2018 Section 6 - Laryngology, Voice Disorders, and Bronchoesophalogy

Fig. 2. Summary receiver operating characteristics plot of videostroboscopy. (Left panel) Circles depict study size. (Right panel) Studies’ reference numbers

recently been questioned. 6,7,16 Colden et al. 7 described a normal VS evaluation in only two of 28 vocal cord lesions with not more than intraepithelial atypia. Addi- tionally, a substantial inter-rater variability in the strob- oscopic evaluation was noted. Djukic et al. 6 demonstrated an absence of mucosal wave by VS in as many as 38.5% of patients with only moderate dysplasia without any signs of invasive lesions. This study did not meet our inclusion criteria, because it only involved patients with premalignant lesions. Correspondingly, Shang et al. 16 reported superior accuracy of diffusion- weighted magnetic resonance imaging compared to VS (96% vs. 76%, respectively) in differentiating laryngeal cancer from precursor lesions, but missing details con- cerning VS findings and histopathological evaluation prevented inclusion in the present study. Furthermore, patients with nonglottic laryngeal cancers were included in the study population, and those were impossible to exclude from the reported data.

diseased population as having the disease. On average, VS correctly identifies 96% of cancers. On the contrary, the average specificity indicates that only 65% of the tested population is correctly identified as not having cancer. Also, the confidence interval around the specific- ity estimate covers the entire possible range, indicating a high level of uncertainty of the estimate. Risk of Bias Across Studies Egger’s test showed that the results of the meta- analysis were not subject to publication bias in the way that smaller studies contribute disproportionately to the overall results ( P 5 .7, data not shown elsewhere). DISCUSSION For years, VS has been recommended for assessing vocal cord lesions, especially when invasive lesions are suspected. 2,3,14,15 However, the value of VS in predicting the degree of dysplasia or malignant invasion has

Fig. 3. Forest plot of sensitivities of individual studies based on Wilson’s scores.

Fig. 4. Forest plot of specificities of individual studies based on Wilson’s scores.

Laryngoscope 126: September 2016

Mehlum et al.: VS and Prediction of Early Glottic Cancer

24