Prevalence of swallowing and speech problems in daily life after chemoradiation for head and neck cancer based on cut-off scores of the patient-reported outcome measures SWAL-QOL and SHI

The study was conducted according to regular procedures of the local ethical committee of the VU University Medical Center, Amsterdam. All patients provided informed consent.

Inclusion criteria comprised primary head and neck carcinoma (all subsites, all stages) and chemoradiation treatment. One hundred and three patients with primary head and neck carcinoma underwent CHRT of which 71 were alive at the time of the study. Seven patients were excluded because of distant metastasis or loco-regional recurrence and 4 patients were excluded because they did not speak Dutch, leaving a study cohort of 60 patients.

Data on age, gender, comorbidity, and tumor and treatment characteristics [site and stage (according to UICC)], radiation technique, time since treatment, placement and removal of gastrostomy tube were collected from the medical records. Comorbidity was assessed with the Adult Comorbidity Evaluation 27 (ACE-27) [20]. The ACE-27 includes 27 comorbid conditions, including cardiovascular, respiratory, gastro-intestinal, renal, endocrine, neurological, immunological, psychiatric and rheumatologic disorders, previous or synchronous malignancy, alcohol abuse and excessive body weight. The ACE-27 was designed specifically for cancer patients and classifies patients into 4 grades of comorbidity [none (grade 0), mild (grade 1), moderate (grade 2), severe (grade 3)].

During a time span of 8 months, 52 out of 60 patients returned the SWAL-QOL (response rate 87 %) and 47, the SHI (response rate 78 %). An overview of patient characteristics is provided in Table 1. Median age of the patients was 58 years (range 36–75). Thirty-five (67 %) of the patients were male. No comorbidity was observed in 16 patients, 24 patients had grade 1, 8 patients grade 2 and 4 patients grade 3. Primary tumor locations were oral cavity (n = 5), oropharynx (n = 30), nasopharynx (n = 4), larynx (n = 10) and hypopharynx (n = 3) and these were categorized into oral cavity/oropharynx/nasopharynx (n = 39) and larynx/hypopharynx (n = 13) for comparison and statistical analyses. Overall tumor stages were II (n = 2), III (n = 18) and IV (n = 32), which were categorized into stage II–III (n = 20) vs. IV (n = 32) for comparison and statistical analyses. Almost all (97 %) patients underwent PEG placement at start of treatment. All patients underwent chemoradiation, in 79 % of the patients consisting of Cisplatin 100 mg/m2 in 3 cycles. Of the remaining patients, 4 % received 1 cycle and 12 % received 2 cycles. Three patients (6 %) underwent intra-arterial Cisplatin 150 mg/m2 in 4 cycles. Chemotherapy was given concomitantly with radiotherapy (2 Gy per fraction, 5 times per week, total dose of 70 Gy). One patient received also 5-Fluor-Uracil because of cisplatin side-effects after 1 cycle. In 31 patients, conventional three-dimensional conformal radiotherapy (3D-CRT) was given and intensity-modulated radiation therapy (IMRT) in 21 patients. Patients treated before 2005 were treated with 3D-CRT, without sparing the salivary glands. In 2005, IMRT was introduced as part of standard care at the department of radiation oncology enabling a significant reduction of the dose to the salivary glands. The gastrostomy tube was removed in 80 % after complete remission and when oral intake was sufficient (in average 3–6 months after end of treatment). Time since treatment ranged from 6 to 58 months with a median of 21 months.

Swallowing and speech impairment was measured via the Swallowing Questionnaire on Quality of Life (SWAL-QOL) and the Speech Handicap Index (SHI).

The SWAL-QOL is a 44-item questionnaire on swallowing-related problems in daily life [17, 18]. Response categories range on a 5-point scale. There are 10 subscales: (1) food selection (2 items); (2) eating duration (2 items); (3) eating desire (3 items); (4) Fear (4 items); (5) burden (2 items); (6) mental health (5 items); (7) social functioning (5 items); (8) communication (2 items); (9) sleep (2 items); and (10) fatigue (3 items). Furthermore, there is an overall symptom scale (14 items). Finally, a total SWAL-QOL Score can be calculated (based on the 23 items of the first 7 scales listed above). All SWAL-QOL scales range from 0 to 100, a higher score indicating more impairment. Three single questions are included regarding nutrition intake (“normal”, “soft”, “pureed”, “liquids only”, “mostly tube feeding” and “tube feeding only”), liquid intake (“all liquids”, “thick liquids”, “very thick liquids”, “thickened liquids” and “no liquids”), and general health (“poor”, “moderate”, “good”, “very good” and “excellent”). The SWAL-QOL was translated and validated for use in Dutch head and neck cancer patients. A cut-off score on the total SWAL-QOL score of 14 points (or higher) (94 % sensitivity and 84 % specificity) indicates swallowing problems in daily life [18].

The SHI is a validated speech-specific quality of life questionnaire and consists of 30-items focusing on speech-related problems in daily life. Response categories range on a 5-point Likert scale (“never”, “almost never”, “sometimes”, “almost always” and “always”). The questionnaire also includes an overall speech quality item, with 4 response categories (“good”, “reasonable”, “poor” and “severe”). A total SHI score is calculated by summing the scores on all 30 items (score range 0–120), with a higher score indicating a higher level of speech-related problems. Two subscales are distinguished: psychosocial function and speech function. A total SHI score ≥6 indicates speech problems in daily life (95 % sensitivity and 90 % specificity) [19].

Descriptive statistics were calculated for the SWAL-QOL and SHI scales. A t test for independent groups was used to test differences between patients regarding gender (male vs. female), tumor site (larynx/hypopharynx vs. oral cavity/oropharynx/nasopharynx), tumor stage (stage I–II vs. stage III–IV) and type of radiotherapy (3D vs. IMRT) on SWAL-QOL and SHI scores. Analysis of variance (ANOVA) was carried out to test differences regarding comorbidity (grade 0, 1, 2 or 3) and food intake (normal, soft diet, pureed diet, tube feeding). In case of overall significant differences, post hoc analyses were carried out to pairwise compare all different groups. Bonferroni correction was applied to correct for multiple testing. Correlation analyses were performed to study the relation between age (Pearson) and time since treatment (Spearman’s rank) and SWAL-QOL and SHI scores, and to study the relation between the total SWAL-QOL scale and the total SHI scale. Those variables that showed univariate significant relations with the outcome measures, were included into the multivariate regression analysis (stepwise) to obtain insight into which sociodemographic (age, gender) and clinical (comorbidity, tumor site and stage, radiation technique, time since treatment) parameters are associated with patient-reported speech and swallowing outcome. For all statistical tests, significance was defined as p < 0.05. All analyses were performed using the IBM Statistical Package for the Social Science (SPSS) version 20 (IBM Corp., Armonk, NY USA).

A deviant SWAL-QOL score (total SWAL-QOL score ≥14) was observed in 79 % of the patients (41/52). Mean scores are presented in Table 2. Relatively high scores were observed on the subscales general burden, food selection, eating duration, eating desire, fatigue and sleep.

Normal food intake was reported by 23 patients (45 %), while 18 (35 %) took a soft diet or pureed food, and 10 (20 %) had tube feeding (data were missing for 1 patient). Patients with normal food intake had significantly better mean scores compared to patients with soft diet or tube feeding on all SWAL-QoL scales, except for the subscales sleep and fatigue; patients with soft diet or tube feeding did not differ from each other significantly on any of the scales (Table 2).

Univariate analyses revealed that age, gender, comorbidity and tumor stage were not significantly related to SWAL-QOL scores. Swallowing outcome was significantly associated with tumor site and radiotherapy technique. Patients treated for a laryngeal or hypopharyngeal tumor had significantly (p < 0.05) better scores compared to patients treated for an oral cavity, oropharyngeal or nasopharynx tumor on the total SWAL-QOL (df = 50; t = −2.10; p = 0.041) and the subscales general burden (df = 50; t = −2.23; p = 0.030), mental health (df = 50; t = −2.30; p = 0.026) and social functioning (df = 50; t = −2.46; p = 0.017). Compared to 3D-CRT, patients after IMRT had significantly better scores on the total SWAL-QOL (df = 49.9; t = 2.22; p = 0.031), food selection (df = 50; t = 2.01; p = 0.05), fear of eating (df = 50; t = 3.34; p = 0.002), sleep (df = 50; t = 2.89; p = 0.006) and social functioning (df = 50; t = 2.46; p = 0.018). Furthermore, a positive correlation (more problems on the long term) was found between time since treatment and the following SWAL-QOL subscales:fear of eating (ρ = 0.57; p < 0.001), sleep (ρ = 0.41; p = 0.002), fatigue (ρ = 0.33; p = 0.015), social function (ρ = 0.33; p = 0.019) and the total SWAL-QOL score (ρ = 0.37; p = 0.007).Because radiotherapy technique was related to time since treatment, correlation coefficients were also calculated in these subgroups (3D-CRT vs. IMRT) and no significant relations between time since treatment and swallowing outcome were found.

For the multivariable regression analysis only tumor site and radiotherapy technique were included in the selection procedure, which revealed that only tumor site was significantly associated with the total SWAL-QOL score (B = 15.5, R ² = 0.081, F = 4.40, p = 0.041). Note that radiotherapy technique is univariate more significantly associated to total SWAL-QOL, but this factor did not enter the multivariate model due to unequal variances of the two groups.

A deviant SHI score (total SHI score ≥6) was observed in 55 % of the patients (26/47). Mean scores are presented in Table 2. Univariate analyses revealed that radiotherapy technique was associated significantly with the subscale SHI psychosocial function: patients treated with IMRT had a better score (df = 39.7; t = 2.48; p = 0.017). Age, gender, comorbidity, tumor stage and site, and time since treatment (corrected for radiotherapy technique) were not significantly associated with SHI total scores. Therefore, no multivariate regression analyses were carried out.