There is little available evidence to demonstrate how cough strength mediates the risk of aspiration-related pneumonia in acute stroke. Our secondary analysis of trial data indicates that risk of pneumonia reduces with increasing peak cough flow (PCF) of voluntary cough (OR 0.994 for each 1 L/min increase in PCF, 95% CI 0.988 to 1.0, p=0.035); and to a lesser degree with increasing PCF of reflex cough (OR 0.998 for each 1 L/min increase in PCF, 95% CI 0.992 to 1.004, p=0.475). These data serve hypothesis generation. Further studies are needed to confirm these findings and validate their clinical utility.
Clinical trial registration number ISRCTN40298220 (post-results).
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Cough is the most immediate defence mechanism against aspiration.1 It is a commonly encountered clinical belief that strong cough offers some protection from aspiration-related pneumonia, although there is little evidence available to support this. Data from our completed trial of respiratory muscle training in acute stroke (ISRCTN40298220) allowed us to examine the association between cough flow and pneumonia risk. We have previously shown that stroke leads to impairment of both voluntary and reflex cough.2 ,3 Here, we present an exploratory secondary analysis of trial data, examining whether higher peak cough flow (PCF) (indicating stronger cough) might be protective against pneumonia in patients with stroke and swallowing problems.
Data from 72 patients were available for this analysis. Study procedures have been detailed previously.4 Briefly, we recruited adults within 2 weeks of stroke, and excluded patients with significant cardiac/pulmonary disease; neurological conditions other than stroke; orthopaedic conditions affecting respiratory mechanics; inability to cooperate or signs of pneumonia at enrolment. Swallowing function was described according to standardised bedside swallow assessment.5 We measured cough flow of volitional and capsaicin-induced reflex cough, using a calibrated pneumotachograph with full face mask.4 Pneumonia was observed for 4 weeks following baseline assessment and determined from documented medical diagnosis.
Our analysis was hypothesis-driven, assuming the data structure of a longitudinal observational study and examining only the predictor PCF for outcome pneumonia. First, we stratified the sample according to aspiration risk and pneumonia, and conducted group comparison tests. Second, we used logistic regression to examine the association between PCF and outcome pneumonia in the unsafe-swallow group. Third, we categorised patients in the unsafe-swallow group in two groups of high and low voluntary PCF, using an arbitrary threshold of 400 L/min; and we calculated the OR for outcome pneumonia according to dichotomised PCF. All analyses were conducted using Stata V.11.2 statistical software.
Analysis of the sample stratified by aspiration risk showed that PCF of voluntary cough was significantly lower in patients who had unsafe swallow and who developed pneumonia (table 1, see online supplementary figures S1 and S2). Full sample characteristics at baseline are given in online supplementary table S2.
Logistic regression showed a statistically significant association between PCF of voluntary cough and pneumonia (OR 0.994 for each 1 L/min increase in PCF, 95% CI 0.988 to 1.0, p=0.035). The association between PCF of reflex cough and pneumonia was smaller and not statistically significant (OR 0.998 for each 1 L/min increase in PCF, 95% CI 0.992 to 1.004, p=0.475). Goodness of fit indicators were adequate (Pearson χ2 and Hosmer–Lemeshow tests, p>0.05). Stata outputs for the logistic regression are given in online supplementary tables S3 and S4.
Categorising patients with unsafe swallow according to a threshold of 400 L/min voluntary PCF resulted in 22 patients in the low-PCF category, out of which nine developed pneumonia; and 11 patients in the high-PCF group, out of which two developed pneumonia. The risk of pneumonia was approximately three times higher for patients in the low-PCF group, although this was not statistically significant (OR 3.12, 95% CI 0.45 to 35.24). The Stata output is given in online supplementary table S5.
Our data lend support to the notion that strong cough protects from aspiration-related pneumonia. This association was stronger for voluntary cough, which leads us to hypothesise that PCF of voluntary cough might serve as a useful predictor of pneumonia risk in acute stroke. Logistic regression showed that each increase in voluntary PCF by 1 L/min reduced the risk of pneumonia by 0.6% (OR 0.994). The equivalent ORs for an increase in voluntary PCF by 50 and 100 L/min are approximately 0.73 and 0.53, respectively.
To illustrate how application of a PCF threshold might inform pneumonia risk in clinical practice, we applied an informed, although somewhat arbitrary cut-off of 400 L/min to categorise patients into those with stronger and those with weaker voluntary cough. The appropriate cut-off for PCF in stroke patients is currently not known. The small sample size is a limitation to this analysis. Although we maximised statistical precision by examining only one association of interest, which was defined a priori, studies with larger sample sizes are required to develop more sophisticated multivariable predictor models, which would also allow adjustment for other known risk factors of poststroke pneumonia.6
Further limitations to this analysis are trial eligibility criteria, which may have introduced selection bias not present in observational studies on consecutive patients. Respiratory muscle training in the intervention group may have affected the incidence of pneumonia, but this is unlikely as the trial showed no effect of these exercises on PCF compared with control patients. Although criteria based, pneumonia was physician diagnosed, but detection bias is unlikely as physicians were masked to allocation and to baseline assessments. Any future study of PCF and pneumonia risk would benefit from robust methods for diagnosing pneumonia.7 In particular, the potential for diagnosis to be influenced by the diagnosing physician's subjective assessment of cough strength needs to be considered.
Despite limitations, the present analysis provides potentially valuable findings in a little researched field. Measurement of cough flow may provide an objective, device-based method to inform pneumonia risk in patients with stroke and unsafe swallow at the bedside. Further studies are needed to confirm these results and validate their clinical application.
Anthony Rudd, David Smithard, Maxine Power, Charles Wolfe and Robin Cant were coapplicants on the study grant. We wish to thank the patients who participated in this study.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Data supplement 1 - Online supplement
Contributors STK: design, data collection, analysis and interpretation, manuscript writing. SSB and GFR: design, data collection, analysis and interpretation. JH: analysis and interpretation. JM: design, analysis and interpretation. LK: conception and design, data analysis and interpretation. All authors critically revised and approved the final manuscript.
Funding This article presents independent research funded by the UK National Institute for Health Research (NIHR), Research for Patient Benefit Programme (grant reference number PB-PG-0408-16096). The views expressed are those of the authors and not necessarily those of the National Health Service, NIHR or Department of Health.
Competing interests None declared.
Patient consent Obtained.
Ethics approval UK National Research Ethics Service, South West London Research Ethics Committee 3, reference 10/H0803/32.
Provenance and peer review Not commissioned; externally peer reviewed.