br Analysis used Stata software
Analysis used Stata 14.2 software.
Eighty-seven patients treated for oral cavity or oropharyngeal cancer had voice recordings: 51 male, 36 female; mean age, 65.8 years (range, 36–87 years).
Thirty-five patients (40.2%) had oral cavity cancer: 16 oral floor, 8 tongue, 6 retromolar and 5 mandibular. Fifty-two (59.8%) had oropharyngeal cancer: 26 tonsillar, 13 tongue-base, 4 velar and 9 unspecified oropharyngeal.
Seventy-three patients (83.9%) underwent surgery with associ-ated Filipin Complex therapy, chemotherapy or neck dissection. Fourteen patients received either exclusive radiation therapy (n = 5) or chemoradiotherapy (n = 9).
3.2. Inter-observer reliability
Table 1 shows ICC values for perceptual evaluations. Agreement was moderate to good (>0.69) on all tasks except
reading intelligibility (0.43).
The panel of 6 speech therapists was thus homogeneous.
Table 2 shows mean score results.
Table 3 shows mean scores according to tumor size and location. Mean scores were lower for signal impairment than for intelli-gibility. There was also a tumor size effect, with lower scores on all tests in advanced (T3, T4) than early tumor (T1, T2: Dwivedi, 2012).
Scores were higher in oropharyngeal than oral cavity cancer.
Distribution analysis showed a ceiling effect for no or slight dis-order, which was stronger on intelligibility than signal impairment scores. Scatter was greater for signal impairment than intelligibility (Fig. 1).
Results for “degree of intelligibility” (on picture description, with continuous values) and for “intelligibility” (discrete values from 0 to 10) were similar. Even so, comparison of means for matched series on non-parametric Wilcoxon test showed a signif-icant difference (P = 0.001), with a higher mean value for degree of intelligibility than for intelligibility (7.96 vs. 7.61).
3.4. Correlations between mean signal impairment and intelligibility scores
Correlations between scores were very high, with Spearman rho systematically > 0.72, and P-values systematically < 0.001 after Bonferroni correction.
For a given task, correlations between signal impairment and intelligibility were at least 0.90 (DESC: 0.91; READ: 0.90).
Signal impairment on READ and DESC showed 0.89 correlation, and intelligibility 0.88.
On different tasks, correlations between signal impairment and intelligibility were between 0.80 and 0.87.
Degree of intelligibility on picture description showed 0.81 correlation with signal impairment on description and 0.89 with intelligibility on description. Correlations were slightly lower but still strong between intelligibility and signal impairment on reading (between 0.73 and 0.74).
3.5. Comparison between READ and DESC task scores
Although signal impairment scores were close in absolute terms between the two tasks (Fig. 2), the differences were significantly different.
There have been few studies of intelligibility and speech dis-order severity in oral cavity and oropharyngeal cancer [14,15]. The present results confirm impact on speech production, notably related to initial tumor size and location: oral cavity location and T4 stage were associated with greater signal impairment and poorer intelligibility. However, data are still lacking to predict treatment impact, and notably on speech-related quality of life.
Regarding speech production disorder severity assessment in clinical practice, the present results showed that all test results, whether reading or semi-spontaneous, correlated in terms both of signal impairment and of intelligibility.
On the other hand, inter-observer reliability was a limita-tion, better for picture description than reading, and better for
Scores attributed by panels.
Mean Median Minimum Maximum Interquartile range
Standard deviation not shown, as none of the distributions were normal: P-value (Shapiro-Wilk test) < 0.01.
Please cite Silurian Period article in press as: Balaguer M, et al. Assessment of impairment of intelligibility and of speech sig-nal after oral cavity and oropharynx cancer. European Annals of Otorhinolaryngology, Head and Neck diseases (2019), https://doi.org/10.1016/j.anorl.2019.05.012
4 M. Balaguer et al. / European Annals of Otorhinolaryngology, Head and Neck diseases xxx (2019) xxx–xxx
Mean scores according to tumor size and location, with standard deviations in brackets.
T stage (TNM classification)
T: T stage on TNM classification; OC: oral cavity; OP: oropharynx.
Fig. 1. Box-plot of scores. Sigread: signal impairment on reading; sigdesc: signal impairment on picture description; inteldeg: degree of intelligibility on picture description