The primary objective of the present study was to find an effective and simple scale to identify patients at high risk of pneumonia after AIS. It was the first study to evaluate the feasibility of the Braden Scale in predicting the occurrence of pneumonia after AIS. In this study, we explored a comprehensive range of demographic and clinical characteristics associated with post-stoke pneumonia. We found that the stroke classification, the NIHSS score and the score of the Braden Scale, especially the six subscales of the Braden Scale between the pneumonia and no pneumonia groups had significant differences.
In this study, pneumonia was found in 13.8% of patients presenting with an AIS, which was similar to the incidence of prior studies ranging from 5% to 26%[12-15]. The post-stoke pneumonia was associated with lower early and long-term survival, longer hospitalization, and higher disability at discharge[4]. So it was very important to prevent post-stoke pneumonia. However, a systematic review on efficacy of early antibiotics prophylaxis after stroke failed to show benefit in patients’ outcome[16]. This might be attributable to inclusion of patients with low risk of developing post-stoke pneumonia in these studies. It was critical to find an effective scale to predict the occurrence of pneumonia in patients after AIS, and to intervene in high risk patients to prevent pneumonia and improve the outcome of patients. The Braden scale was composed of six subscales: sensory perception, skin moisture, activity, mobility, nutrition, and friction and shear, which seemed to be related to the occurrence of pneumonia. In this study, we found that the mean score of the Braden Scale in the pneumonia group was significantly lower than that in the no pneumonia group, and the six subscales of the Braden Scale between the two groups had significant differences. And, the AUC for the Braden scale predicting post-stoke pneumonia was 0.883, which was identified as good accuracy, as shown above. With 18 points as the demarcation score, the sensitivity and the specificity was high.
We also found that the NIHSS score in the pneumonia group was significantly higher than the no pneumonia group. The NIHSS score has been found to be an independent predictor of pneumonia in some prior prediction models[14, 17-19]. The occurrence of pneumonia in patients with a higher NIHSS score may be due to the decreased consciousness or to gastroesophageal reflux because of a supine or recumbent position. This result also suggested that the pneumonia group had a higher neurological deficit. Previous studies had confirmed that patients with cardiogenic embolism tended to have more neurological deficits[20], and our study supported this conclusion that patients with cardiogenic embolism are more likely to be complicated with pneumonia. However, the Braden Scale had a greater advantage in quantifying risk factors and evaluating the incidence of post-stoke pneumonia.
Several post-stoke pneumonia prediction rules have been developed; however, these models have not been widely used in clinical practice. It is not our intention to show superiority of the the Braden Scale in predicting the occurrence of post-stoke pneumonia compared to the earlier scores; however, we want to point out the difference. Kwon et al[17] developed a pneumonia score, including age, sex, mechanical ventilation, NIHSS score, and dysphagia. However, the study was not validated externally and was limited by small sample size. Sellars et al[6] presented several key predictors for post-stroke pneumonia, including older age, dysarthria, low abbreviated mental test score, modified Rankin Scale score >4, and failed water swallowing test. Although the model was informative, some useful predictors were not routinely collected. Chumbler et al[21] presented a three-level scoring system for predicting post-stroke pneumonia, including medical history of pneumonia, increasing NIHSS score, dysphagia, being found down at symptom onset, and older age. Although the model showed acceptable C-statistics, the study was limited by its lacking of validation and retrospective nature.
Our study had some limitations that deserve comment. First, as all observational studies, we cannot rule out the possibility that additional baseline variable (unmeasured confounders) might have some impact on the development of post-stroke pneumonia, such as use of angiotensin receptor blockers or angiotensin-converting enzyme inhibitors[22]. Second, the time course of post-stroke pneumonia was unclear. Because we only have information on new-onset post-stroke pneumonia during hospitalization without documentation of the exact date, our data allowed no conclusion as to whether patients with a longer length of stay were more likely to develop pneumonia or whether diagnosis of pneumonia leaded to a longer hospitalization. Third, the study included only hospitalized patients with AIS, and those patients died shortly after admission, in the emergency department, or treated in outpatient clinics were not included. Fourth, our study was from a single center with limited patients. Finally, the Braden Scale used for predicting the occurrence of post-stoke pneumonia need to be further validated in additional populations.