- Email: [email protected]
Taking Fatigue Seriously, II: Variability in Fatigue Levels in Cancer Patients
Taking Fatigue Seriously, II: Variability in Fatigue Levels in Cancer Patients JOEL E. DIMSDALE, M.D., SONIA ANCOLI-ISRAEL, PH.D. LIAT AYALON, PH.D., TIMOTHY F. ELSMORE, PH.D. WILLIAM GRUEN
Fatigue is a common and distressing complaint of cancer patients. It is typically measured with symptom inventories that reflect the patient’s experience over the previous days or weeks. This study examined short-term variation in fatigue levels in a heterogeneous group of cancer patients over a 3-day period to examine the feasibility of such repeated assessments and to characterize the extent and pattern of fatigue symptoms in cancer patients. Thirty-four cancer outpatients with diverse malignancies wore a prototype fatigue watch monitor for three consecutive 24-hour periods and provided fatigue ratings every hour while awake for the 3 days. Patients completed an average of 40 self-reports over 72 hours. These reports revealed a diurnal variation in fatigue, with increasing levels in the evening. The reports also revealed considerable differences across individuals and within individuals in terms of fatigue ratings. Multiple ratings of fatigue within short periods of time can be obtained and reveal that fatigue levels are quite variable, even within an individual. Cancer patients experience their fatigue as “moderate to extreme” 33% of the time. (Psychosomatics 2007; 48:247–252)
U
nderstanding and controlling fatigue is an enormous challenge for cancer patients and their physicians. Contrasted with pain, depression, and nausea, where innovative treatments have been developed, diagnosing and treating fatigue is still a work-in-progress.1 Fatigue is a major complaint in cancer patients before, during, and after treatment. It is, in fact, the most distressing complaint of cancer patients.2 The extent to which fatigue is related to poor sleep,3 altered circadian rhythms, anemia, circulating cytokines, depression, or pain is still unclear. Like all quality-of-life complaints, fatigue is subjective, and is usually elicited by self-report. Ground-breaking advances have been made by investigators using scales such as the Multidimensional Fatigue Symptom Inventory,4 the Brief Fatigue Inventory,5 the Functional Assessment of Cancer Therapy,6 and the Piper Fatigue Inventory7 to study fatigue in cancer patients. These scales differ in the number of items used to evaluate fatigue and thus the time-requirePsychosomatics 48:3, May-June 2007
ment for scale completion. They also differ in the timeperiod of evaluation (e.g., “recently,” or “in the past week”). The increasing sophistication of such scales does not come without a “cost.” Longer scales have more psychometric stability and allow fine-grained characterization of different dimensions of fatigue. Unfortunately, longer scales require more time to complete and thus pose a potential response burden to patients. The desire to track such self-reports has increased as newer options for cancer therapies have been developed. Although many of these newer treatments increase the likelihood of remission, the effects of such treatment on Received November 29, 2005; revised March 20, 2006; accepted March 31, 2006. From the Dept. of Psychiatry, Univ. of California, San Diego, Activity Research Services, Chula Vista California; and Ambulatory Monitoring, Inc, Ardsley, NY. Send correspondence and reprint requests to Joel E. Dimsdale, M.D., UCSD, 9500 Gilman Dr., La Jolla, CA 920930804. e-mail: [email protected] 䉷 2007 The Academy of Psychosomatic Medicine
http://psy.psychiatryonline.org
247
Fatigue in Cancer Patients quality-of-life measures such as fatigue are challenging to study. Studies are accumulating that track, for instance, variations in fatigue levels during the course of treatment.8 However, these studies are complicated to execute because there is a fundamental difference between asking a patient to report on a discrete, time-delineated event closely tied to pathophysiology (“Did you vomit yesterday?”) and asking about a less well-delineated complaint (“How significant was your fatigue?”). Fatigue levels may vary throughout the day, and a single recall summarizing a treatment interval (“yesterday,” “last week,” “since chemotherapy started”) may well be suspect. Although a single-recall report of a subjective symptom like fatigue is a useful first step, repeated sampling of such symptoms may offer more precision in their measurement. More-frequent sampling at different times of day and in different environments has been referred to as “ecological momentary assessment (EMA).” EMA is a powerful technique that relies on repeated sampling (i.e., “momentary”) assessment of behavior in the actual environment (i.e., “ecological”), as opposed to assessment of behavior retrospectively from memory.9 A more familiar biomedical equivalent of EMA is the Holter monitor, which records heart rhythm continuously in the real-world environment, as opposed to relying on a single ECG in the doctor’s office. To a certain extent, EMA approaches are already being incorporated in managing cancer patients’ pain, through repeated use of simple-to-use, one-item visual-analog scales for pain. However, the application of such approaches for assessing (and, theoretically, thereafter, managing) fatigue has not yet been accomplished. EMA completion rates go down with increased subject burden. Thus, investigators using an EMA approach try to limit the number of queries they ask of subjects so as to maximize data acquisition. This approach may offer benefits for understanding fatigue symptoms. In previous work,10 we described a new device to obtain repeated measures of fatigue in the ambulatory environment. The device measures 4 ⳯ 3 ⳯ 1 cm and is worn on the wrist. An event-button on the device allows patients to enter a self-rating, which is time-locked. In that study, 10 healthy-control subjects were asked to rate their fatigue levels repeatedly on a 5-point scale (anchored from “no fatigue” to “extreme fatigue”) over the course of 72 hours. Normal subjects had no difficulty in using the device and completed numerous self-entries from which to sample fatigue levels. The rating on this single-item index of fatigue correlated well with ratings obtained from readministra248
http://psy.psychiatryonline.org
tions of the 6-item fatigue scale of the Profile of Mood States (POMS).11 In the current study, we intentionally studied a diverse group of cancer outpatients at varying levels of medical acuity. Some were currently receiving chemotherapy or radiation therapy, and others were cancer survivors, status posttreatment. The first aim of this study was to determine whether this device was broadly “acceptable” to cancer patients (whether they would use it as directed, whether they would they find it easy to use). The second aim was to map out the variations of fatigue level over a 3-day interval by use of descriptive statistics and graphics, measuring how levels changed as the day wore on and how much of the time patients indicate their fatigue levels were high. Four questions were asked: 1) What percent of the requested observations would diverse cancer patients complete using the device? 2) How was fatigue affected by time of day? 3) How well would a patient’s single report of fatigue represent the fatigue reports elicited over a 3-day interval? 4) How often did cancer patients experience moderate or severe fatigue? METHOD Subjects Thirty-four patients with various cancers completed written consent to participate in this study, which was approved by the UCSD Human Subjects Committee. This was a convenience sample of patients recruited from public-service advertisements or word-of-mouth referral. All were outpatients and represented a diverse group in terms of malignancies, age, and therapies. About half were currently receiving chemotherapy and/or radiation therapy; the rest had completed their treatments. Table 1 summarizes the sample characteristics. TABLE 1.
Sample Characteristics (Nⴔ34)
Age, years Men, % Treatment status Completed treatment Current treatment Malignancies, N Breast Prostate Non-Hodgkin’s lymphoma Miscellaneous
57 (SD: 10) 29 16 18 17 3 2 12
SD: standard deviation.
Psychosomatics 48:3, May-June 2007
Dimsdale et al. Procedure Patients wore the fatigue-recording device and a wristwatch, which beeped hourly during waking hours to remind them to enter a fatigue rating. The device was worn at all times except when bathing or swimming. Patients were asked to complete a 1-item rating (“How fatigued are you now?”) on a 5-point scale, corresponding to Not At All, A Little, Moderately, Quite A Bit, or Extreme fatigue hourly while awake for 3 days. The 3-day period of observation was scheduled for patient convenience. Patients were given an option of keying in extra fatigue ratings whenever they chose. Information from the device was downloaded into the computer by use of Ambulatory Monitoring, Inc. software (Ardsley, NY). After completing the 3-day study, patients were asked whether they had found the device easy to use. Acceptability and feasibility of use were inferred from replies to the question plus the frequency with which they completed the possible 48 repeated measures.
values ⬍0.01), such that cancer patients showed more fatigue in the evening (Figure 2). Figure 3 shows the detailed self-reports from three patients with very different patterns of fatigue symptoms. The individual in Panel A was fairly consistent in his low reports of fatigue throughout the day, every day. The patient in Panel B reported different levels of fatigue throughout the 3-day interval; however these complaints seemed to worsen as the day wore on. Patient C reported highly variable fatigue levels, but the differences were not clearly related to clock-time. Overall, cancer patients rated their fatigue as “Not At All” 31% of the time; “A Little” at 36%; “Moderate,” 23%; and “Quite A Bit”–“Extreme” 10% of the time. However, FIGURE 1.
14 12
Data Analysis
10 Patients
Descriptive statistics were used to examine the overall use patterns of the fatigue-watch system; 72 hours of data were superimposed onto one 24-hour grid and plotted in order to determine whether there were significant diurnal variations in fatigue levels. These variations were tested with one-way, repeated-measures ANOVA by use of StatView statistical software (Cary, NC).
Frequency of Patients’ Self-Reports of Fatigue Over a 3-Day Interval
8 6 4 2 0
0–10
11–20
RESULTS
Psychosomatics 48:3, May-June 2007
High
FIGURE 2.
31–40
41–50
51–60
>60
Fatigue Ratings by Time of Day
4.0
**
Fatigue Rating
3.5
*
3.0
* 2.5 2.0 1.5 Low
A total of 34 cancer patients completed the study (Table 1). They included patients with remote histories of diagnosis and treatment of malignancy as well as patients currently receiving chemotherapy or radiation therapy. Figure 1 graphs the frequency of self-ratings completed by the patients over a 3-day period. On average, patients completed 40 self-reports over the 3-day period. Patients wore the device for the entire study period and had no complaints regarding the device. ANOVA revealed a significant time-effect on fatigue (F[4,168]⳱13.31; p⬍0.0001). Fisher’s least significant difference (LSD) post-hoc comparisons showed significantly less fatigue during the morning hours (07:00–09:00) than during 13:00–15:00, 16:00–18:00, and 19:00–21:00 (p values: ⬍0.05, ⬍0.05, and ⬍0.01, respectively). Significant differences were also found between the evening hours (19:00–21:00) and all the other time-points (all p
21–30
Number of Fatigue Ratings
1.0 7–9
10–
13–
16–
19–
Time of Day (2-hour blocks)
http://psy.psychiatryonline.org
249
Fatigue in Cancer Patients these ratings, which include all subjects and all time-points, fail to describe the unique patterning of fatigue complaints in individual cancer patients. Figure 4 illustrates a different way of summarizing fatigue profiles for four individual patients. The figure shows the percent of observations in which individual patients reported varying levels of fatigue. Some patients reported their fatigue as moderate-toextreme over half the time that observations were completed. DISCUSSION Cancer patients across a wide range of illness severity readily used the fatigue watch-monitoring device. There were no complaints of discomfort or inconvenience. Patients were asked to enter an hourly rating over a 3-day period. FIGURE 3.
Examples of Fatigue Patterns in Three Patients
BB010 (y = –0.0616x + 1.7124) 5 4 3 2 1 MA011 (y = 5.495x + 0.7218)
Fatigue Rating
5 4 3 2 1 JW028 (y = 0.4929x + 1.8071) 5 4 3 2 1 0:00
4:00
8:00
12:00 Time
250
http://psy.psychiatryonline.org
16:00
20:00
0:00
Assuming 16 hours of wakefulness per day, a total of 48 measures might be expected. Our patients completed an average of 40 assessments, and thus had no difficulty completing multiple self-reports of fatigue. In general, cancer patients reported that their fatigue levels increased throughout the day, being highest in the evening. This finding agrees with the observations of Miaskowski and Lee, using different rating techniques.12 The unique usefulness of the device, however, is suggested by the data portrayed in Figure 3 and Figure 4. When patients are asked, in a clinical visit “How fatigued are you?” what time-frame do they use as a reference point for their response? Figure 3 suggests that, for most patients, fatigue levels vary considerably throughout the day. For such patients, a single fatigue rating may not be representative of the patient’s moment-to-moment symptoms. Fatigue is extraordinarily variable; indeed, in this context, one can question whether it is a “level” at all. A single value for “fatigue” is thus a good starting point, but provides a somewhat limited guide for treatment. It is possible that a more detailed characterization of the time-course of fatigue symptoms may guide treatment—for instance, with light therapy to advance circadian rhythms.13 Figure 4 graphically portrays the fatigue reports of four different cancer patients. The figure examines what percent of each patient’s self-reports of fatigue were “minimal” to “extreme” in nature. Although some patients report only minimal levels of fatigue, it is troubling that other patients reported protracted levels of moderate-to-extreme fatigue throughout their 3 days of monitoring. Patients B, C, and D rated their fatigue as “Moderate-to-Extreme” 60%–90% of the time. The palliative-care movement has educated physicians about the importance of assessing and treating pain. Indeed, if Figure 4 portrayed “pain complaints,” one would conclude that patients’ symptoms were not being adequately treated. Because studies of fatigue are still in their infancy, as compared with studies of pain,14 there is less consensus about how to define fatigue or when to intervene. The first step in treating cancer patients’ symptoms is to assess complaints methodically. We have described a new approach for such characterization. This approach is a variant of EMA, as used to monitor physiological signals with devices such as Holter monitors or measure subjective variables with questionnaires or electronic diaries.15 With such techniques, it is necessary to strike a balance between ease of administration and thoroughness. We assessed fatigue with a single item, administered repeatedly over a 3day interval. Single-item scales are notoriously unstable, Psychosomatics 48:3, May-June 2007
Dimsdale et al. and that instability may have contributed to the variation of fatigue complaints we observed within individuals. However, fatigue levels, like pain levels, do indeed vary throughout the day, and the variability we encountered may be a valid reflection of changing symptoms. There is a very clear attenuation of data entry with greater response burden; for this reason, we relied upon a single-item assessment of fatigue. There exist a number of excellent fatigue inventories (see, for instance, Stein et al.16 with 10–25 items per inventory). It is our contention that patients would not complete such inventories if asked to self-administer them repeatedly (40 times in 3 days), as in this design. Perhaps there is some optimal number of items in a fatigue inventory that will provide greater psychometric consistency while not vitiating the completeness of repeated data entry. Future work may well define how many fatigue-item queries patients can tolerate in order to provide reliable data. EMA approaches provide rich datasets that can be analyzed with sophisticated methods that consider auto-correlation and time-series analyses. In general, this article features descriptive statistics that provide an overview, as opposed a comprehensive statistical analysis. Fatigue ratings are now being integrated into other EMA devices, such as wrist actigraphs, which measure sleep/wake activity.17 Given the relationship between fatigue and sleep, these wrist-worn devices might make data collection, both on sleep habits and fatigue, more of a reality, allowing the examination of the relationship between the two. We studied a heterogeneous group of cancer patients. The point in doing so was to determine whether the technique was generally usable in this population. There are sparse comparable data in healthy-normal subjects to ascertain how they would rate their fatigue symptoms. The “meaning” of fatigue is, of course, at the heart of the problem of knowing what to do about it. Unlike “pain,” where all can agree that it is distressing, there are times when “fatigue” is normative or pleasurable (e.g., after jogging), FIGURE 4.
as distinct from distressing (e.g., as an accompaniment of radiation therapy). Further work on the epistemology and psychometrics of fatigue assessment would be useful in the context of EMA approaches. It might, for instance, be valuable to assess both “fatigue” and “distress” with EMA. Perhaps at the heart of distressing fatigue is the fact that it worsens so readily and improves so marginally, even after rest. The goal of this study was to ascertain whether cancer patients with very different treatment characteristics would be able to complete the repeated fatigue measurements with this device and whether the device might reveal interesting patterns of fatigue that might be worth pursuing in subsequent further studies. The ultimate test of the usefulness of this approach would be to examine symptoms in more homogeneous samples of cancer patients who differ in terms of therapy or coexisting levels of depressive symptoms or anemia. For instance, it would be interesting to compare the pattern of fatigue symptoms in patients with more or less anemia or depressive symptoms. Who are the patients with highly variable fatigue symptoms, and what elicits the variation? The approach described here might allow this characterization, which, in turn, might shed light on appropriate interventions. It would also be important to compare what is learned from this EMA-like evaluation of fatigue with the information revealed by more established inventories. All of the patients were outpatients, and thus it is not known whether hospitalized patients with high levels of acuity would complete this scale or find it more (or less) acceptable than traditional fatigue inventories. In summary, repeated assessments of fatigue in cancer patients revealed a number of interesting and clinically important observations. In this diverse group, patients characterized their fatigue levels as “Extreme” 10% of the time. Complaints of fatigue were highly variable throughout the day, although, for most individuals, fatigue levels increased inexorably as the day went on. Figure 3 portrays the challenge of these data. When we ask patients about their fa-
Representative Patterns of Fatigue in Four Patients Over a 3-Day Interval
“No fatigue” “A little fatigue” “Moderate fatigue”
Patient A
Patient B
Psychosomatics 48:3, May-June 2007
Patient C
Patient D
“Quite a bit of fatigue” to “extreme fatigue”
http://psy.psychiatryonline.org
251
Fatigue in Cancer Patients tigue, we implicitly assume that one rating can characterize an entire day. In reality, across 3 successive days, the average patient describes his or her fatigue symptoms across an entire spectrum. Which number is “right”? Viewing the detailed fatigue analyses in Figure 4, one wonders whether it is clinically acceptable when patients rate their fatigue as “Extreme.” For what percentage of the time should patients tolerate extreme fatigue levels before some in-
tervention is attempted? We suspect that tools such as those described in this article may offer guidance for better-informed future treatments. The authors thank the cancer patients who volunteered for this study. This research was supported by NIH grants CA96631, CA84866, CA112035, CA85264, CBCRP 11IB-0034, HL36005, HL44915, and MH18399.
References
1. Morrow G, Andrews P, Hickok J, et al: Fatigue associated with cancer and its treatment. Support Care Cancer 2002; 10:389–398 2. Stone P, Ream E, Richardson A, et al: Cancer-related fatigue: a difference of opinion? results of a multicentre survey of healthcare professionals, patients, and caregivers. Eur J Cancer Care 2003; 12:20–27 3. Ancoli-Israel S, Moore P, Jones V: The relationship between fatigue and sleep in cancer patients: a review, Eur J Cancer Care 2001; 10:245–255 4. Stein KD, Martin SC, Hann DM, et al: A multidimensional measure of fatigue for use with cancer patients. Cancer Pract 1998; 6:143–152 5. Mendoza TR, Wang XS, Cleeland CS, et al: The rapid assessment of fatigue severity in cancer patients: use of The Brief Fatigue Inventory. Cancer 1999; 85:1186–1196 6. Yellen BY, Cella DF, Webster K, et al: Measuring fatigue and other anemia-related symptoms with the Functional Assessment of Cancer Therapy (FACT) measurement system. J Pain System Manage 1997; 13:63–74 7. Piper BF, Dibble SL, Dodd MJ, et al: The Revised Piper Fatigue Scale: psychometric evaluation in women with breast cancer. Oncol Nurs Forum 1998; 25:677–684 8. Berger AM: Patterns of fatigue and activity and rest during adjuvant breast cancer chemotherapy. Oncol Nurs Forum 1998; 25:51– 62
252
http://psy.psychiatryonline.org
9. Stone A, Shiffman S, DeVries M: Ecological momentary assessment, in Well-Being: The Foundation of Hedonic Psychology. Edited by Kaheneman D, Diener E. New York, Russell Sage Publications Foundation, 1999, pp 26–39 10. Dimsdale J, Ancoli-Israel S, Elsmore T, et al: Taking fatigue seriously, I: variation in fatigue sampled repeatedly in healthy controls. J Med Engineer Tech 2003; 27:218–222 11. McNair D, Lorr M, Droppleman L: Profile of Mood States. San Diego, CA, Educational and Industrial Testing Service, 1971 12. Miaskowski C, Lee KA: Pain, fatigue, and sleep disturbances in oncology outpatients receiving radiation therapy for bone metastasis: a pilot study. J Pain Symptom Manage 1999; 17:320–332 13. Campbell SS, Eastman CI, Terman M, et al: Light treatment for sleep disorders: consensus report, I: chronology of seminal studies in humans. J Biol Rhythms 1995; 10:105–109 14. Cleeland C: Cancer-related fatigue: new directions for research, introduction. Cancer 2001; 92(suppl 6):1657–1661 15. Schwartz JE, Stone AA: Strategies for analyzing ecological momentary assessment data. Health Psychol 1998; 17:6–16 16. Stein KD, Martin SC, Hann DC, et al: A multidimensional measure of fatigue for use with cancer patients. Cancer Pract 1998; 6:143–152 17. Ancoli-Israel S, Cole R, Alessi C, et al: The role of actigraphy in the study of sleep and circadian rhythms. Sleep 2003; 26:342– 392
Psychosomatics 48:3, May-June 2007
Fatigue is a common and distressing complaint of cancer patients. It is typically measured with symptom inventories that reflect the patient’s experience over the previous days or weeks. This study examined short-term variation in fatigue levels in a heterogeneous group of cancer patients over a 3-day period to examine the feasibility of such repeated assessments and to characterize the extent and pattern of fatigue symptoms in cancer patients. Thirty-four cancer outpatients with diverse malignancies wore a prototype fatigue watch monitor for three consecutive 24-hour periods and provided fatigue ratings every hour while awake for the 3 days. Patients completed an average of 40 self-reports over 72 hours. These reports revealed a diurnal variation in fatigue, with increasing levels in the evening. The reports also revealed considerable differences across individuals and within individuals in terms of fatigue ratings. Multiple ratings of fatigue within short periods of time can be obtained and reveal that fatigue levels are quite variable, even within an individual. Cancer patients experience their fatigue as “moderate to extreme” 33% of the time. (Psychosomatics 2007; 48:247–252)
U
nderstanding and controlling fatigue is an enormous challenge for cancer patients and their physicians. Contrasted with pain, depression, and nausea, where innovative treatments have been developed, diagnosing and treating fatigue is still a work-in-progress.1 Fatigue is a major complaint in cancer patients before, during, and after treatment. It is, in fact, the most distressing complaint of cancer patients.2 The extent to which fatigue is related to poor sleep,3 altered circadian rhythms, anemia, circulating cytokines, depression, or pain is still unclear. Like all quality-of-life complaints, fatigue is subjective, and is usually elicited by self-report. Ground-breaking advances have been made by investigators using scales such as the Multidimensional Fatigue Symptom Inventory,4 the Brief Fatigue Inventory,5 the Functional Assessment of Cancer Therapy,6 and the Piper Fatigue Inventory7 to study fatigue in cancer patients. These scales differ in the number of items used to evaluate fatigue and thus the time-requirePsychosomatics 48:3, May-June 2007
ment for scale completion. They also differ in the timeperiod of evaluation (e.g., “recently,” or “in the past week”). The increasing sophistication of such scales does not come without a “cost.” Longer scales have more psychometric stability and allow fine-grained characterization of different dimensions of fatigue. Unfortunately, longer scales require more time to complete and thus pose a potential response burden to patients. The desire to track such self-reports has increased as newer options for cancer therapies have been developed. Although many of these newer treatments increase the likelihood of remission, the effects of such treatment on Received November 29, 2005; revised March 20, 2006; accepted March 31, 2006. From the Dept. of Psychiatry, Univ. of California, San Diego, Activity Research Services, Chula Vista California; and Ambulatory Monitoring, Inc, Ardsley, NY. Send correspondence and reprint requests to Joel E. Dimsdale, M.D., UCSD, 9500 Gilman Dr., La Jolla, CA 920930804. e-mail: [email protected] 䉷 2007 The Academy of Psychosomatic Medicine
http://psy.psychiatryonline.org
247
Fatigue in Cancer Patients quality-of-life measures such as fatigue are challenging to study. Studies are accumulating that track, for instance, variations in fatigue levels during the course of treatment.8 However, these studies are complicated to execute because there is a fundamental difference between asking a patient to report on a discrete, time-delineated event closely tied to pathophysiology (“Did you vomit yesterday?”) and asking about a less well-delineated complaint (“How significant was your fatigue?”). Fatigue levels may vary throughout the day, and a single recall summarizing a treatment interval (“yesterday,” “last week,” “since chemotherapy started”) may well be suspect. Although a single-recall report of a subjective symptom like fatigue is a useful first step, repeated sampling of such symptoms may offer more precision in their measurement. More-frequent sampling at different times of day and in different environments has been referred to as “ecological momentary assessment (EMA).” EMA is a powerful technique that relies on repeated sampling (i.e., “momentary”) assessment of behavior in the actual environment (i.e., “ecological”), as opposed to assessment of behavior retrospectively from memory.9 A more familiar biomedical equivalent of EMA is the Holter monitor, which records heart rhythm continuously in the real-world environment, as opposed to relying on a single ECG in the doctor’s office. To a certain extent, EMA approaches are already being incorporated in managing cancer patients’ pain, through repeated use of simple-to-use, one-item visual-analog scales for pain. However, the application of such approaches for assessing (and, theoretically, thereafter, managing) fatigue has not yet been accomplished. EMA completion rates go down with increased subject burden. Thus, investigators using an EMA approach try to limit the number of queries they ask of subjects so as to maximize data acquisition. This approach may offer benefits for understanding fatigue symptoms. In previous work,10 we described a new device to obtain repeated measures of fatigue in the ambulatory environment. The device measures 4 ⳯ 3 ⳯ 1 cm and is worn on the wrist. An event-button on the device allows patients to enter a self-rating, which is time-locked. In that study, 10 healthy-control subjects were asked to rate their fatigue levels repeatedly on a 5-point scale (anchored from “no fatigue” to “extreme fatigue”) over the course of 72 hours. Normal subjects had no difficulty in using the device and completed numerous self-entries from which to sample fatigue levels. The rating on this single-item index of fatigue correlated well with ratings obtained from readministra248
http://psy.psychiatryonline.org
tions of the 6-item fatigue scale of the Profile of Mood States (POMS).11 In the current study, we intentionally studied a diverse group of cancer outpatients at varying levels of medical acuity. Some were currently receiving chemotherapy or radiation therapy, and others were cancer survivors, status posttreatment. The first aim of this study was to determine whether this device was broadly “acceptable” to cancer patients (whether they would use it as directed, whether they would they find it easy to use). The second aim was to map out the variations of fatigue level over a 3-day interval by use of descriptive statistics and graphics, measuring how levels changed as the day wore on and how much of the time patients indicate their fatigue levels were high. Four questions were asked: 1) What percent of the requested observations would diverse cancer patients complete using the device? 2) How was fatigue affected by time of day? 3) How well would a patient’s single report of fatigue represent the fatigue reports elicited over a 3-day interval? 4) How often did cancer patients experience moderate or severe fatigue? METHOD Subjects Thirty-four patients with various cancers completed written consent to participate in this study, which was approved by the UCSD Human Subjects Committee. This was a convenience sample of patients recruited from public-service advertisements or word-of-mouth referral. All were outpatients and represented a diverse group in terms of malignancies, age, and therapies. About half were currently receiving chemotherapy and/or radiation therapy; the rest had completed their treatments. Table 1 summarizes the sample characteristics. TABLE 1.
Sample Characteristics (Nⴔ34)
Age, years Men, % Treatment status Completed treatment Current treatment Malignancies, N Breast Prostate Non-Hodgkin’s lymphoma Miscellaneous
57 (SD: 10) 29 16 18 17 3 2 12
SD: standard deviation.
Psychosomatics 48:3, May-June 2007
Dimsdale et al. Procedure Patients wore the fatigue-recording device and a wristwatch, which beeped hourly during waking hours to remind them to enter a fatigue rating. The device was worn at all times except when bathing or swimming. Patients were asked to complete a 1-item rating (“How fatigued are you now?”) on a 5-point scale, corresponding to Not At All, A Little, Moderately, Quite A Bit, or Extreme fatigue hourly while awake for 3 days. The 3-day period of observation was scheduled for patient convenience. Patients were given an option of keying in extra fatigue ratings whenever they chose. Information from the device was downloaded into the computer by use of Ambulatory Monitoring, Inc. software (Ardsley, NY). After completing the 3-day study, patients were asked whether they had found the device easy to use. Acceptability and feasibility of use were inferred from replies to the question plus the frequency with which they completed the possible 48 repeated measures.
values ⬍0.01), such that cancer patients showed more fatigue in the evening (Figure 2). Figure 3 shows the detailed self-reports from three patients with very different patterns of fatigue symptoms. The individual in Panel A was fairly consistent in his low reports of fatigue throughout the day, every day. The patient in Panel B reported different levels of fatigue throughout the 3-day interval; however these complaints seemed to worsen as the day wore on. Patient C reported highly variable fatigue levels, but the differences were not clearly related to clock-time. Overall, cancer patients rated their fatigue as “Not At All” 31% of the time; “A Little” at 36%; “Moderate,” 23%; and “Quite A Bit”–“Extreme” 10% of the time. However, FIGURE 1.
14 12
Data Analysis
10 Patients
Descriptive statistics were used to examine the overall use patterns of the fatigue-watch system; 72 hours of data were superimposed onto one 24-hour grid and plotted in order to determine whether there were significant diurnal variations in fatigue levels. These variations were tested with one-way, repeated-measures ANOVA by use of StatView statistical software (Cary, NC).
Frequency of Patients’ Self-Reports of Fatigue Over a 3-Day Interval
8 6 4 2 0
0–10
11–20
RESULTS
Psychosomatics 48:3, May-June 2007
High
FIGURE 2.
31–40
41–50
51–60
>60
Fatigue Ratings by Time of Day
4.0
**
Fatigue Rating
3.5
*
3.0
* 2.5 2.0 1.5 Low
A total of 34 cancer patients completed the study (Table 1). They included patients with remote histories of diagnosis and treatment of malignancy as well as patients currently receiving chemotherapy or radiation therapy. Figure 1 graphs the frequency of self-ratings completed by the patients over a 3-day period. On average, patients completed 40 self-reports over the 3-day period. Patients wore the device for the entire study period and had no complaints regarding the device. ANOVA revealed a significant time-effect on fatigue (F[4,168]⳱13.31; p⬍0.0001). Fisher’s least significant difference (LSD) post-hoc comparisons showed significantly less fatigue during the morning hours (07:00–09:00) than during 13:00–15:00, 16:00–18:00, and 19:00–21:00 (p values: ⬍0.05, ⬍0.05, and ⬍0.01, respectively). Significant differences were also found between the evening hours (19:00–21:00) and all the other time-points (all p
21–30
Number of Fatigue Ratings
1.0 7–9
10–
13–
16–
19–
Time of Day (2-hour blocks)
http://psy.psychiatryonline.org
249
Fatigue in Cancer Patients these ratings, which include all subjects and all time-points, fail to describe the unique patterning of fatigue complaints in individual cancer patients. Figure 4 illustrates a different way of summarizing fatigue profiles for four individual patients. The figure shows the percent of observations in which individual patients reported varying levels of fatigue. Some patients reported their fatigue as moderate-toextreme over half the time that observations were completed. DISCUSSION Cancer patients across a wide range of illness severity readily used the fatigue watch-monitoring device. There were no complaints of discomfort or inconvenience. Patients were asked to enter an hourly rating over a 3-day period. FIGURE 3.
Examples of Fatigue Patterns in Three Patients
BB010 (y = –0.0616x + 1.7124) 5 4 3 2 1 MA011 (y = 5.495x + 0.7218)
Fatigue Rating
5 4 3 2 1 JW028 (y = 0.4929x + 1.8071) 5 4 3 2 1 0:00
4:00
8:00
12:00 Time
250
http://psy.psychiatryonline.org
16:00
20:00
0:00
Assuming 16 hours of wakefulness per day, a total of 48 measures might be expected. Our patients completed an average of 40 assessments, and thus had no difficulty completing multiple self-reports of fatigue. In general, cancer patients reported that their fatigue levels increased throughout the day, being highest in the evening. This finding agrees with the observations of Miaskowski and Lee, using different rating techniques.12 The unique usefulness of the device, however, is suggested by the data portrayed in Figure 3 and Figure 4. When patients are asked, in a clinical visit “How fatigued are you?” what time-frame do they use as a reference point for their response? Figure 3 suggests that, for most patients, fatigue levels vary considerably throughout the day. For such patients, a single fatigue rating may not be representative of the patient’s moment-to-moment symptoms. Fatigue is extraordinarily variable; indeed, in this context, one can question whether it is a “level” at all. A single value for “fatigue” is thus a good starting point, but provides a somewhat limited guide for treatment. It is possible that a more detailed characterization of the time-course of fatigue symptoms may guide treatment—for instance, with light therapy to advance circadian rhythms.13 Figure 4 graphically portrays the fatigue reports of four different cancer patients. The figure examines what percent of each patient’s self-reports of fatigue were “minimal” to “extreme” in nature. Although some patients report only minimal levels of fatigue, it is troubling that other patients reported protracted levels of moderate-to-extreme fatigue throughout their 3 days of monitoring. Patients B, C, and D rated their fatigue as “Moderate-to-Extreme” 60%–90% of the time. The palliative-care movement has educated physicians about the importance of assessing and treating pain. Indeed, if Figure 4 portrayed “pain complaints,” one would conclude that patients’ symptoms were not being adequately treated. Because studies of fatigue are still in their infancy, as compared with studies of pain,14 there is less consensus about how to define fatigue or when to intervene. The first step in treating cancer patients’ symptoms is to assess complaints methodically. We have described a new approach for such characterization. This approach is a variant of EMA, as used to monitor physiological signals with devices such as Holter monitors or measure subjective variables with questionnaires or electronic diaries.15 With such techniques, it is necessary to strike a balance between ease of administration and thoroughness. We assessed fatigue with a single item, administered repeatedly over a 3day interval. Single-item scales are notoriously unstable, Psychosomatics 48:3, May-June 2007
Dimsdale et al. and that instability may have contributed to the variation of fatigue complaints we observed within individuals. However, fatigue levels, like pain levels, do indeed vary throughout the day, and the variability we encountered may be a valid reflection of changing symptoms. There is a very clear attenuation of data entry with greater response burden; for this reason, we relied upon a single-item assessment of fatigue. There exist a number of excellent fatigue inventories (see, for instance, Stein et al.16 with 10–25 items per inventory). It is our contention that patients would not complete such inventories if asked to self-administer them repeatedly (40 times in 3 days), as in this design. Perhaps there is some optimal number of items in a fatigue inventory that will provide greater psychometric consistency while not vitiating the completeness of repeated data entry. Future work may well define how many fatigue-item queries patients can tolerate in order to provide reliable data. EMA approaches provide rich datasets that can be analyzed with sophisticated methods that consider auto-correlation and time-series analyses. In general, this article features descriptive statistics that provide an overview, as opposed a comprehensive statistical analysis. Fatigue ratings are now being integrated into other EMA devices, such as wrist actigraphs, which measure sleep/wake activity.17 Given the relationship between fatigue and sleep, these wrist-worn devices might make data collection, both on sleep habits and fatigue, more of a reality, allowing the examination of the relationship between the two. We studied a heterogeneous group of cancer patients. The point in doing so was to determine whether the technique was generally usable in this population. There are sparse comparable data in healthy-normal subjects to ascertain how they would rate their fatigue symptoms. The “meaning” of fatigue is, of course, at the heart of the problem of knowing what to do about it. Unlike “pain,” where all can agree that it is distressing, there are times when “fatigue” is normative or pleasurable (e.g., after jogging), FIGURE 4.
as distinct from distressing (e.g., as an accompaniment of radiation therapy). Further work on the epistemology and psychometrics of fatigue assessment would be useful in the context of EMA approaches. It might, for instance, be valuable to assess both “fatigue” and “distress” with EMA. Perhaps at the heart of distressing fatigue is the fact that it worsens so readily and improves so marginally, even after rest. The goal of this study was to ascertain whether cancer patients with very different treatment characteristics would be able to complete the repeated fatigue measurements with this device and whether the device might reveal interesting patterns of fatigue that might be worth pursuing in subsequent further studies. The ultimate test of the usefulness of this approach would be to examine symptoms in more homogeneous samples of cancer patients who differ in terms of therapy or coexisting levels of depressive symptoms or anemia. For instance, it would be interesting to compare the pattern of fatigue symptoms in patients with more or less anemia or depressive symptoms. Who are the patients with highly variable fatigue symptoms, and what elicits the variation? The approach described here might allow this characterization, which, in turn, might shed light on appropriate interventions. It would also be important to compare what is learned from this EMA-like evaluation of fatigue with the information revealed by more established inventories. All of the patients were outpatients, and thus it is not known whether hospitalized patients with high levels of acuity would complete this scale or find it more (or less) acceptable than traditional fatigue inventories. In summary, repeated assessments of fatigue in cancer patients revealed a number of interesting and clinically important observations. In this diverse group, patients characterized their fatigue levels as “Extreme” 10% of the time. Complaints of fatigue were highly variable throughout the day, although, for most individuals, fatigue levels increased inexorably as the day went on. Figure 3 portrays the challenge of these data. When we ask patients about their fa-
Representative Patterns of Fatigue in Four Patients Over a 3-Day Interval
“No fatigue” “A little fatigue” “Moderate fatigue”
Patient A
Patient B
Psychosomatics 48:3, May-June 2007
Patient C
Patient D
“Quite a bit of fatigue” to “extreme fatigue”
http://psy.psychiatryonline.org
251
Fatigue in Cancer Patients tigue, we implicitly assume that one rating can characterize an entire day. In reality, across 3 successive days, the average patient describes his or her fatigue symptoms across an entire spectrum. Which number is “right”? Viewing the detailed fatigue analyses in Figure 4, one wonders whether it is clinically acceptable when patients rate their fatigue as “Extreme.” For what percentage of the time should patients tolerate extreme fatigue levels before some in-
tervention is attempted? We suspect that tools such as those described in this article may offer guidance for better-informed future treatments. The authors thank the cancer patients who volunteered for this study. This research was supported by NIH grants CA96631, CA84866, CA112035, CA85264, CBCRP 11IB-0034, HL36005, HL44915, and MH18399.
References
1. Morrow G, Andrews P, Hickok J, et al: Fatigue associated with cancer and its treatment. Support Care Cancer 2002; 10:389–398 2. Stone P, Ream E, Richardson A, et al: Cancer-related fatigue: a difference of opinion? results of a multicentre survey of healthcare professionals, patients, and caregivers. Eur J Cancer Care 2003; 12:20–27 3. Ancoli-Israel S, Moore P, Jones V: The relationship between fatigue and sleep in cancer patients: a review, Eur J Cancer Care 2001; 10:245–255 4. Stein KD, Martin SC, Hann DM, et al: A multidimensional measure of fatigue for use with cancer patients. Cancer Pract 1998; 6:143–152 5. Mendoza TR, Wang XS, Cleeland CS, et al: The rapid assessment of fatigue severity in cancer patients: use of The Brief Fatigue Inventory. Cancer 1999; 85:1186–1196 6. Yellen BY, Cella DF, Webster K, et al: Measuring fatigue and other anemia-related symptoms with the Functional Assessment of Cancer Therapy (FACT) measurement system. J Pain System Manage 1997; 13:63–74 7. Piper BF, Dibble SL, Dodd MJ, et al: The Revised Piper Fatigue Scale: psychometric evaluation in women with breast cancer. Oncol Nurs Forum 1998; 25:677–684 8. Berger AM: Patterns of fatigue and activity and rest during adjuvant breast cancer chemotherapy. Oncol Nurs Forum 1998; 25:51– 62
252
http://psy.psychiatryonline.org
9. Stone A, Shiffman S, DeVries M: Ecological momentary assessment, in Well-Being: The Foundation of Hedonic Psychology. Edited by Kaheneman D, Diener E. New York, Russell Sage Publications Foundation, 1999, pp 26–39 10. Dimsdale J, Ancoli-Israel S, Elsmore T, et al: Taking fatigue seriously, I: variation in fatigue sampled repeatedly in healthy controls. J Med Engineer Tech 2003; 27:218–222 11. McNair D, Lorr M, Droppleman L: Profile of Mood States. San Diego, CA, Educational and Industrial Testing Service, 1971 12. Miaskowski C, Lee KA: Pain, fatigue, and sleep disturbances in oncology outpatients receiving radiation therapy for bone metastasis: a pilot study. J Pain Symptom Manage 1999; 17:320–332 13. Campbell SS, Eastman CI, Terman M, et al: Light treatment for sleep disorders: consensus report, I: chronology of seminal studies in humans. J Biol Rhythms 1995; 10:105–109 14. Cleeland C: Cancer-related fatigue: new directions for research, introduction. Cancer 2001; 92(suppl 6):1657–1661 15. Schwartz JE, Stone AA: Strategies for analyzing ecological momentary assessment data. Health Psychol 1998; 17:6–16 16. Stein KD, Martin SC, Hann DC, et al: A multidimensional measure of fatigue for use with cancer patients. Cancer Pract 1998; 6:143–152 17. Ancoli-Israel S, Cole R, Alessi C, et al: The role of actigraphy in the study of sleep and circadian rhythms. Sleep 2003; 26:342– 392
Psychosomatics 48:3, May-June 2007