Hoosieroncologygroup.org

F l u o x e t i n e V e r s u s P l a c e b o i n A d v a n c e d C a n c e r O u t p a t i e n t s : A
D o u b l e - B l i n d e d T r i a l o f t h e H o o s i e r O n c o l o g y G r o u p
By Michael J. Fisch, Patrick J. Loehrer, Jean Kristeller, Steven Passik, Sin-Ho Jung, Jianzhao Shen, Matthew A. Arquette, Purpose: To determine whether fluoxetine improves
modeling revealed that patients treated with fluoxetine
overall quality of life (QOL) in advanced cancer patients
exhibited a significant improvement in QOL as shown by the
with symptoms of depression revealed by a simple survey.
Functional Assessment of Cancer Therapy–General, com-
Patients and Methods: One hundred sixty-three patients
pared with patients given placebo (P ؍ .01). Specifically, the
with an advanced solid tumor and expected survival be-
level of depressive symptoms expressed was lower in pa-
tween 3 and 24 months were randomly assigned in a
tients treated with fluoxetine (P ؍ .0005), and the subgroup
double-blinded fashion to receive either fluoxetine (20 mg
of patients showing higher levels of depressive symptoms
daily) or placebo for 12 weeks. Patients were screened for
on the two-question screening survey were the most likely
at least minimal depressive symptoms and assessed every 3
to benefit from treatment.
to 6 weeks for QOL and depression. Patients with recent
Conclusion: In this mix of patients with advanced cancer
exposure to antidepressants were excluded.
who had symptoms of depression as determined by a
Results: The groups were comparable at baseline in
two-question bedside survey, use of fluoxetine was well
terms of age, sex, disease distribution, performance status,
tolerated, overall QOL was improved, and depressive
and level of depressive symptoms. One hundred twenty-
symptoms were reduced.
nine patients (79%) completed at least one follow-up as-
J Clin Oncol 21:1937-1943. 2003 by American
sessment. Analysis using generalized estimating equation
Society of Clinical Oncology.
THE CARE of patients with advanced cancer is becoming outpatients.6 An 11-item version of the ZSDS, the Brief Zung increasingly challenging because of the growing numbers Self-Rating Depression Scale (BZSDS), has been found to have of patients living with cancer and the increasing expectations of acceptable levels of criterion validity.7 However, several studies patients and their families for effective palliative care. According have demonstrated the utility of one- or two-item questionnaires to a recent report from the Institute of Medicine, “A major in both primary care and palliative care settings.8,9 We have problem in palliative care is the underrecognition, underdiagno- piloted the use of a two-item questionnaire that assesses the sis, and thus undertreatment of patients with significant distress cardinal symptoms of depressed mood and anhedonia in ad- ranging from existential anguish to anxiety and depression.”1 vanced cancer outpatients and have found it easy to administer Significant depressive symptoms occur in roughly 25% to 35% of cancer patients.2,3 However, there are several inherent For patients who are treated for the syndrome of major difficulties in diagnosing depression in this population. The most depression, 50% to 60% respond to initial therapy with antide- obvious problem is that sadness and grief are normal responses pressants, psychotherapy, or both.11 Compared with placebo, to the changes associated with the diagnosis of cancer and attransitional points in the disease. In addition, the physical signsof depression may be caused by the malignancy or by medica-tions commonly used for cancer patients. Furthermore, patients From the Department of Palliative Care and Rehabilitation, University of and their family caregivers often do not recognize or accept the Texas M.D. Anderson Cancer Center, Houston, TX; Division of Hematology/ Oncology, Indiana University; Walther Cancer Institute, and Division ofBiostatistics, Indiana University School of Medicine, Indianapolis; Depart- A symptom-based approach is commonly used for the man- ment of Psychology, Indiana State University, Terre Haute, IN; Department agement of phenomena such as pain and nausea, as well as of Palliative Care, University of Kentucky, Lexington, KY; Department of psychologic distress in cancer patients. This involves ascertain- Biostatistics and Bioinformatics, Duke University, Durham, NC; and Divi- ing whether the symptom is present, elucidating further details, sion of Medical Oncology, Washington University, St Louis, MO. and deciding on the proper management approach for the patient.
Submitted August 2, 2002; accepted February 20, 2003.
Supported in part by the Mary Margaret Walther Program for Cancer In addition, there are no universally accepted criteria for diag- Care Research, Indianapolis, IN. Fluoxetine, placebo, and the study note- nosing depression in the terminally ill.4 Multiple depression books were provided by the Eli Lilly Company, Indianapolis, IN. questionnaires have been developed for the primary-care popu- Presented in part at the Thirty-Seventh Annual Meeting of the American lation, with instruments ranging in length from one to 30 items Society of Clinical Oncology, San Francisco, CA, May 12-15, 2001. and ranging in administration times from less than 1 to 5 Address reprint requests to Michael J. Fisch, MD, MPH, University of Texas M.D. Anderson Cancer Center, Box 008, Room P12.2911, 1515 minutes.5 In the largest survey of depressive symptoms in Holcombe Blvd, Houston, TX 77030-4009; email: [email protected]. ambulatory cancer patients, the 20-item Zung Self-Rating De- 2003 by American Society of Clinical Oncology. pression Scale (ZSDS) was administered to more than 1,000 Journal of Clinical Oncology, Vol 21, No 10 (May 15), 2003: pp 1937-1943 Information downloaded from jco.ascopubs.org and provided by Indiana University School of Medicine Libraries on July 23, Copyright 2003 by the American Society of Clinical Oncology. All rights reserved. patients with subsyndromal depression or dysthymia also benefit epines) within 6 weeks of the baseline study evaluation; uncontrolled brain or from treatment with an antidepressant or psychotherapy with leptomeningeal disease; bilirubin more than 2.0 mg/dL; current use of amonoamine oxidase inhibitor; enrollment onto another clinical trial with QOL as similar response rates.12 Fluoxetine is a familiar antidepressant the primary end point; recent or active substance abuse; and major depression to many oncologists, and it is the first and most widely studied diagnosed by a psychiatrist in the past 6 months. Informed consent was obtained selective serotonin-reuptake inhibitor. Uncontrolled trials pub- for each patient with appropriate institutional review.
lished in the late 1980s and 1990s indicated that antidepressantsmight also be helpful in selected cancer patients.13-18 Only six published, randomized, placebo-controlled trials have compared Before random assignment to a treatment group, patients were stratified on the an antidepressant drug (a tricyclic antidepressant or serotonin- basis of Eastern Cooperative Oncology Group performance status (0 to 1 v Ն 2).
Patients were then randomly assigned in a double-blind manner to receive either reuptake inhibitor) with placebo for the treatment of depression fluoxetine (20-mg tablets) or an identical placebo tablet. The randomization was in patients with cancer.18-23 The trend in these data shows a performed centrally through a preprinted randomization table, and the study drug modest benefit of an antidepressant compared with placebo, but was sent by overnight mail directly to the patient. The study drug was there is concern about the generalizability of the data because of self-administered by the patient once daily in the morning.
patient dropout and the relative preponderance of women with Longitudinal assessments were performed at baseline and every 3 to 6 weeks thereafter and included the measurement of QOL and depression. The breast or gynecologic malignancies in these studies.
visit interval varied among patients and often depended on the schedule for It is clear that patients with depressive disorders benefit from anticancer therapy. Patients were assessed for 12 weeks, and complete treatment, but it is also evident that there are major barriers to assessment involved three to five sessions of data collection (depending on diagnosing depressive disorders in outpatients with advanced the individual patient’s visit intervals). After the 12-week study period, the cancer. The purpose of this study was to explore the efficacy and patients were given the option to continue the study drug (blinded) for up to9 more months. The flow of study participants is summarized in Fig 1.
feasibility of treating outpatients with advanced cancer with anantidepressant on the basis of the presence of at least minimal depressive symptoms. A randomized, placebo-control design QOL was measured using a reliable and valid 28-item self-report instru- was needed to obtain an accurate assessment of efficacy. The ment called the Functional Assessment of Cancer Therapy–General primary objective was to compare the change in quality of life (FACT-G, version 3).24 Spiritual well-being was measured using a related (QOL) of these patients; the secondary objective was to compare instrument, the Functional Assessment of Chronic Illness Therapy–Spiritual, with 12 items25 (data for the latter will be reported separately). Depressivesymptoms were measured using the 11-item BZSDS.6 Higher scores repre- sent better QOL, better spiritual well-being, and more depressive symptoms(worse health) in these assessment tools, respectively. These assessmenttools were chosen because they can easily be used in outpatient cancer-care settings and because of their face validity.
Between July 1998 and October 2000, 163 adult ambulatory patients with advanced, incurable malignancy were enrolled from one of 15 sites of theHoosier Oncology Group (three academic centers and 12 community sites).
The primary end point of the study was overall QOL as determined by the Patients with an expected survival between 3 and 24 months were eligible for FACT-G. The generalized estimating equation (GEE) method of regression study participation. Expected survival was judged by clinicians on the basis of was used because it is useful for analyzing data with variable measurement the histology, stage, and comorbid status of the patient and any other prognostic times and when there is a high probability of missing data. The GEE method data typically used by the individual clinician (physician or nurse). Clinician- relates predictors to the mean response variable (as in standard regression), predicted survival was 3 to 12 months for 59% of the patients.
but it does not require that the distribution of the response variable or the During the study period, all participating Hoosier Oncology Group correlation structure among repeated measurements be specified.26 Fisher’s clinicians made available to each outpatient a two-item screening survey to exact test was used to analyze differences in the categorical variables assess depressed mood and anhedonia. These data were used for clinical between the study groups. Continuous variables and ordinal baseline vari- purposes and were not limited to patients who were possible candidates for ables were compared between treatment groups using Wilcoxon rank sum this study. The questions were as follows: (1) “During the past month, have tests and independent sample t tests. All hypothesis testing was carried out you often been bothered by feeling down, depressed, or hopeless?” and (2) using a two-sided alternative hypothesis and a 5% type I error rate.
“During the past month, have you often been bothered by having little An additional analysis planned a priori involved a comparison of the interest or pleasure in doing things?” To be eligible for this study, each best-change score between the patient groups. The best-change score was patient had to have a score of 2 or greater on this Two-Question Screening defined as the difference between the baseline FACT-G score and the Survey (TQSS). Each of the items (depressed mood and anhedonia) in the average of the best consecutive scores. The use of summary measures, such survey has five possible responses that were assigned values of 0 to 4 as as the best change score, has been described as a relevant and statistically follows: 0, not at all; 1, a little bit; 2, somewhat; 3, quite a bit; and 4, very valid way to evaluate longitudinal data,27 and it has been used to analyze much. The score on the survey was the sum of the two questions. As it QOL data in clinical cancer research.28 We defined a best-change score of 6 pertains to this trial, the TQSS was not being used as a casefinding points or greater as showing a response. This 6-point change was chosen as instrument; rather, it was being used to eliminate from consideration the response threshold because a 6-point change in the FACT-G is roughly enrollment of patients who have no problems at all with depressed mood or equivalent to a one-level change in the Eastern Cooperative Oncology Group anhedonia. This was an ethical issue raised by study reviewers because these performance status and because, in similar patient populations, it represents asymptomatic patients were believed to have a very low possibility of an effect size of 0.4 to 0.5, which is generally considered a small-to-moderate benefiting from treatment with an antidepressant.
effect.24 For example, a patient with a baseline FACT-G score of 90 who had Patients could be receiving any anticancer therapy or only the best supportive a follow-up score of 100 and then was lost to follow-up would be considered care. Exclusion criteria included the following: serious suicidal risk or psychotic a nonresponder with a best-change score of 5 points. If this same patient had behaviors; inability to swallow oral medications; regular use of antidepressants one more follow-up with a score of 94, the best-change score of 7 points or psychotropic drugs (other than phenothiazine-type antiemetics or benzodiaz- would have been sufficient for the patient to be considered a responder.
Information downloaded from jco.ascopubs.org and provided by Indiana University School of Medicine Libraries on July 23, Copyright 2003 by the American Society of Clinical Oncology. All rights reserved. FLUOXETINE VERSUS PLACEBO IN CANCER PATIENTS Fig 1. Study flow.
Patients in each study arm were comparable at baseline with We estimated that the SD of the FACT-G in advanced cancer patients respect to age, sex, performance status, symptom status regarding would be approximately 15 points. Assuming a correlation of 0.5 between pain and depression, disease distribution, and current treatment with the baseline and 3-month measures and adjusting for the stratification chemotherapy. These patient characteristics are listed in Table 2. A variable of performance status, we estimated that 80 patients per study arm family history of depression in first-degree relatives and current or would provide 82.7% power to detect a 6-point difference in the total score prior psychologic counseling (individual or group) since their between arms. This estimate was deliberately conservative because of ourintent to analyze these data primarily using either a model with patient-level diagnosis was more frequent in patients randomly assigned to random effects or another appropriate regression technique (such as GEE).
receive placebo. The mean interval between visits (in weeks) wassimilar for patients in each treatment arm. The patients’ cancer treatments are listed in Table 3. There were significantly more The baseline questionnaires were completed by 159 patients fluoxetine patients currently receiving radiation therapy, and more (98%), and at least one follow-up assessment was completed by placebo patients had undergone surgery.
129 patients (81%). The baseline characteristics of the 30 Patient compliance to the study drug was not assessed with patients who did not complete at least one follow-up are listed in pill counts but by patient self-report at follow-up. Six patients Table 1. Patients who completed at least one follow-up assess- discontinued the study drug (3.7%). Four patients dropped out of ment were assessable for the primary outcome variable of the fluoxetine arm (two because of daily headaches and two because of nausea or vomiting), and two patients dropped out of Table 1. Baseline Assessments of the Assessable and Inassessable Patients
NOTE. t tests are used to test for differences of means.
Abbreviations: TQSS, Two-Question Screening Survey; ECOG PS, Eastern Cooperative Oncology Group performance status; FACT-G, Functional Assessment of Cancer Therapy–General; BZSDS, Brief Zung Self-Rating Depression Scale.
*Missing baseline TQSS data were the result of a systematic error, but 75% of the data were retrieved by subsequent †Higher scores represent better quality of life.
‡Higher scores represent more severe depression.
Information downloaded from jco.ascopubs.org and provided by Indiana University School of Medicine Libraries on July 23, Copyright 2003 by the American Society of Clinical Oncology. All rights reserved. Table 2. Patient Characteristics
Table 4. Longitudinal Data on Quality of Life and Depressive Symptoms
Abbreviations: FACT-G, Functional Assessment of Cancer Therapy–General; BZSDS, Brief Zung Self-Rating Depression Scale.
*Numbers in parentheses represent the number of patients in the fluoxetine group the study completion visit, compared with two (4.6%) of 43 The longitudinal data regarding QOL are listed in Table 4. The pattern of missing data was similar in the two groups. There were no significant differences in the best-change scores for the FACT-G or BZSDS between the groups or in the proportion of responders whose best-change score exceeded the 6-point threshold for the FACT-G (Table 5). Moreover, the FACT-Gsubscales (functional, social, emotional, and physical well-being) showed similar best-change score findings, as summa- the placebo arm (one because of nonspecific side effects and one because of nausea or vomiting). Fifteen patients had unexpected There was a statistically significant improvement in the total hospitalizations during the study; nine in the fluoxetine arm and FACT-G scores in the fluoxetine group, compared with the six in the placebo arm (P ϭ .59 by two-sided Fisher’s exact test).
placebo group, using the GEE method of regression (P ϭ .05).
To assess a common side effect of fluoxetine, clinicians were This difference remained highly significant after multivariate asked to report at each follow-up the number of times the patient adjustments were made for age, sex, the two-question screening vomited in the past week. There was no difference in the score, performance status, family history of depression, and frequency of emesis for the second, third, and fourth visits, but current cancer therapy (P ϭ .01). The fluoxetine group also the frequency was significantly greater in patients in the fluox- improved significantly, compared with placebo, on the depres- etine arm at the study completion visit; nine (33%) of 27 patients sion scale (P ϭ .0005). The full GEE model characteristics for in the fluoxetine arm reported one or more episodes of emesis at depression and QOL are listed in Table 7. In both the QOL and Table 3. Cancer Treatment Exposure of the Study Population
Information downloaded from jco.ascopubs.org and provided by Indiana University School of Medicine Libraries on July 23, Copyright 2003 by the American Society of Clinical Oncology. All rights reserved. FLUOXETINE VERSUS PLACEBO IN CANCER PATIENTS Table 5. Best-Change Scores for Quality of Life and Depression
of the GEE model, but there were no significant differences in the mean best-change scores for any of the subscales.
The TQSS score was less than 4 in 32% of the patients. In this subgroup, the GEE model showed no treatment effect on QOL (P ϭ .19) or depression (P ϭ .35). However, the treatment effect was more pronounced in the subset of patients whose TQSS score was 4 or greater, with significant improvement in QOL ϭ .05) and depression (P ϭ .0008). This subgroup analysis on the basis of the TQSS was unplanned and is considered explor- atory. The conclusions were the same when the subset analysis was performed using the baseline BZSDS data dichotomized by those patients scoring in the normal range (0 to 21; 64 patients) versus patients with mild or moderate depression at baseline The survival of the treatment groups is illustrated using the NOTE. Higher scores represent better health for the FACT-G, but lower scores (or more negative best-change scores) represent better health (less depressive symptoms) Kaplan-Meier method in Fig 2. An eligibility requirement for this study was a projected survival between 3 and 24 months.
Abbreviations: FACT-G, Functional Assessment of Cancer Therapy–General; Sixty-four percent of patients (102 of 160 patients) had a BZSDS, Brief Zung Self-Rating Depression Scale.
survival in that projected range. Twenty-three percent of patients *The best-change score refers to the difference between the baseline total score (37 of 160 patients) survived fewer than 3 months, and three and the average of the best consecutive scores on the respective instruments (FACT-Gfor quality of life or BZSDS for depression).
patients were lost to follow-up for survival. The median survival of the patients is listed according to the category of clinician- ‡Responders were those patients whose best-change score was Ն 6 points for the This study is the first to explore an oncologist-driven, symp- tom-management solution to the prevalent problem of depressive depression model, female sex was associated with significantly symptoms in advanced cancer patients. To our knowledge, it is better outcomes (P ϭ .004 for QOL and P ϭ .04 for depression).
also the largest study examining the placebo response rate Moreover, the TQSS was a significant predictor of overall QOL associated with the FACT-G instrument for QOL assessment of and depression (P ϭ .001 and P ϭ .01, respectively), having a outpatients with advanced cancer. Our findings show that simple stronger association with both outcomes (QOL and depression) symptom screening followed by antidepressant treatment is both than did performance status. Patients with no family history of feasible and associated with improvement in overall QOL and depression had less depression (P ϭ .02). Two subscales of the FACT-G, social well-being and emotional well-being, showed The overall benefit to QOL in the fluoxetine arm, although statistically significant improvement with fluoxetine on the basis statistically significant, is of uncertain clinical significance be- Table 6. FACT-G Subscales: Best-Change Scores and GEE Results
Abbreviations: FACT-G, Functional Assessment of Cancer Therapy–General; GEE, generalized estimating equations; FWB, functional well-being; SWB, social well-being; EWB, emotional well-being; PWB, physicalwell-being.
*The results are obtained from GEE models for FACT-G subscales with adjustment of a set of covariates. Slope estimates indicate the outcome improvement in the fluoxetine group with the placebo group as a reference.
†P is for test of significance of differences of means.
Information downloaded from jco.ascopubs.org and provided by Indiana University School of Medicine Libraries on July 23, Copyright 2003 by the American Society of Clinical Oncology. All rights reserved. Table 7. Generalized Estimating Equations for the FACT-G and BZSDS
NOTE. Age, TQSS, and ECOG performance status were considered continuous variables in this model.
Abbreviations: FACT-G, Functional Assessment of Cancer Therapy–General; BZSDS, Brief Zung Self-Rating Depression Scale; CL, confidence limits; TQSS, Two-Question Screening Survey; ECOG PS, Eastern Cooperative Oncology Group Performance Status.
cause the proportion of patients with a 6-point difference in their kind of clinical problem. Somerfield et al30 have argued persua- best-change score, indicating clinically important improvement, sively that a more reasonable expectation of an intervention is did not differ between the treatment arms. Determining the that it would be effective at the more proximal target problem (in clinical significance of changes in self-reported health measures this case depression) rather than a more distal, global outcome measured longitudinally is difficult, and this is an area of active such as overall QOL. This was indeed borne out in our study, in research and debate.29 The best-change score represented an which there was a more convincing proximal effect of the attempt to provide a very conservative summary of change of antidepressant (decrease in the depressive symptoms) and a longitudinal data that was both meaningful and simple to statistically significant but less convincing distal effect. It seems calculate. By taking the difference between the baseline measure that the depressive symptoms did not abate as the result of and the average of the best consecutive measures, this method adaptation and regression to the mean but, instead, as a result of gives greater weight to any sustained improvements and dis- a real effect from the antidepressant. In addition, for those who counts the treatment effect when there are missing data. If there might speculate that use of an antidepressant might cause the had been a significant difference between the treatment arms average patient worse problems than the depression itself be- with this best-change score method, the clinical significance of cause of polypharmacy, drug interactions, or antidepressant side the statistical findings would have been convincing, but it is effects, there are reassuring data that the overall QOL was clearly more difficult to interpret a no-difference result. There- fore, more research is needed to validate this method.
Just as overall QOL might be regarded as a distal outcome for In addition to the controversy surrounding the definition of an antidepressant trial, survival would be even more distal.
clinically significant changes, there is also controversy regarding Survival was measured in this trial as a descriptive end point to the appropriate use and interpretation of QOL end points in better understand this patient population and its generalizability studies in which the intervention is expected to affect a specific to other groups of patients. As expected, there was no differencein survival between the treatment arms.
Since United States Food and Drug Administration approval of fluoxetine in 1987, there has been a national trend toward greateruse of psychotropic drugs and greater involvement of physicians inthe management of depression in outpatients.31 Thus, our data maybroaden the comfort zone of oncologists for prescribing antidepres-sants for some patients. A particular factor favoring the use ofantidepressants in patients with advanced cancer is that access tobehavioral health professionals and psychologic interventions is notalways feasible for very sick patients. It is also important that theTQSS proved to be quite easy to adapt to clinical practice as an Table 8. Comparison of Clinician-Expected Survival With Estimated Survival
Fig 2. Estimated survival. The overall median survival of this patient population
was 7.77 months. The median survival in the fluoxetine group was 6.26 months
compared with 9.39 months in the placebo group (P ؍ .50 by the log-rank test).
Information downloaded from jco.ascopubs.org and provided by Indiana University School of Medicine Libraries on July 23, Copyright 2003 by the American Society of Clinical Oncology. All rights reserved. FLUOXETINE VERSUS PLACEBO IN CANCER PATIENTS initial assessment of psychologic distress. The exploratory analysis cancer outpatient research, and it was no worse in this study than revealed that the subset of patients with mild or moderate depressive in other similar studies. Nevertheless, nonrandom reasons for symptoms (on the basis of the baseline TQSS or the baseline missing data introduce possible bias into this study. Although BZSDS) benefited the most from the fluoxetine intervention.
new methods for the analysis of studies with missing data such Therefore, the TQSS, as an example of simple bedside screening, is as this study are being explored, more important innovations for worthy of additional research as a predictor of response to antide- preventing missing data also are being pursued. Better funding pressant therapy in this population.
for symptom research projects and use of telephone assessments Selection bias is one of several important limitations to this may lessen the missing data problem in future research.
study. These 163 patients were enrolled over a period of 30 In summary, this study demonstrated that simple two-question months and represent a small fraction of the total eligible screening for depressive symptoms followed by treatment with a patients. The pace of accrual was affected by the physicians’ lack once-a-day pill proved effective in identifying and treating of familiarity with placebo-controlled symptom research, and depressive symptoms and, hence, improving QOL in patients there was a substantial learning curve in this regard. More important, collecting questionnaire data using existing officestaff is cumbersome and was not always feasible on a day-to-day basis. This group’s willingness to pursue this research is a credit We thank the nurses and physicians of the Hoosier Oncology Group and to oncologists and their drive to learn more about their patients’ study design consultants Ian Tannock, PhD, and Jimmie Holland, MD, and psychologic distress. Another important limitation was the sub- we acknowledge the generosity and courage of the patients who agreed to stantial missing data. This is an inherent problem in advanced 1. Foley KM, Gelband H (eds): Improving Palliative Care for Cancer: 17. Holland J, Romano S, Heiligenstein J, et al: A controlled trial of Summary and Recommendations. Washington, DC, National Academy fluoxetine and desipramine in depressed women with advanced cancer.
2. Massie MJ, Popkin MK: Depressive disorders, in Holland JC (ed): 18. Razavi D, Allilaire J, Smith M, et al: The effect of fluoxetine on Psycho-Oncology. New York, NY, Oxford University Press, 1998, 518-540 anxiety and depression symptoms in cancer patients. Acta Psychiatr Scand 3. Wilson KG, Chochinov HM, de Faye B, et al: Diagnosis and manage- ment of depression in palliative care, in Chochinov HM, Breitbart W (eds): 19. van Heeringen K, Zivkov M: Pharmacological treatment of depression Handbook of Psychiatry in Palliative Care. Oxford, United Kingdom, Oxford in cancer patients: A placebo-controlled study of mianserin. Br J Psychiatry 4. Lloyd-Williams, M: Screening for depression in palliative care. Am J 20. Eija K, Tiina T, Pertti NJ: Amitriptyline effectively relieves neuro- pathic pain following treatment of breast cancer. Pain 64:293-302, 1996 5. Williams JW Jr, Noel PH, Cordes JA, et al: Is this patient clinically 21. Purohit DR, Navlakha PL, Modi RS, et al: The role antidepressants in depressed? J Am Med Assoc 287:1160-1170, 2002 hospitalised cancer patients: A pilot study. J Assoc Physicians India 6. Passik SD, Dugan W, McDonald MV, et al: Oncologists’ recognition of depression in their patients with cancer. J Clin Oncol 16:1594-1600, 1998 22. Costa D, Mogos I, Toma T: Efficacy and safety of mianserin in the 7. Passik SD, Kirsh K, Donaghy KB, et al: An attempt to employ the Zung treatment of depression of women with cancer. Acta Psychiatr Scand Self-Rating Depression Scale as a “lab test” to trigger follow-up in ambulatory oncology clinics: Criterion validity and detection. J Pain Symp- 23. Musselman DL, Lawson DH, Gumnick JF, et al: Paroxetine for the prevention of depression induced by high-dose interferon alfa. N Engl J Med 8. Whooley M, Avins A, Miranda J, et al: Case-finding instruments for depression: Two questions are as good as many. J Gen Intern Med 24. Cella DF, Tulsky DS, Gray G, et al: The Functional Assessment of Cancer Therapy Scale: Development and validation of the general measure.
9. Chochinov H, Wilson K, Enns M, et al: “Are you depressed?” Screening for depression in the terminally ill. Am J Psychiatry 154:674-676, 1997 25. Brady MJ, Peterman AH, Fitchett G, et al: A case for including 10. Fisch M, Marx R, Brames M, et al: Use of a two-question screening spirituality in quality of life measurement in oncology. Psychooncology survey for depression in outpatients with advanced cancer. Proc Am Soc Clin 26. Liang KY, Zeger SL: Longitudinal data analysis using generalized 11. Whooley MA, Simon GE: Managing depression in medical outpa- 27. Matthews JN, Altman DG, Campbell MJ, et al: Analysis of serial 12. Williams JW Jr, Barrett J, Oxman T, et al: Treatment of dysthymia measurements in medical research. Br Med J 300:230-235, 1990 and minor depression in primary care: A randomized controlled trial in older 28. Tannock IF, Osoba D, Stockler MR, et al: Chemotherapy with adults. J Am Med Assoc 284:1519-1526, 2000 mitoxantrone plus prednisone or prednisone alone for symptomatic hormone- 13. Breitbart W: Psycho-oncology: Depression, anxiety, delirium. Semin resistant prostate cancer: A Canadian randomized trial with palliative end 14. Evans DL, McCartney CF, Haggerty JJ Jr, et al: Treatment of 29. Sloan JA, Cella D, Frost M, et al: Assessing clinical significance in mea- depression in cancer patients is associated with better life adaptation: A pilot suring oncology patient quality of life: Introduction to the symposium, con- tent overview, and definition of terms. Mayo Clin Proc 77:367-370, 2002 15. Massie M, Holland J: Depression and the cancer patient. J Clin 30. Somerfield M, Jatoi A, Nguyen PL, et al: Hazards of quality of life data for clinical decision making. J Clin Oncol 19:594-595, 2001 16. Mermelstein H, Lesko L: Depression in patients with cancer. Psy- 31. Olfson M, Marcus SC, Druss B, et al: National trends in the outpatient treatment of depression. J Am Med Assoc 287:203-209, 2002 Information downloaded from jco.ascopubs.org and provided by Indiana University School of Medicine Libraries on July 23, Copyright 2003 by the American Society of Clinical Oncology. All rights reserved.

Source: http://www.hoosieroncologygroup.org/resources/00005083-200402150-00033.pdf