Cough peak flow to predict extubation outcome: a systematic review and meta-analysis

This systematic review was designed to assess the usefulness of cough peak flow to predict the extubation outcome in subjects who passed a spontaneous breathing trial.

The search covered the scientific databases MEDLINE, Lilacs, Ibecs, Cinahl, SciELO, Cochrane, Scopus, Web of Science and gray literature. The Quality Assessment of Diagnostic Accuracy Studies was used to assess the methodological quality and risk of study bias. The statistical heterogeneity of the likelihood (LR) and diagnostic odds ratios were evaluated using forest plots and Cochran's Q statistic, and a crosshair summary Receiver Operating Characteristic plot using the multiple cutoffs model was calculated.

We initially retrieved 3,522 references from the databases; among these, 12 studies including 1,757 subjects were selected for the qualitative analysis. Many studies presented an unclear risk of bias in the "patient selection" and "flow and time" criteria. Among the 12 included studies, seven presented "high risk" and five "unclear risk" for the item "reference standard." The diagnostic performance of the cough peak flow for the extubation outcome was low to moderate when we considered the results from all included studies, with a +LR of 1.360 (95%CI 1.240 - 1.530), -LR of 0.218 (95%CI 0.159 - 0.293) and a diagnostic odds ratio of 6.450 (95%CI 4.490 - 9.090). A subgroup analysis including only the studies with a cutoff between 55 and 65 L/minute showed a slightly better, although still moderate, performance.

A cough peak flow assessment considering a cutoff between 55 and 65L/minute may be useful as a complementary measurement prior to extubation. Additional well-designed studies are necessary to identify the best method and equipment to record the cough peak flow as well as the best cutoff.