Introduction
Randomised, controlled trials (RCTs) are the ‘gold standard’ of medical research, because they enable the prediction of outcomes for given interventions. However, RCT methodology may be less relevant to some patient groups because of what has been termed ‘diffuse technology’ – an acknowledgement that some treatments are not easy to prescribe in experimental terms.
One example of this is the management of head and neck cancer patients.Reference Ferguson, Franks, Harrison, Baker and Kirk1 The complexity and heterogeneity of their condition, and the impact of both the disease and its treatment, mean that RCT investigation of relevant interventions may be neither appropriate nor possible to conduct satisfactorily. Head and neck cancer patients' treatment, therefore, may be best managed not through rigid adherence to clinical protocol, but instead through incorporating into clinical decision-making a wide range of patient-centred factors.
This position is supported by the research on clinical judgement. The bedrock of treatment decision-making, clinical judgement is defined as the fusion of technical judgement (encompassing scientific and evidence-based information) and humane judgement (relying on the health care professional's interpretation and insight into the patient and their symptoms).Reference Downie, Macnauhgton, Downie and Macnauhgton2 It is clear from this definition that clinicians need to use, and to interpret, a constellation of objective, empirical data as well as subjective, experiential and interpersonal data.
Traditionally, the technical aspect of clinical judgement has relied on research. In head and neck cancer patients, research has focussed almost solely on survival and loco-regional recurrence rates as outcome measures.Reference Norris, Busse and Clark3, Reference Bernier, Domenge, Ozsahin, Matuszewska, Lefèbvre and Greiner4 However, this takes little account of the comprehensive impact which head and neck cancer and its treatment may have on the quality of all aspects of the patient's life – an impact which, given the complex and diverse nature of the condition, should be central to humane judgements regarding treatment options.
More recently, an acknowledgment of this fact has resulted in an exponential growth in research relating to the quality of life (QOL) of patients with head and neck cancer.Reference Rogers, Ahad and Murphy5 The typically poor prognosis, the wide variation in reported problems, and the highly individual response to the condition and its management all mean that treatment decisions must be predicated on both technical and humane aspects of clinical judgement.
However, if humane judgements are being used in the management of this patient group, it is important to establish whether the health care professionals involved in treatment decisions share an understanding of the physical and psycho-social implications of the condition for the patient. Without a common understanding of what is relevant and why, clinical judgements may be ill-conceived and randomly applied, rather than informed and systematic. Health care professionals therefore need to be aware of, and responsive to, a range of subtle and often immeasurable psycho-social cues and patient concerns, and must be able to interpret this information in a way that enhances treatment decisions. Although this clearly challenges the evidence-based health care culture, it may, nonetheless, be highly relevant in the management of a condition which, because of its diversity, does not readily lend itself to randomised, controlled scrutiny.Reference Brown, Crawford, Hicks, Brown, Crawford and Hicks6
Research evidence suggests that the site and size of head and neck tumours may enable some reasonably reliable predictions regarding a range of outcomes; this would mean that such research can inform technical judgement and aid clinical decision-making. For example, the site of a tumour has been reported to influence the type and severity of pre-treatment swallowing difficulties; in addition, the site and stage of a tumour have been shown to have more impact than any other clinical features of the disease on health-related QOL.Reference Stenson, MacCracken, List, Haraf, Brockstein and Weichselbaum7, Reference Hammerlid, Bjordal, Ahlner-Elmqvist, Boysen, Evensen and Bijorklund8 Similarly, prior to treatment, a relationship between tumour burden, speech and swallowing has been demonstrated for oral and oropharyngeal cancer patients; these authors concluded that the percentage of oral tongue affected significantly reduced articulation and swallowing efficiency.Reference Colangelo, Logemann and Rademaker9
From such objective clinical evidence, it may be possible to formulate reasonably accurate predictions about how patients will report their condition and how it will affect their functioning and QOL. To this end, reported psycho-social and QOL information may be highly relevant to the management of an individual patient's condition.Reference Funk, Karnell, Smith and Christensen10
Multi-disciplinary teams (MDTs) play a crucial role in improving both the quality of treatment decisions and the coordination of cancer care; this means that their interpretation of the totality of the patient's condition will be pivotal.Reference Tattersall11 Consequently, the way in which MDTs use patient data to interpret the patient's condition and to make humane judgements may be worthy of investigation.
Therefore, the aim of this study was to investigate whether the MDT was influenced by the size of a patient's tumour when making judgements relating to QOL issues, prior to treatment. The research question was, ‘do health care professionals use their knowledge of the patient's condition (i.e. tumour size) consistently, to inform their humane judgements about the patient's QOL?’
Method
Study design
We used a between-subject design to compare the responses of two separate groups of health care professionals. The aim was to establish whether tumour size (i.e. the independent variable) affected the health care professionals' assumptions about a patient's QOL and other psycho-social issues (i.e. the dependent variables).
The study used a variant of Asch's central trait theory, with tumour size operating as a central trait.Reference Asch12 Asch's original study demonstrated that, by interchanging one adjective from two otherwise identical descriptive lists, participants would form a different impression of a person. Asch's findings were interpreted as suggesting that past experiences become organised as frames of reference and therefore affect current judgements. The theory also suggests that, in making a judgement about a person, a participant will eliminate any incompatible perspectives, since rarely are attitudes half-formed or absent. Evidence for central trait theory exists within the field of health care.Reference Darbyshire13
Based on this research, this pilot study used the former descriptors of tumour size to test whether manipulation of this variable affected health care professionals' assumptions about other aspects of a patient's condition.
Sample
A convenience sample of health care professionals was invited to participate. All worked in a MDT within a large head and neck cancer centre which was involved in the management of patients with oral tumours. Participants were selected if they were known to have regularly worked with head and neck cancer patients. Seventy-nine participants were approached, from a range of professional backgrounds (see Table I).
Table I Participants' professional details
Data represent number of participants. T = tumour size stage; H&NC = head & neck cancer
Materials
Materials comprised a questionnaire specifically designed for the purpose. This questionnaire was split into three parts.
The first part focussed on the participant's basic biographical details.
The second part provided a description of a fictitious head and neck cancer patient, giving clinical and social information which would be typical of a patient presenting to the participants' unit. The information about the fictitious patient was presented in the style and content format routinely used within the MDT weekly meetings, and in language used on the unit to describe patients (both in discussion and in written reports sent out to referring agents).
It was the researchers' deliberate intention that the hypothetical patient's gender be unspecified, and that the patient's job might be seen as one requiring some verbal communication on a daily basis. A carcinoma of the tongue was chosen, so that the location of the tumour would be clear and unambiguous. The description of the patient was identical for each group, apart from tumour size. To ensure clarity and avoid ambiguity in participants' understanding of the presenting clinical problem, the labels T2 (i.e. tumour size larger than 2 cm but smaller than 4 cm) or T4 (i.e. larger than 4 cm and invading adjacent structures) were used.
The third part of the questionnaire asked participants to rate 20 statements relating to a range of clinical and psycho-social outcomes which the patient described in part two might typically experience. These outcomes were derived from a thematic analysis of relevant research, including work involving the European Organisation for Research and Treatment of Cancer quality of life C30 questionnaire and head and neck specific questionnaire, both of which have been validated and are intended for use with head and neck cancer patients.Reference Aaeronson, Ahmedzai, Bergman, Bullinger, Cull and Duez14–Reference Sherman, Simontown, Camp-Adams, Vuval, Owens and Hanna17
Twenty domains were derived from this research, including: tension, energy, sleep, appetite, memory, depression, limitations on work, oral pain, choking, worry, oral communication, sexual enjoyment, swallowing diet, social eating, saliva consistency, use of supplements, sore mouth and dry mouth.Reference Colangelo, Logemann and Rademaker9, Reference Funk, Karnell, Smith and Christensen10 Participants were required to score 20 corresponding statements on a 10 cm visual analogue scale (such scales have been demonstrated to have good reliability and validity as measures of subjective experience).Reference McCormack, Horne and Sheather18
Procedure
Pre-pilot
A pilot form of the study was tested on three people to assess whether the questionnaire and procedure had any specific problems not identified by the investigators. No difficulties were identified.
Main pilot study
Questionnaires were distributed to each participant by hand. Participants were asked to complete the form on their own, without discussion with others. Of the 79 participants, 39 randomly received questionnaires containing the T2 variable, and the remaining 40 randomly received questionnaires containing the T4 variable. In all other respects, the questionnaire and the procedure were identical. The allocation to either variable followed a stratified random sampling to ensure that there was equal distribution of the two types of questionnaire between the participant groups. This procedure involved making a prerequisite number of envelopes to match the numbers of the various health care professionals involved, and then randomly allocating slips of paper to each envelope marked ‘T2’ or ‘T4’. The envelopes were then shuffled and distributed to the appropriate type of health care professional. All the questionnaires were completed anonymously. The data gathered were entered into the Statistical Package for the Social Sciences for Windows version 14 software, and analysed using descriptive and inferential statistical techniques.
Ethical considerations
This research was part of a much larger study approved by the ethics committee of South Birmingham. Participants were able to discuss with the investigator any queries about the questionnaire before completing it, and were able to withdraw without giving any reason.
Results and analysis
Seventy-two of the 79 selected participants completed the questionnaire according to the instructions; of these 72, 36 completed the T2 version and 36 completed the T4 version.
Table I gives details of the participants' professions and their years of experience working with head and neck cancer patients.
Pre-analysis check of questionnaire's internal reliability
To test the internal reliability of the questionnaire, respondents' answers for those variables that were either clinically or psychologically similar were analysed using a Pearson's product-moment correlation. Correlation coefficients were calculated for the following variables: social eating and difficulty eating solids; tension and worry; and health-related QOL and QOL. All showed significant positive correlations at a level of <0.005 (one-tailed), thereby suggesting that the instrument has some internal reliability.
To test the main hypothesis, independent t-tests were used to compare the responses from the T2 and T4 groups on each of the 20 variables in section three of the questionnaire. The results are presented in Table II.
Table II Statistical analysis* of participant group responses: T2vs T4
* Independent samples t-test. †Dependent variable. ‡One-tailed. **Significant at p≤0.05. T = tumour size stage; df = degrees of freedom; HRQOL = health-related quality of life; QOL = general quality of life
The effect sizes for the significant results were as follows: tension = 0.041 (i.e. small to moderate effect); ability to swallow diet = 0.04 (i.e. small to moderate effect); and social eating = 0.053 (i.e. small to moderate effect).Reference Cohen19 Participants' responses indicated that these three domains were thought to be significantly more difficult for a T4 patient than for a T2 patient. The remaining 17 domains yielded no significant differences between the participant groups.
To establish whether there was agreement within each participant group, Kendall coefficients of concordance were calculated, as follows.
T4 group
Calculation of Kendall's coefficient of concordance (using the data S = 337553.71, W = 0.39167, n = 36, N = 20; chi square = 26.791, 19 degrees of freedom) indicated that results were insignificant. That is, there was no significant agreement amongst participants as to which symptoms were most likely to be present in the described T4 tumour patient.
T2 group
Calculation of Kendall's coefficient of concordance (using the data S = 262729.80, W = 0.3858, n = 32, N = 20; chi square = 23.457, 19 degrees of freedom) indicated that the results were insignificant. This means that there was no significant agreement amongst participants as to which symptoms were most likely to be present in the described T2 tumour patient.
These results suggest that there was no agreement within either group regarding the psycho-social and physical effects of tumour size.
Discussion
Insofar as the participants were able to complete the task, it would seem that health care professionals associate an oral tumour with a wide range of emotional, cognitive and physical difficulties. However, these associations were not consistent for the two tumour sizes used. Moreover, there were no significant differences between the two participant groups regarding their assumptions about 17 of the 20 associated patient problems described. It can be concluded that, for this group of experienced health care professionals, tumour size was not demonstrated to be a reliable central trait. Only three variables produced any significant differences: eating socially, swallowing diet and tension were considered to be worse for the T4 patient.
Overall, these results do not agree with the available research, which suggests there is a consistent relationship between tumour size and a range of symptoms.Reference Stenson, MacCracken, List, Haraf, Brockstein and Weichselbaum7–Reference Colangelo, Logemann and Rademaker9 It is also worth noting that there was no consistency of results either between participants of the same professional group, or between those with a similar duration of clinical experience with head and neck cancer. Appropriate analysis of variance models indicated no significant differences for any of the 20 domains based on years of clinical experience, and only two significant differences based on professional group. These latter findings suggested that, compared with doctors and other health care professionals, nurses were more likely to believe that worry and depression would be a significant problem irrespective of tumour size.
There are several possible explanations for these results.
Firstly, it is conceivable that tumour size is not pivotal for health care professionals in their prediction of patients' symptoms, and that other factors (e.g. psycho-social factors and the nature of the patient) may be more influential in determining the perceived impact of the disease on patients' QOL.
Secondly (and related to the above), tumour size may be more variable in its clinical manifestations than the available research would suggest. The inherent heterogeneity of head and neck cancer patients means that the implications of tumour size will be confounded by a range of individual variations regarding psychological and/or medical response, and regarding the patient's personal and social circumstances.
Thirdly, the lack of consistency among the health care professionals studied may be a genuine reflection of their attempt not to stereotype patients, but rather to treat them on an individual basis.
Finally, the health care professionals involved may simply have been unaware of the research evidence relating to the impact of tumour size.
Taking the first two, interrelated explanations, it is conceivable that health care professionals consider head and neck cancer to be so inconsistent, subjective and unpredictable that there is little real clinical use in relying merely on the clinical evidence of tumour size when making predictions about disease impact. Certainly, anecdotal evidence from within the unit suggested that, within a six-month period, it was unlikely that there would be even two patients presenting with similar objective clinical criteria (e.g. age, gender, site, size and pathology). Such variability would mean not only that consistent clinical pictures were scarce, but also that health care professionals might not be able to build up a critical mass of experience on which to base their technical and humane judgement. Exposure to such diversity might also spawn a philosophy (much in favour with the current trend towards patient-centred care) that each case should be managed individually, rather than imposing a condition-driven protocol irrespective of the patient's real needs and problems. The uniqueness of the presenting problem would then preclude the generation of stereotyped assumptions on which to base treatment decisions. While this may be laudable in its objective, the lack of consensus within the groups regarding the clinical and QOL implications for their hypothetical patient may mean that, where treatment decisions are influenced by the humane judgements of health care professionals, arbitrary management may result. Such a situation would compromise target-driven outcomes.
The final possible explanation of our study results suggests that the health care professionals may simply not have been aware of the relevant research. While oncology research has, over the last 20 years, begun to acknowledge QOL as an important outcome which needs to be incorporated into treatment decisions, and one which needs time and experience to assimilate, synthesise and subsequently integrate, it is nonetheless conceivable that health care professionals simply lack knowledge of the relevant literature.Reference Frost and Sloan20 This contention is reinforced by the fact that there is a worrying under-use of QOL questionnaires in clinical practice nationally, despite the number of instruments available and the demonstrable value of the information they provide.Reference Kanatas and Rogers21 As the unit participating in the current study did not routinely collect QOL data, such contextual research may simply be unknown. Allied to this is the lack of research investigating patients' pre-treatment QOL symptoms – the focus of the current study.Reference Stenson, MacCracken, List, Haraf, Brockstein and Weichselbaum7, Reference Hammerlid, Bjordal, Ahlner-Elmqvist, Boysen, Evensen and Bijorklund8, Reference Scharloo, Baatenburg de Jong, Langeveld Ton, Els van Velzen-Verkaik, Doorn-op den Akker and Kaptein22 This may explain in part the typical focus on measurable end-points such as length of time to treat, length of stay, morbidity and disease-free survival. A recent paper suggested that the recent government preoccupation with health care reform has meant that the practical and mundane tasks necessary to improve the quality and outcome of complex services will always be secondary.Reference Haward23
• Health care professionals can draw upon a wide range of data when making clinical judgements
• Assessment of head and neck cancer patients must take account of many variables, in order to arrive at the best treatment options
• Quality of life is recognised as an important outcome measure when reviewing patient management
• In the health care professionals studied, tumour size did not generally appear to aid the assessment of a hypothetical patient's pre-treatment condition
• Study results were not influenced by participants' professional group or duration of clinical experience
• At the pre-treatment stage, patient input directly reflecting their own views should be sought, rather than relying on health care professionals' experiences and assumptions
In absolute terms, participants in both groups tended to attribute high degrees of worry, anxiety and allied problems to their hypothetical patient, irrespective of the severity of the presenting tumour. This statement may be supported by a review of health care professionals' general perceptions of cancer patient care,Reference Lampic and Sjoden24 which suggested that health care professionals tended to ascribe high levels of anxiety and distress to all patients with a cancer diagnosis, and to overestimate patient distress and under-estimate QOL. In practice, this may mean that patients are receiving care from health care professionals who have a more negative interpretation of their disease than do the patients themselves. The interpersonal dynamics of the professional–patient relationship may act as a conduit for these negative suppositions, which in turn may adversely affect the patient's pre-treatment psychological state. Clearly, it would be desirable for the emotional interaction between health care professionals and patients to be based on the facts of the individual patient's situation, rather than on assumptions made from health care workers previous experiences of management. This pilot study needs to be replicated before any generalisations can be drawn, and in addition, any impact of health care professionals' assumptions on the actual delivery of care needs to be addressed.
Conclusion
This study has demonstrated that, in a small, convenience sample at least, there was very little shared understanding of the physical and psycho-social implications of different oral tumour sizes, despite the available evidence. While there may be many reasons for this, such variation in views may adversely affect the humane judgements used to inform the management of this complex and diverse disease. Consequently, such lack of agreement regarding relevant patient factors may result in inconsistent decision-making, unless the patient is more actively involved in the process. The systematic, routine use of psychometrically sound QOL measures could be a means to better incorporate patients' experiences and desires into their care plans.
Acknowledgements
The authors are grateful for the cooperation and participation of the staff of the Head and Neck Unit, University Hospital Birmingham NHS Foundation Trust.
