Keywords

49.1 Introduction: What Is QOL?

Traditionally, the outcome of cancer care was assessed in terms of survival and/or tumor response. As early as 1948, Karnofsky recognized that other outcomes were important to patients. In his study, he used a performance status scale, showing that “subjective improvement was indicated by the patient’s feeling of well-being, his increased appetite and strength, and the relief of specific complaints…” [1]. In the intervening 65 years, his initial concept of patient well-being has evolved into our modern, patient-reported concept of quality of life (QOL). QOL is now recognized as an important outcome of cancer care.

49.1.1 Definition

Broadly speaking, QOL is a measure of an individual’s overall personal well-being. Three aspects critical to the concept are subjectivity (only the individual truly knows his or her own internal state), multidimensionality, and sociocultural context.

49.1.2 QOL and “Health-Related” QOL

Overall QOL is impacted by issues such as income and adequacy of housing, which cannot typically be influenced by the health-care system. In the context of health care, QOL measures are often used to measure the effect of disease, illness, and treatment on the patient and family. For this purpose, issues which are not expected to change based on these effects become measurement “noise” and reduce the ability of questionnaires to detect actual changes. For this reason, the more limited concept of “health-related” QOL is usually applied. The World Health Organization (WHO) has defined it as: “an individual’s perception of their position in life in the context of the culture and value systems in which they live and in relation to their goals, standards and concerns. It is a broad ranging concept affected in a complex way by the person’s physical health, psychosocial state, level of independence, social relationships, and their relationships to salient features of their environment.” [2] When the term “quality of life” is used in the context of health care (and in the remainder of this chapter), it is usually health-related QOL which is meant.

49.1.3 Domains and Multidimensionality

Human beings are complex: the overall human experience reflects many underlying functions and roles. Under the stress of illness, that experience is influenced as well by specific symptoms. Such complexity may be addressed by two very different methods. The first method attempts to explicitly address the many dimensions of experience by constructing specific “domains” within a questionnaire, including cognitive, emotional, social, spiritual, role and physical functioning, and specific symptoms. This approach results in long questionnaires with multiple items organized into separate subscales relating to each domain. The alternative method is to rely upon the respondent’s ability to internally integrate his or her experience and to report overall QOL as a single-item index. One example would be the use of visual analogue scales, such as the “feeling thermometer,” originally developed in 1964 by the US National Election Services to allow voters to rate their feelings toward political candidates but more recently adapted as a health utility instrument [3, 4]. Some instruments use a mixture of both methods; for example, “overall QOL” may be included as a single item, along with more specific domains. Typically, multi-item instruments are more reliable and more sensitive to change over time than single items; however, they require more time to complete.

49.1.4 Patient-Reported Outcomes

In 2006, the US Food and Drug Administration issued a draft guidance document, subsequently finalized in 2009, addressing the use of patient-reported outcomes (PROs) in support of drug-labeling claims [5]. This document was received as both a strong recognition of the importance of QOL research and as a controversial perspective, especially due to several methodological recommendations. Nevertheless, although QOL research is conducted for many reasons beyond the development of new drugs, it has been influential around the world. The term PRO, which has become increasingly popular as a result of the guidance, covers both QOL and other outcomes which may be solicited directly from the patient, such as adherence to therapy, satisfaction with treatment, and direct symptom ratings.

49.1.5 Health Utilities

Utility measures are intended to quantify not only health but also its value to the individual. They are derived from utility theory to address preference under conditions of uncertainty [6]. Direct utility assessment uses one of two methods: the standard gamble, in which the respondent must accept a risk of immediate death to gain QOL, or the time trade-off, in which he or she gives up time in order to gain QOL. Both methods use the concept of “perfect health,” defined by the WHO as “a state of complete physical, mental and social well-being, not merely the absence of disease and infirmity” [7]. Utilities have the advantage that they may be used directly as quality weights to determine “quality adjusted life years” (QALYs) for use in decision analyses, thus allowing integration of quality and quantity of life. Direct utility measurement requires abstract thinking and an understanding of probabilities, so it can be cognitively challenging for some patients and cumbersome to use in busy clinical settings [8]. More feasible alternatives include ratings scales, such as the feeling thermometer, or multi-attribute utility scales (questionnaires for which response options have known utility weights).

49.2 Dominant QOL Issues in Head and Neck Cancer Patients

Specific concerns such as xerostomia, pain, dysphagia, and speech disruption often dominate the posttreatment QOL experience of head and neck cancer (HNC) patients. Only a brief review can be provided below. More detail is available in a recent review article [9]. Additionally, under the auspices of the National Cancer Institute (NCI) of the USA, a “Toolbox” was recently published, suggesting commonly used and acceptable measures for assessing these issues [10].

49.2.1 Pain

Pain in HNC patients arises as a result of many factors: tumor-related ulceration, pressure effect, or nerve infiltration; acute treatment-related pain due to radiation and/or chemotherapy mucositis and postoperative wounds; and late treatment-related effects such as shoulder dysfunction, trismus, chronic edema, or osteoradionecrosis (bone necrosis due to radiotherapy). The quality and timing of pain can differ for each responsible mechanism so that a full characterization of pain may require detailed questioning. QOL instruments for HNC will typically include 2–3 pain-related questions: one on general pain, one specific to pain in the mouth or throat, and perhaps one related to shoulder discomfort [11]. A PRO specifically focused on pain, such as the Brief Pain Inventory (BPI), may complement QOL instruments when pain relief is a focus of treatment, such as in the palliative care setting [12]. International efforts to harmonize the classification of pain are ongoing [13].

49.2.2 Xerostomia

Xerostomia, or dry mouth, is a complex problem. Acute and late phases of xerostomia differ in both their pathophysiology and their response to preventive strategies [14]. Salivary fraction from the parotid glands, submandibular/submental glands, and minor salivary glands may play different roles in baseline dryness and in eating-related difficulties. Similarly, swallowing and speech performance have been shown to be impaired in xerostomic patients [15, 16]. Evidence to link xerostomia prevention strategies to reduction in late complications such as dental caries, osteoradionecrosis, and chronic malnutrition is lacking.

The relationship between reduced salivary flow, patient-reported dry mouth, and overall QOL is complex. Reduction in salivary flow to ≤ 25 % of baseline has been arbitrarily classified as xerostomia [17]. Physician-rated outcomes include the Radiation Therapy Oncology Group (RTOG)/EORTC grading scale; the Late Effects in Normal Tissue—Subjective, Objective, Management, and Analytic (LENT-SOMA); and the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE version 4) systems [18, 19, 20, 21]. These measures have rarely been validated against salivary flow or PRO data. All common HNC-specific QOL instruments include at least one item related to xerostomia; however, non-QOL PROs specific to xerostomia have also been developed. Two popular instruments have been a six-item linear analogue scale (LAS) [22] and the eight-item University of Michigan XQ [23]. Though less rigorously developed and validated than most HNC QOL questionnaires, these instruments have performed well in research use.

Clinical strategies to reduce the risk of xerostomia for patients treated with radiotherapy (RT) have included the use of the drugs pilocarpine [24] and amifostine [25], intensity-modulated RT (IMRT) [26, 27], and surgical salivary gland transfer [28–30]. A recent review discusses the literature supporting each strategy, with the latter two approaches having the best evidence of effectiveness [9].

49.2.3 Speech

Of HNC patients who undergo surgical treatment, speech will be affected in most patients immediately after surgery [31] and continues to be affected in over a third (37 %) of patients at 3 months post-surgery [32]. Measures that target speech include the Swedish Self-Evaluation of Communication Experiences after Laryngeal Cancer (S-SECEL) [33] and the voice handicap index (VHI) [34]. A linear analogue self-assessment (LASA) tool has also been developed [35].

49.2.4 Swallowing

Swallowing relies on complex coordination of function and is frequently disrupted by both surgical treatment and RT. After head and neck surgery, short-term dysphagia is common, with about half of the patients experiencing dysphagia at 3 years [36]. Post-RT dysphagia may be worsened with concurrent chemotherapy [31] and may increase in severity over the years [37]. There is good evidence that impairment of both swallowing and speech significantly reduce overall QOL [38]. Fortunately, more than 75 % of selected patients with dysphagia may return to oral intake with swallowing rehabilitation [39].

The gold standard for assessment of dysphagia is the videofluoroscopic (VFS) assessment. A popular clinician-rated performance status measure, the Performance Status Scale for Head and Neck Cancer (PSS-HN) [40], focuses on the impact of dysphagia. Patient-reported QOL measures targeting dysphagia include the MD Anderson Dysphagia Inventory (MDADI) [41], the swallowing quality of life (SWAL-QOL) [42–44], and the swallowing quality of care (SWAL-CARE) [42–44]. Patients’ perceptions of their swallowing problems are not always consistent with their physiological swallowing ability. Some patients with normal VFS may perceive swallowing difficulties, whereas silent aspiration leading to pneumonia can occur in others [45].

49.3 Measurement of QOL: Basic Methodology

QOL instruments measure a subjective concept, but their measurement properties are based on sound scientific principles. Psychometrics, the science of indirect measurement through questionnaires and other related instruments, evolved in educational and psychology research over the course of the twentieth century. It has been applied to health-related questionnaires and PROs for over 20 years [46]. Instruments chosen for use in clinical research should adhere to the principles outlined below.

Item generation should incorporate information about the issues of importance to patients from literature review, health professional expertise, and direct input from patients similar to the instrument’s target population. Questions should be written at an appropriate educational level; grade 6 is often recommended [47]. Items should be formatted in a standard way, including both positively and negatively worded items, and avoid jargon, skip formats, and double-barreled questions. Utilization in other languages and cultural groups requires a formal process of cultural adaptation, including forward and back translation and pilot and field testing in the new language/culture [48].

Item reduction is often required to produce a questionnaire of practical length that remains sufficiently sensitive to change over time for evaluative (longitudinal) use. Direct testing in patients is typically carried out to identify the items most frequently endorsed by patients and ranked as being of the greatest importance. Statistical methods may also be used to identify items which are most informative [49].

Questionnaire design includes principles of readability and clarity. Questionnaires should include a large proportion of white space, with font size and type which is easy to read. Special requirements for the target group need to be considered (e.g., the visually impaired, young children, low-literacy populations, etc.) [47].

49.3.1 Indices and Profiles

Controversy exists regarding the relative preferability of indices or profiles for QOL measurement. Different individuals may apply personal weights to aspects of their quality of life, so summation of scores over multiple domains, as is done for indices, may impose the developer’s values inappropriately on the patient. Exploration of individual, patient-assigned weighting has proven cumbersome and is rarely used. Other instruments present scores separately for each domain (profiles), without summation. Popular questionnaires of both types are currently in use.

Reliability refers to the reproducibility of scores. It may be assessed by repeated administration of the instrument to a population with stable QOL (test-retest reliability) or by correlation of items within a questionnaire (internal consistency). Higher levels of reliability coefficients are conventionally required for evaluative use (to measure change in individuals over time) than in discriminative use (to measure difference between groups of patients), typically 0.8 and 0.7, respectively, for internal consistency [50, 51].

Validity refers to the ability of a questionnaire score to reflect the actual concept of interest. It is important that a “QOL questionnaire” is actually related to the patient’s overall well-being during a defined period (e.g., one week) and not his or her momentary comfort or passing mood. Questionnaire validation lacks a gold standard, so validity is defined by hypothesis testing with respect to convergence or divergence from other findings (concurrent validity). For example, QOL scores might be expected to be better in patients with better performance status and to improve over time in patients who were gaining weight posttreatment. An HNC-specific QOL questionnaire would also be expected to show a moderate correlation with other, more general, QOL or utility instruments. It is important that validation studies included patients similar to those for whom the instrument will be used; a questionnaire validated exclusively in surgically treated patients may not exhibit the same measurement properties in chemoradiation patients.

Responsiveness is the sensitivity of the instrument to changes over time in an individual patient. Responsiveness is correlated with instrument length and the specificity of items. A very detailed, HNC-specific QOL instrument would be highly responsive, whereas a short, general QOL instrument would be less responsive, to change in a HNC patient. Prospective evaluation is required to determine instrument responsiveness.

Minimal clinically important difference (MID) is defined as the smallest change in value on a measurement instrument, which, from the point of view of the patient, represents an important rather than trivial change. In practice, it has been estimated for groups by the use of the minimal detectable difference, that is, the smallest difference which is detectable by the average patient [52]. It is important to differentiate this clinical concept from statistically significant differences, which reflect only the likelihood of observing a given difference, not what it may or may not mean to a patient. Ideally, MID should be determined for every new instrument; however, several studies suggest that a change of 5–10 % of instrument range may represent the MID for many instruments [53, 54].

49.4 Types of QOL Questionnaires

General QOL instruments can be applied to the general population, as well as to those suffering from various types of illness. Popular examples include the SF-36 [55, 56] and the EQ-5D [57].

Disease-specific QOL instruments have been developed in patients with specific types of illness, such as cardiac or respiratory disease, or of course cancer. Two of the most popular cancer-specific QOL instruments are the EORTC QLQ-C30 [58] and the FACT-G [59]. Their questions are better suited to the difficulties of cancer patients, resulting in better validity and responsiveness as compared to general instruments; however, the trade-off is increased difficulty in comparing results with those from healthy people.

Symptom-specific QOL instruments have been developed for several symptoms of importance to HNC patients, such as dysphagia (e.g., MD Anderson Dysphagia Index or MDADI) [41].

Treatment-specific QOL instruments exist for many cancer treatments which are not specific to HNC (e.g., FACT-Taxane) [60]. The author has recently developed an instrument for HNC patients with prophylactic feeding tubes, originally called the QOL-EF and now FACT-EF [61].

Oncology and disease site-specific QOL instruments are a subset of cancer-specific instruments designed for a specific cancer site, such as HNC. Two structured literature reviews have evaluated such instruments for HNC [11, 62]. Several of these instruments are modular, incorporating a cancer-specific instrument and a disease site-specific module (e.g., EORTC, FACT, and QOL-RTI instruments). A selection of the more popular, well-validated questionnaires is mentioned below, with a summary of their characteristics in Table 49.1. Other PROs which are designed for HNC but which focus on performance status, symptoms, specific treatments, or functional issues (e.g., dysphagia, voice, disfigurement, xerostomia) may be complimentary to these HNC QOL instruments.

Table 49.1 Selected characteristics of HNC-specific QOL questionnaires
Full size table

49.5 Popular H&N Cancer-Specific QOL Instruments

EORTC QLQ-C30/HN35 [63, 64] is the most commonly used instrument [65], as well as the longest. It has been translated and cross-culturally validated in many languages. FACT-H&N [40, 59] is another modular instrument which has been translated into many languages; it has been popular in North America. Several English-language instruments have been developed at American universities: the popular, surgically oriented UW-QOL [66, 67] at the University of Washington; the HNQOLQ [68] at the University of Michigan; the HNCI [69] at the University of Iowa; and a modular instrument designed for RT patients, the QOL-RTI [70], at the University of South Florida. Finally, the HNRQ [71] was developed with a specific focus on acute QOL in patients with advanced HNC receiving RT or chemoRT but has been used infrequently.

49.6 Interpretation of QOL Results

Each individual conceptualizes QOL in a personal way. Life experience, optimism or pessimism, and psychological state all contribute to the perception of QOL. Consequently, cross-sectional comparisons among individuals are subject to measurement “noise” which should be less problematic when patient scores are self-controlled, by calculating one individual’s change in QOL over time in a longitudinal study. For this reason, if QOL is to be used as an outcome of a treatment in a clinical trial, prospective measurement at multiple time points is preferred. However, it is important to realize that the baseline administration usually occurs soon after a patient has received a cancer diagnosis or has been found to have disease recurrence or progression. Thus, the “baseline QOL” does not reflect that person’s QOL when healthy. QOL scores that return to baseline over a period of time cannot be interpreted as indicating a resolution of tumor- and treatment-related effects; in many cases, the patient may, in fact, have exchanged tumor-related impairments for different problems induced by treatment.

49.6.1 Response Shift

An additional important consideration in the interpretation of longitudinal QOL data relates to response shift or changing internal standards [72]. Over time, an individual confronted with critical illness may modify his or her values, or standards of measurement, and may also reconceptualize QOL entirely. Response shift may play a role in some initially unexpected findings, such as the fact that patients with serious illness will routinely rate their own QOL as better than the ratings applied to them by surrogates (e.g., family members or health-care professionals). Response shift may be viewed as a beneficial adaptive process; however, it also introduces an additional source of measurement error. Methods of quantifying response shift exist but are labor intensive. One approach in descriptive studies is to compare QOL results with population norms drawn from healthy individuals [73]. Once again, the randomized trial design is favored for studies with QOL outcomes, since it is hoped that unmeasured covariates such as response shift should be balanced between the arms by chance.

49.6.2 Compliance and Missing Data

Results of any study must be assessed for two types of validity: internal validity (does the study measure what it says it does?) and external validity (generalizability). In QOL studies, compliance with planned questionnaires and missing data can threaten both types of validity. Patients self-select study participation, which influences external validity (i.e., study results are applicable only to the type of patients who agreed to participate). Once enrolled in the study, participants determine whether or not they complete requested evaluations. Certain questions or even pages of a given questionnaire may not be completed, or the entire questionnaire may have been missed, either because the patient did not attend a scheduled appointment or because he or she attended but did not complete the QOL instrument. Missed questionnaires threaten both types of validity, since reported results do not really reflect the experience of all patients in the study. Specifically, it has been shown that healthier patients are more likely to comply with QOL assessments [74]. While statistical methods exist to attempt to correct for missing data, they require the assumption that data is missing at random, which is known to be unlikely in QOL studies. Consequently, every effort should be made to maximize compliance. Strategies to do so include adequate resources, education and feedback for those administering the questionnaires, real-time monitoring of compliance, and backup methods of administering questionnaires if an error is detected within an acceptable time window [75].

49.6.3 Mean Changes Versus Response Analyses

Longitudinal studies may report mean change in an overall group; however, this can overestimate longer-term QOL due to “survivor effect”: data from all patients will be included at baseline, but only patients who survive and continue to comply with assessments are included in the follow-up. In comparative two-arm trials, it is even possible that the QOL may appear to be better in the arm with fewer survivors, since a more toxic treatment may selectively eliminate those with poorer QOL. One alternative is to prespecify the QOL hypothesis and MID and analyze QOL response. Each participant is categorized according to “improved,” “stable,” or “worsened” QOL, and arms are compared for proportion of patients with a QOL benefit [76]. This approach also allows calculation of a number needed to treat (NNT) statistic [77].

49.6.4 Knowledge Translation

The concept of knowledge translation refers to the gap between evidence and practice [78]. Awareness, agreement, adoption, and adherence have been proposed as the necessary steps required before clinicians will use new knowledge. A prerequisite of both awareness and agreement is that information must be presented in a manner which is interpretable and usable. This has been a challenge for QOL data [79]. Two user’s guides have been published to assist the clinician with evaluating and interpreting QOL results [80, 81]. The International Society for Quality of Life Research (ISOQOL) has recently published guidelines on methodologic details which should be included in publications of QOL results [82]. However, a review showed that in recent publications of oncology clinical trials, recommended information items were included in only 10–70 % of cases; a trend to improvement of most data points was seen over time [83]. In a survey of academic oncologists in Canada, the UK, and Australia, 73 % reported that published QOL data were useful; however, only 45 % integrated such data in consultations with most of their patients; there was a call for consistent, improved reporting standards, and discussion of clinical implications in reports of QOL results [84]. Additional research is needed to help bridge the current gap between QOL researchers and oncologists in the clinical setting.

49.7 Research and Clinical Applications

49.7.1 Clinical Trials

The concept of levels of evidence for medical decision-making applies to QOL research just as it does to studies with survival outcomes. However, the field of QOL research is newer, and few phase III randomized controlled trials (RCTs) in HNC have yet reported QOL results. Two recent systematic reviews searched for high-quality studies of QOL for HNC patients receiving definitive (chemo)radiotherapy [85] and primary surgical treatment with or without adjuvant treatment [86]. The first found 18 papers, 8 of them prospective, while the second found 26 and 9, respectively; no phase III randomized controlled trial (RCT) results were identified among papers that met quality criteria. However, RCTs using valid instruments have begun to appear [87], and most current ongoing phase III trials include a QOL component. Publication of QOL results from several large RCTs led from Australia/New Zealand (TROG 02.02 HeadSTART), Canada (NCIC-CTG HN.6), and the USA (NRG/RTOG 1016 and 0129) is expected in 2016.

A few useful results have been reported for definitive therapy. QOL after RT for nasopharynx cancer (NPC) was found to be superior after parotid sparing with IMRT than with conventional RT [27], and HNC patients had better experience with selected domains of QOL (especially dry mouth and mouth opening) with IMRT versus 3D conformal RT [88]. QOL was found to return to baseline 12 months after treatment of locoregionally advanced HNC, regardless of whether patients received RT or RT plus cetuximab [87], and sequential versus alternating chemotherapy and radiation [89]. A comparison of intravenous versus intra-arterial cisplatin given with concurrent RT showed only a transient worsening of nausea and vomiting reported in the intravenous group, with no differences in QOL at 1 or 5 years [90]. Induction TPF chemotherapy before RT, relative to PF, did not worsen and may have improved global QOL, while showing better survival and toxicity [91].

In the adjuvant setting, the addition of subcutaneous mistletoe extract to surgery +/− postoperative RT had no effect on QOL as compared to no further therapy [92].

In the palliative setting, injection of a cisplatin/epinephrine gel into HNC tumors, as compared to placebo, did not appear to alter QOL, although QOL compliance was poor [93]. No differences in QOL change from baseline were seen between methotrexate and gefitinib for recurrent/metastatic HNC [94].

Other RCTs have focused on supportive care for HNC patients. A 12-month nurse-led psychosocial intervention, relative to usual care, improved multiple QOL domains with persistent benefits out to 24 months for emotional functioning and fatigue [95]. Low-level laser application during radiotherapy showed short-term QOL improvements, likely related to reduced mucositis [96–98]. Placement of a prophylactic enteral feeding tube prior to (chemo)RT improved 6-month QOL scores on multiple domains, relative to usual care [99]. Two RCTs of pilocarpine showed no QOL benefit over placebo in patients with post-RT xerostomia [24, 100], and three others showed no QOL benefit from a lozenge intended to reduce mucositis [101], from a cream intended to reduce dermatitis [102], or from subcutaneous GM-CSF during radiotherapy [103]. More such studies are needed and, indeed, anticipated.

Cancer rehabilitation is attracting increasing interest and attention. Three small randomized trials have assessed the QOL effects of exercise interventions in HNC patients. In two studies of exercise during (chemo)RT, QOL declined less in patients randomized to resistance exercise versus usual care [104] and improved with a personalized exercise program (versus a decline with usual care) [105]. HNC patient groups randomized to 12 weeks of progressive resistance training either 2 or 5 months post-RT both had physical benefit, but QOL improved more with the earlier intervention [106]. Larger studies are needed to demonstrate the feasibility and acceptability of such programs to HNC patients in general.

49.7.2 Prognostic Applications

In cancer generally, baseline QOL is among the strongest available prognostic factors. A confounding variable is defined as a covariate which is associated with both the predictor and the outcome; for baseline QOL and survival, there are many potential confounders. Less baseline comorbidity, lack of ongoing tobacco and alcohol use, higher socioeconomic status and education levels, better social supports, a more optimistic outlook, and less extensive disease have all been associated with both higher QOL and improved survival. Nonetheless, several studies have shown the independent value of baseline QOL in multivariable analyses. Figure 49.1 shows the overall survival by baseline global QOL on the EORTC QLQ-C30 in a study of RT versus RT and cetuximab; only Karnofsky performance status was a stronger predictor of survival [87].

Fig. 49.1
figure 1

Kaplan-Meier survival curves stratified by baseline EORTC QLQ-C30 global health status/quality of life scores [Reprinted from Curran D, Giralt J, Harari P, et al. Quality of life in head and neck cancer patients after treatment with high-dose radiotherapy alone or in combination with cetuximab. J Clin Oncol. 2007;25(16):2191–2197. © 2008 American Society of Clinical Oncology. All rights reserved]

Full size image

A retrospective analysis of prospectively collected QOL data categorized HNC patients as short- (<1 year), intermediate- (1–3 years), or long- (>3 years) term survivors and found significant differences in QOL at all time points, including baseline [107]. The RTOG has published a combined analysis of two HNC randomized trials using FACT-H&N, which showed baseline QOL to be predictive of locoregional control but not overall survival [108]. Finally, a recently completed study of concurrent chemoRT with or without the hypoxic cell sensitizer tirapazamine has shown QOL to be a strong predictor of overall survival, even independently of p16 status in oropharyngeal cancer patients [109]. These results suggest that baseline QOL may be useful as a future tool to assist in selecting patients for differing treatment intensities or for additional supportive care measures, but such a strategy has not yet been explored.

49.7.3 Routine Clinical Application

The use of QOL instruments in clinical trial protocols has become widely accepted; however, their use in general clinical practice is just beginning. An overview of RCTs allocating patients or physicians to use versus not to use QOL data in routine practice showed mixed results, with some studies showing benefits in patient satisfaction or process of care but others failing to show such benefits [110]. Two oncology RCTs have suggested positive effects: Velikova et al. found more frequent discussion of symptoms and improved emotional well-being in patients for whom QOL data was provided to the physician before a visit compared to those for whom it was not [111]. Detmar et al. found more frequent discussion of QOL issues by physicians who had been provided QOL data for patients receiving palliative chemotherapy [112]. Although neither of these studies demonstrated a change in patient management, a recent focus on patient-centered care in many jurisdictions has led to some early initiatives. For example, the province of Ontario (Canada) has incorporated patient-reported outcomes (the Edmonton Symptom Assessment System) throughout all cancer centers, while its largest center (the Princess Margaret) has added patient-reported screening for depression, anxiety, and social difficulties, as well as a head and neck cancer-specific instrument (MDASI-HN), at every HNC patient visit [113]. A recent review identified 33 systems in use to track such data [114]. No published studies evaluating the impact of routine QOL measurement for HNC patients have yet been identified.

49.7.4 Special Challenges (Compliance, Education, and Communication)

The incident population of HNC patients is undergoing a period of rapid change. The traditional risk factors of smoking and alcohol use translated into a patient population with lower than average socioeconomic status and educational levels [115]. For QOL measurement, this led to special considerations, including the need for instruments that were short, easy to read, and not excessively intellectually complex. In many studies, compliance with questionnaire completion was low; moreover, lack of social support, alcohol abuse, and lower levels of education have all been shown to correlate with lower QOL [116]. More recently, however, 66 % of oropharynx cancer patients in Toronto, Canada, were shown to have HPV/p16-associated cancers; such patients often lack the risk factors of smoking or alcohol use and tend to be younger with higher socioeconomic status and education levels [117]. HPV-associated cancer has an improved prognosis and a unique QOL trajectory [109]. Results from clinical trials designed to test less toxic treatment approaches (such as the use of EGFR inhibitors or trans-oral resection) in such patients, employing QOL and specific symptoms such as dysphagia as central outcomes, are anticipated.

49.8 A Glimpse into the Future

Computer-adaptive tests (CATs) use technology to deliver questionnaires in a logical manner and can significantly reduce respondent burden by producing high reliability and validity with far fewer questions. The approach combines the capability of computers to adapt using if/then algorithms, with the application of item response theory (IRT) to individual questions. IRT is a statistical method which uses mathematical modeling to characterize the ability of each individual item to discriminate differences depending on the level of a patient’s problem. Together, this type of system allows the computer to present questions which are most likely to produce a reliable and valid characterization of the underlying trait of interest. For example, a CAT test might begin with an item such as “do you have pain?”, which does not make any assumptions about pain level. However, a respondent who answers “yes” would receive follow-up questions regarding pain severity, whereas one who answers “no” might have confirmatory question such as, “does discomfort interfere with your ability to participate in sports?”.

49.8.1 PROMIS and CaPS

The Patient-Reported Outcomes Measurement Information System (PROMIS) is a large-scale project sponsored by the National Institute of Health (NIH) in the USA [118]. The goal of PROMIS is to develop a comprehensive bank of items with known IRT characteristics, drawn from existing PRO instruments. These items may then be used for CAT or combined in new ways to create fixed-length PROs for specific purposes [49]. Significant progress has been made for the general health bank, and a prototype online CAT administration tool now exists [119]. Validation of the PROMIS instruments is ongoing. The Cancer PROMIS Supplement (CaPS) has been funded to insure that PROMIS adequately meets the needs for PRO measurement in cancer patients; currently available PROMIS-Cancer instruments exist to measure fatigue, pain interference, and physical function.

49.9 Summary

Improving the survival outcomes of HNC patients remains the primary goal of most advances in therapy. However, the importance of QOL to patients cannot be overemphasized. Both tumor characteristics and treatment strategy influence QOL in the short and long term. The measurement science of QOL and other PRO tools is well developed, and these instruments have been increasingly incorporated into clinical trials. Evidence from prospective, phase III trials is emerging. Future questions include the potential value of using QOL questionnaires in routine clinical care, the best strategies for translating QOL knowledge to clinicians, and the role of CAT administration of PROs.