children's attitudes toward physical activity within the
school environment. 26
Supported by the Danish Health Insurance Foundation,
the Danish Health Services Development Foundation, the
Danish Heart Foundation, the Health Insurance Foundation
of Denmark, the Danish Medical Research Council, the
Funen Prevention Council, the Danish Sports Research
Council, and the Rosalie Petersen Foundation.
I WHO Scientific Group. Primary prevention of essential hypertension. WHO
Tech Rep Ser 1983;686:1-40.
2 The 1988 report of the Joint National Committee on Detection, Evaluation,
and Treatment of High Blood Pressure. Arch Intern Med 1988;148:1023-38.
3 Nelson L, Jennings GL, Esler MD, Korner PI. Effect of changing levels of
physical activity on blood pressure and haemodynamics in essential
hypertension. Lancet 1986;ii:473-6.
4 Jennings G, Nelson L, Nestel P, Esler M, Korner P, Burton D, et al. The
effects of changes in physical activity on major cardiovascular risk factors,
hemodvnamics, sympathetic function, and glucose utilization in man: a
controlled study of four levels of activity. Circulation 1986;73:30-40.
5 Seals DR, Hagberg JM. The effect ofexercise training on human hypertension:
a review. Med Sci Sports Exerc 1984;16:207-15.
6 Subcommittee on Nonpharmacological Therapy. Nonpharmacological
approaches to the control of high blood pressure. Hypertension 1986;8:
4-4-67.
7 Report of the Second Task Force on Blood Pressure Control in Children-
1987. Pediatrics 1987;79:1-25.
8 Fagard R. Habitual physical activity, training, and blood pressure in normo-
and hypertension. Int-J Sports Med 1985;6:57-67.
9 Montoyc HJ. Physical activity, physical fitness, and heart disease risk factors
in children. In: Stull GA, Eckert HM, eds. Effects of physical actizvity on
children. Champaign, Illinois: Human Kinetics Publishers, 1986:127-52.
(American Academy of Physical Education Papers, No 19.)
10 Vaccaro P, Mahon AD. The effects of exercise on coronary heart disease risk
factors in children. Sports Med 1989;8:139-53.
11 Hansen HS, Hyldebrandt N, Nielsen JR, Froberg K. Blood pressure
distribution in a school-age population aged 8-10 years-the Odense
schoolchild study.tHvpertens 1990;8:641-6.
12 Fanner JM, ed. Gr(rwth at adolescence. Oxford: Blackwell Scientific Publica-
tions, 1962.
13 Hanscn HS, Froberg K, Nielsen JR, Hyldebrandt N. A new approach to
assessing maximal aerobic power in children: the Odense school child
study. Eurj Appl Phystol 1989;58:618-24.
14 Cole IWL, Grizzle IE. Applications of multivariate analysis of variance to
repeated measures experiments. Biometrics 1966;22:810-28.
15 O'Brien RG, Kaiser M1K. MANOVA method for analyzing repeated measttres
designs: an extensive primer. Psychol Bull 1985;97:316-33.
16 SPSS Inc. SPSSX user's guide. New York: McGraw-Hill Book Company,
1983:487-8.
17 Gunsollev JC, Chinchilli VM, Koertge TE, Palcanis KG, Sarbin AG, Brooks
CN. The use of repeated measures analysis of variance for plaque and
gingival indices. J Clin Periodontol 1989;16:156-63.
18 Eriksson BO, Koch G. Effect of physical training on haemodynamic response
during submaximal and maximal exercise in 11-13-year old boys. Acta
PhvsiolScand 1973;87:27-39.
19 Dwyer T, Coonan WE, Leitch DR, Hetzel BS, Baghurst RA. An investigation
of the effects of daily physical activity on the health of primary school
students in South Australia. Intj Epidemiol 1983;12:308-13.
20 Linder CW, DuRant RH, Mahoney OM. The effect of physical conditioning
on serum lipids and lipoproteins in white male adolescents. Med Sci Sports
Exerc 1983;15:232-6.
21 Stone EJ. School-based health research funded by the National Heart, Lung,
and Blood Institute. J Sch Health 1985;55:168-74.
22 Puska P, Vartiainen E, Pallonen U, Salonen JT, Poyhia P, Koskela K, et al.
The North Karelia Youth Project: evaluation of two years of intervention on
health behavior and CVD risk factors among 13 to 15-year-old children.
PrevMed 1982;11:550-70.
23 Tell GS, Vellar OD. Non-communicable disease risk factor intervention in
Norwegian adolescents: the Oslo youth study. In: Hetzel BS, Berenson GS,
eds. Cardiovascular risk factors in childhood: epidemiologv and prevention.
Amsterdam: Elsevier, 1987:203-17.
24 Walter HJ, Hofman A, Vaughan AD, Wynder EL. Modification of risk factors
for coronary heart disease. N Englj Med 1988;318: 1093-1 100.
25 Berenson GS, Shear CL, Chiang YK, Webber LS, Voors AW. Combined low-
dose medication and primary intervention over a 30-month period f'or
sustained high blood pressure in childhood. Amj Med Sci 1990;299:79-86.
26 Demarco T, Sidney K. Enhancing children's participation in physical activity.
J Sch Health 1989;59:337-40.
(Accepted 43Julv 1991)
Fluoxetine and suicide: a meta-analysis of controlled trials of
treatment for depression
Charles M Beasley Jr, Bruce E Dornseif, Janet C Bosomworth, Mary E Sayler, Alvin H Rampey Jr,
John H Heiligenstein, Vicki L Thompson, David J Murphy, Daniel N Masica
Division of Clinical
Neurosciences, Lilly
Research Laboratories, Eli
Lilly and Company,
Indianapolis, Indiana
46285, USA
Charles M Beasley Jr, MD,
clinical resource physician
Bruce E Dornseif, PHD,
research scientist
Janet C Bosomworth, BS,
statistician
Mary E Sayler, MS,
statistician
Alvin H Rampey Jr, PHD,
senior statistician
John H Heiligenstein, MD,
clinical research physician
Vicki L Thompson, MSN,
clinical research administrator
David J Murphy, BBA,
senior information analyst
Daniel N Masica, MD,
international research adviser
Correspondence to: Dr
Charles M Beasley Jr, Lilly
Research Laboratories, Eli
Lilly and Company, Lilly
Corporate Center 2128,
Indianapolis, Indiana 46285,
USA.
B.J 1991;303:685-92
Abstract
Objective-A comprehensive meta-analysis of
clinical trial data was performed to assess the
possible association of fluoxetine and suicidality
(suicidal acts and ideation).
Design-Retrospective analysis of pooled data
from 17 double blind clinical trials in patients with
major depressive disorder comparing fluoxetine (n=
1765) with a tricyclic antidepressant (n=731) or
placebo (n= 569), or both.
Main outcome measures-Multiple data sources
were searched to identify patients with suicidal acts.
Suicidal ideation was assessed with item 3 of
the Hamilton depression rating scale, which
systematically rates suicidality. Emergence of
substantial suicidal ideation was defined as a change
in the rating of this item from 0 or 1 at baseline to 3 or
4 during double blind treatment; worsening was
defined as any increase from baseline; improvement
was defined as a decrease from baseline at the last
visit during the treatment.
Results-Suicidal acts did not differ significantly
in comparisons of fluoxetine with placebo (0-2% v
0-2%, p=0 494, Mantel-Haenszel adjusted incidence
difference) and with tricyclic antidepressants (0-7%
v 0-4%, p=0-419). The pooled incidence of suicidal
acts was 0*3% for fluoxetine, 0-2% for placebo, and
0-4% for tricyclic antidepressants, and fluoxetine did
not differ significantly from either placebo (p=0 533,
Pearson's x2) or tricyclic antidepressants (p=0 789).
Suicidal ideation emerged marginally significantly
less often with fluoxetine than with placebo (0-9% v
2-6%, p=0.094) and numerically less often than with
tricyclic antidepressants (1.7% v 3-6%, p=0-102).
The pooled incidence of emergence of substantial
suicidal ideation was 1-2% for fluoxetine, 2-6% for
placebo, and 3-6% for tricyclic antidepressants. The
incidence was significantly lower with fluoxetine
than with placebo (p=0042) and tricyclic anti-
depressants (p=0-001). Any degree of worsening of
suicidal ideation was similar with fluoxetine and
placebo (15-4%v 17-9%, p=0- 196) and with fluoxetine
and tricyclic antidepressants (15-6% v 16-3%,
p=0-793). The pooled incidence of worsening of
suicidal ideation was 15-3% for fluoxetine, 17*9% for
placebo, and 16-3% for tricyclic antidepressants.
The incidence did not differ significantly with
fluoxetine and placebo (p=0-141) or tricyclic anti-
depressants (p=0542). Suicidal ideation improved
significantly more with fluoxetine than with placebo
(72-0% v 54-8%, p<0001) and was similar to the
improvement with tricyclic antidepressants (72-5% v
69-8%, p=0294). The pooled incidence of improve-
ment of suicidal ideation was 72-2% for fluoxetine,
54-8% for placebo, and 69-8% for tricyclic anti-
depressants. The incidence with fluoxetine was
significantly greater than with placebo (p<0-001) and
did not differ from that with tricyclic antidepressants
(p=0296).
Conclusion-Data from these trials do not show
that fluoxetine is associated with an increased risk of
suicidal acts or emergence of substantial suicidal
thoughts among depressed patients.
Introduction
Because depression is an important risk factor for
suicide'3 there is a need to study the effects of
BMJ VOLUME 303 21 SEPTEMBER 1991 685
antidepressants on suicidalitv (suicidal acts and suicidal
ideation) in patients with major depressive illnesses.
Though it is generally expected that any treatment that
improves the depression is also likely to reduce
suicidality, one study did not support this view.4 It
has not been suggested until fairly recently that,
paradoxically, worsening of suicidality might in a small
subset of patients be associated with the use of
antidepressants. "' Five reports"' (one subsequently
retracted") have hypothesised that the use of fltioxetine
(a serotonin uptake inhibitor) might lead to the
emergence or worsening of suicidal ideation in a very
small proportion of patients taking this drug. Patients
receiving different classes ofantidepressants, including
fluoxetine, have shown no differences in rates of
suicidalityv' while greater improvement with respect
to suicidalitv has occurred with serotonin uptake
inhibitors than with comparative drugs in depressive
illnesses."-' Recently Rouillon et al reported that
maprotiline (a noradrenaline uptake inhibitor) was
associated with a greater incidence of suicidal acts than
placebo during long term treatment of depression.'2
To examine any possible relation of fluoxetine to
the emergence of suicidality, we carried out a com-
prehensive meta-analysis of all relevant clinical trial
data.
Methods
TYPES AND SOURCES OF DATA
The meta-analysis was carried out on the United
States investigational new drug depression clinical trial
database for fluoxetine; this consists of all double
blind, randomised trials of fluoxetine in depression
controlled against placebo or tricyclic antidepressants.
Trials that had been completed and analysed up to the
end of December 1989 were included. The exclusions
were: depression trials that had not used a comparative
drug, trials that had used a comparator other than
placebo or a tricyclic antidepressant, non-blind
extensions of controlled trials, non-blind compas-
sionate trials, trials for other potential indications, and
pharmacokinetic trials.
For these analyses the clinical trials were organised
into five analysis groups: (1) placebo controlled trials
(five trials); (2) trials controlled with tricyclic anti-
depressants (10 trials); (3) trials controlled with
placebos and tricyclic antidepressants (two trials); (4)
analysis group 1 and the fluoxetine and placebo arms of
analysis group 3; and (5) analysis group 2 and the
fluoxetine and tricyclic antidepressant arms of analysis
group 3. The specific protocols included in each
analysis group and the characteristics of the patients
studied are summarised in the Appendix.
Potential cases of suicidal acts were first identified
electronically by searching two sources: (a) clinical
report form data from the trials (for adverse events,
reasons for trial discontinuation, Hamilton depression
rating scale item 3 scores,'" and free text comments)
and (b) data from the drug experience network for.
adverse events and outcomes. The drug experience
network database contains reports of all serious adverse
events (as defined by United States Food and Drug
Administration criteria) that have occurred in clinical
trials, as well as all adverse events voluntarily reported
as part of the manufacturer's (Eli Lilly and Company)
post-marketing surveillance. '9 Those clinical com-
ments that had not been transferred to computer were
examined by the research staff, and all cases in which it
was clear that there had been no suicidal act were
eliminated. All remaining cases were then reviewed
independently by two Eli Lilly and Company psy-
chiatrists, who were blind to the drug that had been
used, to determine whether or not a suicidal act had
occurred.
D)EFINITIONS
A suicidal act was defined as any behaviour under-
taken purposefully from which the outcome was likely
to be self harm, and where no explicit data suggested
that suicide had not been intended.' Actions that might
be described as suicidal gestures were not excluded. A
suicidal act had to have occurred before or during the
day following the last day of double blind treatment, in
compliance with the trial protocol. This time limit was
adopted for three reasons: post-discontinuation data
had not been collected as part of the trials; the end of
participation in the trial or withdrawal of the study
treatment, or both, might have influenced an event
occurring after discontinuation; and other drugs might
have been started after the end of the study treatment.
Suicidal ideation was identified by the scores on item
3 of the Hamilton depression rating scale. The
scale defines these as: 0=absence of such ideation;
1=doubtful or trivial ideation; 2=mild ideation; 3=
active suicidal ideation and suggestive behaviours; and
4= severe ideation, usually involving a suicidal act.
Emergence of substantial suicidal ideation was
defined as a change in score on item 3 of the Hamilton
depression rating scale from 0 or 1 at baseline to 3 or 4
at any time during the double blind treatment.
Emergence of substantial suicidal ideation was
evaluated only for those patients whose score was 0 or 1
at baseline.
Worsening of suicidal ideation was defined as
any increase in item 3 score from baseline at any
time during double blind treatment. Worsening was
evaluated only for those patients who could worsen
during double blind treatment-that is, those whose
score was less than 4 at baseline.
Improvement of suicidal ideation was defined as any
decrease in item 3 score from baseline to the last
evaluation while the patient was in double blind
treatment. Improvement was evaluated only for those
patients who could improve during double blind
treatment -that is, those whose score was greater than
0 at baseline.
DESIGN
Data were analysed from 17 single centre and
multicentre randomised, double blind trials including
3065 patients (1765 receiving fluoxetine, 731 receiving
tricyclic antidepressants (amitriptyline, desipramine,
doxepin, imipramine, nortriptyline), and 569 receiving
placebo). Five trials compared fluoxetine with placebo,
10 compared fluoxetine with a tricyclic antidepressant,
and two compared fluoxetine with a tricyclic anti-
depressant and a placebo.
Fluoxetine doses ranged from 20 mg to 80 mg a day
(except in one trial where the range was 5 mg to 40 mg a
day); in 15 trials the fluoxetine doses were individually
adjusted, and in two the patients were randomly
assigned to one of several fixed doses. Doses of tricyclic
antidepressants were adjusted individually within
current manufacturers' guidelines.
In 16 trials the patients met criteria for non-
psychotic major depressive disorder (three trials used
Research Diagnostic Criteria,0 nine trials used the
Diagnostic and Statistical Manual of Mental Disorders,
3rd edition (DSM-III) criteria for symptoms of one
month's duration, and four trials required DSM-III
criteria"'). Most patients had a score ¢20 on the 21 item
Hamilton depression rating scale (one trial used a score
18 and one trial included a stratum of patients with
scores between 14 and 19), and most did not improve
their score by 20% or more during the lead in period
(approximately one week of single blind placebo
treatment). In one trial the patients were diagnosed by
DSM-III criteria as having bipolar disorder and being
depressed; they had a baseline score :20 on the
21 item Hamilton depression rating scale and failed to
BMJ VOLUME 303 21 SEPTEMBER 1991686
show a 20010 or greater improvement in this score
during placebo lead in.
Written informed consent was given appropriately
in all cases.
Exclusion criteria included a history of substance
misuse within one vear; psychotic or organic mental
disorder; serious suicidal risk as clinically assessed by
the investigator (suicidal ideation was not a criterion;
two inpatient trials did not have an explicit exclusion
criterion based on serious suicidal risk); and unstable
medical conditions or any medical condition precluding
use of one of the drugs used in these studies.
Trials lasted five or six weeks with evaluations about
once a week, except in one inpatient trial where there
were two evaluations a week for the first two weeks of
double blind treatment and one outpatient trial
where there were two evaluations during the first week
and three during the second week of double blind
treatment.
ANALYTICAL AND STATISTICAL METHODS
Descriptive statistics, including the incidence of
suicidal acts and the emergence of substantial suicidal
ideation (the primary measures) and worsening of
suicidal ideation and improvement of suicidal ideation,
were conmputed for all individual trials, for analysis
groups 4 and 5, and for all trials combined. Inferential
statistical analyses were performed for analysis group 4
(fluoxetine z' placebo), analysis group 5 (fluoxetine v
tricyclic antidepressants), and all trials combined. All
3065 randomised patients were included in the analysis
of suicidal acts. A total of 1999 patients with a baseline
score on item 3 of the Hamilton depression rating scale
of 0 or 1 and at least one post-baseline score were
included in the analysis of emergence of substantial
suicidal ideation. The incidence of worsening of
suicidal ideation was based on 2995 patients with a
baseline score <4 and at least one post-baseline score.
The incidence of improvement of suicidal ideation was
based on 2053 patients with a baseline score >0 and at
least one post-baseline score.
The incidence difference and corresponding 95%
confidence interval was used to compare the incidence
of the four outcome variables (suicidal acts, emergence
of substantial suicidal ideation, worsening of suicidal
ideation, improvement of suicidal ideation) between
treatments for individual clinical trials. The incidence
difference22 was defined as the incidence in patients
treated with fluoxetine minus the incidence in patients
treated with the comparator. An incidence difference
greater than 0 implies a numerically higher incidence
with fluoxetine than with the comparator; an incidence
difference less than 0 implies a numerically higher
incidence with the comparator than with fluoxetine;
and an incidence difference equal to 0 implies equal
incidence.
Owing to potential heterogeneity of trials the
adjusted incidence difference, which stratifies by trial,
was used to compare the incidence of the four
outcome variables between treatments in analysis
groups 4 and 5. The individual incidence differences
TABLE I -Characteristics of patietnts at baselinle
MNean (S)) score on Hamilton
deprcssion rating scalc
MNleni w)romen MNedian ratigei
No of patients (years) 21 Itcms Item 3
Analysis group 4:
Fluoxetinc 1322 39 (61) 38 (13-70) 23-5 )5.3) (09(099)
Placebo 569 41:59) 37(12-70) 25 5 (5 5) 1 1(0 9)
Ynalvsis group 5:
Fluoxetine 720 34)66) 43 (19-90() 27-2(5-4) 1-3(1-0)
Tricvclics 731 36(64) 45 (18-88) 27-2 (5 5) 1-4(1-0
All trials:
Fluoxetine 1765 38 62) 40 (13-90) 24 4 (5 6) 1 0 (1 0)
Tricyclics 731 36(64) 45(18-88) 27-2)(5-5) 14)(1-0)
Placebo 569 41 (59) 37 12-70) 25 5 (5-5) 1.1 )0 9)
were combined across the clinical trials to form the
adju
本文档为【Fluoxetine and suicide】,请使用软件OFFICE或WPS软件打开。作品中的文字与图均可以修改和编辑,
图片更改请在作品中右键图片并更换,文字修改请直接点击文字进行修改,也可以新增和删除文档中的内容。
该文档来自用户分享,如有侵权行为请发邮件ishare@vip.sina.com联系网站客服,我们会及时删除。
[版权声明] 本站所有资料为用户分享产生,若发现您的权利被侵害,请联系客服邮件isharekefu@iask.cn,我们尽快处理。
本作品所展示的图片、画像、字体、音乐的版权可能需版权方额外授权,请谨慎使用。
网站提供的党政主题相关内容(国旗、国徽、党徽..)目的在于配合国家政策宣传,仅限个人学习分享使用,禁止用于任何广告和商用目的。