When Is it Justified to Treat Symptoms? Measuring Symptom Burden

ABSTRACT: Symptom burden is certainly not a new concept in the literature on disease and treatment, but recent developments in our understanding of how to measure symptoms and their impact make it possible to cast symptom burden as a reasonable summary measure of both disease- and treatment-outcome status. We discuss the use of symptom burden as an alternative to quality-of-life measures or as a supplement to these measures. We examine the potential responsiveness and relative ease of interpretation of a symptom-burden outcome. We review studies of the prevalence and negative impact of multiple symptoms on health outcomes among cancer patients. By way of example, we illustrate the potential utility of a measure of symptom burden in the epidemiologic studies of multiple symptoms as well as an outcome measure in clinical studies. Finally, we discuss potential research challenges and offer recommendations on future research directions. [ONCOLOGY 16(Suppl 10):64-70, 2002]

It is now common forcliniciansand patients to face choicesamong treatments that are highly similar in efficacy and even in prolongingsurvival. As a result, differences in treatment toxicity and in the patient’sstatus during the survival period have become critical variables in makingtreatment choices and in developing new therapies. Paramount in evaluating bothtoxicity and quality of survival time is the inclusion of patient-reportedoutcomes as important measures of treatment differences. These outcomes can takea variety of forms, including measures of the differences in symptom severity,perception of daily functioning, feelings of well-being, global impressions ofthe impact of treatment on daily life, satisfaction with treatment, andhealth-related quality of life (QOL).

Because these patient-outcome variables are becoming moreimportant relative to "hard" end points such as survival or time torecurrence, decisions about which patient-outcome measures to use and how tointerpret them for patients, health-care professionals, and policy makers willneed to be made. Almost all of these outcome measures will ultimately depend onthe patient’s perception of his or her status, before and after becoming illand receiving treatment. The patient’s representation of this change forms thebasis for making practical judgments about how treatment should progress.

As treatments that depend on such subjective outcomes emerge,how such treatments will be "valued" becomes a major issue. Willpatients, given subjective outcome information about the treatments, opt topursue them? Will physicians use this outcome information to recommend andprescribe treatments? When such treatments are being developed, will approvalagencies decide they are beneficial? Will agencies (public and private healthfunders) decide that such treatments are worth paying for?

Thus, when a treatment offers little or no survivaladvantage, yet seems to improve how patients feel, judgments about the worth ofthe treatment will be made by several audiences who will ultimately determinewhether the treatment is effective, should be approved, and will be used. Herewe discuss the concept of symptom burden as a way to portray patient status, andoffer a definition of symptom burden as an outcome measure.

Symptom Burden:An Operational Definition

We can trace the origin of the word "symptom" fromthe Greek "symptoma," which means "anything that has befallenone."[1] A more useful definition, however, is provided by Webster,[2] andstates that symptom "is the subjective evidence of disease or physicaldisturbance observed by a patient." Implicit in this definition is thesubjective and negative nature of symptoms. Symptoms are an observation by thepatient; ie, the person experiencing the evidence of disease or physicaldisturbance. Symptoms can be the subjective expression of the disease itself, orthe products of disease treatment, the latter often referred to as side effectsor toxicities. Symptom burden can be thought of as the subjective counterpart ofsummary expressions of disease such as tumor burden.

Patients typically experience multiple concurrent symptoms,due either to the disease or its treatment. We propose that a measure of symptomburden be a summative indicator of

• The severityof the symptoms that are most associated with a disease or treatment

• A summary ofthe patient’s perception of the impact of these symptoms on daily living,including activity, mood, ability to work, and ability to relate to others.

Symptom Burden and Quality of Life

Most clinical trials currently include one or more measuresof quality of life. In 1996, the American Society of Clinical Oncology statedthat quality of life was an end point secondary only to survival.[3] The US Foodand Drug Administration (FDA) recommended that quality of life is a moreimportant measure of treatment efficacy than most other traditional end pointsfor drugs that do not have an impact on survival.[4]

Clearly, improving a patient’s quality of life is a goal ofall medical treatment. Based on the World Health Organization definition ofhealth,[5] quality of life includes psychological and social functioning as wellas physical functioning, and incorporates positive aspects of well-being as wellas negative aspects of disease and infirmity. Among health researchers, a commonconsensus is that quality of life is a multidimensional construct composed of atleast four dimensions:

• Physicalfunction (ie, daily activities, self-care, etc)

• Psychologicalfunction (ie, emotional/mental status, mood)

• Socialfunction (ie, social interactions, family dynamics), and

• Disease/treatmentsymptoms (ie, pain, nausea, fatigue).

The comprehensive nature of quality of life, however, is bothone of its attractions and its difficulties as an outcome measure. Quality oflife is best viewed as a subjective evaluation of life as a whole.[6] In thissection, we present the potential advantages of symptom-burden measure,highlighting its potential utility as a supplement to QOL measures.

Clinical Conditionand Quality of Life

QOL measures may be relatively insensitive to obvious changesin clinical condition. As a multidimensional construct, a measure of quality oflife may include aspects that go beyond the impact of cancer or its treatment.Among bone marrow transplant patients, studies have shown patients rating theirquality of life as above average despite the persistence of significant physicaland psychological symptoms. For example, Bush and colleagues[7] conducted adescriptive study of quality of life, psychological distress, demands oflong-term recovery, and health perceptions of 125 bone marrow transplantsurvivors. They were no different from individuals sampled from the generalpopulation with regard to their perceived current health and health outlook.However, 10 or more years posttransplantation, long-term survivors continued toexperience a moderate incidence of lingering complications and demands,including emotional and sexual dysfunction, fatigue, eye problems, sleepdisturbance, general pain, and cognitive dysfunction.

Molassiotis et al[8] found that long-term autologous bonemarrow transplant survivors rated their quality of life as good to excellent,although 20% reported symptoms of anxiety, 10% had signs of clinical depression,and 20% had not returned to full-time employment. McQuellon et al[9] found thatquality of life and mood improved slightly in patients following autologous bonemarrow transplant, despite the fact that 30% had problems with sexuality,fatigue, and depressive symptoms.

These studies show that patients may report their quality oflife as improved or unchanged, despite changes or deterioration in health. QOLmeasures in such situations are liable to yield discrepant results compared withthe proposed symptom- burden measure. Further, results indicating that patients’quality of life did not deteriorate may also imply that intervention need not beundertaken, possibly masking the issue of specific residual symptoms thatrequire intervention.

Symptom-Burden Measures

Symptom burden is correlated with quality of life, and manyQOL measures (eg, European Organization for Research and Treatment of Cancer [EORTC],Functional Assessment of Cancer Therapy [FACT]) have components that measuresymptoms. Symptom severity is a strong predictor of scores on QOL measures[10]suggesting the potential of symptom- burden measures as an immediate indicatorof deteriorating quality of life. For example, Bull and colleagues[11] studiedquality of life among women with recurrent breast cancer. Results showed thatself-reported physical symptoms were a strong predictor of postrecurrenceratings of overall quality of life. In a study of 206 adult patients withmultiple myeloma, Poulos and colleagues[12] looked at the relationship betweenpain and mood disturbance, and the factors that influence quality of life. Theyfound that pain and mood disturbance scores were significant predictors ofquality of life in this group of patients.

Symptom-burden measures may be more informative about truedifferences in patient status when comparing treatments. A randomized clinicaltrial compared the quality of life of patients treated with single-agentpaclitaxel vs doxorubicin as first-line therapy for advanced breast cancer.[13]Quality of life was measured using the EORTC Quality of Life Questionnaire (QLQ)-C30and the Rotterdam Symptom Checklist. Results showed that no statisticallysignificant differences were observed in either the functional scales or theglobal QOL scores during the third cycle of chemotherapy. However,cross-sectional analyses of data at the end of cycle 3 showed that patients inthe doxorubicin arm had significantly more nausea/vomiting and loss of appetitethan those in the paclitaxel arm. The authors concluded that there is a trendtoward greater burden of disease and treatment and lower general health statusassociated with doxorubicin therapy.

When using QOL measures, components assessing symptom burdenare more responsive indicators of changes in patient status. For example, a 1985randomized phase III study compared patients with low-grade cerebral glioma whoreceived high-dose (59.4 Gy in 6.5 weeks) vs low-dose (45 Gy in 5 weeks)radiotherapy with conventional techniques.[14] A QOL questionnaire, consistingof 47 items that assessed a range of physical, social, psychological, andsymptom domains, was included in the trial to measure the impact of treatmentover time. No difference in survival was observed between the two treatmentstrategies.

The authors also report that, in general, the results suggestno major differences in quality of life between the two treatment arms. However,patients who received high-dose radiotherapy reported significantly morefatigue/malaise and insomnia immediately after radiotherapy, and pooreremotional functioning at 7 to 15 months postrandomization. For the remaining QOLdomains, no statistically significant differences between the two treatment armswere found. The authors concluded that conventional radiation therapy forlow-grade cerebral glioma using an administration schedule of 45 Gy over 5 weeksis at least as good as prolonged treatment in terms of clinical efficacy,survival, and quality of life.

‘Response Shift’

Why might more general QOL measures fail to reflectincreasing or persistent symptoms and associated distress? This paradox has beenaddressed by several investigators. For example, it has been proposed that suchfindings may be due to the phenomenon of "response shift."[15]Response shift is described as occurring because of changes in patients’internal standards and conceptualization of quality of life over the course ofdisease, which is an inherent process of adjusting to the illness. Severalattempts have been made to suggest ways of correcting for this response shift.The studies cited above, however, suggest that simple symptom measures may beless sensitive to this attempt to redefine standards by which patients judgetheir overall quality of life.

Symptom burden may have advantages compared with QOL measureswhen interpreting potential outcomes for patients. In clinical practice, it maybe difficult to make sense of improvement in QOL scores. For example, thefinding that a treatment is associated with a 10-point improvement in quality oflife as opposed to a 15-point improvement is not intuitive to either clinicianor patient. Because of this, interpreting findings presented as improvements inquality of life for patients brings with it challenges when discussing treatmentoptions. In contrast, both patient and clinician should have an easier decisionto make when given the information that, although two treatments produce asimilar survival benefit, one is associated with significantly less pain,fatigue, and sleep disturbance.

Prevalence and Adverse Impact of Multiple Symptoms

Studies have assessed the prevalence of symptoms at differentstages of the disease. Among cancer patients, symptoms may be the result of thedisease itself or side effects from cancer therapies. For example, even whentheir disease is in remission or cured, cancer survivors continue to experiencephysical, affective, or cognitive symptoms. These symptoms may be due to thephysiologic changes associated with prior treatments, delayed side effects oftreatment, or the long-term consequences of the disease.

Most symptoms and side effects (pain, vomiting, fatigue) havebeen studied in isolation, with the possible crude ratings of"toxicities" associated with new therapies. The study of theprevalence and severity of sets of symptoms (sometimes referred to as symptomclusters[16]) is relatively new, even though the co-occurrence of multiplesymptoms is obvious to both cancer patients and clinicians. Understanding theimpact of symptoms acting together is critical for the construct of symptomburden.

Portenoy et al assessed multiple symptoms in a random sampleof inpatients and outpatients with cancer.[17] The most frequently reportedsymptoms were lack of energy, worry, sadness, and pain. In a prospective studyof 1,000 patients with advanced cancer, Donnelly et al[18] found that pain,fatigue, and anorexia were among the 10 most prevalent symptoms in all 17primary sites studied. When pain, anorexia, weakness, anxiety, lack of energy,easy fatigue, early satiety, constipation, and dyspnea were present, a majorityof patients rated them as moderate or severe.

A second prospective study evaluated 1,635 cancer patientsreferred to a pain clinic[19]. In addition to pain, patients reported an averageof more than three additional symptoms. The most common were insomnia, anorexia,and constipation. One in five patients also reported sweating, nausea andvomiting, dyspnea, dysphagia, and neuropsychiatric symptoms. Similarly, aprospective study of cancer patients in palliative care centers in Europe,Australia, and the United States found that pain (57%) and weakness (51%) werethe most frequently reported symptoms.[20] Weight loss, anorexia, constipation,nausea, and dyspnea were also common. These studies suggest the high prevalenceof symptom burden among patients with a given disease or treatment.

Common symptoms of cancer and cancer treatment have also beenfound to adversely affect patients’ daily functioning, and have become a majorhealth problem in their own right. The adverse impact of pain, for example, onhealth outcomes such as function,[21-23] affective status,[24-26] and quality oflife[27,28] among patient populations is well documented.

Symptoms that are unrecognized by treatment teams may alsobecome so severe that emergency room visits or hospitalization are required formanagement, adding substantially to the cost of treatment and to disruption ofthe patients’ routines and those of their families. Untreated symptoms mayalso negatively influence treatment efficacy by interrupting therapy.[29]Multiple and severe symptoms present a significant challenge to the resources ofthose who care for and manage cancer patients. For example, intensive cancertherapies produce severe and sometimes life-threatening side effects, resultingin the inability of patients to care for themselves and in total dependence oncaregivers.[30]

Measuring Symptomsand Symptom Burden

Patient outcome measures, especially when they must berepeated several times, need to be easy to use and intuitive. Symptom scaleseasily meet these requirements. An ideal tool for measuring symptom burdenshould focus on the most prevalent and distressing symptoms for the group ofpatients who will use the scale. Long scales with exhaustive symptom lists tirepatients and reduce their interest in completing them. Symptom scales shouldalso assess the patient’s perspective of the collective impact of symptoms ondifferent activities.

Several symptom scales have been developed for use withcancer patients. Any of these scales, if the salient symptoms and the degree towhich these symptoms interfere with patient functioning are represented, couldserve as the basis for measuring symptom burden. Such scales include the SymptomDistress Scale,[31] the Memorial Symptom Assessment Scale,[32] and the RotterdamSymptom Checklist.[33]

The M. D. AndersonSymptom Inventory

We have developed the M. D. Anderson Symptom Inventory, abrief measure of the severity and impact of cancer-related symptoms.[34] It isbased on our previous efforts in assessing the severity and interference ofsingle symptoms, including the Brief Pain Inventory and the Brief FatigueInventory.[35] The "core" M. D. Anderson Symptom Inventory (seeappendix 1) is a 19-item scale consisting of 13 symptoms frequently reported bycancer patients: pain, fatigue, nausea, disturbed sleep, emotional distress,shortness of breath, lack of appetite, drowsiness, dry mouth, sadness, vomiting,remembering, and numbness or tingling. It also contains six items that describehow much the symptoms have interfered with different aspects of the patient’slife over the past 24 hours: general activity, mood, walking ability, normalwork (including both work outside the home and housework), relations with otherpeople, and enjoyment of life.

Both the symptom and interference items are rated on an11-point scale, with 0 indicating "does not interfere" and 10indicating "completely interferes." The validation process for the M.D. Anderson Symptom Inventory has been reported elsewhere.[34]

Interpreting SymptomScale Results

The use of any of the measurement scales described aboveshould facilitate and improve the symptom- assessment process, help directtreatment choices, and supply information about the effectiveness of treatment.Providing patients and clinicians with additional information to help theminterpret the relative severity of a score should be helpful. Based on ourexperience with pain, we have found that the relationship between symptomseverity and the degree of the symptom’s interference in functions such asactivity and mood may help determine the level of symptom severity, which may,in turn, direct treatment. In particular, we have found that pain intensityratings have a corresponding degree of interference with function. Mild,moderate, and severe pain differentially affect the function of cancerpatients[22]: A rating of 1-4 on a 0-10 scale corresponded to mild pain, 5-6to moderate pain, and 7-10 to severe pain. Cut-off points such as these arenow incorporated into several guidelines for the management of pain.

Recent studies also indicate that the severity ofcancer-related fatigue can be classified in the same fashion, suggesting that apatient’s report of fatigue severity might directly contribute totreatment recommendations.[35] A similar methodology might be applied to othercancer-related symptoms (ie, sleep disturbance, negative affect, cognitiveimpairment) to help determine levels of severity that may help guide treatmentdecisions.

Studies of Symptom Burden

Symptom burden is not a new concept in studies of disease andtreatment. Desbiens and colleagues[36] report the "symptom burden" andassociated factors in a cohort of 1,582 seriously ill patients admitted at fivetertiary care academic centers in the United States. They found that pain,dyspnea, anxiety, and depression caused the greatest symptom burden. Accountingfor 67.3% of all symptoms that were at least moderately severe at least half ofthe time were

• Dyspnea(19.2%)

• Pain (17.6%)

• Pain/dyspnea(7.6%)

• Anxiety (5.9%)

• Depression(5.0%)

• Anxiety/depression(4.4%)

• Anxiety/depression/pain/dyspnea(3.9%), and

• Nausea (3.7%).

Hospital, male gender, disease category, more comorbidities,more dependencies in activities of daily living prior to illness, and poorerquality of life were associated with greater symptom burden.

In their evaluation of the prevalence of and the QOL factorsassociated with self-reported depression in 987 patients with inoperable lungcancer, Hopwood and Stephens found that depression was self-rated in 322patients (33%) before treatment and persisted in more than 50% of patients.[37]Pretreatment physical symptom burden and clinician-rated physical symptoms wereindependent predictors. Physical symptom burden was measured as the number ofphysical items reported in the Rotterdam Symptom Checklist.

Clinical Benefit Response

Changes in the severity and impact of single symptoms, suchas pain, vomiting, and sleep disturbance, are often primary outcome measures forclinical trials targeted at reducing or preventing these symptoms. Recently,some trials have used combinations of symptoms and other outcomes in an attemptto approximate ecologically valid clinical outcomes. One such attempt is theclinical benefit response developed by Burris et al.[38] Clinical benefitresponse has been used in trials of agents in cancers with an extremely poorprognosis, such as pancreatic and lung cancer. In a study of gemcitabine(Gemzar) for pancreatic cancer, Burris et al[38] used clinical benefit responseas a primary outcome measure. Clinical benefit response was defined as adichotomous outcome, with a positive response being defined as (1) animprovement in one or more symptoms, (2) a reduction in analgesic usage, and (3)a specified improvement in performance status.

In order to be counted as a positive responder, improvementhad to be sustained for at least 4 weeks. A specified weight gain could be addedas a secondary measure.[38] In the first major trial using clinical benefitresponse, the measure was presented as a synopsis of expected palliative changesfrom previous trials. Clinical benefit response was used as a primary outcomemeasure even though the investigators acknowledged that no prospectiveevaluation (or validation) of the clinical benefit outcome had been undertaken.

Gemcitabine was approved by the FDA as well as in the UnitedKingdom, based primarily on a modest survival advantage. The clinical benefitresponse was not recognized as a validated outcome.[39]

Patients with lung cancer also receive only a modest survivalbenefit from cancer therapies, so palliation of symptoms becomes a majortherapeutic goal. A recent trial comparing gemcitabine and cisplatin/vindesinein advanced non-small-cell lung cancer illustrates an even moresymptom-focused outcome.[40] The primary outcome measure for this trial was alsoa combined clinical benefit measure, this time composed of (1) the mean ofweekly symptom ratings of six symptoms from the Lung Cancer Symptom Scale,[41](2) improved, stable, or worsening Karnofsky performance status, and (3) weight.The time to worsening of clinical status was defined as the time fromrandomization to the first evidence of worsening postbaseline.

Symptom-Specific Measures

The effects of a treatment may occasionally be reflected inboth global measures of quality of life as well as symptom-specific measures.For example, recent large phase IV community trials of erythropoietin in anemiccancer patients have shown that an increase in hemoglobin has been associatedwith improvements in ratings on measures of quality of life. These improvementswere noted on both linear analog scales[42,43] and the FACT-Anemia subscale(FACT-An, see Appendix).[43] One of the scales demonstrating improvement was alinear analog rating of energy, suggesting that patients were less fatigued.

A recent incremental analysis[44] of the second data setsuggested that there was a nonlinear relationship between increasing hemoglobinand quality of life. An incremental analysis is a regression technique thatattempts to identify maximum benefit (in this case improvement in quality oflife) as it relates to another variable (in this case hemoglobin). This analysisindicated that the relationship was nonlinear, with the maximum benefitoccurring when hemoglobin levels reached 12 g/dL.

The FACT-An contains several symptom-specific items, and itwould be interesting to identify other areas of symptom improvement that mayhave benefited from the correction of anemia. The literature on the treatment ofanemia in dialysis patients has documented the association between theresolution of anemia and the relief of several symptoms, including sleepdisturbance, depressed affect, and appetite[45] as well as improved daytimealertness[46] and cognitive test performance.[47,48] In the anemia of cancerstudies, supplementing QOL measures with multiple symptom scales could yieldinformation about specific symptom areas that might be expected to improve, andhow much symptomatic improvement might be expected. Such information would beuseful to patients, physicians, and those who make decisions about funding orapproving therapies.

Treating Symptom Burden: When Is it Justified?

The development of treatments that produce symptom relief ora QOL outcome as their primary focus generates the need to consider how thesetherapies might be deployed and accepted by the audiences who will be the targetof clinical data concerning these treatments. Symptom burden may provide aconceptual focus for decisions about treatment allocations, researchprioritization, and health policy.

The treatment of symptoms, such as fatigue, emotionaldistress, sleep disturbance, and cognitive slowing, presents a challenge fromboth assessment and interpretation perspectives. Just how much symptomimprovement is worth the effort, in terms of research, development of newagents, the approval process, establishing practice, and reimbursementstandards?

Over the last 20 years, changes in the treatment ofcancer-related pain have been significant. Although several issues still lead toundertreatment of pain, the treatment of cancer-related pain presents a modelfor how to think of the evolution and acceptance of the treatment of othersymptoms.

• First, theadverse impact of pain on patient comfort and function is now well established.Several studies have demonstrated that cancer pain is pervasive, especially whenpatients have metastatic disease, and that moderate-to-severe pain impairsfunction. Studies have also documented pain that was undertreated with availablemethods, and that oncologists recognized that it was undertreated. The positiveeffects of treating pain on patient comfort and function have now been soclearly demonstrated that few would deny that patients are entitled totreatment for cancer-related pain.

• Second, in thecase of pain, a variety of effective treatment strategies are available, manywith acceptable side-effect profiles.

• Third, severalconsensus-based and evidence-based guidelines that provide treatmentrecommendations for pain management have become available.

• Fourth,effective pain treatment is valued not only by patients and their families butalso by health-care professionals, policy makers, and the general public.

• Finally, inrecognition of the need to manage pain, accrediting bodies such as the JointCommission for the Accreditation of Hospital Organizations have developedstandards for both the assessment and management of pain.[49]

The approach to the treatment of other symptoms in cancer,such as fatigue, depression, and sleep disturbance is less systematicallydeveloped than the treatment of pain. For further development in this area,researchers must first establish that a single symptom or combined set ofsymptoms is prevalent and severe enough to cause distress as well as loss offunction to significant numbers of patients. Second, evidence that supports theeffectiveness of treatment for these symptoms must be demonstrated. Third,effective symptomatic treatments, since they can be viewed as elective, need tominimize negative side effects. For example, a treatment for fatigue that causessignificant sleep disturbance would be unacceptable. Guidelines for both theassessment of these symptoms (when to treat) and treatment (how to treat) needto be in place, and established as standards. Finally, data describing the costand benefit of symptom reduction need to be available for public debate and theconsideration of approval and funding agencies; how much is an improvement(small, medium, large) in symptom burden worth?

Future Development of Symptom Burden as an Outcome Measure

The acceptance of symptom burden as a patient outcome measurefor clinical trials, clinical treatment, and policy decisions will requirefurther conceptual elaboration and clinical research. The evolution of researchin cancer-related pain suggests how symptom-burden research might evolve. Forexample, as seen above, studies that describe the prevalence and severity ofmultiple symptoms among cancer patients are relatively new. These studies havedepended on the development of scales that can reliably capture differences (inquality, severity, and impact) between cancer-related symptoms and symptomsexperienced by others who do not have cancer (eg, Mendoza et al[35]).

In terms of symptom burden, the development of methods forits measurement will give us the tools necessary to conduct the descriptive andepidemiologic studies that define who will experience severe symptom burden,what symptoms contribute most to symptom distress, and what factors correlatewith its development. This should lead to a subset of symptoms (such as pain,cognitive problems, fatigue, depression, and sleep disturbance) that account forthe bulk of symptom burden and become the target of the most aggressivetherapeutic development and testing. An additional need is to determine thechronicity of both disease and treatment-related symptoms. As with pain, we needto determine the prevalence, severity, and impact of symptom burden throughoutthe course of disease: at diagnosis, during treatment, when cancer is inremission or cured, and, for some patients, near the end of life. More chronicsymptoms will obviously have greater weight in overall symptom burden.

Now that we have scales that successfully capture multiplesymptoms and the distress they cause, we are very much in need of longitudinalstudies of symptom clusters or patterns in large numbers of patients. We need tobe able to specify risk factors (disease, stage, type of treatment) that areassociated with increased symptom burden, as well as "host" factors(such as gender, age, ethnicity) that may influence symptom-burden status. Weneed to continue to look at potential causes, both through clinical studies thatexamine correlates of symptom burden and those that examine what might improvesymptoms.

Data on potential biologic correlates of the mechanismsunderlying symptoms (such as hemoglobin, albumin, and the inflammatory process)might be collected at the same intervals as symptom-burden measures. These datacould be a rich source of hypotheses about potential new therapeutic approachesto symptom management. It would be helpful to standardize symptom-burdenmeasures, cognitive and affective assessments, and measures of potentiallycorrelated biological variables across institutions in order to develop largedatabases that would best yield common patterns, enabling the identification oftherapies that benefit more than one symptom at the same time.

With further development, symptom burden has many potentialbenefits as a patient-outcome summary measure. Symptom burden should beresponsive to appropriate treatment. It should be sensitive to differences intreatment toxicities. It should be easy for patients, health-care professionals,and policy makers to understand and use as the basis for treatment choices andpriorities. Finally, if symptom-burden data are collected along with biologicmarkers of potential mechanisms that may cause symptoms, measures of symptomburden have the potential for identifying new treatment strategies.

References:

1. Haubrich WS: Medical Meanings: A Glossary of WordOrigins. Philadelphia, American College of Physicians,1997.

2. Webster’s Third International Dictionary.Springfield, Massachusetts, Merriam and Company, 1966.

3. Outcomes of cancer treatment for technology assessment andcancer treatment guidelines. J Clin Oncol 14:671-679, 1996.

4. Bietz J, Gnecco C, Justice R: Quality of life endpoints incancer clinical trials: The US Food and Drug Administration perspective. JNatl Cancer Inst Monogr 20:7-9, 1996.

5. World Health Organization: Constitution of the WorldHealth Organization, in Handbook of Basic Documents. Geneva, 1948.

6. De Haes JCJM: Quality of life: Conceptual and theoreticalconsiderations, in Watson M, Greer S, Thomas C (eds): Psychosocial Oncology,pp 61-70. Oxford, Pergamon Press, 1988.

7. Bush NE, Haberman M, Donaldson G, et al: Quality of lifeof 125 adults surviving 6-18 years after bone marrow transplantation. Soc SciMed 40(4):479-490, 1995.

8. Molassiotis A, van den Akker OB, Milligan DW, et al:Quality of life in long-term survivors of marrow transplantation: Comparisonwith a matched group receiving maintenance chemotherapy. Bone MarrowTransplant 17(2):249-258, 1996.

9. McQuellon RP, Craven B, Russell GB, et al: Quality of lifein breast cancer patients before and after autologous bone marrowtransplantation. Bone Marrow Transplant 18(3):579-584, 1996.

10. Chan YM, Ngan HY, Li BY, et al: A longitudinal study onquality of life after gynecologic cancer treatment. Gynecol Oncol83(1):10-19, 2001.

11. Bull AA, Meyerowitz BE, Hart S, et al: Quality of life inwomen with recurrent breast cancer. Breast Cancer Res Treat 54:47-57,1999.

12. Poulos AR, Gertz MA, Pankratz VS, et al: Pain, mooddisturbance, and quality of life in patients with multiple myeloma. OncolNurs Forum 28(7):1163-1171, 2001.

13. Kramer JA, Curran D, Piccart M, et al: Randomised trialof paclitaxel vs doxorubicin as first-line chemotherapy for advanced breastcancer: Quality of life evaluation using the EORTC QLQ-C30 and the RotterdamSymptom Checklist. Eur J Cancer 36:1488-1497, 2000.

14. Kiebert GM, Curran D, Aaronson NK, et al: Quality of lifeafter radiation therapy of cerebral low-grade gliomas of the adult: Results of arandomised phase III trial on dose response (EORTC trial 22844). EORTCRadiotherapy Co-operative Group. Eur J Cancer 34(12):1902-1909, 1998.

15. Sprangers MA, Moinpour CM, Moynihan TJ, et al: Assessingmeaningful change in quality of life over time: A users’ guide for clinicians.Mayo Clin Proc 77(6):561-571, 2002.

16. Dodd MJ, Miaskowski C, Paul SM: Symptom clusters andtheir effect on the functional status of patients with cancer. Oncol NursForum 28(3):465-470, 2001.

17. Portenoy RK, Thaler HT, Kornblith AB, et al: Symptomprevalence, characteristics and distress in a cancer population. Qual LifeRes 3(3):183-189, 1994a.

18. Donnelly S, Walsh D, Rybicki L: The symptoms of advancedcancer: Identification of clinical and research priorities by assessment ofprevalence and severity. J Palliat Care 11(1):27-32, 1995.

19. Grond S, Zech D, Diefenbach C, et al: Prevalence andpattern of symptoms in patients with cancer pain: A prospective evaluation of1635 cancer patients referred to a pain clinic. J Pain Symptom Manage9(6):372-382, 1994.

20. Vainio A, Auvinen A: Prevalence of symptoms amongpatients with advanced cancer: An international collaborative study. SymptomPrevalence Group. J Pain Symptom Manage 12(1):3-10, 1996.

21. Cleeland CS, Gonin R, Hatfield AK, et al: Pain and itstreatment in outpatients with metastatic cancer. N Engl J Med330(9):592-596, 1994.

22. Serlin RC, Mendoza TR, Nakamura Y, et al: "When iscancer pain mild, moderate or severe? Grading pain severity by its interferencewith function." Pain 61(2):277-284, 1995.

23. Given CW, Given B, Azzouz F, et al: Comparison of changesin physical functioning of elderly patients with new diagnoses of cancer. MedCare 38(5):482-493, 2000.

24. Given CW, Given BA, Stommel M: The impact of age,treatment, and symptoms on the physical and mental health of cancer patients. Alongitudinal perspective. Cancer 74(7 suppl):2128-2138, 1994.

25. Turk DC, Okifuji A, Scharff L: Chronic pain anddepression: Role of perceived impact and perceived control in different agecohorts. Pain 6:93-101, 1995.

26. Rudy TE, Kerns RD, Turk DC: Chronic pain and depression:Toward a cognitive-behavioral mediation model. Pain 35:129-140, 1988.

27. Ahmedzai S: Recent clinical trials of pain control:Impact on quality of life. Eur J Cancer 31A(suppl 6):S2-7, 1995.

28. Wang XS, Cleeland CS, Mendoza TR et al: The effects ofpain severity on health-related quality of life: A study of Chinese cancerpatients. Cancer 86(9):1848-1855, 1999.

29. Borden EC, Parkinson D: A perspective on the clinicaleffectiveness and tolerance of interferon-alpha. Semin Oncol 25(suppl1):3-8, 1998.

30. Sarna L, McCorkle R: Burden of care and lung cancer. CancerPract 4:245-251, 1996.

31. McCorkle R, Young K: Development of a symptom distressscale. Cancer Nurs 1:373-378, 1978.

32. Portenoy RK, Thaler HT, Kornblith AB, et al: The MemorialSymptom Assessment Scale: An instrument for the evaluation of symptomprevalence, characteristics, and distress. Eur J Cancer 30A:1326-1336,1994b.

33. de Haes JC, van Knippenberg FC, Neijt JP: Measuringpsychological and physical distress in cancer patients: Structure andapplication of the Rotterdam Symptom Checklist. Br J Cancer 62:1034-1038,1990.

34. Cleeland CS, Mendoza TR, Wang XS, et al: Assessingsymptom distress in cancer patients: The M.D. Anderson Symptom Inventory. Cancer89(7):1634-1646, 2000.

35. Mendoza TR, Wang XS, Cleeland CS, et al: The rapidassessment of fatigue severity in cancer patients: Use of the Brief FatigueInventory. Cancer 85:1186-1196, 1999.

36. Desbiens NA, Mueller-Rizner N, Connors AF Jr, et al: Thesymptom burden of seriously ill hospitalized patients. SUPPORT Investigators.Study to Understand Prognoses and Preferences for Outcome and Risks ofTreatment. J Pain Symptom Manage 17(4):248-255, 1999.

37. Hopwood P, Stephens RJ: Depression in patients with lungcancer: Prevalence and risk factors derived from quality-of-life data. J ClinOncol 18(4):893-903, 2000.

38. Burris HA 3rd, Moore MJ, Andersen J, et al: Improvementsin survival and clinical benefit with gemcitabine as first-line therapy forpatients with advanced pancreas cancer: A randomized trial. J Clin Oncol15:2403-413, 1997.

39. Steer CB, Marx GM, Harper PG: Is there quality inclinical benefit? Ann Oncol 12(9):1191-1193, 2001.

40. Vansteenkiste JF, Vandebroek JE, Nackaerts KL, et al:Clinical-benefit response in advanced non-small-cell lung cancer: A multicentreprospective randomised phase III study of single agent gemcitabine versuscisplatin-vindesine. Ann Oncol 12(9):1221-1230, 2001.

41. Hollen PJ, Gralla RJ, Kris MG, et al: Quality of lifeassessment in individuals with lung cancer: Testing the Lung Cancer SymptomScale (LCSS). Eur J Cancer 29A(suppl 1):S51-8, 1993.

42. Glaspy J, Bukowski R, Steinberg D, et al: Impact oftherapy with epoetin alfa on clinical outcomes in patients with nonmyeloidmalignancies during cancer chemotherapy in community oncology practice. ProcritStudy Group. J Clin Oncol 15(3):1218-1234, 1997.

43. Demetri GD, Kris M, Wade J, et al for the Procrit StudyGroup. Quality-of-life benefit in chemotherapy patients treated with epoetinalfa is independent of disease response or tumor type: Results from aprospective community oncology study. J Clin Oncol 16:3412-3425, 1998.

44. Crawford et al. Manuscript Submitted.

45. Evans RW, Rader B, Manninen DL, and the CooperativeMulticenter EPO Clinical Trial Group: The quality of life of hemodialysisrecipients treated with recombinant human erythropoietin. JAMA 263:825-830,1990.

46. Benz RL, Pressman MR, Hovick ET, et al: A preliminarystudy of the effects of correction of anemia with recombinant humanerythropoietin therapy on sleep, sleep disorders, and daytime sleepiness inhemodialysis patients (The SLEEPO study). Am J Kidney Dis34(6):1089-1095, 1999 .

47. Grimm G, Stockenhuber F, Schneeweiss B, et al:Improvement of brain function in haemodialysis patients treated witherythropoietin. Kidney Int 38:1-7, 1990.

48. Temple RM, Langan SJ, Deary IJ, et al: Recombinanterythropoietin improves cognitive function in chronic haemodialysis patients. NephrolDial Transplant 7(3):240-245, 1992.

49. Available at:http://www.jcaho.org/standard/stds2001_mpfrm.html. Joint Commission onAccreditation of Healthcare Organizations Pain Standards for 2001.