|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2001 American Society for Clinical Oncology Methodologic Lessons Learned From Hot Flash StudiesByFrom the Mayo Clinic, Rochester, and CentraCare Clinic, St Cloud, MN; and Ann Arbor Regional Community Clinical Oncology Program, Ann Arbor, MI. Address reprint requests to Charles L. Loprinzi, MD, Division of Medical Oncology, Mayo Clinic, 200 1st St SW, Rochester, MN 55905; email: cloprinzi{at}mayo.edu
PURPOSE: In the course of conducting a series of prospective clinical trials devoted to defining new treatment opportunities for hot flashes in cancer survivors, considerable experience has been acquired with related methodologic issues. This article has been written in response to many queries regarding this methodology. PATIENTS AND METHODS: A series of seven different clinical trials that involved 968 patients was used for this work. Reliable and valid definitions of hot flash intensity were developed from patient-reported descriptions. Concomitant validity and reliability assessment of patient-completed diaries was undertaken to compare hot flash data with toxicity and quality-of-life (QOL) end points and to examine consistency across patient groups using variability analysis and correlation procedures. Parametric data from this meta-analysis was used to examine relative power considerations for the design of phase II and phase III clinical trials. RESULTS: Daily diaries used in these studies exhibited consistency and reliability and had few missing data. Hot flash frequency and hot flash score (frequency multiplied by average severity) variables produced almost identical end point results. For phase III placebo-controlled studies, 50 patients per treatment arm seem appropriate to provide sufficient power specifications to detect a clinically meaningful change in hot flash activity. For phase II trials, 25 patients per trial seem to provide reasonable estimates of eventual hot flash efficacy to screen potential agents for more definitive testing. CONCLUSION: Given the data gained from these experiences, we can plan and carry out more efficient trials to identify efficacious agents for the reduction of hot flash activity.
DURING THE LAST several years, hot flashes have been recognized as a cause of major morbidity in postmenopausal women in general and particularly in breast cancer survivors. Prevalence rates for bothersome hot flash activity are approximately 90% in perimenopausal women.1-3 Understanding this, in 1989 the North Central Cancer Treatment Group (NCCTG) began to develop clinical trials to evaluate potential remedies for hot flashes in breast cancer survivors, women with a concern about taking estrogen for fear of breast cancer, and men with prostatic cancer on androgen-deprivation therapy. This work led to a series of seven clinical trials, six placebo-controlled and one pilot study, which involved 968 patients (820 women and 148 men) (Table 1).
During the past few years, a number of research teams have also developed an interest in hot flashes in breast cancer survivors, partially because of the recent recognition that newer antidepressants have efficacy against hot flashes.4-6 As a result of this increased interest in research on hot flash activity, our research team has received multiple requests regarding information for methodologies and instruments for conducting such studies. This is understandable because the measurement of intangible, subjective phenomena, such as hot flashes, can present unique methodologic challenges. We have developed instrumentation and analytic methods that have proven to be successful in assessing the efficacy of agents that reduce hot flash activity and have stood the test of time during more than a decade of work on hot flash studies. This article was developed to help describe our methodology and instruments so that others will be able to use this information to further the science and provide new options for patients with hot flashes.
A series of seven clinical trials that involved a total of 968 patients was used for this work. Reliable and valid definitions of hot flash intensity were developed from patient-reported descriptions. Concomitant validity and reliability assessment of patient-completed diaries was undertaken to compare hot flash data with toxicity and quality-of-life (QOL) end points and to examine consistency across patient groups using variability analysis and correlation procedures. Equilibration of individual hot flash data to a percentage of baseline values was the basis for analytic processes. Parametric data from this meta-analysis was used to examine relative power considerations for designing phase II and phase III clinical trials.
Instrument Development and Hot Flash Definition When we initiated our first hot flash study7 in 1989, there was no validated instrument that we could find to measure hot flashes in our patients. Indeed, to our knowledge, there is still no validated instrumentation in cancer survivors beyond what we have carried out in our studies. A well-validated psychometric approach is to ask patients to record their perceptions of the frequency and intensity of observable and understandable clinical events in a diary,8-12 using examples of symptoms, such as pain and fatigue. Thus, we decided to use patient diaries to gather subjective patient assessments of hot flash activity. To do this, we developed an instrument that used concrete, straightforward language and a common-sense approach. This instrument gathered information from patients on a daily basis regarding how many total hot flashes they had per day as well as information regarding the severity of each of these hot flashes (mild, moderate, severe, or very severe). A current version of this instrument is contained in Appendix A. The use of self-report diaries for data collection has long been established as a valid approach to obtaining data on subjective phenomena such as patient-reported symptoms and perceptions.13-16 Diaries have been used successfully to provide months of trustworthy information and are a staple of modern oncology clinical trials.17 There are cautionary tales about the use of such diaries, but, in general, the method is accepted as being able to produce valid and reliable data under reasonable circumstances and even to produce data with greater detail and accuracy than objective measures in many situations.18-20 Although missing data are seen as a challenge to such an approach, we have a dedicated staff and an established track record of being able to successfully collect patient data within the NCCTG. Prospective diaries of hot flash activity can obtain information on the subjective experience of each hot flash as well as the simple recording of hot flash frequency. Completion rates for the NCCTG patient-reported hot flash activity diaries have remained consistently at approximately 90% for all studies undertaken. We used our initially developed instrument in our first protocol without validating it in advance. Although this might seem unusual at first, it is clearly defensible given the lack of existing valid measures. We have undertaken the philosophy of concurrently validating new instruments while investigating new things rather than spending substantial time and funds on instrument development before proceeding to intervention trials. We have not only done this for hot flashes but also for anorexia/cachexia and mucositis studies. On the basis of the results of our first couple of studies, our hot flash instrument was modified slightly to better clarify patient responses. Reviewers of one of our early studies asked us to define hot flash severities for our patients. We could not find any good definitions to use for such and so took the position that: "If a woman told us she was having a severe hot flash, then we did not believe it was appropriate for us to tell her that it was only a mild hot flash!" Given that the reviewers did not have a response to this statement, they allowed us to proceed with our study without providing strict definitions. Nonetheless, to respond to our reviewers and provide construct validity for our hot flash measurements, we asked women who had participated in one hot flash trial21 to describe to us what they considered to be mild, moderate, severe, and very severe hot flashes. We descriptively analyzed these severity definitions. Through this effort, it became relatively easy to categorize these definitions into clear, congruent representations of severity.22 We performed a similar process in men who participated in one of our hot flash studies.23 This work demonstrated that patients could indeed describe a hot flash and delineate the various component symptoms that comprised the experience of having a hot flash. It also proved they could provide definitions that differentiated degrees of hot flash severities. In our current hot flash studies, we still allow each patient to determine independently the severity of each hot flash, but we provide them with definitions of what others have used to define their hot flash severities (these definitions for women and men are provided as Appendices B and C). We have no data to demonstrate that providing these hot flash definitions to patients actually impacts the results in our clinical trials because the patient definitions were relatively congruent. Nonetheless, we believe that the definitions may help a minority of patients who might benefit from clearly differentiated definitions of symptom severity. Recently, there have been attempts to validate electronic monitoring devices as hot flash measurement instruments based on skin temperature.2,24 Interestingly, these monitors have to date been used in only a few studies with small sample sizes so, despite their being an electronic device, questions remain regarding their relative validity in measuring hot flash activity.25 Further, despite the weak power associated with these studies, there is evidence to suggest that results from subjective diaries correlate well with those obtained via electronic monitoring.2 We maintain that the patient’s subjective experience and opinion regarding the presence of a hot flash takes precedence over an objective skin temperature reading. After all, if a patient does not perceive that she is having a hot flash, is it important that she or others declare that she is having one?
Characterization of Hot Flash Activity Distributions
Instrument Validity and Reliability Do our hot flash diaries demonstrate validity and reliability? We believe that the answer to this question is yes. Convincing evidence of concurrent and discriminant validity as well as reliability is shown with data from our series of hot flash clinical trials. Concurrent validity is demonstrated through correlation analyses with toxicity, QOL, and preference data. First, we gathered toxicity data throughout our series of hot flash studies. The relationship between the amount of hot flash activity and the toxicities reported at baseline (such as the abnormal sweating, insomnia, and so on that were recorded in the format seen in Appendix A) is investigated in Table 4. Patients who report toxicities that theoretically should be related to hot flash activity (such as abnormal sweating, trouble sleeping, and sleepiness) had significantly higher levels of hot flash activity than those who did not report such toxicities. For toxicities that were unrelated to hot flash activity (eg, appetite loss or nausea), the hot flash activity was the same between patients who reported such toxicities and those who did not. Correlation coefficient analysis produced the same results (data not shown). The reduction in the percentage of patients who experienced various hot flash–related symptoms, such as abnormal sweating, was consistent with the proportion of patients who experienced a reduction in hot flash activity (data not shown).
Second, evidence of concurrent validity is provided by the comparison of the hot flash activity reported by patients who claim no change in their QOL scores versus those who report a change in QOL from baseline to after treatment. We asked patients a recent hot flash study5 whether the hot flash activity had an impact on their QOL. We dichotomized their QOL answers based on the median change of an increase of less than 6 points versus an increase of six or more points on a 0- to 100-point scale. We have used this cutpoint as a conservative estimate of a minimally clinically significant change in QOL.26 The patients who reported at least a six-point increase in hot flash–related QOL reported a concurrent drop in hot flash activity of 3.5 hot flashes and 7.0 hot flash score units per day. Patients whose QOL increased by less than six points averaged a drop of 2.7 hot flashes and 5.5 hot flash units per day (P = .02 and .03, respectively). These results provide evidence that patients who perceive a change in QOL related to hot flash activity report scores on the patient diaries that are reflective of this perceived change in hot flash activity. Further evidence of validity comes from the following question asked of all patients participating in the five crossover clinical trials: "Which treatment period (first or second) do you think was most effective in reducing your hot flashes?" Patients who expressed a preference of treatment periods reported a median of 1.3 fewer hot flashes per day and 2.6 fewer hot flash score units per day when receiving the treatment that they reported as being efficacious. For patients who did not express a preference, their hot flash activity differed between treatment periods by only 0.4 hot flashes per day and 0.9 hot flash score units per day (P = .01 and .009, respectively). Statistically significant differences between the two study arms and over time on a clinical protocol provide evidence of discriminant validity. Using this instrument, we have reported statistically significant differences between placebo-controlled arms of studies compared with active drug arms investigating clonidine, megestrol acetate, venlafaxine, vitamin E, and fluoxetine, thus providing discriminant validity. Further evidence of reliability comes from data generated in a recent placebo-controlled trial that looked at three different doses of venlafaxine for the treatment of hot flashes.5 This four-arm, variable-dose study had approximately 50 patients per arm. All three of the venlafaxine arms began with an identical venlafaxine dose (37.5 mg/d) for the first week. The two higher dose arms were titrated upward in subsequent weeks. Figure 2 illustrates the mean hot flash scores during the first treatment week compared with a baseline week for all three of the arms that received identical doses of venlafaxine and for the placebo arm. The superimposable data points for the first week’s scores certainly serve to help validate the entire study methodology, including validation of the daily hot flash diary.
Finally, confirmatory information comes from the virtually identical study results from independently conducted studies that used the same basic instrument, which provides construct validity for the methodology. Two separate studies that tested clonidine for the alleviation of hot flashes in breast cancer survivors came to the same conclusion, indicating a moderate amount of efficacy for clonidine.7,27
Hot Flash Frequency Versus Hot Flash Score The advantages of using the hot flash score are that it does take into account hot flash severity. If a patient started off with 15 very severe hot flashes per day and had these decrease down to 15 mild hot flashes per day, then this would be a theoretical therapeutic advantage that would be measured by the hot flash score metric, but not by the hot flash frequency metric. A change in the score can be interpreted as a change in the number of mild hot flashes or in half the number of moderate hot flashes experienced, and so on. For example, if the hot flash score changes by six units, this could reflect a reduction of six mild hot flashes, three moderate hot flashes, two severe hot flashes, or a combination of these. This can render the score readily interpretable in a clinically relevant context. To date, relatively comparable results have been obtained whether hot flash scores or hot flash frequencies are used for analyzing the data in the studies we have undertaken. Figure 3 displays comparative plots drawn from two studies: venlafaxine in Fig 3A 5 and megestrol acetate in Fig 3B,33 which indicate that the efficacy results for the frequency and score end points are almost superimposable. Figure 4 displays a plot of 968 standardized frequency and score values reported by patients during the seven clinical trials we conducted. The plot indicates that the two scores are closely related and almost always within a half SD of each other. The Spearman correlation coefficient between the two end points is .95, a level generally regarded as indicative of redundancy.34 In summary, it can be concluded that hot flash score is all that is needed on which to base subsequent analyses because it encompasses frequency and severity in a single end point.
Placebo Effect We have observed on multiple occasions a substantial placebo effect in hot flash studies.4,5,7,32,35,36 On average, hot flash frequencies and hot flash scores diminish by approximately 20% to 30% with 4 weeks of a placebo. A notable minority of patients who receive a placebo (typically 15%) report profound decreases in hot flash activity, ie, greater than 75% hot flash reduction. Roughly one out of every four patients who receives a placebo report that their hot flash activity is reduced by at least 50%. Table 5 presents the raw and percentage reduction in hot flash activity for the 375 patients treated with placebo during seven clinical trials. Patients who receive a placebo average a reduction of 1.5 hot flashes per day (24% reduction) and 2.8 score units (26% reduction) from baseline to the end of the first 4 weeks of treatment. Hot flash score reductions in the placebo arms of five previous NCCTG trials (Table 1) illustrate a consistent effect in Fig 5. This well-described placebo effect needs to be considered when anecdotal reports are heard regarding new potential therapies.
Presenting the Results From Randomized Trials Different study arms will not necessarily have identical baseline hot flash frequencies nor hot flash scores when the data are analyzed, because of variability among patient baseline hot flash activity levels. Graphing results may be hard to decipher, given the different baseline values. An illustration of this is shown in Fig 6A.
To make things more easily understood, we present our data by equilibrating baseline information (counting the baseline hot flash information as 100% of total for that arm) and illustrating percent changes from baseline in our figures. Figure 6B illustrates the clarity provided with the same information presented this way as opposed to that presented in Fig 6A. The percentage of baseline results, which are more easily interpreted by clinicians, are valid if the baseline distributions are reasonably comparable and there are few patients with values of zero at baseline. Another potentially useful method for presenting results of hot flash studies is a graphic representation that we have come to call a stream plot that displays the hot flash activity reported by each individual patient (Fig 7A). The advantage of this graphic is the identification of the current within the stream on which the majority of patients are traveling beyond what is achievable by simple inspection of average values. Figure 7A demonstrates the strong effects in hot flash activity reduction reported by the majority of individuals in response to megestrol acetate. Side-by-side comparisons of treatment arms can be made by juxtaposing stream plots based on percentage of baseline values (Fig 7B). The symmetry or lack thereof displayed between the two halves of the graphic indicates the degree of treatment efficacy. Figure 7B illustrates the superiority of venlafaxine over placebo to produce reductions in hot flash activity.
We commonly do exploratory subset analyses in our hot flash studies. With these, we are able to ask questions as to whether there is a difference in hot flash score reductions from baseline on the basis of whether a patient has relatively few baseline hot flashes versus many. These subset analyses illustrate that the percent of reduction in hot flashes looks similar regardless of the baseline hot flash severity. Likewise, we have not noted different outcomes with 4 weeks of therapy on the basis of whether patients were receiving tamoxifen or not. In addition, we have seen similar results in men (who had hot flashes as a result of androgen deprivation therapy for prostate cancer) and women when we examined the efficacy of megestrol acetate33 and venlafaxine,5 but we were unable to see a statistically significant hot flash reduction with clonidine in men,35 although we were able to see this in women.7
Required Baseline Information
We reran analyses of all seven Mayo Clinic and NCCTG clinical trials using only 1 day of baseline activity (repeated seven times for each study to look at each baseline day) for comparison with the original results, which used an entire week of baseline data. P values for the test of treatment efficacy were comparable in virtually all cases (data not shown). We also looked for trends in reporting across days over time (ie, day 1 v day 2, and so on). There was no consistent or significant pattern across the days (Table 6). We also ran these analyses with simulated comparison studies of 50 patients per treatment group and pilot single-arm trials of 25 patients. Results remained consistent whether a single day or an entire week of baseline data was used. For example, Table 6 lists estimated 95% confidence intervals for the mean that would result from a randomized trial of 50 patients per treatment group. The results indicate virtually overlapping confidence intervals across all days of the week and in relation to the use of an entire week of baseline data. Hence, the results for the test of significance of treatment effect would be no different regardless of which day of the week was used or whether the entire baseline week was used. A potential weakness in the use of the individual baseline days appears if there are zero values among patient-reported hot flash activity. If a patient does not experience any hot flashes in a given day, then it is not possible to calculate the percentage of baseline hot flash activity. This happened, however, in fewer than 4% for any single day of our data across the seven studies. Nonetheless, this is a consideration in the use of individual days of baseline data because the daily baseline activity will be overestimated if a considerable number of zero–hot flash days were observed. Hence, the use of a single day of baseline data needs to be considered only in situations where the level of hot flash activity is sufficient to preclude the possibility of many zero–hot flash days.
Thus, the bulk of data above suggests that a single daily baseline hot flash period would suffice as opposed to a complete baseline week. In further considering this, we took 20 random samples of patient baseline data with a sample size of 50 patients each and another 20 samples with sample sizes of 25 patients each. We looked at the variability of individual baseline days, compared with the entire baseline week. As opposed to the 4% difference we saw with the entire sample of 968 patients (Table 6), we saw
How Many Patients Are Needed for a Phase III Placebo-Controlled Clinical Trial? The first way to determine an appropriate sample size is to investigate classic power analysis characteristics. Data from the placebo arms of the seven previous studies indicated that differences in hot flash activity between baseline and the end of the first treatment period had a SD of two hot flashes and five score units per patient per day. Fifty patients per group provides 80% power to detect differences in average hot flash activity of 0.58 SDs via a standard two-sample t test using a two-sided type I error rate of 5%, which is a moderate effect size.37 Hence, 50 patients per treatment arm provides 80% power to detect an average shift of 1.2 hot flashes per day or a hot flash score of three units per day. Smaller differences are probably not clinically important. Power for crossover design linear model approaches will be greater than these specifications as a result of each patient serving as his or her own control. In essence, each power specification effect size above could be divided by the square root of 2 (1.414) to get the appropriate effect sizes for the linear model hypothesis testing. Further support for the sample size of 50 patients per group can be found in simulation studies. Figure 8 demonstrates the reproducibility of the results with approximately 50 patients per arm using 500 bootstrapped samples on the basis of our database of observed patient hot flash activity on placebo versus an agent that has been declared active, with an average reduction in hot flash activity of greater than 50% (venlafaxine 75 mg/d, which had an associated 61% hot flash score reduction). The figure portrays two separate populations with no overlap between the two treatment groups after having simulated for 500 trials. The virtually identical results seen in the three identical treatment arms of approximately 50 patients each, represented in Fig 2, also support the theory that 50 patients per arm are appropriate.
How Many Patients Are Needed for Pilot Studies? When we started performing clinical trials that evaluated different therapies for hot flashes, we considered that, given the marked placebo effect, randomized placebo-controlled trials were necessary, as opposed to pilot studies. Although we still believe that was a reasonable assertion to make at the time, we now believe that the available data coupled with the methodology we have been using allows for appropriate nonrandomized pilot studies to be done to screen potential hot flash therapies. We illustrate this with information from our pilot trial of venlafaxine.4 Data from various numbers of patients (starting with five patients) are illustrated in Fig 9. This figure suggests that results for the study end point are apparent on this clinical trial with a total of 10 patients, with similar results to the results we eventually saw when there were 45 patients on this clinical trial.
Figure 10 provides further information to help evaluate how many patients will be sufficient to test the efficacy of a promising agent in a single-arm pilot study. This figure indicates the differences, on the basis of variable patient numbers, from the final single-arm results for various agents tested in our previous seven trials. This figure demonstrates that 25 to 30 patients provide a reasonably close estimate to the final results, which suggests that this is an appropriate patient number for pilot trials.
We carried out formal power analyses for pilot studies using data drawn from the seven NCCTG studies for parametric assumptions. First of all, we know that a placebo will reduce hot flash activity by approximately 25% (ie, to 75% of baseline) and that the SD of hot flash percentage reduction is 25%. Given that we now have nonhormonal agents that reduce hot flashes by more than 50%, it seems reasonable that we would be most interested in identifying agents that have a capability of reducing hot flashes by 50% or more. With a hot flash score reduction of 25% in 25 patients, the chances are small (0.001%) that the agent in question would truly have a hot flash reduction capability of  50%. Figure 11 illustrates, in both tabular and graphical format, the probabilities of different hot flash reductions seen in 25 pilot patients, associated with a true hot flash reduction potential of at least 50%. This information suggests that we would want at least a 45% reduction in hot flash activity among 25 patients to further study a particular drug. Given the data available regarding sample sizes of 25, treatments with a hot flash score reduction of less than 40% would be consistent with a placebo effect. Patients with a hot flash score reduction of greater than 50% should be considered an indication that the reduction in hot flash activity is greater than would be expected with a placebo. Agents with intermediate results, ie, with hot flash score reductions between 40% and 50%, may be active agents against hot flashes, but the extent of the effect would likely be marginal at best.
The goal of discovering and testing new agents for the reduction of hot flash activity has proven to be an achievable one. More than a decade ago, we started without validated measurement instrumentation or existing analytic methodology. After seven trials involving close to 1,000 patients, much has been learned. Our measurement approach is valid. Patients can not only accurately report the level of hot flash activity, they do it in a remarkably complete and consistent manner. The wealth of data gained from the running of these trials has taught us about required sample sizes and effect sizes needed to declare an experimental agent promising or efficacious.
The appendices are available online at www.jco.org.
APPENDIX A Patient-Reported Hot Flash Diary
APPENDIX B Patient-Reported Hot Flash Definitions: Female Reprinted by permission of Mayo Foundation. 
APPENDIX C Patient-Reported Hot Flash Definitions: Male Reprinted by permission of Mayo Foundation.
1. Feldman BM, Voda A, Gronseth E: The prevalence of hot flash and associated variables among perimenopausal women. Res Nurs Health 8: 261-268, 1985[Medline] 2. Carpenter JS, Andrykowski MA, Freedman RR, et al: Feasibility and psychometrics of an ambulatory hot flash monitoring device. Menopause 6: 209-215, 1999[Medline] 3. Carpenter JS, Andrykowski MA: Menopausal symptoms in breast cancer survivors. Oncol Nurs Forum 26: 1311-1317, 1999[Medline] 4. Loprinzi CL, Pisansky TM, Fonseca R, et al: Pilot evaluation of venlafaxine hydrochloride for the therapy of hot flashes in cancer survivors. J Clin Oncol 16: 2377-2381, 1998[Abstract] 5. Loprinzi CL, Kugler JW, Sloan JA, et al: Venlafaxine in management of hot flashes in survivors of breast cancer: A randomized controlled trial. Lancet 356: 2059-2063, 2000[Medline]
6.
Stearns V, Isaacs C, Rowland J, et al: A pilot trial assessing the efficacy of paroxetine hydrochloride (Paxil) in controlling hot flashes in breast cancer survivors. Ann Oncol 11: 17-22, 2000 7. Goldberg RM, Loprinzi CL, O’Fallon J, et al: Transdermal clonidine for ameliorating tamoxifen-induced hot flashes. J Clin Oncol 12: 155-158, 1994[Abstract] 8. Cleeland CS, Syrjala KL: How to assess cancer pain, in Turk DC, Melzack R (eds): Handbook of Pain Assessment. New York, NY, The Guilford Press, 1992, pp 360-387 9. Cleeland CS, Mendoza TR, Wang XS, et al: Assessing symptom distress in cancer patients: The M.D. Anderson Symptom Inventory. Cancer 89: 1634-1646, 2000[Medline] 10. de Wit R, van Dam F, Hanneman M, et al: Evaluation of the use of a pain diary in chronic cancer pain patients at home. Pain 79: 89-99, 1999[Medline] 11. Groutz A, Blaivas JG, Chaikin DC, et al: Noninvasive outcome measures of urinary incontinence and lower urinary tract symptoms: A multicenter study of micturition diary and pad tests. J Urol 164: 698-701, 2000[Medline] 12. Mendoza TR, Wang XS, Cleeland CS, et al: The rapid assessment of fatigue severity in cancer patients: Use of the brief fatigue inventory. Cancer 85: 1186-1196, 1999[Medline] 13. Maunsell E, Allard P, Dorval M, et al: A brief pain diary for ambulatory patients with advanced cancer: Acceptability and validity. Cancer 88: 2387-2397, 2000[Medline] 14. Sherliker L, Steptoe A: Coping with new treatments for cancer: A feasibility study of daily diary measures. Patient Educ Couns 40: 11-19, 2000[Medline] 15. Richardson A: The health diary: An examination of its use as a data collection method. J Adv Nurs 19: 782-791, 1994[Medline] 16. Lippa R, Donaldson SI: Self-monitoring and idiographic measures of behavioral variability across interpersonal relationships. J Pers 58: 465-479, 1990
17.
Ross MM, Rideout EM, Carson MM: The use of the diary as a data collection technique. West J Nurs Res 16: 414-425, 1994 18. Haythornthwaite JA, Hegel MT, Kerns RD: Development of a sleep diary for chronic pain patients. J Pain Symptom Manage 6: 65-72, 1991[Medline] 19. Smith SL: Physical exercise as an oncology nursing intervention to enhance quality of life. Oncol Nurs Forum 23: 771-778, 1996[Medline] 20. Edwards JE, McQuay HJ, Moore RA, et al: Reporting of adverse effects in clinical trials should be improved: Lessons from acute postoperative pain. J Pain Symptom Manage 18: 427-437, 1999[Medline] 21. Barton D, Loprinzi CL, Quella SK, et al: Prospective evaluation of vitamin E for hot flashes in breast cancer survivors. J Clin Oncol 16: 495-500, 1998[Abstract] 22. Finck G, Barton DL, Loprinzi CL, et al: Definitions of hot flashes in breast cancer survivors. J Pain Symptom Manage 16: 327-333, 1998[Medline] 23. Quella S, Loprinzi CL, Dose AM: A qualitative approach to defining "hot flashes" in men. Urol Nurs 14: 155-158, 1994[Medline] 24. Carpenter JS, Andrykowski MA, Sloan P, et al: Post-mastectomy pain: Prevalence, characteristics, correlates, and relation to quality of life. J Clin Epidemiol 51: 1285-1292, 1998[Medline] 25. Swartzman LC, Edelberg R, Kemmann E: The menopausal hot flush: Symptom reports and concomitant physiological changes. J Behav Med 13: 15-30, 1990[Medline] 26. Sloan JA, O’Fallon JR, Suman VJ, et al: Incorporating quality-of-life measurement in oncology clinical trials. Proc Am Stat Assoc : 282-287, 1998 27. Pandya KJ, Raubertas RF, Flynn PJ, et al: Oral clonidine in postmenopausal patients with breast cancer experiencing tamoxifen-induced hot flashes: A University of Rochester Cancer Center Community Clinical Oncology Program study. Ann Intern Med 32: 788-793, 2000 28. McCorkle R, Benoliel JQ: Symptom distress, current concerns, and mood disturbances after diagnosis of life threatening disease. Soc Sci Med 17: 431-438, 1983 29. Donovan MI: Nursing assessment of cancer pain. Semin Oncol Nurs 1: 109-115, 1985[Medline] 30. Pollack MM, Patel KM, Ruttimann U, et al: Frequency of variable measurement in 16 pediatric intensive care units: Influence on accuracy and potential for bias in severity of illness assessment. Crit Care Med 24: 74-77, 1996[Medline] 31. Martin J, Elliot D: Creating an overall measure of severity of disability for the Office of Population Censuses and Surveys Disability Survey. J R Stat Soc A 155: 121-140, 1992 32. Lancar R, Kramar A, Haie-Meder C: Non-parametric methods for analysing recurrent complications of varying severity. Stat Med 14: 2701-2712, 1995.[Medline]
33.
Loprinzi CL, Michalak JC, Quella SK, et al: Megestrol acetate for the prevention of hot flashes. N Engl J Med 331: 347-352, 1994 34. Nunnally JC, Berenstein IH: Psychometric Theory (ed 3). New York, NY, McGraw-Hill, 1994 35. Loprinzi CL, Goldberg RM, O’Fallon JR, et al: Transdermal clonidine for ameliorating post-orchiectomy hot flashes. J Urol 151: 634-636, 1994[Medline]
36.
Quella SK, Loprinzi CL, Barton D, et al: Evaluation of soy phytoestrogens for treatment of hot flashes in breast cancer survivors: A North Central Cancer Treatment Group trial. J Clin Oncol 18: 412-420, 2000 37. Cohen J: Statistical power analysis for the behavioral sciences. Hillsdale, NJ, Lawrence Erlbaum Associates, 1988 Submitted April 16, 2001; accepted July 10, 2001.
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2001 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
22% variability in an individual baseline day compared with the baseline week. For this reason, when sample sizes of 
