Cnossen Maryse C, Polinder Suzanne, Andriessen Teuntje M, van der Naalt Joukje, Haitsma Iain, Horn Janneke, Franschman Gaby, Vos Pieter E, Steyerberg Ewout W, Lingsma Hester
1Center for Medical Decision Making, Department of Public Health, Erasmus Medical Center, Rotterdam, The Netherlands.2Department of psychology, RIVAS Healthcare Group, Gorinchem, The Netherlands.3Department of Neurology, University Medical Center Groningen, Groningen, The Netherlands.4Department of Neurosurgery, Erasmus Medical Center Rotterdam, Rotterdam, The Netherlands.5Department of Intensive Care Medicine, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands.6Department of Anesthesiology, VU University Medical Center Amsterdam, Amsterdam, The Netherlands.7Department of neurology, Slingeland Hospital, Doetinchem, The Netherlands.
Crit Care Med. 2017 Apr;45(4):660-669. doi: 10.1097/CCM.0000000000002263.
Although guidelines have been developed to standardize care in traumatic brain injury, between-center variation in treatment approach has been frequently reported. We examined variation in treatment for traumatic brain injury by assessing factors influencing treatment and the association between treatment and patient outcome.
Secondary analysis of prospectively collected data.
Five level I trauma centers in the Netherlands (2008-2009).
Five hundred three patients with moderate or severe traumatic brain injury (Glasgow Coma Scale, 3-13).
We examined variation in seven treatment parameters: direct transfer, involvement of mobile medical team, mechanical ventilation, intracranial pressure monitoring, vasopressors, acute neurosurgical intervention, and extracranial operation.
Data were collected on patient characteristics, treatment, and 6-month Glasgow Outcome Scale-Extended. Multivariable logistic regression models were used to assess the extent to which treatment was determined by patient characteristics. To examine whether there were between-center differences in treatment, we used unadjusted and adjusted random effect models with the seven treatment parameters as dependent variables. The influence of treatment approach in a center (defined as aggressive and nonaggressive based on the frequency intracranial pressure monitoring) on outcome was assessed using multivariable random effect proportional odds regression models in those patients with an indication for intracranial pressure monitoring. Sensitivity analyses were performed to test alternative definitions of aggressiveness. Treatment was modestly related to patient characteristics (Nagelkerke R range, 0.12-0.52) and varied widely among centers, even after case-mix correction. Outcome was more favorable in patients treated in aggressive centers than those treated in nonaggressive centers (OR, 1.73; 95% CI, 1.05-3.15). Sensitivity analyses, however, illustrated that the aggressiveness-outcome association was dependent on the definition used.
The considerable between-center variation in treatment for patients with brain injury can only partly be explained by differences in patient characteristics. An aggressive treatment approach may imply better outcome although further confirmation is required.
尽管已经制定了指南以规范创伤性脑损伤的治疗，但治疗方法在不同中心之间的差异仍屡有报道。我们通过评估影响治疗的因素以及治疗与患者预后之间的关联，来研究创伤性脑损伤治疗的差异。
对前瞻性收集的数据进行二次分析。
荷兰的五家一级创伤中心（2008 - 2009年）。
503例中度或重度创伤性脑损伤患者（格拉斯哥昏迷量表评分为3 - 13分）。
我们研究了七个治疗参数的差异：直接转运、移动医疗队的参与、机械通气、颅内压监测、血管升压药、急性神经外科干预和颅外手术。
收集了患者特征、治疗情况以及6个月的扩展格拉斯哥预后量表的数据。多变量逻辑回归模型用于评估治疗在多大程度上由患者特征决定。为了检验各中心之间治疗是否存在差异，我们使用了以七个治疗参数为因变量的未调整和调整后的随机效应模型。在有颅内压监测指征的患者中，使用多变量随机效应比例优势回归模型评估中心的治疗方法（根据颅内压监测频率定义为积极和非积极）对预后的影响。进行敏感性分析以测试积极性的替代定义。治疗与患者特征有适度关联（Nagelkerke R范围为0.12 - 0.52），即使在病例组合校正后，各中心之间的差异仍然很大。在积极治疗中心接受治疗的患者预后比在非积极治疗中心接受治疗的患者更有利（比值比，1.73；95%置信区间，1.05 - 3.15）。然而，敏感性分析表明，积极性与预后的关联取决于所使用的定义。
脑损伤患者治疗在不同中心之间存在的显著差异，只能部分地由患者特征的差异来解释。积极的治疗方法可能意味着更好的预后，尽管还需要进一步证实。
