• Yongfan Xie, Yongyi Wang, Yong Zhou, Mingxing Liu, Shengli Li, Yue Bao, Wenbo Jiang, Siwei Tang, Fangbao Li, Hao Xue, Luo Li, Xingyuan Gong, Yongliang Liu, Weimin Wang, and Tong Li.
• Department of Neurosurgery, Qingdao Municipal Hospital (Headquarters), No. 1 Jiaozhou Road, Qingdao, 266011, Shandong, People's Republic of China.
• Neurocrit Care. 2022 Apr 1; 36 (2): 421-433.

ObjectivesAcute respiratory failure (ARF) is a common medical complication in patients with cervical traumatic spinal cord injury (TSCI). To identify independent predictors for ARF onset in patients who underwent cervical TSCI without premorbid respiratory diseases and to apply appropriate medical supports based on accurate prediction, a nomogram relating admission clinical information was developed for predicting ARF during acute care period.MethodsWe retrospectively reviewed clinical profiles of patients who suffered cervical TSCI and were emergently admitted to Qingdao Municipal Hospital from 2014 to 2020 as the training cohort. Univariate analysis was performed using admission clinical variables to estimate associated factors and a nomogram for predicting ARF occurrence was generated based on the independent predictors from multivariate logistic regression analysis. This nomogram was assessed by concordance index for discrimination and calibration curve with internal-validated bootstrap strategy. Receiver operating characteristic curve was conducted to compare the predictive accuracy between the nomogram and the traditional gold standard, which combines neuroimaging and neurological measurements by using area under the receiver operating characteristic curve (AUC). An additional 56-patient cohort from another medical center was retrospectively reviewed as the test cohort for external validation of the nomogram.Results162 patients were eligible for this study and were included in the training cohort, among which 25 individuals developed ARF and were recorded to endure more complications. Despite the aggressive treatments and prolonged intensive care unit cares, 14 patients insulted with ARF died. Injury level, American Spinal Injury Association Impairment Scale (AIS) grade, admission hemoglobin (Hb), platelet to lymphocyte ratio, and neutrophil percentage to albumin ratio (NPAR) were independently associated with ARF onset. The concordance index of the nomogram incorporating these predictors was 0.933 in the training cohort and 0.955 in the test cohort, although both calibrations were good. The AUC of the nomogram was equal to concordance index, which presented better predictive accuracy compared with previous measurements using neuroimaging and AIS grade (AUC 0.933 versus 0.821, Delong's test p < 0.001). Similar significant results were also found in the test cohort (AUC 0.955 versus 0.765, Delong's test p = 0.034). In addition, this nomogram was translated to a Web-based calculator that could generate individual probability for ARF in a visualized form.ConclusionsThe nomogram incorporating the injury level, AIS grade, admission Hb, platelet to lymphocyte ratio, and NPAR is a promising model to predict ARF in patients with cervical TSCI who are absent from previous respiratory dysfunction. This nomogram can be offered to clinicians to stratify patients, strengthen evidence-based decision-making, and apply appropriate individualized treatment in the field of acute clinical care.© 2021. The Author(s).

15 keywords...

hide…