Purpose
The purpose of this study was to identify early predictors of pneumonia complications in patients with acute stroke requiring assistance for mobility.
Methods
We included 111 patients who suffered acute stroke (63 with cerebral infarction and 48 with cerebral hemorrhage). We retrospectively collected data from medical records to analyze their baseline and medical characteristics at admission, the presence or absence of pneumonia complication during their hospitalization, the status of physical therapy interventions, physical function at initial ambulation and discharge, and outcomes at discharge.  We performed logistic regression analysis of factors with a significance level of <0.2 as independent variables for the comparison between the two groups (presence or absence of pneumonia complication).  In addition, cutoff value was calculated using receiver operating characteristic (ROC)  curves.
Results
There were 17 patients (15.3%) with pneumonia complications. In logistic regression analysis, the NIHSS at admission (OR 1.066, P = 0.024 ), a history of orthopedic disease (OR 4.212, P = 0.025), and the presence of dementia (OR 6.697, P = 0.026) were identified as predictors of pneumonia complications. The cutoff value of pneumonia complications with ROC curves was NIHSS 12.5 points at admission (sensitivity 58.8%, specificity 66.0%).
Conclusion
Concomitant diseases, such as orthopedic diseases and dementia, are important factors in addition to the severity of stroke and may predict pneumonia complications in patients requiring assistance with basic mobility immediately after admission.

PREDICTORS OF PNEUMONIA COMPLICATION IN PATIENTS WITH ACUTE STROKE REQUIRING ASSISTANCE FOR MOBILITY