OBJECTIVE: To investigate variability between hand-written and computerized anesthesia records and evaluate any associated bias. DESIGN AND MEASUREMENTS: A computer system that was used to collect intraoperative data for a study of hemodynamic management during coronary artery bypass graft surgery is described. The system collected and recorded hemodynamic data automatically downloaded from the anesthesia monitor as well as surgical events and drug administration data entered through menu options. The system then combined, summarized, and graphed the data as well as formatted it for export to a commercially available database program. In a sample of 14 patients, blood pressure data collected by the computer system was compared with the blood pressure data charted in the hand-written anesthesia record. MAIN RESULTS: Although general linear models controlling for within-patient variation and randomization assignment for mean arterial pressure range on cardiopulmonary bypass showed a significant relationship; low R2 values indicated that much of the variability could not be explained and that there was, therefore, poor agreement between the two records. Furthermore, a systematic bias in the hand-written anesthesia record was found when the computer system record was compared with the hand-written record and to the difference of the two records, so that extremes seen in the computer system record tended to be minimized in the hand-written anesthesia record. CONCLUSIONS: Because of the lack of explained variability between the computer system and hand-written anesthesia records and the bias in the hand-written anesthesia record, the hand-written anesthesia record should not be relied on as a source of accurate data for research purposes.
OBJECTIVE: To investigate variability between hand-written and computerized anesthesia records and evaluate any associated bias. DESIGN AND MEASUREMENTS: A computer system that was used to collect intraoperative data for a study of hemodynamic management during coronary artery bypass graft surgery is described. The system collected and recorded hemodynamic data automatically downloaded from the anesthesia monitor as well as surgical events and drug administration data entered through menu options. The system then combined, summarized, and graphed the data as well as formatted it for export to a commercially available database program. In a sample of 14 patients, blood pressure data collected by the computer system was compared with the blood pressure data charted in the hand-written anesthesia record. MAIN RESULTS: Although general linear models controlling for within-patient variation and randomization assignment for mean arterial pressure range on cardiopulmonary bypass showed a significant relationship; low R2 values indicated that much of the variability could not be explained and that there was, therefore, poor agreement between the two records. Furthermore, a systematic bias in the hand-written anesthesia record was found when the computer system record was compared with the hand-written record and to the difference of the two records, so that extremes seen in the computer system record tended to be minimized in the hand-written anesthesia record. CONCLUSIONS: Because of the lack of explained variability between the computer system and hand-written anesthesia records and the bias in the hand-written anesthesia record, the hand-written anesthesia record should not be relied on as a source of accurate data for research purposes.