SIP SMART is a pre-treatment swallowing intervention for newly diagnosed head and neck cancer patients. All four main criteria specified in the protocol to determine success from the feasibility trial were satisfied: the target recruitment (32 patients) was achieved at an average rate of four patients a month; the MDADI was identified as a suitable primary outcome for which sample size was estimated; patient responses to the questionnaire on acceptability and randomisation was mainly positive; and the minimum reported adherence for the intervention group was attained.
The recruitment strategy of using the weekly multidisciplinary meeting list for pre-screening worked well and would be relatively straightforward to implement in a larger multi-centre trial. Mapping the different treatment pathways that exist at each centre, and clarifying how these will impact eligibility for inclusion in the trial will help to avoid over-estimating recruitment potential. The researcher and research nurse found it useful to identify potential patients to the treating consultant at the start of the clinic, and to ensure a study card was attached to the front of the medical notes so that the researcher could be alerted when the patient was being seen. Whilst this worked for the current study, we recognise that many hospitals are moving toward electronic health record systems, which may make it easier for potential participants to be identified. Given the narrow window of opportunity to enrol patients into this pre-treatment intervention study, these strategies should be fully explored for each site involved in a future trial ensuring full MDT support to achieve good recruitment. Patients without adequate proficiency in English were excluded from the current study given the nature of the intervention. However, a sizable number of the patients who were not eligible (11/26, 42%) were excluded due to language barriers, making it important to consider whether a future trial should cost for interpreter services.
Participant characteristics were broadly reflective of the population of patients who present with HNC. There was a higher incidence of males compared to females, although the male:female ratio across oral and oropharyngeal cancers is approximately 3:1 (18–20). The average age (57 years) reflects the typical patient demographic. The sample included a higher proportion of patients with oropharyngeal cancer that tends to be associated with the human papillomavirus and often manifests before 60 years of age (18). It is possible that this demographic may differ in parts of the country where other etiological factors (for example, chewing betel products) play a more dominant role in cancer incidence (21). In our sample, there was a higher proportion of patients who received chemo-radiation therapy compared with surgery as first line treatment as this is the standard care for the tumour site (mainly oropharyngeal) and tumour stage (Stage III/IV) mostly found in our sample (22). The placement of a prophylactic feeding tube for individuals with advanced tumours undergoing chemo-radiation is in keeping with the practice guidelines at the institution.
Choosing the type and timing of primary outcome measure was an important outcome for this feasibility trial. Patients treated for HNC usually show a decline in swallowing function in the immediate post-treatment phase with gradual recovery over the following months until more stable functioning is achieved at around 6-months after treatment (23,24). The pattern of data for all swallowing outcome measures in this feasibility study showed a similar trend. It may therefore be appropriate to consider whether the research process for the future trial would be better served by having just two time-points (baseline and 6-month) for data collection, thereby ensuring a higher percentage of data completeness as recorded in this trial. Guidelines suggest that a 70% completion rate for studies with 6-month or longer follow-up would be satisfactory (16).
A known minimum clinically important difference (MCID) is helpful in estimating sample size for a primary outcome. Published data suggest that an actual difference of 5-12 points on the FACT-HN (25) and 10 points on the MDADI and PSS-HN(Performance Status Scale) are generally regarded as clinically meaningful (26). The FACT-HN demonstrated the best effect size but may be too generic and more suitable as a secondary outcome. MIO focuses on jaw opening only and is therefore too narrow to capture swallowing function but could provide useful information as a secondary measure. MIO is perhaps the easiest measure by which to observe changes that may occur if patients adhere well to their exercises. While the MBS composite score might be the most appropriate measure of swallow physiology, the scoring system is yet to be validated, and it was much more challenging to obtain complete data on this measure. The decision to select the MDADI as the outcome measure for sample size estimation was therefore informed by several factors: it is a validated patient reported measure, has a known MCID, shown to have excellent completion rates in the feasibility study and has an achievable sample size for a definitive trial of SIP SMART.
The majority of patients reported positive views towards participation in the trial and were not worried by the idea of randomization. The method used for randomisation was well received by both patients and clinicians, as it was transparent and done immediately after consent. Previous research has indicated that patients may decline participation in trials if they have a preference for one arm of the trial or if they are worried about the idea of randomisation (14). Based on questionnaire responses, this was not a major concern for SIP SMART. Patients seemed to have understood the concept of chance and were satisfied that either the new intervention or usual care would be suitable. In general, the reasons reported for participation in this trial were consistent with those reported in other cancer trials (14). On the whole, patients newly diagnosed with HNC are willing to participate in this type of trial often for altruistic reasons. It is therefore expected that patients would be no less likely to agree to participate in a future larger trial, despite recruitment occurring at a difficult time pre-treatment.
This feasibility study made use of a study specific form to capture information on exercise adherence. The target was to obtain a minimum of 35% of patients achieving at least satisfactory to good adherence to the intervention based on figures reported in a previous similar trial (15). There remains much debate about how to measure adherence to swallowing exercises. Some researchers have suggested that adherence should be measured in terms of both frequency and intensity of exercises, highlighting the challenges for measuring adherence in clinical trials (27,28). It is possible that the measurement method used in this study was too generalized, and that adherence was overestimated. In the absence of any formal measure, the study specific form devised for SIP SMART used a combination of questions (reflecting frequency and intensity) to classify responses into satisfactory to good versus poor to no adherence. Further work may be necessary to ascertain whether a more suitable method may be available to capture adherence in a larger trial or whether the current method provides a good enough reflection of adherence given that most current methods rely on patient self-report. Researcher effort and patient burden involved in obtaining more detailed and specific adherence information will be important considerations when making this decision in a future trial.
While no direct safety concerns or harms arose from the feasibility trial, it is important to recognize that patients participating in such a behavioural intervention may feel under immense pressure if they are unable to adhere to their swallowing exercises. Feelings of guilt or hopelessness may surface, and it is imperative that mechanisms are in place to identify and counteract these feelings. A problem solving, facilitative approach was built into the intervention design to mitigate such feelings, and regular appointments with the team acted as an important safeguard. The modified barium swallow procedure was identified in the protocol as a possible source of harm if contrast material was inhaled. However, given that the modified barium swallow protocol starts with very small amounts of contrast (5mL) compared with that given during a standard barium swallow, patients at risk for airway compromise can usually be identified before they are asked to swallow any substantial volumes.
Study limitations must be acknowledged. This study was conducted as part of a fellowship in which the intervention developer (RG) was also involved in recruitment and was therefore a key driver in the study. As an invested recruiter, this may have increased success with recruitment. For this success to be replicated, it will be crucial to identify key individuals (principal investigators) to drive the project and enlist a similar level of multidisciplinary support at other sites. Failure to ensure this could be met with poorer recruitment in a larger trial.
Our inclusion/exclusion criteria was as per protocol, and we therefore cannot be sure of how the results may differ if the excluded groups are included. Furthermore, our sample had small numbers for some tumour sub-sites such as nasopharynx. Given that this was a feasibility trial, a larger trial would be necessary to consider differences in different cancer sub-groups.
For the feasibility trial, implementation of the intervention was done solely by the clinical researcher and was therefore well controlled. The researcher was able to accommodate performing the pre-treatment MBS that was part of the new intervention at short notice. Patients in the usual care group did not receive a baseline MBS, as it was not part of usual care. This would have required additional resources from the clinical team. Practical implementation of the protocol will therefore need to be carefully planned for each site taking into account available resources. Further to this, it will be necessary for some training in the new intervention so that it can be delivered with good fidelity. These aspects have not been addressed by the current feasibility study.
Patients, clinicians and the researcher were all aware of group allocation as blinding was not attempted in this feasibility trial. The researcher was not involved with the collection of outcome data (questionnaires and clinical measures were collected by the research nurse/other SLTs) and made every effort to maintain distance until the completion of the study. Improved processes to ensure blinding should be further explored prior to a definitive trial.
Practical constraints imposed limitations on the study design and procedures. For example, conducting an in-depth exploration of patient experiences after participation could have provided useful insights for further refinements of the intervention itself and/or the study processes. However, given the time constraints this could not be accommodated as part of the current programme of work. It may be possible to do this as part of further pilot work.
Three patients died (approximately 10%) over the duration of this trial, an important consideration given the nature of the disease and expected survival. Whilst none of these deaths were related to the SIP SMART intervention, attrition due to death is an important consideration for the future trial. To our knowledge, all patients who provided final outcome data at the end of the study were tumour-free, based on their last multidisciplinary team review. This study did not distinguish between cancers that were HPV positive (better prognosis and survival) and those that were not. A study sample with greater numbers of advanced cancer due to non-HPV positive disease may therefore show greater attrition due to generally poorer survival.
Several study-related forms were devised for the feasibility study and most could be used in a larger trial. Some may require modification. The case report form, for example, was detailed, and it may be possible to reduce the amount of information collected. In this study, information on smoking and alcohol could be reduced to simpler yes/no questions without the level of detail included in the current case report form.
Many previous studies of swallowing exercise interventions in the head and neck population have been reported as definitive trials drawing conclusions about efficacy despite frequent inadequate sample sizes (29). The focus of our study has been on feasibility outcomes that will optimise the planning and conduct of a future trial, and therefore does not report on intervention efficacy at this stage. It is encouraging to note that since our initial work on this topic, a more recent meta-analysis of swallowing exercise interventions in the head and neck population has demonstrated that exercises both before and after chemoradiation treatment show improvements in jaw opening and swallow function, with early interventions also having a positive effect on airway protection during swallowing (29). Although the review authors found no evidence for improved quality of life outcomes for either early or late interventions, their meta-analysis and recommendations render further support for more well-designed trials on this subject.