Clinical and lung-function variables associated with vocal cord dysfunction.

BACKGROUND: Vocal cord dysfunction (VCD) is difficult to diagnose. Laryngoscopy while the patient is symptomatic is the accepted standard method to establish a diagnosis of VCD, but patient characteristics and spirometry values are thought to be useful for predicting VCD. We sought to identify clinical and spirometric variables that suggest VCD.

METHODS: We performed 2 parallel studies. First, 3 staff pulmonologists (who were blinded to the laryngoscopy results), scored 3 flow-volume loops from each PFT session on the likelihood that the inspiratory curve indicated VCD. We also performed a cross-sectional study of clinical characteristics and spirometric data from all patients who underwent laryngoscopy for any indication, including suspected VCD, over a 3-year period. We compared the laryngoscopy findings to the clinical characteristics, spirometry results, and the pulmonologists' assessments of the flow-volume loops. We used multivariate logistic regression to identify independent predictors of VCD.

RESULTS: The pulmonologists agreed about which flow-volume loops predicted VCD (quadratic kappa range 0.55-0.76), but those ratings were not predictive of laryngoscopic diagnosis of VCD. During the study period, 226 patients underwent laryngoscopy. One hundred (44%) were diagnosed with VCD. Independent predictors of VCD included female sex (odds ratio 2.72, 95% confidence interval 1.55-4.75) and obesity (body mass index > 30 kg/m(2)) (odds ratio 2.06, 95% confidence interval 1.12-3.80). With spirometric data from the effort that had the best forced-vital-capacity, multivariate analysis found the ratio of the forced inspiratory flow at 25% of the inspired volume to forced inspiratory flow at 75% of the inspired volume (FIF(25%/75%)) predictive of VCD (odds ratio 1.97, 95% confidence interval 1.12-3.44). The diagnostic performance of these characteristics was poor; the area under the receiver-operating-characteristic curve was 0.68. With the spirometric data from the effort that had the subjectively determined best inspiratory curve, and after controlling for the reproducibility of the inspiratory curves, multivariate analysis found none of the spirometric variables predictive of VCD.

CONCLUSIONS: VCD remains difficult to predict with spirometry or flow-volume loops. If VCD is suspected, normal flow-volume loop patterns should not influence the decision to perform laryngoscopy.