led us to review the records of all patients for whom blood was crossmatched in the emergency unit of our hospital in 1977. Of 378 patients who had blood crossmatched for 1,230 units, only 105 (28%) received blood transfusions (331 units total). Variables showing relationship to transfusion of blood were consolidated into four criteria: 1) shock; 2) hematocrit less than 30%; 3) observed blood loss of at least 500 ml or grossly visible gastrointestinal bleeding; and 4) emergency surgery with anticipated blood loss. Application of these criteria identified 55% of patients who did not receive transfusions and 41% of units unnecessarily crossmatched, while failing to identify only three patients received single-unit transfusions. Use of the criteria would have reduced the crossmatch/transfusion ration from 3.7:1 to 2.6:1.