A cross sectional quality of life study of 116 recurrence free head and neck cancer patients. The first use of EORTC H&N35 in Danish

Abstract

The quality of life questionnaire EORTC C30 with the head and neck specific module H&N35 has been validated in many languages and cultural settings, but the H&N35 module has not been formally translated to and validated in Danish. This validation was the purpose of the current study. In a cross sectional study 116 of 120 (97%) recurrence free head and neck cancer patients returned a valid questionnaire. The patients were attending follow up after radical radiotherapy (n=83), surgery (n=33) for cancer of the larynx (n=44), pharynx (n=34) or oral cavity (n=38). The previously described psychometric properties of the questionnaire were confirmed with the Danish translation. Nevertheless, there were some foreseeable problems with heavily skewed endpoints. Several scales of the questionnaire were sensitive to patient, tumour and treatment related factors: Good performance, high age, male gender, laryngeal cancer, low tumour stage and surgery correlated with a low score on the symptom scales or high score on the function scales. Symptom intensity increased with time since therapy in patients who had surgery and decreased in the irradiated patients. In conclusion, the current validation study confirmed the psychometric properties of the EORTC H&N35 questionnaire. The questionnaire detected correlations between clinical factors (performance status, gender, age, stage, site, time since therapy, treatment) and a large number of QoL factors. EORTC H&N35 in conjunction with EORTC C30 is a valid and informative tool in assessing quality of life, also in Danish head and neck cancer patients.

Cancer and its treatment often have detrimental and life long impact on quality of life (QoL) of patients [1]. With the increasing number of treatment possibilities with comparable efficiency, knowledge of the impact on objective side effects, such as mucocitis or fibrosis, as well as quality of life aspects is becoming ever more imperative.

The EORTC Quality of Life questionnaires (QLQ) are informative and valid tools in exploring the health related quality of life in several groups of cancer patients. The questionnaires consist of the global part (C30) and several site-specific questionnaires; H&N35 for head and neck cancer patients. Both questionnaires have been extensively tested in several populations, in several languages and cultural settings [2–6]. One of the authors (ABJ) took part in the development of the predecessor of EORTC H&N35–the H&N37 [7], but no formal translation of any of the questionnaires has been made in Danish. This paper presents the results of the translation and validation of the EORTC H&N35.

Materials and methods

The questionnaires consist of questions describing an item (e.g. vomiting). More items can make up a scale (or factor–the terms are used in parallel [8]).

The EORTC C30 questionnaire consists of 30 questions grouped in six function scales and nine symptom scales. EORTC H&N35 consist of 35 questions grouped in 18 symptom scales.

The patient is asked to rate the presence of a symptom or a limitation of a function on a Likert like scale from one to four (not at all, a little, quite a bit or very much) or on a scale from one to seven (very poor to excellent regarding the two questions about global quality of life). Five questions of the EORTC H&N35 are answered with a yes or no.

The factors consist of one to five questions. They either describe a symptom or a function. The values of symptom scales are means of the value of answered questions on a scale normalized to a number from 0–100 with 100 representing the highest degree of symptom. The function scales are also represented by a number from 0–100, but here 100 describes the best function. In the yes/no questions of EORTC H&N35 the score 100 indicates the presence of the symptom. At least half of the questions of a scale must be answered otherwise the scale value is missing in that patient. In the following, the questions will be denoted as C plus question number for questions from the C30 questionnaire and HN plus question number, e.g. HN4 for question number four from the H&N35 questionnaire. Scales will be denoted Pain as the pain score from C30 and HN Pain as the pain score from the H&N35 etc.

The primary forward and backward translation of the head and neck module QLQ H&N35 was done by the EORTC Quality of Life group. KJ and ABJ did the secondary translation before the questionnaire was presented to the patients. Only minor corrections were made. A pilot study was carried out using the QLQ H&N35 without the QLQ C30. A structured interview was performed, according to the pilot test instructions provided by EORTC [9].

The pilot study was carried out in the beginning of July 2003. Eleven patients, nine men and two women with cancers of the head and neck participated. The age distribution was 36–89 years with a mean of 62 years. The patients had been treated with either surgery or radiotherapy and were interviewed at planned or extraordinary follow-up or during radiotherapy. They represented all tumour sites. Ten of them had no theoretical education apart from primary school. In general, patients judged the questionnaire as easy and fast to complete. Some were however insecure about the meaning of a few questions. When the interviewer stressed that the patient only had to consider the past week this insecurity was solved. No questions were reported to be offending or upsetting. The completion time was 2–10 minutes (mean 5.4 minutes). No changes were made in the tested second translation since no more than one patient was uncertain or had remarks about the same question.

The validation study was performed in our follow up clinic between September and December 2003. Eligibility criteria were: oral cavity, pharyngeal or laryngeal carcinoma, no prior treatment of recurrence or other head and neck cancer, no signs of recurrence, only one treatment modality used no psychiatric or physical illness that prohibited filling out the questionnaire. The patient should speak and understand Danish.

Before inclusion the WHO performance status (PS) and mental status was assessed by one of the authors, and the patient was asked to participate in the study after a brief introduction about the study purpose and the questionnaires. The questionnaire was given to the patient in a prepaid and addressed envelope together with a brief note explaining the purpose of the study. If the patient failed to respond after 14 days, he or she was phoned and asked for a reason for not responding. After this, all patients but one answered. Patients were not contacted to collect missing data or correct inconsistencies. The local ethics committee approved the study. A returned questionnaire was considered as written informed consent.

Patient accrual and selection is presented in Figure 1. Accrual was continued until 100 patients had answered irrespective of the number of unreturned questionnaires. Of the 120 patients asked for participation, 116 returned a valid questionnaire (97%). The characteristics of non-responders was not analysed due to the limited numbers. Characteristics of the 116 patients returning a valid questionnaire are presented in Table I. In general, oral cavity tumours were managed primarily with surgery, whereas pharyngeal and laryngeal tumours were treated with primary radiotherapy as the only treatment modality.

Figure 1.  Accrual and selection from the outpatient clinic program of the 116 participating recurrence free patients treated for cancer of the pharynx, larynx and oral cavity.

Table I.  Characteristics of the 116 recurrence free head and neck cancer patients who returned a valid questionnaire. Percentages are of the groups within one treatment modality if nothing else is stated.

	Surgery	Radiotherapy
	n=33	n=83	All
	28% of all patients	72% of all patients	n=116
Mean age (years) (range)	65 (45–92)	62 (36–90)	63 (36–92)
Mean time since therapy (months) (range)	21 (1–47)	20 (2–65)	20 (1–65)
WHO Performance Status (PS)
0	23 (70%)	42 (51%)	65 (56%)
1	10 (30%)	40 (48%)	50 (43%)
2	0 (0%)	1 (1%)	1 (1%)
Gender
Women	15 (45%)	27 (33%)	42 (36%)
Men	18 (55%)	56 (67%)	74 (64%)
Tumour Site
Larynx	0 (0%)	44 (53%)	44 (38%)
Pharynx	2 (6%)	32 (39%)	34 (29%)
Oral cavity	31 (94%)	7 (8%)	38 (33%)
Stage
I	20 (60%)	19 (23%)	39 (34%)
II	6 (18%)	26 (31%)	32 (28%)
III	5 (15%)	21 (25%)	26 (22%)
IV	2 (6%)	17 (20%)	19 (16%)

The QoL data were ordinal and, in this data set, non-normal distributed and non-parametric tests were therefore used: Kruskal-Wallis/Wilcoxon-Mann-Whitney when comparing categorical data and Spearmans rho test was used when testing for correlations. Nevertheless, data are presented as means in graphs and tables. Normal distribution is assumed for exploratory purposes in order to make post hoc comparison using the Bonferoni method to correct for multiple testing. Internal consistency was examined using Cronbachs α. Scales were checked for construct validity by comparing intra and interscale correlation of items. Items showing possible scaling errors (items with higher correlation coefficients to other scales than it's own after correction for overlap) were checked for definitive scaling errors by comparing correlation coefficients using the approximation method described by Fayers and Machin [8]. A p-value <0.05 was regarded as significant and all tests were two tailed. SPSS 11.0 for Windows was used for the statistical analysis.

Results

Missing answers

Seventy-seven questionnaires of the 116 questionnaires returned (66%) were complete, and fifteen (13%) were complete except one question. Eighty-eight (75%) were complete except HN30 and/ or HN29 (see below). All together, 125 answers were missing (1.6% of all items). Both answers and missing answers were often explained by comments or even letters attached to the questionnaire.

Three or more patients did not answer the following questions: C26 (six patients): “Has your physical condition or medical treatment inferred with your family life”. One patient answered the question (“Not at all”) but stated that he had no family, indicating a possible explanation. HN4 (four patients): “Have you had a painful throat”. In our translation we used an equivalent phrasing to “Have you had a sore throat”. This was pointed out, by one of the patients in the pilot study as a potential confusing phrasing. HN25 (four patients) “Have you had trouble having social contact with family”. Again having no family could be a confusing factor. HN29 (16 patients) “Have you felt less interest in sex” and HN30 (24 patients) “Have you felt less sexual enjoyment”. No patients answered HN30 without answering HN29. Not answering these questions was significantly correlated with age (data not shown) and several patients had added small messages about being widowed or of old age. Among patients answering both HN29 and HN30 compared with the other patients there was a significant difference in seven QoL endpoints, among them Physical functioning, Role functioning and HN Social contact with the non-respondents having more symptoms and lower function.

Score distribution

Mean values of the function and symptom scores are presented in Table II. The scores were heavily skewed toward having few symptoms and high level of function: Sixteen of the 27 symptom scales had a median score of 0. Two of the six function scales had a median value of 100. Dry mouth was the most disturbing symptom in all groups, including the surgical patients, with an overall mean score of 51. Twenty-seven percent of all patients indicated that they suffered very much from dry mouth.

Table II.  Overall mean score of EORTC C30 and H&N35 and mean score dependent on site and treatment among 116 recurrence free head and neck cancer patients.

	Overall	Larynx	Pharynx	Oral cavity	Surgery	Radiotherapy
Global health status/QoL	66	65	59	74	77	62
Physical Function	82	79	79	88	90	79
Role Function	75	71	69	86	88	70
Emotional Function	75	70	70	84	84	71
Cognitive Function	82	82	76	87	88	80
Social Function	80	76	76	89	90	76
Fatigue	31	37	39	18	14	38
Nausea/vomiting	8	9	10	4	2	10
Pain	22	23	27	16	16	24
Dyspnoea	14	20	15	8	5	18
Insomnia	23	20	39	13	10	29
Appetite loss	23	23	35	12	9	29
Constipation	15	17	24	4	4	19
Diarrhoea	14	13	21	11	11	16
Financial problems	16	16	20	12	8	19
HN Pain	21	17	29	17	14	23
HN Swallowing	19	16	30	13	9	23
HN Senses	26	24	36	21	17	30
HN Speech	23	30	24	14	11	28
HN Social eating	21	14	35	17	12	25
HN Social contact	12	14	13	9	7	14
HN Sexuality	35	43	43	18	16	43
HN Teeth	30	29	47	17	12	37
HN Opening mouth	21	9	37	21	17	23
HN Dry mouth	51	44	70	43	39	56
HN Sticky saliva	40	38	58	25	20	48
HN Coughed	29	33	27	26	24	31
HN Felt ill	15	18	20	8	6	19
HN Pain killers	39	35	53	32	31	43
HN Nutritional supp.*	25	17	35	24	18	27
HN Feeding tube	10	5	24	3	0	13
HN Weight loss	21	23	21	18	9	26
HN Weight gain	26	31	32	16	9	33

*Nutritional supplement.

Validity and reliability

All interscale correlations were of expected direction. A highly significant correlation was observed between all multi-item scales and strong correlation was seen between scales with logical relationship (Table III). The factors consisting of a single item were in general characterized by a lower degree of correlation (data not shown). A strong correlation, with an absolute correlation coefficient ≥0.6, was observed between Appetite loss and HN Social eating and between HN Sticky saliva and HN Dry mouth and between HN Feel ill and QoL, Role function, Emotional function, Fatigue, Social function and Pain.

Table III.  Correlation coefficient between scales of EORTC C30 and H&N35 among 116 recurrence free head and neck cancer patients.

	QOL	PF	RF	EF	CF	SF	FA	NV	PA	HNPA	HNSW	HNSE	HNSP	HNSO	HNSC	HNSX
Global health status/QoL	1.00
Physical Function	0.67	1.00
Role Function	0.64	0.67	1.00
Emotional Function	0.71	0.50	0.53	1.00
Cognitive Function	0.50	0.58	0.59	0.59	1.00
Social Function	0.62	0.50	0.56	0.68	0.59	1.00
Fatigue	−0.75	−0.73	−0.70	−0.70	−0.62	−0.62	1.00
Nausea/vomiting	−0.45	−0.45	−0.33	−0.46	−0.40	−0.38	0.56	1.00
Pain	−0.61	−0.41	−0.49	−0.62	−0.45	−0.53	0.56	0.43	1.00
HN Pain	−0.49	−0.26	−0.38	−0.44	−0.37	−0.47	0.45	0.38	0.68	1.00
HN Swallowing	−0.49	−0.43	−0.44	−0.50	−0.39	−0.41	0.51	0.46	0.54	0.62	1.00
HN Senses	−0.42	−0.44	−0.45	−0.44	−0.50	−0.39	0.54	0.34	0.41	0.38	0.47	1.00
HN Speech	−0.45	−0.48	−0.54	−0.43	−0.34	−0.48	0.51	0.33	0.36	0.41	0.47	0.50	1.00
HN Social eating	−0.48	−0.46	−0.58	−0.55	−0.52	−0.51	0.53	0.42	0.53	0.59	0.76	0.60	0.50	1.00
HN Social contact	−0.48	−0.44	−0.60	−0.62	−0.45	−0.57	0.52	0.33	0.47	0.46	0.43	0.36	0.58	0.59	1.00
HN Sexuality	−0.52	−0.52	−0.52	−0.52	−0.57	−0.46	0.64	0.35	0.37	0.28	0.36	0.49	0.45	0.38	0.52	1.00

Abbreviations in the column headlines refer to the corresponding scale name in the rows. Absolute correlation coefficient ≥0.6 bold, ≥0.7 also underlined. All significant p < 0.0005

All questions had an acceptable, according to Fayers [8], absolute correlation of >0.4, with their own scale except C5 (correlation coefficient −0.28) and HN4 (correlation coefficient 0.35) after correcting for overlap. C5 “Do you need help for eating, dressing or washing” had a clear floor effect with only three patients not responding “not at all” to the question. C14 (Nausea), C20 (Concentration), C25 (Memory) and HN4 (Painful throat) correlated widely with other scales to a higher degree than their own scale. HN16 (Hoarseness) seemed to correlate better with HN Social eating, HN Coughing and HN Feeling ill than HN Speech (other “throat-function” related questions). HN23 (Trouble talking to other people) correlated better with HN Social contact than HN Speech. No definitive scaling errors were found.

Overall internal consistency was acceptable to excellent with a Cronbachs α of 0.71–0.93 (Table IV). The two-item scales, Cognitive function and Nausea/vomiting, was characterized by a floor effect and were the only scales with a Cronbachs α below 0.8. Low internal consistency has previously been reported and accepted since the scales are clinically sensible [2]. Cronbachs α tends to be higher in heterogeneous samples. A subset analysis on sites was therefore performed in spite of the low numbers. No additional information was gained. Internal consistency in subgroups was high except in cases with a high degree of floor effect.

Table IV.  Internal Consistency of the multi item scales of EORTC C30 and H&N35 among 116 recurrence free head and neck cancer patients.

	Cronbachs α
Scale	Overall	Oral cavity	Pharynx	Larynx
Global Health/QoL	0.91	0.92	0.94	0.89
Physical function	0.86	0.75	0.89	0.87
Role function	0.90	0.94	0.85	0.90
Emotional function	0.85	0.79	0.66	0.93
Cognitive function	0.71	0.58	0.55	0.85
Social function	0.83	0.67	0.79	0.91
Fatigue	0.88	0.88	0.87	0.87
Nausea/ vomiting	0.75	0.29	0.77	0.81
Pain	0.84	0.87	0.84	0.83
HN Pain	0.84	0.83	0.83	0.84
HN Swallowing	0.82	0.79	0.81	0.85
HN Senses	0.83	0.92	0.83	0.78
HN Speech	0.82	0.80	0.87	0.80
HN Social eating	0.92	0.95	0.91	0.92
HN Social Contact	0.89	0.92	0.94	0.87
HN Sexuality	0.93	0.98	0.83	0.95

The scales of the questionnaires were compared between known groups to assess criterion validity in the absence of a gold standard.

Patients with WHO Performance status 0 had fewer symptoms and higher function than patients with PS ≥ 1, except for the scale HN Weight gain. The difference was significant in 20 of the 33 scales.

Age correlated significantly with eight factors; Social Function, Insomnia, Diarrhoea, Financial Problems, HN Pain, HN Opening Mouth, HN Dry Mouth, HN Feeding Tube. The absolute correlation coefficient was very low (<0.36). However when significant, higher age was invariantly negatively correlated with the symptom scales and positively correlated with the function scale.

Gender only correlated significantly with HN Social contact, HN Painkiller and HN Nutritional supplement, with females having the most symptoms/use of intervention.

Tumour site was significantly correlated with 17 of the 33 factors. In order to separate the influence of site from that of treatment modality, the surgical patients were excluded. For the radiotherapy patients, the seven factors presented in Figure 2 were significantly correlated with stage. Univariat analysis of variance with post hoc testing was performed, and the results are shown in Figure 2. The score results obtained from oral cavity cancer patients were not significantly different presumably because of low numbers. The stage distribution was not equal between sites; pharynx cancer patients had a higher stage than the two other groups. When looking only at the stage III + IV patients treated with radiotherapy, there were no differences in any of the QoL endpoints. Figure 3 shows the mean score of issues of EORTC C30 and H&N35 that were significantly dependent on tumour site in 45 patients with stage I + II disease. Eight factors were significantly different between tumour sites. No meaningful analysis of tumour site could be made in the group of surgical patients, since 29 of the 31 patients were treated for oral cancer.

Figure 2.  Mean scores of issues from the EORTC C30 and H&N35 significantly dependent on tumour site in 83 patients after radiotherapy for cancer of the pharynx, larynx and oral cavity. “{”Indicate significant differences between single groups with post hoc testing correcting for multiple testing.

Figure 3.  Mean score of issues of EORTC C30 and H&N35 that were significantly dependent on tumour site in 45 stage I-II patients after radiotherapy for cancer of the pharynx, larynx and oral cavity. “{”Indicate significant differences between single groups with post hoc testing correcting for multiple testing.

Clinical stage was significantly and positively correlated with nine scales overall, and with seven scales in the radiotherapy group. The effect of stage was similar in subgroups based on site and treatment and a few items were significantly correlated with stage in all subgroups.

Compared to surgery the radiotherapy patients invariably had the lowest function score and the highest symptom scores; the differences were significant for 23 of the 33 factors. The exceptions were Cognitive Function, Pain, Diarrhoea, Financial Problems, HN Pain, HN Social Contact, HN Opening Mouth, HN Coughed, HN Pain Killer and HN Nutritional supplement.

An analysis was performed with time after initial therapy as a dependent factor for the quality of life issues. This was done with an exploratory purpose since the present study was not prospective. The observation time after therapy was independent of tumour site, stage or treatment. Figure 4 shows that mean HN Pain score dropped with time after radiotherapy, whereas the pain score increased with time after surgery. The same improvement was seen for QoL, Physical Function, Emotional Function, Fatigue, Appetite Loss, Constipation, HN Pain, HN Swallowing, HN Senses, HN Social eating, HN Sexuality, HN Sticky Saliva and HN Feeding Tube. Among the surgical patients Pain, HN Pain, Constipation, Pain Killer and HN Nutritional supplement scores were significantly higher in the patients with the longest follow up. No symptom scores were lower in this group.

Figure 4.  Mean HN Pain differ between groups with different time after therapy dependent on treatment modality. Results from a cross sectional study in 116 recurrence free oral cavity, pharyngeal and laryngeal cancer patients. Error bars represent±standard error.

Discussion

This study represents the first use of the EORTC H&N35 in a Danish patient population. The patients were willing to participate in the study and were eager to explain their symptoms.

Missing answers were a relatively small problem apart from the items concerning sexuality. It can be questioned whether it is possible to draw any conclusion concerning sexuality in this patient group of high age without further follow-up on missing items, as we must assume that the missing answers are not randomly distributed. Patients who did not answer sexuality items had an inferior score in seven QoL endpoints, which suggests that missing values probably will produce a systematic error when analysing sexuality. Similar problems with missing answers have been encountered in other studies of head and neck cancer patients [3] and breast cancer patients [10]. The two items concerning sexuality has an inherent limitation, as there is no obvious answer for those sexually inactive. The present phrasing has been chosen because asking specifically about activity would decrease the acceptability [2]. As long as the possible answers are not exhaustive the patients will answer in an unpredictable way to questions not relevant for them (sticky saliva in case of total dryness of the mouth, problems with family life in those without contact to family, problems with sexuality in the sexual inactive and problems with teeth in the edentulous).

Dry mouth was also in this study a very disturbing symptom for head and neck cancer patients with major implications for the overall quality of life [1], [11]. Many of the other disturbing symptoms, like nutrition, teeth problems, speech and even insomnia, are also secondary to salivary gland dysfunction. This confirms that dry mouth has a major impact on the social- and professional life and general well being of head and neck cancer patients.

Significant correlation was found in the present study among various scales, even above 0.7, which is regarded as an indication of overlapping constructs. A previous study in heterogeneous patient populations have also reported this finding [12] in contrast to a report of a more homogenous populations especially with respect to time from treatment [2]. This finding could be explained by the mathematical nature of correlations that tend to be higher in heterogeneous datasets. The high and low correlating scales of the present study were in agreement with the two mentioned studies.

The factor scores were heavily skewed towards few symptoms and high level of functions for most scales, but responses covered the full range of possible scores. This indicates a sensible categorization of the items and scales. The items with a low interscale correlation was not different from those previously described [2], [3], [5], [6], [13]. This has not been perceived as problematic since these scales are clinimetrically sensible [2], [14]. It is nevertheless difficult to differentiate some of the scales, e.g. HN Social Contact with HN Social Eating and HN Speech both conceptually in understanding the construct and psychometrically since many items are highly correlated. The problems are even larger in the more general constructs of C30.

The observation that patients of higher age had higher function score and lower symptom score could be an indication of decreasing expectations to physical function with increasing age. In a validation study of EORTC C30 and H&N35 in head and neck cancer patients before, during and after treatment, Sherman et al [12] also reported a negative correlation between symptoms of the C30 questionnaire and age. However, in contrast to our findings, site specific symptoms (H&N35) increased with age. In a prospective study by de Graeff [15] using the EORTC questionnaires, only limited but negative influence of age was found. Ringash [16] reported no correlation using the quite similar FACT-HN questionnaire in a cross sectional study of a population of laryngeal cancer patients with a high mean QoL score. Explanations for the above findings could be that the influence of age is limited, and that age is positively correlated with symptoms in prospective studies including scoring during treatment where the more vulnerable patients of old age could suffer more from side effects e.g. because of increasing co-morbidity with age [17]. Previous studies of toxicity have shown no major influence of age on both acute and late endpoints but increased susceptibility to e.g. dehydration [18], [19].

In our study, Social Contact and interventions were influenced by gender, but generally, QoL has been found to be only minimally different between male and female patients with head and neck cancer, except in one clinical study, where female patients had poorer QoL [15].

The severity of various symptoms according to tumour site was in good agreement with previous finding [3]. HN Speech was higher in the population a laryngeal cancer patients, but, contrary to our expectation, not significantly related to site as previously described in populations also including operated larynx cancer patients [2]. The reason for this could be that all irradiated patients in this study had elective or high dose irradiation to the larynx and that none of the patients with laryngeal cancer also had surgery.

Time of assessment after radiotherapy has previously been shown to influence quality of life endpoints with decreasing QoL and increasing symptoms during treatment, but recovery in many endpoints after treatment [11], [20], [21]. In surgically treated patients, a similar pattern is generally observed: the QoL and pain scores worsens compared to pre-treatment level until three months after treatment and improves thereafter [22–25]. In a surgical study it has previously been shown that pain may increase due to pain in the arm or shoulder [20] and in a study comparing different techniques of neck dissection radical neck dissection seemed to result in increased pain over time [26]. In the current study we found that pain in the head and neck region (and most other items directly related to pain) was more pronounced in patients with longer follow up. Time since therapy was independent of tumour and patient related factors, and no systematic change in surgical treatment technique was observed in the observation period and our patients received only one treatment modality. The finding should be confirmed in a larger prospective study. Nevertheless, the influence of single or multi-modality treatment on the dynamics of problems should be examined since the intensity and frequency of complaints may differ between treatment modalities dependent on time of assessment.

In conclusion, the current validation study has confirmed the psychometric properties of the EORTC H&N35 questionnaire. The questionnaire detected correlations between clinical factors (gender, age, site, stage, time since therapy, treatment) and a large number of QoL factors. EORTC H&N35 in conjunction with EORTC C30 is a valid and informative tool in assessing quality of life, also in Danish head and neck cancer patients.

The authors would like to thank the EORTC quality of Life Group for assistance and the use of questionnaires. The Danish Cancer Society, Radiumstationens Forskningsfond, Agnes Niebuhr Anderssons Fond and William Nielsens Fond are thanked for financial support.