TRANSFUSION PRACTICE IN MEDICAL PATIENTS

Objective: Blood transfusion raises serious issues of safety and econo mics. We therefore examined blood usage and its characteristics in med ical inpatients, given the relative scarcity of existing data. Design: During a 1-year period, transfusion episodes on two medical services were reviewed by five specialists for justifiability on the basis of g enerally agreed on guidelines. Setting: The study was conducted at two institutions, a municipal teaching hospital where house staff deliver care and a community hospital where patients are under the direct car e of private physicians. Patients: Four hundred thirty-eight randomly selected transfusion episodes on the medical services of the two insti tutions were reviewed. Main Outcome Measures: The prevalence of unjust ifiable transfusions based only on the information available to the ma naging physician at the time of transfusion. Results: Eighteen percent of the 438 randomly selected transfusion episodes were viewed as not justifiable by at least four of five reviewers; another 17% were class ified as equivocal because two or three reviewers judged them to be no t justifiable. The most striking observation was the greater prevalenc e of nonjustifiable transfusion episodes at the community hospital (26 % vs 16% at the teaching institution; P=.0121). Other observations inc luded a tendency for physicians to prescribe transfusions by the numbe rs (at least 11% of nonjustifiable transfusions) and to overtransfuse. The routineness with which transfusion was viewed by managing physici ans was also identifiable by the absence of written transfusion notes in 39% of all episodes reviewed, which incidentally raises questions a bout the adequacy of the medical chart's documentary functions today. Conclusions: The rate of nonjustifiable or equivocal transfusion on me dical services may be as high as 35%. Reliance on numbers rather than clinical status seems to be a major problem. Education is obviously a critical issue and should also target private practitioners, who seeme d to perform less well than physicians in training. Transfusion guidel ines that use specific hematocrit values also need to be reexamined.