Jennett B, Teasdale G, Braakman R, Minderhoud J, Knill-Jones R
Lancet. 1976 May 15;1(7968):1031-4. doi: 10.1016/s0140-6736(76)92215-7.
Clinical data available in the first few days after severe head injury have been collected prospectively on 600 patients from Glasgow and the Netherlands and stored on computer; the patients in the two countries were similar in initial severity and in their outcome on a defined scale at 6 months. Calculated predictions of outcome were made in 200 randomly selected cases, using Bayesian statistics to compare the data from each patient with those from the 400 remaining cases whose outcome was known. Confident predictions (greater than 0-97 probability) were made in 44% of cases within the first 24 hours of coma, and in 52-61% when data up to 3 days were available. The higher confidence-rate occurred when prediction was limited to two outcomes (death or survival). Comparison of predicted with actual outcomes showed that 96-98% of confident predictions were correct. More logical clinical decisions should be possible when these predictions are available.
对来自格拉斯哥和荷兰的600名严重颅脑损伤患者在伤后的头几天收集了前瞻性临床数据,并存储在计算机中;这两个国家的患者在初始严重程度以及6个月时按既定量表得出的结果方面相似。在200例随机选择的病例中进行了预后的计算预测,使用贝叶斯统计将每位患者的数据与其余400例已知预后的病例的数据进行比较。在昏迷后的头24小时内,44%的病例做出了可靠预测(概率大于0.97),当有3天的数据时,这一比例为52%至61%。当预测仅限于两种结果(死亡或存活)时,置信率更高。预测结果与实际结果的比较表明,96%至98%的可靠预测是正确的。当有这些预测时,应该能够做出更符合逻辑的临床决策。