Quality-of-Life Assessment in the Symptom Management Trials of the National Cancer Institute-Supported Community Clinical Oncology Program
http://www.100md.com
《临床肿瘤学》
the Community Oncology and Prevention Trials Research Group, Division of Cancer Prevention, National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Bethesda, MD
ABSTRACT
METHODS: All QOL research objectives, rationales, assessment instruments, symptoms treated, and types of interventions from the CCOP symptom management portfolio of clinical trials were extracted and analyzed.
RESULTS: QOL assessments were proposed in 68 (52%) of the 130 total CCOP symptom management trials initiated since 1987. A total of 22 global QOL instruments were identified. Both the frequency of symptom management trials and the frequency of QOL assessment have increased significantly over time. The Functional Assessment of Cancer Therapy and Uniscale instruments were the most widely used QOL instruments, included in 55% of trials assessing QOL. The conceptual framework for QOL inclusion was limited to univariate relationships between symptom relief and global improvements in QOL. No consistent associations were found between QOL assessment and either the symptoms targeted or types of interventions.
CONCLUSION: To advance the state of the science, research protocols need to provide more explicit rationales for assessing QOL in symptom management trials and for the selection of the QOL instrument(s) to be used. Conceptual frameworks that specify the hypothesized links between the specific symptom(s) being managed, interactions with other symptoms, different domains of QOL, and global QOL also need to be more precisely described. Methodologic and conceptual advances in QOL symptom management trials are critical to fulfill the promise of alleviating suffering and improving the QOL of cancer patients.
INTRODUCTION
Although QOL assessment in cancer treatment trials has received considerable attention,2-9 its utility in symptom management trials has yet to be determined. In cancer treatment trials, survival is the most important outcome,1 and QOL assessment has demonstrated clear research value as a secondary outcome measure that can discriminate between treatments with equivalent survival rates.10 In symptom management trials, the primary end point is the relief of targeted symptoms caused by cancer and its treatment. Unlike treatment trials, QOL assessment in symptom management trials is not anchored to objective indicators but only to subjective, patient-reported symptoms.
Symptom management research has grown in significance in recent years. As cancer patients live longer, researchers have documented that many survivors experience significant negative effects from the cancer and/or treatment on the quality of their daily lives, long after the completion of therapy.11-14 In 2001, the Institute of Medicine reported that at least half of the 550,000 people who die from cancer each year suffer from a spectrum of symptoms including pain, labored breathing, distress, nausea, confusion, and other conditions that seriously diminish the quality of their survivorship.13 With the rapid growth of symptom management trials, it is timely and important to examine the contribution of QOL assessment to this line of research.
The purpose of the research described in this report was to investigate how researchers prospectively conceptualized, defined, and measured QOL in the symptom management clinical trials conducted by the Community Clinical Oncology Program (CCOP). The CCOP has been conducting symptom management trials with QOL end points since 1987. The CCOP is a distinct clinical trials network supported by the National Cancer Institute (NCI). The CCOP funds cooperative groups and select cancer centers as research bases to design and conduct clinical trials in cancer control and prevention, and it funds community physicians to accrue patients to these trials. Currently, eight cooperative groups and six cancer centers are funded as CCOP research bases (Table 1). We examined the CCOP experience to assess progress to date and to identify future directions to advance the state of the science of QOL assessment in symptom management trials.
METHODS
Data Extraction
The analysis focused on the symptom management trials and excluded all other types of cancer control and prevention research. To identify the symptom management research protocols, all CCOP cancer control clinical trials (N = 241) were independently categorized by two of the authors (A.M.O. and L.M.M.) as symptom management, prevention, screening and early detection, or other trials, and the few discrepancies were resolved through consensus discussions.
the original research protocols, the research objectives, assessment instruments, study rationale, and plan of analysis were extracted and analyzed to determine how the investigators prospectively conceptualized the purpose of QOL assessment and the relationship between symptom management and QOL. In addition, data on the symptom treated, type of intervention, research design, sample size, and cancer type(s) and stages were also retrieved.
Coding
All research instruments were reviewed by the entire research team for coding and classification. After an initial review, the instruments were grouped into the following three broad categories: global QOL instruments, symptom instruments, and a miscellaneous other category (eg, alcohol and other drug use, social support, and so on). On the basis of the definition by Goodwin et al,10 the criterion used for classifying questionnaires as global QOL instruments was "self-report of the impact of cancer and its treatment on some aspect of function (eg, physical, role, emotional, or social)." All instruments were coded independently by the research team, and disagreements regarding classifications were resolved through consensus discussions. To determine whether there were any changes over time, the studies were grouped into three time periods, corresponding to the early, middle, and most recent phases of the CCOP experience. To provide sufficiently large denominators that would not be unduly influenced by the results of one or two studies and that would allow readily interpretable comparisons, we purposely selected cutoff points that resulted in roughly equal groups of approximately 40 studies for each time period.
RESULTS
Definitions and Conceptual Frameworks
In the 68 studies that included QOL assessment, QOL was most frequently referred to as a phenomenon that was subjective, multidimensional, and recognized as an increasingly important outcome in addition to objective indicators of tumor response and survival time. QOL was operationally defined by the specification of the particular instrument to be used in the study. In terms of a conceptual framework, QOL was identified as a secondary end point in all but seven studies (see Study End Points section), with an explicit or implicit linear relationship indicated between symptom improvement (primary end point) and QOL improvement.
Trials were powered to detect the estimated effect size on the primary end point. When power estimates for determining changes in QOL were presented, it was usually on the basis of the sample size determined by the primary end point. For example, "With a sample size of 180 patients, we will have 70% power to detect a two-point change (or 0.3 standard deviations) in QOL and 80% power to detect a three-point change." Justifications for attaining a given effect size on QOL were rarely presented, and when they were, they were most often treated as equivalent to the effect size on the primary end point. For example, if trials were powered to detect a two-point decrease in nausea, then the plan of analysis would indicate sufficient power to detect a two-point improvement in QOL, with no prospective plan to correlate the two.
As shown in Table 2, between 1987 and 1994, 42 symptom management studies were activated (an average of approximately five studies per year); between 1995 and 2000, 39 studies were activated (approximately six and a half studies per year); and between 2001 and 2004, 49 symptom management trials were activated (approximately 12 studies per year). In the early period, only 26% of the trials included QOL assessment; in the middle period, 54% of the trials included QOL assessment; and, in the most recent period, 74% of trials included QOL assessment.
Instruments
In the CCOP symptom management portfolio, 22 different global QOL assessment instruments were identified. A breakdown of the QOL instruments used in each time period is shown in Table 3. The Functional Living Index: Cancer15 was the most frequently used QOL instrument in the first 8 years of the CCOP experience; it has not been used since that time. The Functional Assessment of Cancer Therapy16 (FACT), Uniscale,17 and Spitzer Quality of Life Index18 were the most frequently used instruments between 1995 and 2000 and were included in almost three quarters of the trials in which QOL was assessed. In the most recent 2001 to 2004 period, although the total number of QOL instruments increased substantially, the FACT and Uniscale were used in 59% of the trials that measured QOL. Over the entire 18-year experience, the FACT and Uniscale were selected in 55% of the trials that proposed QOL assessment.
Study End Points
Of the 68 trials that included QOL assessment, global QOL was the primary end point in only seven trials. It was a secondary end point in the remaining 59 trials (note, two trials that included QOL measures did not identify QOL improvement as a primary or secondary end point).
The primary end points for 130 symptom management clinical trials for each time period are listed in Table 4. The frequency of symptoms targeted in the CCOP studies differed over time. Pain, cachexia/anorexia, and stomatis/mucosistis were the most frequent symptoms targeted for amelioration in the 1987 to 1994 period. During that early period, there was only one study of cognitive functioning and no studies of fatigue or osteoporosis. During the 1995 to 2000 period, the most frequently addressed symptoms were stomatitis/mucositis, cachexia/anorexia, and nausea and vomiting. Trials targeting fatigue and hot flashes became more common in this period. However, there was only one study of pain relief and no studies of cognitive functioning or osteoporosis between 1995 and 2000. In the most recent period, cachexia/anorexia, fatigue, hot flashes, and cognitive functioning were the most frequently targeted symptoms, with a declining number of studies addressing stomatitis/mucositis. Overall, cachexia/anorexia was the most frequently targeted symptom, followed by pain, stomatitis/mucositis, hot flashes, and nausea and vomiting.
The number of trials that included QOL assessment, broken down by targeted symptom, is shown in Table 5. In the early CCOP period, QOL assessment was included most frequently in studies targeting psychological symptoms (depression and anxiety) and pain, in addition to the one study specifically designed to improve overall QOL. During the middle period, QOL was included most frequently in trials targeting cachexia, skin toxicities, hot flashes, nausea and vomiting, and stomatitis. In the most recent period, QOL was incorporated most frequently in studies of hot flash, diarrhea, pain, cachexia, and fatigue.
Over the entire 18-year period, the frequency at which QOL assessment was included in these symptom management trials by symptom ranged from 21% for stomatitis to 67% for hot flashes and fatigue. However, both of these findings may be artifacts of the period in which they were studied. That is, mucositis was investigated most frequently in the early period when QOL was least frequently measured; hot flashes and fatigue were most frequently addressed in the most recent period when QOL was most frequently included. In general, the results show that QOL was included in roughly half of the trials across all symptoms, indicating no consistent pattern or rationale for its inclusion or omission in trials targeting any given symptom.
Interventions
Tables 6 and 7 show the frequency of the major types of interventions used in the CCOP symptom management trials and the frequency of QOL assessment by intervention type for each time period. The majority of these phase III symptom management trials investigated the efficacy of pharmaceutical interventions. The proportion of trials that investigated the efficacy of pharmacologic interventions remained fairly constant over time, accounting for approximately two thirds of all studies in each time period. The proportion of behavioral interventions dropped, from 24% of the earliest trials to only 6% of the most recent trials. In contrast, the percentage of complementary and alternative medicine trials increased, from 5% of the trials in the 1987 to 1994 period to 21% of the most recent trials.
As with the symptoms targeted, there was no consistent pattern evident in the inclusion or omission of QOL assessment by intervention type. Over the entire CCOP experience, QOL assessment was included in slightly more than half of the trials for each major intervention type.
DISCUSSION
One major limitation of this research is that the results may be biased because of the sample that was investigated. Because the data were limited to the CCOP portfolio, the findings may reflect the interests and research capacities of the CCOP research groups, rather than the universe of symptom management studies. For example, we speculate that the evident changes in symptoms addressed over time is the result of a complex mix of symptoms perceived to be important and the availability of new agents to test, rather than cumulative progress in finding effective treatments for one symptom and then moving on to the next. Therefore, caution should be exercised in making generalizations or drawing conclusions from these data.
With these caveats in mind, given the results of this research, we make three recommendations to improve the quality and utility of QOL assessment in symptom management trials. The recommendations are presented in the following order of decision-making steps in protocol development: (1) the decision to include QOL assessment or not; (2) the conceptual framework to be used for determining the relationship between symptom relief and its impact on QOL; and (3) the selection of a particular QOL instrument. On the basis of these recommendations, the prospective conceptualization and definition of QOL should drive research design, measures, and analysis.
First, investigators need to present a more compelling rationale for the inclusion of global QOL measures in any symptom management trial under consideration. Because cancer patients usually suffer multiple symptoms, it is not surprising that studies designed to relieve single symptoms have largely failed to demonstrate significant improvements in global QOL.19 Yet, as Jatoi et al20 point out, if the results show that a particular symptom was alleviated, but the study participants did not show improvements in overall QOL, it would be foolish to conclude that the treatment should not, therefore, be recommended. Thus, the need to measure QOL in all symptom management studies is not self-evident.
Investigators need to make a better case for hypothesizing that the relief of the symptom under investigation will be sufficient to yield improvements in global QOL or provide other reasons for its inclusion. For example, one possible rationale for assessing QOL might be to provide a validation measure of the perceived importance that patients ascribe to the relief of a particular symptom. Another rationale might be to validate a symptom assessment instrument (eg, is a hypothetical two-point change on a given fatigue scale clinically meaningful). Still another rationale might be to gain better diagnostic information; for instance, does the relief of one symptom (eg, fatigue) lead to an increase in perceived discomfort from another (eg, nausea) Clearly, ameliorating distressing symptoms is in itself a worthy goal. Investigators need to show how and why the inclusion of QOL assessment adds research value. If the researchers cannot show how evidence of an effect on QOL would improve the evaluation of an intervention, then its measurement may be unnecessarily burdensome.
Second, if the measurement of global QOL is warranted, then researchers need to present a conceptual framework that specifies relationships between the symptom(s) under investigation, interactions with other variables, the various domains of QOL, and global QOL. In the CCOP studies that included QOL measures, the conceptual framework was limited to a univariate linear relationship between single symptom reduction and global improvements in QOL. To flesh out more detailed conceptual frameworks, investigators might start by specifying hypothetical relationships between symptom relief and various domains of QOL (eg, nausea reduction may lead to improvements in the physical domain of QOL; and depression reduction may lead to improvements in the emotional domain). To elaborate the model further, there is a need for more explicit hypotheses about how improvements in one particular domain (eg, physical) might be expected (or not) to produce improvements in other domains (eg, emotional) of QOL.
In addition, hypothesized relationships between different symptoms need to be more explicitly articulated.21 For example, several studies have suggested a relationship between fatigue, depression, and QOL, but the direction and magnitude of effects have not been explicitly identified. Are reductions in fatigue hypothesized to decrease depression, or is it vice versa Are reductions in fatigue hypothesized to reduce depression, which is then expected to lead to improvements in (the emotional domain of) QOL Or is there a direct effect of fatigue reductions on QOL To illustrate, a study of the effects of Paxil (GlaxoSmithKline, Research Triangle Park, NC) found that depressive symptoms were decreased, but there was no effect on fatigue.22 A follow-up study using modafinil to reduce fatigue is being conducted with the same study population and the same assessment instruments to tease out these relationships. The potential impact of intermediary, intervening, or confounding variables, such as social support or cultural background, also needs to be explicitly laid out in such conceptual frameworks. Methodologically, in the CCOP studies reviewed, QOL was virtually always identified as a secondary end point, and the trials were powered to assess the impact of the intervention on the primary end point, symptom relief. If QOL is included, then the trials need to be powered sufficiently to evaluate the effect on global QOL and/or one (or more) of its dimensions, taking these multivariate relationships into account.
To begin building better conceptual models, we recommend that a flowchart illustrating the hypothesized linkages among the specific symptoms being managed, particular domains of QOL, other symptoms, mediating variables, and global sense of QOL be presented in future research proposals. To initiate this process, the NCI will convene a collaborative working group of symptom management and QOL investigators to develop guidelines for building such conceptual frameworks.
Third, investigators need to provide a rationale for the selection of a particular QOL instrument. As we found, there are now a large number of validated measures of QOL available to investigators (although it is important to note that these instruments were validated on patients undergoing cancer treatment and not on patients being treated for symptom relief). The different instruments assess different dimensions (physical, mental, emotional, social, spiritual, and so on) of QOL and have different degrees of sensitivity in particular domains. The selection of the QOL instrument needs to be tailored to the specific research objectives and validated for the specific patient population.23,24 For example, if the investigators hypothesize the greatest impact in a particular domain, then the QOL instrument with the greatest sensitivity in that domain should be selected. In contrast, if the relief of a particular symptom for a particular population is unlikely to result in changes in their work satisfaction, for example, then the selection of a QOL instrument that measures the impact of job satisfaction on QOL may not be appropriate. Along these lines, Ganz25 has recommended that QOL instruments need to be selected and appropriately tailored to the stage of cancer. Researchers must indicate why the particular instruments selected are the best suited to their research objectives.
In conclusion, attention to health-related QOL has grown in cancer clinical trials over the past two decades. QOL assessment in cancer treatment trials has now demonstrated clear research and clinical utility in resolving the relative advantage of different treatment regimens with equivalent survival rates.10 One clear advantage of QOL assessment in treatment trials is that it can be anchored to objective indicators of tumor response and survival time.26 In symptom management research, however, QOL assessment can generally only be compared with other subjective patient-reported indicators, such as pain relief, nausea and fatigue reduction, and changes in mood states. The added research value of QOL assessment in symptom management research has yet to be established. We have identified a number of issues that need to be addressed to advance the state of the science and to demonstrate the value of QOL assessment in symptom management trials. With more fully articulated conceptual models and refinements in QOL instrument selection, QOL assessment in symptom management trials holds the potential for demonstrating and documenting that suffering can be alleviated and the QOL for cancer patients can be maintained, restored, and even improved as a result of effective symptom relief.
Authors' Disclosures of Potential Conflicts of Interest
Acknowledgment
We thank Carol Moinpour, PhD, and William Anderson, MD, for their helpful comments on earlier drafts of this article, and Cynthia Whitman for her help in retrieving data from the archives.
NOTES
Supported by the Cancer Prevention Research Fellowship program at the National Cancer Institute, Bethesda, MD.
Presented in part at the 39th Annual Meeting of the American Society of Clinical Oncology, Chicago, IL, May 30-June 3, 2003 and at the 10th Annual Meeting of the International Society for Research on Quality of Life, Prague, Czech Republic, November 12-15, 2003.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
1. American Society of Clinical Oncology: Outcomes of cancer treatment for technology assessment and cancer treatment guidelines. J Clin Oncol 14:671-679, 1996
2. Johnson J, Temple R: Food and Drug Administration requirements for approval of new anticancer drugs. Cancer Treat Rep 69:1155-1157, 1985
3. Soni MK, Cella D: Quality of life and symptom measures in oncology: An overview. Am J Manag Care 8:S560-S573, 2002 (suppl 18)
4. Sloan JA, Cella D, Frost M, et al: Assessing clinical significance in measuring oncology patient quality of life: Introduction to the symposium, content overview, and definition of terms. Mayo Clin Proc 77:367-370, 2002
5. Cella D, Chang C, Lai J, et al: Advances in quality of life measurement in oncology patients. Semin Oncol 29:60-68, 2002
6. Gotay C: Trial-related quality of life: Using quality of life assessment to distinguish among cancer therapies. J Natl Cancer Inst Monogr 20:1-6, 1996
7. Efficace F, Bottomley A: Do quality-of-life randomized clinical trials support clinicians in their decision-making J Clin Oncol 20:4126-4127, 2002
8. Levine M, Ganz P: Beyond the development of quality-of-life instruments: Where do we go from here J Clin Oncol 20:2215-2216, 2002
9. Moinpour C: Do quality of life assessments make a difference in the evaluation of cancer treatments Control Clin Trials 18:311-317, 1997
10. Goodwin P, Black J, Bordeleau L, et al: Health-related quality-of-life measurement in randomized clinical trials in breast cancer: Taking stock. J Natl Cancer Inst 95:263-281, 2003
11. Ferrell B, Dow K, Grant M: Measurement of quality of life in cancer survivors. Qual Life Res 4:523-531, 1995
12. Gotay C, Muraoka M: Quality of life in long-term survivors of adult-onset cancers. J Natl Cancer Inst 90:656-667, 1998
13. Institute of Medicine: Improving Palliative Care for Cancer Patients. Washington DC, National Academy Press, 2001
14. Sepulveda C, Marlin A, Yoshida T, et al: Palliative care: The World Health Organization's global perspective. J Pain Symptom Manage 24:91-96, 2002
15. Schipper H, Clinch J, McMurray A, et al: Measuring the quality of life of cancer patients: The Functional Living Index-Cancer—Development and validation. J Clin Oncol 2:472-483, 1984
16. Cella D, Tulsky D, Gray G, et al: The Functional Assessment of Cancer Therapy Scale: Development and validation of the general measure. J Clin Oncol 11:570-579, 1993
17. Sloan J, Sargent D, Mahoney M, et al: When less is more: The role of single item global QOL assessments in oncology clinical trials. Proc Int Soc Qual Life Res 7:593, 1998 (abstr)
18. Spitzer W, Dobson A, Hall J, et al: Measuring the quality of life of cancer patients: A concise QL-index for use by physicians. J Chronic Dis 34:585-597, 1981
19. Somerfield M: Hazards of quality-of-life data for clinical decision making. J Clin Oncol 21:82-83, 2003
20. Jatoi A, Kumar S, Sloan J, et al: On appetite and its loss. J Clin Oncol 18:2930-2932, 2000
21. Patrick DL, Ferketich SL, Frame PS, et al: National Institutes of Health State-of-the-Science Conference Statement: Symptom management in cancer—Pain, depression, and fatigue, July 15-17, 2002. J Natl Cancer Inst 95:1110-1117, 2003
22. Hickok J, Raubertas R, Hynes H, et al: Effect of an SSRI antidepressant on fatigue and depression in 738 cancer patients treated with chemotherapy: A URCC CCOP study. Proc Am Soc Clin Oncol 20:384a, 2001 (abstr 1531)
23. Sprangers M, Moinpour C, Moynihan T, et al: Assessing meaningful change in quality of life over time: A users' guide for clinicians. Mayo Clin Proc 77:561-571, 2002
24. Sloan J, Aaronson N, Cappelleir J, et al: Assessing the clinical significance of single items relative to summated scores. Mayo Clin Proc 77:479-487, 2002
25. Ganz P: Quality of life across the continuum of breast cancer care. Breast J 6:324-330, 2000
26. Guyatt G, Osoba D, Wu A, et al: Methods to explain the clinical significance of health status measures. Mayo Clin Proc 77:371-383, 2002(David R. Buchanan, Ann M.)
ABSTRACT
METHODS: All QOL research objectives, rationales, assessment instruments, symptoms treated, and types of interventions from the CCOP symptom management portfolio of clinical trials were extracted and analyzed.
RESULTS: QOL assessments were proposed in 68 (52%) of the 130 total CCOP symptom management trials initiated since 1987. A total of 22 global QOL instruments were identified. Both the frequency of symptom management trials and the frequency of QOL assessment have increased significantly over time. The Functional Assessment of Cancer Therapy and Uniscale instruments were the most widely used QOL instruments, included in 55% of trials assessing QOL. The conceptual framework for QOL inclusion was limited to univariate relationships between symptom relief and global improvements in QOL. No consistent associations were found between QOL assessment and either the symptoms targeted or types of interventions.
CONCLUSION: To advance the state of the science, research protocols need to provide more explicit rationales for assessing QOL in symptom management trials and for the selection of the QOL instrument(s) to be used. Conceptual frameworks that specify the hypothesized links between the specific symptom(s) being managed, interactions with other symptoms, different domains of QOL, and global QOL also need to be more precisely described. Methodologic and conceptual advances in QOL symptom management trials are critical to fulfill the promise of alleviating suffering and improving the QOL of cancer patients.
INTRODUCTION
Although QOL assessment in cancer treatment trials has received considerable attention,2-9 its utility in symptom management trials has yet to be determined. In cancer treatment trials, survival is the most important outcome,1 and QOL assessment has demonstrated clear research value as a secondary outcome measure that can discriminate between treatments with equivalent survival rates.10 In symptom management trials, the primary end point is the relief of targeted symptoms caused by cancer and its treatment. Unlike treatment trials, QOL assessment in symptom management trials is not anchored to objective indicators but only to subjective, patient-reported symptoms.
Symptom management research has grown in significance in recent years. As cancer patients live longer, researchers have documented that many survivors experience significant negative effects from the cancer and/or treatment on the quality of their daily lives, long after the completion of therapy.11-14 In 2001, the Institute of Medicine reported that at least half of the 550,000 people who die from cancer each year suffer from a spectrum of symptoms including pain, labored breathing, distress, nausea, confusion, and other conditions that seriously diminish the quality of their survivorship.13 With the rapid growth of symptom management trials, it is timely and important to examine the contribution of QOL assessment to this line of research.
The purpose of the research described in this report was to investigate how researchers prospectively conceptualized, defined, and measured QOL in the symptom management clinical trials conducted by the Community Clinical Oncology Program (CCOP). The CCOP has been conducting symptom management trials with QOL end points since 1987. The CCOP is a distinct clinical trials network supported by the National Cancer Institute (NCI). The CCOP funds cooperative groups and select cancer centers as research bases to design and conduct clinical trials in cancer control and prevention, and it funds community physicians to accrue patients to these trials. Currently, eight cooperative groups and six cancer centers are funded as CCOP research bases (Table 1). We examined the CCOP experience to assess progress to date and to identify future directions to advance the state of the science of QOL assessment in symptom management trials.
METHODS
Data Extraction
The analysis focused on the symptom management trials and excluded all other types of cancer control and prevention research. To identify the symptom management research protocols, all CCOP cancer control clinical trials (N = 241) were independently categorized by two of the authors (A.M.O. and L.M.M.) as symptom management, prevention, screening and early detection, or other trials, and the few discrepancies were resolved through consensus discussions.
the original research protocols, the research objectives, assessment instruments, study rationale, and plan of analysis were extracted and analyzed to determine how the investigators prospectively conceptualized the purpose of QOL assessment and the relationship between symptom management and QOL. In addition, data on the symptom treated, type of intervention, research design, sample size, and cancer type(s) and stages were also retrieved.
Coding
All research instruments were reviewed by the entire research team for coding and classification. After an initial review, the instruments were grouped into the following three broad categories: global QOL instruments, symptom instruments, and a miscellaneous other category (eg, alcohol and other drug use, social support, and so on). On the basis of the definition by Goodwin et al,10 the criterion used for classifying questionnaires as global QOL instruments was "self-report of the impact of cancer and its treatment on some aspect of function (eg, physical, role, emotional, or social)." All instruments were coded independently by the research team, and disagreements regarding classifications were resolved through consensus discussions. To determine whether there were any changes over time, the studies were grouped into three time periods, corresponding to the early, middle, and most recent phases of the CCOP experience. To provide sufficiently large denominators that would not be unduly influenced by the results of one or two studies and that would allow readily interpretable comparisons, we purposely selected cutoff points that resulted in roughly equal groups of approximately 40 studies for each time period.
RESULTS
Definitions and Conceptual Frameworks
In the 68 studies that included QOL assessment, QOL was most frequently referred to as a phenomenon that was subjective, multidimensional, and recognized as an increasingly important outcome in addition to objective indicators of tumor response and survival time. QOL was operationally defined by the specification of the particular instrument to be used in the study. In terms of a conceptual framework, QOL was identified as a secondary end point in all but seven studies (see Study End Points section), with an explicit or implicit linear relationship indicated between symptom improvement (primary end point) and QOL improvement.
Trials were powered to detect the estimated effect size on the primary end point. When power estimates for determining changes in QOL were presented, it was usually on the basis of the sample size determined by the primary end point. For example, "With a sample size of 180 patients, we will have 70% power to detect a two-point change (or 0.3 standard deviations) in QOL and 80% power to detect a three-point change." Justifications for attaining a given effect size on QOL were rarely presented, and when they were, they were most often treated as equivalent to the effect size on the primary end point. For example, if trials were powered to detect a two-point decrease in nausea, then the plan of analysis would indicate sufficient power to detect a two-point improvement in QOL, with no prospective plan to correlate the two.
As shown in Table 2, between 1987 and 1994, 42 symptom management studies were activated (an average of approximately five studies per year); between 1995 and 2000, 39 studies were activated (approximately six and a half studies per year); and between 2001 and 2004, 49 symptom management trials were activated (approximately 12 studies per year). In the early period, only 26% of the trials included QOL assessment; in the middle period, 54% of the trials included QOL assessment; and, in the most recent period, 74% of trials included QOL assessment.
Instruments
In the CCOP symptom management portfolio, 22 different global QOL assessment instruments were identified. A breakdown of the QOL instruments used in each time period is shown in Table 3. The Functional Living Index: Cancer15 was the most frequently used QOL instrument in the first 8 years of the CCOP experience; it has not been used since that time. The Functional Assessment of Cancer Therapy16 (FACT), Uniscale,17 and Spitzer Quality of Life Index18 were the most frequently used instruments between 1995 and 2000 and were included in almost three quarters of the trials in which QOL was assessed. In the most recent 2001 to 2004 period, although the total number of QOL instruments increased substantially, the FACT and Uniscale were used in 59% of the trials that measured QOL. Over the entire 18-year experience, the FACT and Uniscale were selected in 55% of the trials that proposed QOL assessment.
Study End Points
Of the 68 trials that included QOL assessment, global QOL was the primary end point in only seven trials. It was a secondary end point in the remaining 59 trials (note, two trials that included QOL measures did not identify QOL improvement as a primary or secondary end point).
The primary end points for 130 symptom management clinical trials for each time period are listed in Table 4. The frequency of symptoms targeted in the CCOP studies differed over time. Pain, cachexia/anorexia, and stomatis/mucosistis were the most frequent symptoms targeted for amelioration in the 1987 to 1994 period. During that early period, there was only one study of cognitive functioning and no studies of fatigue or osteoporosis. During the 1995 to 2000 period, the most frequently addressed symptoms were stomatitis/mucositis, cachexia/anorexia, and nausea and vomiting. Trials targeting fatigue and hot flashes became more common in this period. However, there was only one study of pain relief and no studies of cognitive functioning or osteoporosis between 1995 and 2000. In the most recent period, cachexia/anorexia, fatigue, hot flashes, and cognitive functioning were the most frequently targeted symptoms, with a declining number of studies addressing stomatitis/mucositis. Overall, cachexia/anorexia was the most frequently targeted symptom, followed by pain, stomatitis/mucositis, hot flashes, and nausea and vomiting.
The number of trials that included QOL assessment, broken down by targeted symptom, is shown in Table 5. In the early CCOP period, QOL assessment was included most frequently in studies targeting psychological symptoms (depression and anxiety) and pain, in addition to the one study specifically designed to improve overall QOL. During the middle period, QOL was included most frequently in trials targeting cachexia, skin toxicities, hot flashes, nausea and vomiting, and stomatitis. In the most recent period, QOL was incorporated most frequently in studies of hot flash, diarrhea, pain, cachexia, and fatigue.
Over the entire 18-year period, the frequency at which QOL assessment was included in these symptom management trials by symptom ranged from 21% for stomatitis to 67% for hot flashes and fatigue. However, both of these findings may be artifacts of the period in which they were studied. That is, mucositis was investigated most frequently in the early period when QOL was least frequently measured; hot flashes and fatigue were most frequently addressed in the most recent period when QOL was most frequently included. In general, the results show that QOL was included in roughly half of the trials across all symptoms, indicating no consistent pattern or rationale for its inclusion or omission in trials targeting any given symptom.
Interventions
Tables 6 and 7 show the frequency of the major types of interventions used in the CCOP symptom management trials and the frequency of QOL assessment by intervention type for each time period. The majority of these phase III symptom management trials investigated the efficacy of pharmaceutical interventions. The proportion of trials that investigated the efficacy of pharmacologic interventions remained fairly constant over time, accounting for approximately two thirds of all studies in each time period. The proportion of behavioral interventions dropped, from 24% of the earliest trials to only 6% of the most recent trials. In contrast, the percentage of complementary and alternative medicine trials increased, from 5% of the trials in the 1987 to 1994 period to 21% of the most recent trials.
As with the symptoms targeted, there was no consistent pattern evident in the inclusion or omission of QOL assessment by intervention type. Over the entire CCOP experience, QOL assessment was included in slightly more than half of the trials for each major intervention type.
DISCUSSION
One major limitation of this research is that the results may be biased because of the sample that was investigated. Because the data were limited to the CCOP portfolio, the findings may reflect the interests and research capacities of the CCOP research groups, rather than the universe of symptom management studies. For example, we speculate that the evident changes in symptoms addressed over time is the result of a complex mix of symptoms perceived to be important and the availability of new agents to test, rather than cumulative progress in finding effective treatments for one symptom and then moving on to the next. Therefore, caution should be exercised in making generalizations or drawing conclusions from these data.
With these caveats in mind, given the results of this research, we make three recommendations to improve the quality and utility of QOL assessment in symptom management trials. The recommendations are presented in the following order of decision-making steps in protocol development: (1) the decision to include QOL assessment or not; (2) the conceptual framework to be used for determining the relationship between symptom relief and its impact on QOL; and (3) the selection of a particular QOL instrument. On the basis of these recommendations, the prospective conceptualization and definition of QOL should drive research design, measures, and analysis.
First, investigators need to present a more compelling rationale for the inclusion of global QOL measures in any symptom management trial under consideration. Because cancer patients usually suffer multiple symptoms, it is not surprising that studies designed to relieve single symptoms have largely failed to demonstrate significant improvements in global QOL.19 Yet, as Jatoi et al20 point out, if the results show that a particular symptom was alleviated, but the study participants did not show improvements in overall QOL, it would be foolish to conclude that the treatment should not, therefore, be recommended. Thus, the need to measure QOL in all symptom management studies is not self-evident.
Investigators need to make a better case for hypothesizing that the relief of the symptom under investigation will be sufficient to yield improvements in global QOL or provide other reasons for its inclusion. For example, one possible rationale for assessing QOL might be to provide a validation measure of the perceived importance that patients ascribe to the relief of a particular symptom. Another rationale might be to validate a symptom assessment instrument (eg, is a hypothetical two-point change on a given fatigue scale clinically meaningful). Still another rationale might be to gain better diagnostic information; for instance, does the relief of one symptom (eg, fatigue) lead to an increase in perceived discomfort from another (eg, nausea) Clearly, ameliorating distressing symptoms is in itself a worthy goal. Investigators need to show how and why the inclusion of QOL assessment adds research value. If the researchers cannot show how evidence of an effect on QOL would improve the evaluation of an intervention, then its measurement may be unnecessarily burdensome.
Second, if the measurement of global QOL is warranted, then researchers need to present a conceptual framework that specifies relationships between the symptom(s) under investigation, interactions with other variables, the various domains of QOL, and global QOL. In the CCOP studies that included QOL measures, the conceptual framework was limited to a univariate linear relationship between single symptom reduction and global improvements in QOL. To flesh out more detailed conceptual frameworks, investigators might start by specifying hypothetical relationships between symptom relief and various domains of QOL (eg, nausea reduction may lead to improvements in the physical domain of QOL; and depression reduction may lead to improvements in the emotional domain). To elaborate the model further, there is a need for more explicit hypotheses about how improvements in one particular domain (eg, physical) might be expected (or not) to produce improvements in other domains (eg, emotional) of QOL.
In addition, hypothesized relationships between different symptoms need to be more explicitly articulated.21 For example, several studies have suggested a relationship between fatigue, depression, and QOL, but the direction and magnitude of effects have not been explicitly identified. Are reductions in fatigue hypothesized to decrease depression, or is it vice versa Are reductions in fatigue hypothesized to reduce depression, which is then expected to lead to improvements in (the emotional domain of) QOL Or is there a direct effect of fatigue reductions on QOL To illustrate, a study of the effects of Paxil (GlaxoSmithKline, Research Triangle Park, NC) found that depressive symptoms were decreased, but there was no effect on fatigue.22 A follow-up study using modafinil to reduce fatigue is being conducted with the same study population and the same assessment instruments to tease out these relationships. The potential impact of intermediary, intervening, or confounding variables, such as social support or cultural background, also needs to be explicitly laid out in such conceptual frameworks. Methodologically, in the CCOP studies reviewed, QOL was virtually always identified as a secondary end point, and the trials were powered to assess the impact of the intervention on the primary end point, symptom relief. If QOL is included, then the trials need to be powered sufficiently to evaluate the effect on global QOL and/or one (or more) of its dimensions, taking these multivariate relationships into account.
To begin building better conceptual models, we recommend that a flowchart illustrating the hypothesized linkages among the specific symptoms being managed, particular domains of QOL, other symptoms, mediating variables, and global sense of QOL be presented in future research proposals. To initiate this process, the NCI will convene a collaborative working group of symptom management and QOL investigators to develop guidelines for building such conceptual frameworks.
Third, investigators need to provide a rationale for the selection of a particular QOL instrument. As we found, there are now a large number of validated measures of QOL available to investigators (although it is important to note that these instruments were validated on patients undergoing cancer treatment and not on patients being treated for symptom relief). The different instruments assess different dimensions (physical, mental, emotional, social, spiritual, and so on) of QOL and have different degrees of sensitivity in particular domains. The selection of the QOL instrument needs to be tailored to the specific research objectives and validated for the specific patient population.23,24 For example, if the investigators hypothesize the greatest impact in a particular domain, then the QOL instrument with the greatest sensitivity in that domain should be selected. In contrast, if the relief of a particular symptom for a particular population is unlikely to result in changes in their work satisfaction, for example, then the selection of a QOL instrument that measures the impact of job satisfaction on QOL may not be appropriate. Along these lines, Ganz25 has recommended that QOL instruments need to be selected and appropriately tailored to the stage of cancer. Researchers must indicate why the particular instruments selected are the best suited to their research objectives.
In conclusion, attention to health-related QOL has grown in cancer clinical trials over the past two decades. QOL assessment in cancer treatment trials has now demonstrated clear research and clinical utility in resolving the relative advantage of different treatment regimens with equivalent survival rates.10 One clear advantage of QOL assessment in treatment trials is that it can be anchored to objective indicators of tumor response and survival time.26 In symptom management research, however, QOL assessment can generally only be compared with other subjective patient-reported indicators, such as pain relief, nausea and fatigue reduction, and changes in mood states. The added research value of QOL assessment in symptom management research has yet to be established. We have identified a number of issues that need to be addressed to advance the state of the science and to demonstrate the value of QOL assessment in symptom management trials. With more fully articulated conceptual models and refinements in QOL instrument selection, QOL assessment in symptom management trials holds the potential for demonstrating and documenting that suffering can be alleviated and the QOL for cancer patients can be maintained, restored, and even improved as a result of effective symptom relief.
Authors' Disclosures of Potential Conflicts of Interest
Acknowledgment
We thank Carol Moinpour, PhD, and William Anderson, MD, for their helpful comments on earlier drafts of this article, and Cynthia Whitman for her help in retrieving data from the archives.
NOTES
Supported by the Cancer Prevention Research Fellowship program at the National Cancer Institute, Bethesda, MD.
Presented in part at the 39th Annual Meeting of the American Society of Clinical Oncology, Chicago, IL, May 30-June 3, 2003 and at the 10th Annual Meeting of the International Society for Research on Quality of Life, Prague, Czech Republic, November 12-15, 2003.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
1. American Society of Clinical Oncology: Outcomes of cancer treatment for technology assessment and cancer treatment guidelines. J Clin Oncol 14:671-679, 1996
2. Johnson J, Temple R: Food and Drug Administration requirements for approval of new anticancer drugs. Cancer Treat Rep 69:1155-1157, 1985
3. Soni MK, Cella D: Quality of life and symptom measures in oncology: An overview. Am J Manag Care 8:S560-S573, 2002 (suppl 18)
4. Sloan JA, Cella D, Frost M, et al: Assessing clinical significance in measuring oncology patient quality of life: Introduction to the symposium, content overview, and definition of terms. Mayo Clin Proc 77:367-370, 2002
5. Cella D, Chang C, Lai J, et al: Advances in quality of life measurement in oncology patients. Semin Oncol 29:60-68, 2002
6. Gotay C: Trial-related quality of life: Using quality of life assessment to distinguish among cancer therapies. J Natl Cancer Inst Monogr 20:1-6, 1996
7. Efficace F, Bottomley A: Do quality-of-life randomized clinical trials support clinicians in their decision-making J Clin Oncol 20:4126-4127, 2002
8. Levine M, Ganz P: Beyond the development of quality-of-life instruments: Where do we go from here J Clin Oncol 20:2215-2216, 2002
9. Moinpour C: Do quality of life assessments make a difference in the evaluation of cancer treatments Control Clin Trials 18:311-317, 1997
10. Goodwin P, Black J, Bordeleau L, et al: Health-related quality-of-life measurement in randomized clinical trials in breast cancer: Taking stock. J Natl Cancer Inst 95:263-281, 2003
11. Ferrell B, Dow K, Grant M: Measurement of quality of life in cancer survivors. Qual Life Res 4:523-531, 1995
12. Gotay C, Muraoka M: Quality of life in long-term survivors of adult-onset cancers. J Natl Cancer Inst 90:656-667, 1998
13. Institute of Medicine: Improving Palliative Care for Cancer Patients. Washington DC, National Academy Press, 2001
14. Sepulveda C, Marlin A, Yoshida T, et al: Palliative care: The World Health Organization's global perspective. J Pain Symptom Manage 24:91-96, 2002
15. Schipper H, Clinch J, McMurray A, et al: Measuring the quality of life of cancer patients: The Functional Living Index-Cancer—Development and validation. J Clin Oncol 2:472-483, 1984
16. Cella D, Tulsky D, Gray G, et al: The Functional Assessment of Cancer Therapy Scale: Development and validation of the general measure. J Clin Oncol 11:570-579, 1993
17. Sloan J, Sargent D, Mahoney M, et al: When less is more: The role of single item global QOL assessments in oncology clinical trials. Proc Int Soc Qual Life Res 7:593, 1998 (abstr)
18. Spitzer W, Dobson A, Hall J, et al: Measuring the quality of life of cancer patients: A concise QL-index for use by physicians. J Chronic Dis 34:585-597, 1981
19. Somerfield M: Hazards of quality-of-life data for clinical decision making. J Clin Oncol 21:82-83, 2003
20. Jatoi A, Kumar S, Sloan J, et al: On appetite and its loss. J Clin Oncol 18:2930-2932, 2000
21. Patrick DL, Ferketich SL, Frame PS, et al: National Institutes of Health State-of-the-Science Conference Statement: Symptom management in cancer—Pain, depression, and fatigue, July 15-17, 2002. J Natl Cancer Inst 95:1110-1117, 2003
22. Hickok J, Raubertas R, Hynes H, et al: Effect of an SSRI antidepressant on fatigue and depression in 738 cancer patients treated with chemotherapy: A URCC CCOP study. Proc Am Soc Clin Oncol 20:384a, 2001 (abstr 1531)
23. Sprangers M, Moinpour C, Moynihan T, et al: Assessing meaningful change in quality of life over time: A users' guide for clinicians. Mayo Clin Proc 77:561-571, 2002
24. Sloan J, Aaronson N, Cappelleir J, et al: Assessing the clinical significance of single items relative to summated scores. Mayo Clin Proc 77:479-487, 2002
25. Ganz P: Quality of life across the continuum of breast cancer care. Breast J 6:324-330, 2000
26. Guyatt G, Osoba D, Wu A, et al: Methods to explain the clinical significance of health status measures. Mayo Clin Proc 77:371-383, 2002(David R. Buchanan, Ann M.)