Patient preference for commonly‐used, head and neck cancer‐specific quality of life questionnaires in the follow‐up setting (Determin): A multi‐centre randomised controlled trial and mixed methods study

Quality of life (QoL) assessment forms an integral part of modern cancer care and research. The aim of this study is to determine patients' preferences and willingness to complete commonly used head‐and‐neck cancer (HNC) QoL questionnaires (QLQs) in routine follow‐up clinics.

patients preferred EORTC QLQ-HN35. Reasons for the reluctance to complete questionnaires require elucidation. Many of the validated QoL questionnaires (QLQs) in HNC patients have shown the ability to differentiate well between patients with tumours at different sub-sites, disease stages, and different treatment modalities. [4][5][6] They have also shown responsiveness over time, and even the ability to predict future QoL and survival. 7 However, there are a plethora of published QLQs for head and neck cancer (HNC) patients. 8,9 While the majority of clinicians indicate they find some QLQs useful, many clinicians are unsure which questionnaire to use, which limits the routine use of QLQs in HNC clinics. [8][9][10][11] Two of the most commonly-used questionnaires (EORTC QLQ-HN35 and FACT-HN) have demonstrated similar score characteristics, reliability, sensitivity and construct validity, but they appear to measure somewhat different aspects of health-related QoL. 12 That emphasises the importance of QLQ selection for the required setting and context.
The preferences of patient sub-groups may therefore be an important factor to determine the QoL instruments to be used in clinical practice.
However, this potentially important determinant of QLQ choice has rarely been explored before. We systematically searched MED-LINE and EMBASE databases for studies examining HNC patients' preference for QLQs, using a comprehensive search strategy with no language or publication date restrictions (search updated on 17 September 2022). A total of 730 articles were screened and only one relevant article by Mehanna and Morton 13 identified, which was a pilot study for this current trial.
Aiming to facilitate the routine use of QoL instruments in clinical practice, this study determined patients' preference and willingness to use four different QLQ in routine follow-up clinics.

| Study design
This was a multi-centre, cross-sectional, blinded randomised controlled trial (RCT), with an additional qualitative analysis. It was undertaken in 17 HNC secondary care clinics between 2008 and 2011 (Table S1). The study followed CONSORT reporting guidelines. 14 The study was given ethics approval by the Coventry and Warwickshire National Ethics Committee (06/Q2802/101).

| Participants
Patients attending for their HNC follow-up at the participating centres were assessed for eligibility. We included patients who had curative treatment for oral, oropharyngeal or laryngeal cancer, and were undergoing follow-up at least 1 month, and no more than 5 years, post-treatment. Patients with confirmed or suspected recurrence were excluded.

| Procedures
Study instrument and information leaflets were posted to patients before their clinic appointment. On attendance to the clinic, if willing to participate, patients gave written informed consent, and were asked to complete the questionnaires before meeting the clinician. The study instrument consisted of four HNC-specific QLQs, and a previously-piloted survey 13 that prompted patients to compare the questionnaires, Figure S1.

| Selection of study questionnaires
The four QLQs included in the study were: the Functional Assessment of Cancer Therapy Head and Neck questionnaire (FACT-HN) version

Key points
• This is the first randomised controlled trial to systematically examine head and neck cancer (HNC) patients' preferences for commonly used quality of life questionnaires (QLQs).
• Most HNC patients find QLQs of benefit in communicating with their clinicians.
• Patients overwhelmingly favour structured questionnaires over non-structured patient-generated lists.
• Only over half of HNC patients favour using the QLQs routinely at the clinic. Women and younger patients favoured FACT-HN and EORTC QLQ-HN35, respectively.  The FACT-HN, EORTC QLQ-HN35 and the UW-QOL were selected because they are the QLQs most widely used by HNC clinicians at the time of the study. The EORTC QLQ-HN35 has since been updated to EORTC QLQ-HN43. 19 FACT-HN comprised of 27 generic questions, and 12 Additional Concerns for Head and Neck questions. The UW-QOL comprised of 10 head-and neckspecific questions with 3 global health-related QoL questions, and one free-text question. Due to its length, we used the EORTC QLQ-HN35 alone, and added only two global QoL questions from EORTC QLQ-C30. The PCL questionnaire is an unstructured unvalidated questionnaire, that asked patients to list up to 10 health concerns in their own language on a blank sheet, and rank them in order of priority. For further details of instrument characteristics, see Table S2.

| Randomisation and masking
All patients received all four QLQs to complete, but the order of questionnaire presentation was randomised in equal ratio to avoid habituation and tiredness bias. Study packs were collated centrally according to one of 24 randomised combinations. using a computer generated random-permuted block algorithm design. Consecutive patients received sequentially numbered, sealed opaque envelopes, according to their stratification by disease site (oral or oropharyngeal versus laryngeal) and stage (Stage I, II vs. Stage III, IV). The randomisation allocation sequence was concealed from the patients and local researchers.

| Outcomes
The primary outcome was to determine the most helpful QLQ in describing the patients' health problems ( Figure S1). Secondary outcomes included: odds ratio (OR) for willingness to complete a questionnaire in the clinic; OR for the least helpful instrument in describing their health problems; the reasons for patients' preferences; and patients' preferences for the characteristics of an ideal instrument for routine use in the clinic.

| Statistical analysis
To detect a 50:25:25 true division of preferences, at 90% power and 5% significance, a total of 126 patients who expressed a preference were required in each stratum. To account for incomplete reporting and missing data, the total sample size was increased by 10% to 560 patients.
Descriptive methods were used to summarise the patient characteristics and instrument preference. Outcomes were analysed using logistic regression models with and without multivariate analysis fitted to generate the OR of patients' preference of each instrument.
Thematic analysis of the questionnaire responses was undertaken. Coding theory was used to explore the qualitative data from the unordered list to identify themes of concerns most important to patients. Using the NVIVO software, coding was independently conducted by Ben Carter, who had no knowledge of the content of the QLQs and was naïve of any likely codes. Common Terminology Criteria for Adverse Events (CTCAE) version 4.0 was used to provide definitions for medical concerns and aid coding of the data. 20 Codes were developed into themes through multiple discussions between both coders and a clinical author Hisham Mehanna.    Table 2).

| Head-and neck-specific versus combined general and specific questionnaires
After adjustment for variables (Table 2), the most common preference was for any structured questionnaire, that is, no specific preference for a particular one (OR 5.3, 95% CI 2.3-8.0, p = .001). Amongst those who expressed a specific preference, the three validated questionnaires were equally preferred, with no statistically significant differences, suggesting that the two head and neck-specific questionnaires (EORTC QLQ-HN35 and UW-QOL) were as good as the combined general and specific questionnaire (FACT-HN).

| Thematic qualitative analysis
Sixteen themes were reported by patients in the open unstructured list of concerns (Table 3). Of these themes, the EORTC QLQ-HN35 did not identify two head and neck specific complaints: periodontal complaints and dysgeusia. UW-QOL and FACT-HN, in addition to these, also did not cover cough complaints, and the latter also did not include dry mouth. Patients consistently reported the least preference for the unstructured patient-generated list.

| Willingness to complete QoL instruments routinely at the follow-up clinic
The advent of computer tablets and smartphone technology may provide an opportunity to facilitate the individualization of patient QoL assessment. 24 However, minimal clinically-important differences and trigger thresholds for interventions need to be determined for the different questionnaires to derive the maximum benefit from them. 25 Research into all these factors would be important, and into the development of algorithms to 'translate' changes in scores between the different questionnaires, thereby allowing clearer and easier understanding of the clinical implications.

| CONCLUSIONS
We found that HNC patients value the use of QoL instruments to aid communicating their health needs to their doctors. They preferred structured disease-specific instruments, but a large proportion also found all the questionnaires useful. The simple patient-generated unstructured list was the least preferred. Only half of our patients expressed preference to complete QLQs routinely in the clinic, which is significantly influenced by their age and gender. F I G U R E 2 Relative risk and 95% confidence interval for patients wanting to complete the questionnaire at the clinic (red triangle); and patients reporting that longer questionnaires (≥21 questions) were most desirable (blue square).