Factors related to swallowing recovery for nil per os patients after stroke and prediction model
10.3969/j.issn.1006-9771.2022.04.012
- VernacularTitle:非经口进食吞咽障碍脑卒中患者预后的相关因素及预测模型
- Author:
Bo WANG
1
,
2
;
Yongxue YUAN
1
,
2
;
Qingsu ZHANG
1
,
2
Author Information
1. Beijing Bo'ai Hospital, China Rehabilitation Research Center, Beijing 100068, China
2. Capital Medical University School of Rehabilitation Medicine, Beijing 100068, China
- Publication Type:Journal Article
- Keywords:
dysphagia, oral feeding, prognostic model, Logistic regression
- From:
Chinese Journal of Rehabilitation Theory and Practice
2022;28(4):453-460
- CountryChina
- Language:Chinese
-
Abstract:
Objective To explore the factors related to the recovery of nil per os (NPO) patients after stroke by retrospective data analysis, and to establish a predictive model.Methods The information of demographics, evaluation and treatment of 141 stroke patients admitted to the Hearing and Language Department in Beijing Bo'ai Hospital from April, 2017 to November, 2020 were selected. The predictive model was established by univariate analysis and Logistic regression. The fitting degree and discriminant validity of the model were evaluated by Hosmer-Lemeshow (H-L) test and receiver operating characteristic (ROC) curve. Other 121 patients with post-stroke dysphagia from December, 2020 to November, 2021 were used as the validation set to verify the model.Results For univariate analysis, National Institute of Health Stroke Scale (NIHSS) score, drinking water test results, autonomous cough ability, cough after swallowing, movement ability of tongue and jaw, and electrical stimulation treatment were significantly associated with the outcome (H=65.803, χ2 > 4.623, P<0.05). Multivariate Logistic regression analysis showed that NIHSS score (X1, OR=0.772, 95%CI 0.64 to 0.82, P<0.001), spontaneous cough ability (X2, OR=5.116, 95%CI 1.28 to 20.41, P=0.021), and electrical stimulation during treatment (X3, OR=94.718, 95%CI 5.65 to 1589.26, P=0.002) were independent factors for the outcome of swallowing function. Thus, the predictive model was P=11+e−(2.368−0.325X1+1.632X2+4.551X3)
P
=
1
1
+
e
-
(
2.368
-
0.325
X
1
+
1.632
X
2
+
4.551
X
3
)
, which was well fitting (P=0.845), with the largest area under curve (0.884). The overall accuracy of the model in the validation set was 91.7%.Conclusion The patients with dysphagia would like to recover well if he/she was with lower NIHSS scores and normal autonomous cough ability; meanwhile, the addition of electrical stimulation therapy in comprehensive rehabilitation may be helpful. A predictive model has been established, which needs a further research.
