You are on page 1of 7

Fluoxetine Versus Trazodone in

Depressed Geriatric Patients


William E. Falk, MD; Jerrold F. Rosenbaum, MD; Michael W. Otto, PhD;
Paul M. Zusky, MD; Jeffrey B. Weilburg, MD; Ralph A. Nixon, MD

Abstract .
A total of 27 subjects began active treatment in this double-blind study comparing the efficacy and safety of trazodone and
fluoxetine in geriatric depressed patients, but only 13 completed 6 weeks on study medication. Both agents were effective
according to weekly and endpoint analyses, and there was no evidence of significant effects on blood pressure, pulse, or weight.
Separate analysis of patients who had received an adequate trial of medication indicated a trend toward relatively more
fluoxetine-treated patients meeting clinical criteria for resolved depression. (
J Geriatr Psychiatry Neurol 1989;2:208-214.)

treatment of major depression in the elderly tion, and less postural hypotension. Its low anti-
he is more complicated than it is for younger pa-
tients. When prescribing antidepressants to geriatric
cholinergic profile diminishes the likelihood of cogni-
tive deficits, narrow angle glaucoma, urinary retention,
patients, physicians must take into account age-related constipation, and other uncomfortable or potentially
changes in drug metabolism, altered pharmacokinetic serious side effects. In addition, its relative safety in
and pharmacodynamic effects, and greater susceptibil- overdose is attractive for an antidepressant agent. All of
ity to side effects.’ Furthermore, since elderly patients these factors encourage the use of trazodone in the eld-
frequently require medications for other medical prob- erly, but this agent is not without potential side effects,
lems, the risk of drug interactions is increased. including an increase in preexisting ventricular irritabil-
Cyclic antidepressants vary in the intensity of ity, priapism, sedation, gastrointestinal distress, and
anticholinergic effects., postural hypotension, and di- symptomatic orthostasis.5 As such, the consideration of
rect cardiac effects produced, but all have some degree the relative efficacy of other antidepressant agents for
of these unwanted characteristics. Monoamine oxidase the elderly is relevant.
inhibitors, while low in anticholinergic and cardiac ef- ,
Fluoxetine is one such agent. Fluoxetine is a selec-
fects, are frequently associated with postural hypoten- tive inhibitor of serotonin reuptake that has virtually no
sion and have the potential for severe drug or food affinity for muscarinic, noradrenergic, or dopaminergic
interactions. Despite these drawbacks and risks, both receptors, and although the drug does down-regulate
classes of agents are used effectively in treating elderly 5-HT1 receptors over time, it does not by itself appear
depressed patients. to down-regulate (3-receptors.b These pharmacody-
Among currently available antidepressants, trazo- namic characteristics are of potential importance for the
done is one that is considered to be well tolerated in side effect profile of this medication. Fluoxetine ap-
geriatric depressed patients.2 In studies comparing it to pears to be relatively benign in overdose and does not
imipramine3 and amitriptyline,4 trazodone-treated pa- appear to lower seizure threshold. In addition, fluoxe-
tients had fewer cardiovascular side effects, including tine has not been associated with cardiovascular
no increase in heart rate, little effect on cardiac conduc- changes, postural dizziness, or sedation. Fluoxetine,
like trazodone, is free of anticholinergic effects but is
associated with increased reports of nausea and
Received April 24, 1989. Received revised July 7, 1989. Ac-
cepted for publication Sept 6, 1989. headaches.
From the Department of Psychiatry, Harvard Medical School, Fluoxetine has proven effective in double-blind,
and the Massachusetts General Hospital, Boston, MA.
Address correspondence to Dr William E. Falk, Clinical placebo controlled trials comparing it to imipramine
and amitriptyline.7-9 In a study of nongeriatric de-
Psychopharmacology Unit, ACC 715,Massachusetts General Hospi-
tal, Boston, MA 02114. pressed patients, its efficacy was comparable to that of

208

Downloaded from jgp.sagepub.com at CARLETON UNIV on June 15, 2015


trazodone, although onset of benefit was somewhat (entry) and visit 2 (baseline). From visit 2 onward, all
delayed.10 It was also found to be as beneficial as patients received active drug by random assignment.
doxepin geriatric depressed patients.ll Doses as low
in Each subject was given study medication in three bot-
as 20 mg have been demonstrated to be effective in tles marked for morning, noon, or evening administra-
some patients, while other patients have improved on tion. All capsules were identical and contained either 20
doses of 40 to 80 mg.12 Side effects appear to increase at mg of fluoxetine, 50 mg of trazodone, or placebo. In the
the higher dosages. fluoxetine-treated patients, noon and evening bottles
The clinical efficacy of fluoxetine, as well as its side actually contained placebo, while for the trazodone- -
effect profile, suggests it may be an attractive agent for treated group, morning medication was placebo.
the treatment of depression in geriatric patients, and For the first 3 weeks of active drug treatment, the
raises questions whether it may offer clinical benefits daily dosage regimen could not be adjusted. Starting at
similar to or beyond that of trazodone. For these rea- week 4 of active drug treatment (visit 5), the daily dos-
sons, we undertook a study to compare the safety, effi- age could be adjusted to achieve maximum therapeutic
cacy, and tolerability of fluoxetine and trazodone in response. For fluoxetine, dosage through week 3 was
older depressed patients. kept at 20 mg, during week 4 it could range from 20 to
40 mg, and at weeks 5 and 6 it could range from 20 to
Methods 60 mg. For trazodone, 100 mg. was administered on
Male and female outpatients at least 62 years old were days 1 through 3, 150 mg on days 4 through 7, 200 mg
on days 8 through 10, and 250 mg on days 11 through
invited to participate in a study comparing fluoxetine to
trazodone. To be included, a subject had to meet criteria 21. During weeks 4 through 6 of active drug treatment,
for Diagnostic and Statistical Manual o f Mental Disorders, dosage could be titrated between 50 and 400 mg.
ed. 3 (DSM-III) unipolar major depressive disorder, ei- Weekly assessments included the 21-item HAM-D
ther single episode or recurrent, and to have experi- and patient- and physician-rated Clinical Global Im-
enced the present episode for at least 4 weeks. Subjects provement (CGI),14 7-point scale ranging from
a

also had to achieve a score of at least 20 on the 21-item marked improvement (1) to markedly worse (7), with
Hamilton Rating Scale for Depression (HAM-D).13 no change denoted by 4. Side effects were evaluated
Exclusion criteria were serious medical illness that using the Treatment Emergent Side-effect Scale (TESS).
could be adversely affected by the study drugs, glau- Continuous response variables were analyzed by t-
coma, unstable cardiac arrhythmias, seizure disorder,
tests, and categorical response variables by Fisher exact
tests,15 A significance level of .05 was used for all t-
antihypertensive agents (ie, guanethidine, reserpine, tests. As the Fisher exact tests were one-tailed, a signifi-
clonidine, and methyldopa), and history of allergy to
cance level of .025 was adopted for these tests. Alpha
either drug. Psychiatric exclusions included patients
with severe psychosis, suicidal symptoms, or DSM-III probabilities between .025 and .05 were termed mar-
diagnoses of schizophrenia, schizoaffective illness, bi- ginally significant.
polar disorder, organic mental disorder, substance
abuse disorders within the past year, or paranoid Results
disorders.
Patients were excluded if they took either study Patient Attrition
drug in the month preceding entry, monoamine oxidase Thirty-six patients were entered for the single-blind,
inhibitors in the prior 14 days, or other antidepressants placebo washout phase of the study. By the end of this
at the time of entry. Aside from the use of benzo- phase, nine patients dropped out or were excluded
diazepines or chloral hydrate for sleep, no other from further participation: HAM-D scores decreased to
psychotropic agent could be taken during the study. less than the criteria of 20 for two patients, medical
After providing informed consent, each subject re- problems precluded participation for three patients,
ceived a medical and psychiatric examination, as well and four patients discontinued because of &dquo;side effects,&dquo;
as laboratory testing (complete blood count, chemistry concerns of family members, or unwillingness to re-
profile, urinalysis, and ECG). Physical examination and main free of excluded medications. Of the 27 patients
laboratory testing were repeated at the time of study continuing in the protocol, four men and 10 women
exit. Blood pressure, pulse, and weight were obtained were assigned to fluoxetine treatment, and three men

weekly throughout the course of the study. and 10 women were assigned to trazodone treatment.
The study duration was 7 weeks, during which There was no significant difference in age between the
there were eight weekly visits. All patients were given treatment groups: 69.1 ± 6.2 years for the fluoxetine-
placebo capsules during the first week between visit 1 treated patients and 67.5 ± 4.0 years for the trazodone-

209

Downloaded from jgp.sagepub.com at CARLETON UNIV on June 15, 2015


treated patients. Likewise, there were no significant TABLE 1
differences between treatment groups in age at onset of Reasons for Discontinuation of Medication
first episode, number of episodes, or duration of the

present episode.
One patient from each treatment group failed to re-
turn at visit 3, leaving a total of 25 patients available for
outcome analysis. Of these, five patients on trazodone
and two on fluoxetine dropped out at visit 3 (after one
week of active drug treatment). Five additional patients
dropped out before termination, leaving only 13 pa-
tients completing the entire protocol, three on trazo-
done and ten on fluoxetine (Table 1). The average dose
of medication at visit 8 was 350 mg for trazodone and
48 mg for fluoxetine. The number of completers was
significantly different between the two groups (P =

.012, one-tailed), but the difference in dropouts due to


complaints of adverse effects (five in the trazodone After 1 week of active drug treatment.
group and two in the fluoxetine group) was not signifi-
cant according to Fisher exact test analysis.
Treatment outcome was assessed by HAM-D scores
and physician-rated CGI scores. Mean scores for these
Depression variables are presented in Table 2. Two statistical com-
The high dropout rate found in this sample of elderly parisons were of interest: (1) a within-subjects compari-
patients presents problems for the analysis of between- son of the significance of changes in these depression
treatment differences. Few subjects are left for compari- ratings occurring since the baseline period for the
son at the last treatment visit, leaving little power for groups pooled together, and (2) a between-subjects
the detection of differences in treatment efficacy. comparison of whether these changes were signifi-
Nonetheless, as a high rate of attrition is common to cantly different between the treatment groups. In addi-
studies of geriatric depressives, 16 the use of endpoint tion to endpoint analysis, these within-subjects
analysis (carrying the last visit forward) does provide a comparisons were examined at each week, and
realistic index of the average improvement that a sam- between-subjects comparisons were examined when-
ple of geriatric patients entering treatment may achieve ever the overall sample size was 16 subjects or more (to
over a short-term drug trial. visit 7). Figures 1 and 2 illustrate these comparisons of

TABLE 2
HAM-D and CGI Scores for All Patients at Each Visit and Endpoint

HAM-D =
Hamilton Rating Scale for Depression; CGI =
Clinical Global Improvement.
&dquo;Baseline.
tLast visit carried forward.

210

Downloaded from jgp.sagepub.com at CARLETON UNIV on June 15, 2015


FIGURE 1
Change in Hamilton Rating Scale for Depression scores between baseline and each subse-
quent visit. Numbers within the columns refer to sample size. No between-group differ-
ences were significant. At all visits, the within-subjects change from baseline scores was

significant (P < .001) for the treatment groups pooled together.

FIGURE 2
Change in Clinical Global Improvement scores between baseline and each subsequent visit.
Numbers within the columns refer to sample size. No between-group differences were sig-
nificant. At all visits, the within-subjects change from baseline scores was significant (P <
.01 for visit 3, P < .001 for all other visits) for the treatment groups pooled together.

211

Downloaded from jgp.sagepub.com at CARLETON UNIV on June 15, 2015


change (baseline visit minus treatment visit)
scores more patients in either treatment group) are listed in
across the treatment
period. Table 3. According to Fisher exact test analysis, there
Within-subjects analysis indicated a significant im- was a trend toward relatively more reports of insomnia

provement in depression across treatment as assessed in the fluoxetine-treated group (P .046, one-tailed)
=

by changes in HAM-D scores at endpoint (t [23] = 6.77, and relatively more reports of constipation in the
P < .001) and as assessed individually at weeks 3 trazodone-treated group (P = .039, one-tailed). No
through 8.* No differences between the
significant other side effects approached significance for compari-
treatment groups obtained at endpoint, (t[23] _
were sons between treatment groups. None of the side ef-

1.02, P < .32) or at any point during the study. Parallel fects were so severe as to be medically dangerous to any
results were obtained for the CGI assessment of patient.
depression.
These comparisons, however, do not fully capture a
variable of clear interest to the clinician: the number of Discussion
Using &dquo;most generous&dquo; criteria, Gerson and colleagues
patients in each treatment condition who received an were able to find only 25 controlled studies of antide-
adequate dose and duration of medication and
achieved resolution of depression. To examine this vari- pressant usage in elderly patients between 1964 and
1986.18 They and others have noted that a high attrition
able, resolution of depression was defined -as a greater
than 50% reduction in HAM-D score and a final rate is a common problem in studies of geriatric pa-
HAM-D score less than 10, criteria that have been de- tients. Forexample, in a study by Feighner and Cohn,11
scribed elsewhere. 17 Of the six patients treated with 61 % of doxepin-treated patients did not complete the
trazodone at a dose of 300 mg or more, only one met study, while 47% of fluoxetine-treated patients termi-
these criteria. In contrast, eight of 11 patients treated nated early. Spivak and colleagues reported that all
with 40 mg of fluoxetine met these criteria. The differ- eight elderly depressed patients with medical problems
ence in the treatment groups was marginally significant
whom they treated with trazodone at a dose of 300 mg
or less had to discontinue the drug due to clinically sig-
by a one-tailed Fisher exact test (P = .043). nificant side effects.44
In our study, 75% of trazodone-treated patients did
Side-Effects not complete the study, while 23% of fluoxetine-
Three variables from the physical examination were ex-
amined : postural blood pressure change, pulse, and treated patients terminated early. A principal reason for
our high dropout rate appeared to be the intolerance of
weight change. For data analysis, postural blood pres- even low-intensity side effects, which were not consid-
sure change was defined as the change in systolic blood
ered sufficient reason for discontinuation by the rater.
pressure between supine and standing assessments and
was examined pretreatment, at visit 3 (allowing the
Given our high dropout rate and the apparent preva-
lence of such problems in the study of this population,
comparison of the greatest number of subjects at the we came to appreciate the need for involving signifi-
same time period), and at endpoint (last visit carried

forward). Resting heart rate was also examined at these cant others, suchspouses and children, in the de-
as

times. No significant differences were found between cision to participate in the study and in ongoing
or within subjects for the change in these variables be-
treatment decisions.
tween baseline and visit 3 or between baseline and
TABLE 3
endpoint evaluations. Treatment-Emergent Side Effects Reported by Three or More
To avoid contamination by length of treatment,
Patients on the TESS
weight change was assessed in completers only. The
___

mean weight change was almost identical in each treat-


ment group, -2.7 ± 3.0 pounds for the fluoxetine-
treated and -2.8 ± 1.8 pounds for the trazodone-
treated group. This apparent decrease in weight across
treatment was not
statistically significant.
Complaints of treatment-emergent side effects in-
dicated by the TESS (noted only if reported by three or

*As equal population variances could not be assumed and sam-


ple sizes were unequal, critical values for these within-subject tests
and the corresponding tests of the CGI data were corrected using an TESS = Treatment-Emergent Side-Effect Scale.
adjustment to degrees of freedom as outlined by Hayes.l5 *P < .05 and > .025, one-tailed test.

212

Downloaded from jgp.sagepub.com at CARLETON UNIV on June 15, 2015


In addition to being a general problem for the study is undoubtedly complicated by many factors affecting
of geriatric depression, the high dropout rates raise cau- the older patient, but further studies are essential to de-
tions for the interpretation of the data. As such, we termine optimal strategies for drug treatment in the
view our conclusions as tentative but important, given elderly.&dquo;19 Our findings encourage further investigation
the difficulties with research with this population. of the use of fluoxetine as a useful and safe alternative
_
Both trazodone and f luoxetine were associated to trazodone in depressed geriatric patients.
with clinical improvement in depression as measured
by endpoint analysis and analysis of changes from
>
baseline evident at each visit. Part of this apparent anti- Acknowledgments
depressant effect may be attributed to placebo re- This research was supported by a grant from Eli Lilly and Com-
sponse, as no placebo control group was used. Yet, pany. The authors appreciate the clinical and administrative support
rather than replicate previous examinations of the ab- of Laura Pill, BS, manuscript preparation of Margaret O’Brien, and
solute efficacy of these medications, 2-4, 7-9 this study statistical assistance of Bruce Dornseif, PhD.
was designed to compare the relative efficacy and side
effect profiles of fluoxetine and trazodone in a geriatric
population. Although statistical comparisons were References
hampered by small sample size, examination of mean 1. Greenblatt DJ, Sellars EM, Shader RI: Drug disposition in old
scores at each week of treatment did not indicate that age. N Engl J Med 1982;306:1081-1088.
2. Gerner RH: Geriatric depression and treatment with trazodone.
fluoxetine was differentially effective compared to
trazodone. Separate analysis of patients receiving an Psychopathology 1987;20(Suppl 1) :82-91.
3. Gerner RH, Estabrook W, Stemer J, Jarvik L. Treatment of geriat-
adequate trial of medication indicated a trend toward ric depression with trazodone, imipramine and placebo. J Clin
relatively more patients on fluoxetine than on trazo- Psychiatry 1980;41:216-220.
done meeting a clinical criteria for resolved 4. Ather SA, Ankier SI, Middleton RS: A double-blind evaluation
of trazodone in the treatment of depression in the elderly. BrJ
depression. Clin Pract 1985;39:192-199.
Given this evidence that fluoxetine is at least as ef- 5. Spivak B, Radvan M, Meltzer M: Side effects of trazodone in a
fective as trazodone in this geriatric population, atten- geriatric population.J Clin Psychopharmacol 1985;9:62-63.
tion is turned to the side effect profile for each 6. Stark P, Fulle RW, Wong DT: The pharmacologic profile of
medication. The present study did not indicate cardiac fluoxetine.J Clin Psychiatry 1985;46:7-13.
effects for either medication as determined by physical 7. Bylerly WF, Reimherr FW, Wood DR, Grosser BI: Fluoxetine, a se-
lective serotonin uptake inhibitor, for the treatment of outpa-
examination data, ECG, and from patient reports. Like- tients with major depression. J Clin Psychopharmacol 1988;
wise, analysis of weight change indicated no significant 8:112-115.
change across treatments. No side effects were associ- 8. Stark P, Hardison CD: A review of multicenter controlled clinical
ated with medical risk. Of minor adverse effects, there studies of fluoxetine vs imipramine in outpatients with major de-
were trends toward relatively more complaints of in- pressive disorder. J Clin Psychiatry 1985;46:53-58.
9. Chouinard G: A double-blind controlled clinical trial of
somnia with fluoxetine and more complaints of consti- fluoxetine and amitriptyline in the treatment of outpatients with
pation with trazodone. Overall, these data indicate that major depressive disorder. J Clin Psychiatry 1985;46:32-37.
fluoxetine had a side effect profile that was roughly 10. Perry PJ, Garvey MH, Kelly MW, et al: A comparative trial of
fluoxetine vs trazodone in outpatients with major depressive dis-
comparable to that of trazodone and. did not differen- order. J Clin Psychiatry, in press.
tially affect attrition rates or the health of the patient. 11. Feighner JP, Cohn JB: Double-blind comparative trials of
This is an important finding, as trazodone has been fluoxetine and doxepin in geriatric depressed with major depres-
identified as a generally well-tolerated antidepressant sive disorder. J Clin Psychiatry 1985;46:20-25.
in the treatment of the elderly depressed patient. 12. Wernicke JF, Dunlop SR, Dornseif BE, Zerbe RL: Fixed dose
fluoxetine therapy for depression. Psychopharmacol Bull 1987;
23:164-168.
Conclusion 13. Hamilton M: A rating scale for depression. J Neurol Neurosurg

Although conclusions are tentative given the small Psychiatry 1960;12:52-62.


14. Guy W (ed): ECDEU Assessment Manual for Psychopharmacology,
sample size, fluoxetine demonstrated efficacy in this revised. Bethesda, MD, US Department of Health, Education
geriatric population that was at least comparable to that and Welfare, 1976.
of trazodone in terms of both antidepressant action and 15. Hayes WL: Statistics, ed 3. New York, Holt Rinehart & Winston,
1981.
tolerability. As indicated by the high dropout rate even 16. Zimmer AW, Calkins E, Hadley E, et al: Conducting clinical re-
during the placebo washout phase of this study, geriat- search in geriatric populations. Ann Intern Med 1985;
ric depressed patients remain a difficult population to 103:276-283.
study. As such, our study is consistent with the conclu- 17. Brotman AW, Falk WE, Gelenberg AJ: Pharmacologic treatment
sions of Rockwell and colleagues that I1geriatrtc research of acute depressive subtypes, in HY Meltzer (ed): Psychophar-

213

Downloaded from jgp.sagepub.com at CARLETON UNIV on June 15, 2015


macology: The Third Generation of Progress. New York, Raven Psychopharmacol 1988;3:311-322.
Press, 1987, pp 1031-1040. 19. Rockwell E, Lam RW, Zisook S: Antidepressant drug studies in
18. Gerson SC, Plotkin DA, Jarvik LF: Antidepressant drug studies, the elderly. Psychiatr Clin North Am 1988;11:215-233.
1964 to 1986: Empirical evidence for aging patients. J Clin

214

Downloaded from jgp.sagepub.com at CARLETON UNIV on June 15, 2015

You might also like