Klaske van Sluis

56 3.4. Discussion 3.3.3 Correlations between outcome measurements When outcomes for the multidimensional assessment methods assessed over time are pooled, strong correlations are found between the dimensions of voice related outcomes. Correlations between the multidimensional voice related out- comes are statistically significant (see appendix I). No statistically significant correlation is found between AVQI and EQ-5D-5L ( p =.228). Correlations between the outcome measurement instruments for the post- surgery time-points (T1-T3) show statistically significant correlations between voice related outcomes (see Appendix II). Statistical significance is lost for voice related outcomes with QoL, seen in the correlation between VHI-10 and EQ-5D- 5L ( p =.021), and AVQI and EQ-5D-5L ( p =.467). AVQI and VHI-10 ( p =.017) still correlate strongly but this is not statistically significant with statistical level set at p ≤ .01. With investigation of the correlation between VHI-10 and AVQI for the post-surgery time points (T1-T3) an AVQI cut-off score of 6 would be indicative for a VHI-10 score >11, indicating an unsatisfactory voice after total laryngectomy. 3.3.4 Predictors of voice outcome We created linear mixed effects models to explore the effect of Treatment, Reconstruction, and Time studied for the main voice outcomes VHI-10, AVQI, Perc. Voice SLP, and Perc. Voice Pt. Time is indicated as a significant factor in predicting AVQI score ( p ≤ .001), Perc. Voice SLP ( p =.015), Perc. Voice Pt. ( p =.002) but not for VHI-10 ( p =.368). Modelling predicted outcomes for the groups for a) primary surgical treatment vs. salvage surgical treatment and b) primary closure vs. major reconstruction did not reach statistical significance. 3.4 Discussion In our study, quality of life, measured with EQ-5D-5L, is lowest before surgery. It is known that levels of anxiety and self-care are severely impacted before as well as up to 14 days post-surgery [7, 30, 31]. We observe improvement at the three months post-surgery assessment, it is likely that patients have begun to adjust to their condition compared to 14 days post-surgery. In the long term, at 12 months post-surgery, mean score for the total group, is comparable to reference values for the age group above 60 years old [23]. This positive result might be influenced by drop-out of patients who were excluded due to mortality, nevertheless, it indicates that the remaining patients are fairly well adjusted to their condition. This general course of worsening after surgery and gradual improvement over time, corresponds to findings of earlier studies assessing the course of QoL [7, 8]. Mean values for VHI-10, were at all assessment time points above 11, which is indicated as having a voice handicap [25, 26]. This is in line with earlier