Dahdah Marie N, Hofmann Melissa, Pretz Christopher, An Viktoriya, Barnes Sunni A, Bennett Monica, Dreer Laura E, Bergquist Thomas, Shafi Shahid
Baylor Institute for Rehabilitation, Dallas, Texas (Dr Dahdah); Baylor Regional Medical Center Plano, Plano, Texas (Dr Dahdah); North Texas TBIMS, Baylor Institute for Rehabilitation, Dallas (Drs Dahdah and Shafi); National Data & Statistical Center/Craig Hospital, Englewood, Colorado (Drs Hofmann, Pretz, and An); Office of the Chief Quality Officer, Baylor Scott & White Health, Dallas, Texas (Drs Barnes, Bennett, and Shafi); University of Alabama at Birmingham Traumatic Brain Injury Model System (Dr Dreer); and Department of Psychiatry and Psychology and Department of Physical Medicine and Rehabilitation, Mayo Clinic, Rochester, Minnesota (Dr Bergquist).
J Head Trauma Rehabil. 2017 Jul/Aug;32(4):E1-E10. doi: 10.1097/HTR.0000000000000270.
To examine differences in patient outcomes across Traumatic Brain Injury Model Systems (TBIMS) rehabilitation centers and factors that influence these differences using hierarchical linear modeling (HLM).
Sixteen TBIMS centers.
A total of 2056 individuals 16 years or older with moderate to severe traumatic brain injury (TBI) who received inpatient rehabilitation.
Multicenter observational cohort study using HLM to analyze prospectively collected data.
Functional Independence Measure and Disability Rating Scale total scores at discharge and 1 year post-TBI.
Duration of posttraumatic amnesia (PTA) demonstrated a significant inverse relationship with functional outcomes. However, the magnitude of this relationship (change in functional status for each additional day in PTA) varied among centers. Functional status at discharge from rehabilitation and at 1 year post-TBI could be predicted using the slope and intercept of each TBIMS center for the duration of PTA, by comparing it against the average slope and intercept.
HLM demonstrated center effect due to variability in the relationship between PTA and functional outcomes of patients. This variability is not accounted for in traditional linear regression modeling. Future studies examining variations in patient outcomes between centers should utilize HLM to measure the impact of additional factors that influence patient rehabilitation functional outcomes.
使用分层线性模型(HLM)研究创伤性脑损伤模型系统(TBIMS)康复中心患者预后的差异以及影响这些差异的因素。
16个TBIMS中心。
共有2056名16岁及以上的中重度创伤性脑损伤(TBI)患者接受了住院康复治疗。
采用HLM的多中心观察性队列研究,对前瞻性收集的数据进行分析。
出院时及创伤性脑损伤后1年的功能独立性测量和残疾评定量表总分。
创伤后遗忘症(PTA)的持续时间与功能结局呈显著负相关。然而,这种关系的强度(PTA每增加一天功能状态的变化)在各中心之间有所不同。通过将每个TBIMS中心PTA持续时间的斜率和截距与平均斜率和截距进行比较,可以预测康复出院时及创伤性脑损伤后1年的功能状态。
HLM显示出由于PTA与患者功能结局之间关系的变异性而产生的中心效应。这种变异性在传统线性回归模型中未得到体现。未来研究中心间患者结局差异时应利用HLM来衡量影响患者康复功能结局的其他因素的影响。