Participants

In total 54 participants were included, the majority of whom were male (64.8%) with a mean age of 67 years (SD 11.2). Baseline swallowing function measured using FOIS, MASA-C and EAT-10 are presented in Table 1. All groups had a median FOIS 5 (IQR 1) at baseline. MASA-C severity scores at baseline were mild to moderate for group 1 and group 2 + 3 across all IDDSI levels. The lowest scores were observed for IDDSI 6.

Table 1 Participant, disease and therapy characteristics of the study population

Patient, disease and therapy characteristics are presented in Table 1. One patient immediately dropped out after baseline measures and was not included in the efficacy analyses. Two patients dropped out after 4 weeks of training. Timing and reasons for drop-out in each group are defined in Table 1.

1st research question. Combining strength and skill training with (sham) HD-tDCS versus isolated strength trainingFunctional Oral Intake Scale (FOIS)

Our results showed a significant main effect of time (Χ2 (3) = 42.28, p < 0.001) over 8 weeks for patients who received isolated strength training (group 1) as well as patients who received strength and skill training with (sham) HD-tDCS (group 2 + 3). All patients showed significant improvements on the FOIS over 8 weeks of training. Additionally, a significant group-by-time interaction effect was observed (Χ2 (3) = 10.9, p = 0.012). Patients in group (2 + 3) had a significantly higher chance of a better FOIS outcome after 8 weeks of training (OR 13.47, 95%CI 5.95 – 30.49, p < 0.001) compared to group 1 (OR 3.04, 95%CI 1.44 – 6.44, p = 0.004) (Fig. 1). Post hoc analysis with Bonferroni-Holm correction showed no significant changes on the FOIS between baseline and 4 weeks in both groups. However, significant increases on the FOIS were observed in both groups between baseline and 8 weeks of training. All post hoc comparisons are shown in Table 2.

Fig. 1

Changes in Functional Oral Intake Scale (FOIS) levels over time across both groups. FOIS levels are presented in Table 3. * p < 0.05, ** p < 0.001

Table 2 Post hoc comparisons FOISConfounding Variables

We investigated the potential impact of confounding variables on the evolution of functional oral intake. Frailty had a significant effect on the FOIS outcome with a negative coefficient (OR 0.23, 95%CI 0.09 – 0.59, p = 0.007), indicating that patients who were frailer, were less likely to achieve higher FOIS scores. However, correcting for frailty in the model doesn’t change the conclusion on treatment effect. Other potential confounders (center, TNM-T v8, TNM-N v8, xerostomia, time post-C/RT, RT vs CRT and age) did not show a significant effect on the model.

Swallowing Function

For IDDSI 0 and 6 the analysis revealed a significant main effect of time (Χ2 (3) = 12.52, p = 0.006; Χ2 (3) = 20.25, p < 0.001), indicating that MASA-C scores improved over training in both groups. However, there was no significant effect of group (Χ2 (1) = 0.81, p = 0.37; Χ2 (1) = 1.48, p = 0.22) or group-by-time interaction (Χ2 (3) = 1.71, p = 0.63; Χ2 (3) = 1.04, p = 0.79), suggesting that the changes over time were similar across groups (Fig. 2a and c). In contrast, for IDDSI 3, no significant effects were found for time (Χ2 (3) = 6.98, p = 0.073), group (Χ2 (1) = 0.22, p = 0.64) or interaction (Χ2 (3) = 2.10, p = 0.55) (Fig. 2b). Secondary analysis for time effects indicated significant changes from baseline to 4 and 8 weeks for IDDSI 0 and 6 for both groups (Table 4). Detailed post hoc comparisons per group and timepoint are presented in Table 5.

Fig. 2

Evolution of mean MASA-C scores for IDDSI 0 (a), IDDSI 3 (b) and IDDSI 6 (c). Abbreviations: MASA-C = Mann Assessment for Swallowing Ability – Cancer, IDDSI = International Dysphagia Diet Inventory. * p < 0.05 evolution over time in both groups combined, error bars: 95% CI

Quality of Life

Our results demonstrated a significant time effect (F(3,130) = 18.205, p < 0.001) and group-by-time interaction effect for DHI scores (F(3,130) = 3.326, p = 0.022), indicating that changes in QOL scores varied between both groups (Fig. 3). Post hoc analyses revealed a significant reduction in DHI scores, indicating an improvement in QOL, from baseline to 4 and 8 weeks in group 2 + 3. In group 1, however, no significant changes were observed over time after correction for multiple testing. All post hoc results are presented in Table 8.

Fig. 3

Evolution of mean Dysphagia Handicap Index (DHI) scores over time for each group. ** p < 0.001, error bars: 95% CI

Muscle Strength

For MIPa the model showed a significant main effect of time (F(3,133) = 55.98, p < 0.001) and group-by-time interaction (F(3,133) = 2.83, p = 0.041). No significant effect of group was found (F(1,51) = 0.38, p = 0.542), indicating that both groups improved over time, with some differences in the pattern of improvement between them (Fig. 4a). Regarding mean MIPp, a significant time effect was also observed (F(3,130) = 34.31, p < 0.001), but no significant group (F(1,51) = 1.46, p = 0.232) or group-by-time interaction effect (F(3,130) = 1.24, p = 0.299) (Fig. 4b). Post hoc analyses for time with Bonferroni-Holm correction confirmed significant increases in MIPa and MIPp from baseline to 4 and 8 weeks in both groups. No significant detraining effects were observed between 8 and 12 weeks follow-up in both groups. Detailed post hoc comparisons are provided in Table 6.

Fig. 4

Evolution of muscle strength measurements over time across both groups. Mean Maximum Isometric Tongue Pressures in (a) the anterior position (MIPa) and (b) the posterior position (MIPp), tongue pressure during saliva swallows in (c) the anterior position (PswalA) and (d) the posterior position (PswalP) and evolution of (e) mean Maximum Expiratory Pressures (MEP), * p < 0.05, ** p < 0.001, error bars: 95% CI

Regarding anterior and posterior tongue strength during saliva swallows, a significant time effect (PswalA: F(3,132) = 15.77, p < 0.001, PswalP: F(3,128) = 7.17, p < 0.001) but no group (PswalA: F(1,52) = 2.49, p = 0.121; PswalP: F(1,51) = 0.84, p = 0.363) or interaction effect was observed (PswalA: F(3,132) = 0.167, p = 0.918; PswalP: F(3,128) = 0.169, p = 0.917). Post hoc analyses for time with Bonferroni-Holm correction showed significant increases in PswalA after 4 and 8 weeks compared to baseline for both group 1 and group 2 + 3. There were no further improvements in PswalA or PswalP after 4 weeks of training and no significant detraining effects between 8 and 12 weeks follow-up in both groups (Fig. 4c and d). All post hoc comparisons are shown in Table 6.

For the evolution in mean MEPs, our results showed a significant main effect of time (F(3,125) = 15.80, p < 0.001) but not group (F(1,51) = 2.58, p = 0.114) or group-by-time interaction (F(3,125) = 1.25, p = 0.293) (Fig. 4e). Post hoc analyses for time showed significant gains in mean MEPs from baseline to 4 and 8 weeks in both groups. No significant detraining effects were observed between 8 and 12 weeks follow-up in both groups. All post hoc comparisons are shown in Table 6.

Percentage of progress for each strength measurement between baseline and 4 to 8 weeks are presented in Table 7.

Patients’ Perceived Efficacy

The results showed no statistically significant difference in patients’ perceived efficacy between the groups, χ2(1) = 0.36, p = 0.550 (Fig. 5). The majority of the participants reported a strong improvement in both groups. None of the participants reported that their condition had worsened as a result of the exercises.

Fig. 5

Scores per group for patients’ perceived efficacy (“rate the extent to which you have changed in therapy so far”) after 8 weeks of training based on the Dutch version of the Work Alliance Inventory (WAV-NL) likert scale

2nd research question. Additional effect of real versus sham HD-tDCS?Functional Oral Intake Scale (FOIS)

A significant main effect of time (χ2 (3) = 33.58, p < 0.001), but not group (χ2 (1) = 0.61, p = 0.43) or group-by-time interaction (χ2 (3) = 2.21, p = 0.53) was observed, indicating that both groups showed improvement in FOIS over 8 weeks of training. Secondary analyses of time effects across both groups combined showed a significant improvement in FOIS scores after 8 weeks therapy compared to baseline (OR 12.39, 95% CI 5.11 – 30.03, p < 0.001). Post hoc analysis between groups showed no significant changes in FOIS between baseline and 4 weeks in group 2 (n = 18) or group 3 (n = 18). Significant increases in FOIS were observed in both groups between baseline and 8 weeks of training with no detraining effects at the 12 weeks follow-up. Results are shown in Table 8.

Swallowing Function

For IDDSI 0 the analysis showed no significant effect of time (Χ2 (3) = 3.67, p = 0.30), group (Χ2 (1) = 0.004, p = 0.95), or group by time interaction (Χ2 (3) = 0.53, p = 0.91). Similar results were observed for IDDSI 3 with no significant effect of time (Χ2 (3) = 3.81, p = 0.28), group (Χ2 (1) = 0, p = 0.99), or group by time interaction (Χ2 (3) = 0.29, p = 0.96). For IDDSI 6, however, a significant effect of time was observed (Χ2 (3) = 0.41, p = 0.49), but not group (Χ2 (1) = 0.16, p = 0.69) or group by time interaction (Χ2 (3) = 2.41, p = 0.49), indicating that both groups showed improvements in MASA-C scores for solid consistencies over 8 weeks of training. Post hoc analysis showed significant improvements in MASA-C scores for IDDSI 6 after 8 weeks compared to baseline in group 3 (Table 8).

Muscle Strength

For the evolution in mean MIPa and MIPp gain, a significant effect of time (MIPa: F(3,86) = 31.82, p < 0.001, MIPp: F(3,85) = 22.37, p < 0.001) but not group (MIPa: F(1,33) = 0.745, p = 0.394, MIPp: F(1,33) = 0.18, p = 0.677) or group-by-time interaction was observed (MIPa: F(3,86) = 0.49, p = 0.688, MIPp: F(3,85) = 0.97, p = 0.411). Post hoc analyses for time across groups showed significant increases in MIPa and MIPp after 4 weeks and 8 weeks compared to baseline in both groups. There were no significant detraining effects between 8 and 12 weeks follow-up in both groups. All post hoc comparisons are presented in Table 9.

Regarding anterior and posterior tongue strength during saliva swallows, a significant time effect (PswalA: F(3,85) = 9.85, p < 0.001, PswalP: F(3,84) = 3.63, p = 0.016) but no group (PswalA: F(1,34) = 0.29, p = 0.865; PswalP: F(1,34) = 0.009, p = 0.923) or interaction effect was observed (PswalA: F(3,85) = 0.88, p = 0.453; PswalP: F(3,84) = 0.743, p = 0.529). Post hoc analyses for time showed significant increases in PswalA after 4 and 8 weeks compared to baseline in both group 2 and group 3. Regarding PswalP, group 2 showed no significant improvements over time, group 3 however showed significant improvements from baseline to 4 and 8 weeks. There were no improvements in PswalA or PswalP between 4 and 8 weeks of training and no significant detraining effects between 8 and 12 weeks follow-up in both groups. All post hoc comparisons are shown in Table 9.

Regarding the evolution in MEPs, the model showed a significant time effect (F(3,80) = 8.716, p < 0.001) but no group (F(1,32) = 0.32, p = 0.859) or interaction effect (F(3,80) = 0.347, p = 0.792). Post hoc analyses for time, showed significant increases in MEP from baseline to both 4 and 8 weeks in both groups. Between 8 and 12 weeks follow-up no significant detraining effects were observed. All comparisons are reported in Table 9.