Letters to the Editor
Am J Psychiatry 157:6, June 2000
Retreatment With Clozapine
After Erythromycin-Induced Neutropenia
Agranulocytosis and neutropenia are significant adverse
effects of clozapine treatment (1). We report the case of a pa-
tient who was retreated with clozapine several years after de-
veloping neutropenia under unusual circumstances.
Antonio, a 17-year-old adolescent with childhood-onset
schizophrenia, was admitted for clozapine treatment. De-
spite adequate trials of chlorpromazine, thioridazine, tri-
fluoperazine, molindone, and haloperidol, he had chronic
hallucinations and delusions as well as aggressive behav-
ior that necessitated hospitalization for over a year. Over 6
weeks, his dose of clozapine was increased to 600 mg/day
with good response and no significant side effects.
Over the following 5 years Antonio continued to im-
prove at this dose. He was able to live with his parents and
attended an outpatient day program. At age 21 he devel-
oped a fever and sore throat and was treated with eryth-
romycin. Before he began taking the antibiotic, his WBC
count was 14,000 cells/µl; it had consistently been above
7,400 cells/µl in the previous month. Within 2 weeks his
absolute neutrophil count had dropped to 918 cells/µl. He
also had a WBC count of 5,100 cells/µl, so clozapine ther-
apy was discontinued.
In the subsequent 2 years Antonio underwent gross de-
terioration, again resulting in chronic hospitalization. De-
spite appropriate trials of haloperidol, risperidone, and
olanzapine, he remained profoundly impaired.
At age 24 Antonio was retreated with clozapine, the
only antipsychotic that had given him significant benefit;
the dose was slowly titrated to 400 mg/day. During this pe-
riod he developed a fever with no apparent focus of infec-
tion. There was an appropriate rise in WBC and absolute
neutrophil counts, so conservative management aimed at
fever reduction was employed. Within several days both
his temperature and WBC count had returned to normal.
He was discharged with clozapine therapy, 400 mg/day.
Six months later he was doing well clinically and was again
living at home.
This report illustrates the case of a patient with refractory
schizophrenia who responded well to clozapine treatment.
The addition of erythromycin therapy precipitated neutrope-
nia and prevented the continuation of clozapine treatment.
However, the patient was retreated 2 years later without diffi-
culty. Because neutropenia almost invariably occurs within 3
months of the initiation of clozapine treatment (1), the devel-
opment of neutropenia after 5 years is unusual. This suggests
that the interaction of clozapine with erythromycin was the
precipitating factor for the patient’s blood dyscrasia, particu-
larly since erythromycin-induced agranulocytosis has been
reported (2). Erythromycin has also been associated with in-
creased clozapine plasma concentrations and a subsequent
In view of clozapine’s efficacy in patients with schizophre-
nia, and because of its adverse effects and drug interactions,
the addition of other medications to an established treatment
regimen of clozapine should be undertaken carefully. Addi-
tionally, in patients who develop neutropenia under unusual
circumstances, retreatment with clozapine should be consid-
ered if previous response has been good.
1. Alvir JM, Lieberman JA: Agranulocytosis: incidence and risk fac-
tors. J Clin Psychiatry 1994; 55(suppl B):S137–S138
2. Tanaka M, Tao T, Kaku K, Kaneko T: Agranulocytosis induced
by macrolide antibiotics (letter). Am J Hematol 1995; 48:133
3. Funderburg LG, Vertrees JE, True JE, Miller AL: Seizure following
addition of erythromycin to clozapine treatment (letter). Am J
Psychiatry 1994; 151:1840–1841
SASHA I. USISKIN, M.D.
ROB NICOLSON, M.D.
MARGE LENANE, M.S.W.
JUDITH L. RAPOPORT, M.D.
Exacerbation of Psychosis by
We report the case of a patient initially treated for metham-
phetamine psychosis whose psychotic symptoms recurred
while she was taking phenylpropanolamine.
Ms. A was a 31-year-old woman with a history of meth-
amphamine abuse who was initially admitted to the inpa-
tient psychiatry unit with a depressed mood, command
hallucinations, and paranoid delusions. She had been us-
ing methamphetamine heavily up until 3 weeks before
admission. Her depressive symptoms, including anhe-
donia, decreased energy, increased sleep, suicidal ide-
ation, and tearfulness, worsened 2 months before admis-
sion. In addition, her psychotic symptoms had been
present for many weeks but had been getting progres-
sively more severe. In the hospital, treatment with parox-
etine was initiated because it had reportedly been effec-
tive previously in spite of episodic methamphetamine
use. Ms. A was also treated with perphenazine. Her de-
pressive symptoms improved, and her psychotic symp-
toms disappeared over 3 weeks. At discharge, she was no
Six months later Ms. A was readmitted to the inpatient
psychiatry unit after a 2-week recurrence of command
hallucinations and paranoia. The results of a urine drug
screening test were negative for methamphetamine;
there was no historical evidence of resumed use. Approxi-
mately 3 weeks before admission Ms. A was prescribed a
combination of phenylpropanolamine, 75 mg, and
guaifenesin, 400 mg, for congestion. She had also been
taking cimetidine for 1 month for gastritis. In the hospital
she continued to received paroxetine, perphenazine, and
oral contraceptive pills, which she had been taking as an
outpatient. The congestion medication and cimetidine
were stopped. Within 3 days the voices and paranoia had
disappeared, and she was discharged.
It is commonly accepted that methamphetamine use can
cause psychosis. It has also been reported that chronic expo-
sure to methamphetamine may alter the response to stimuli,
resulting in a recurrence of psychotic symptoms in abstinent
patients (1, 2). For example, there may be an increased sensi-
tivity to psychosis after exposure to stimulant amines (3). To
our knowledge, psychosis activated by phenlypropanolamine
has not previously been documented. We report that phenyl-
propanolamine can cause a recurrence of psychotic symp-
toms that were initially precipitated by previous meth-
amphetamine use. It is unclear whether the concurrent
Am J Psychiatry 157:6, June 2000
LETTERS TO THE EDITOR
administration of cimetidine, an inhibitor of hepatic metabo-
lism, elevated the patient’s serum phenylpropanolamine
level, although she did not demonstrate peripheral signs of
toxicity. Once treatment with cimetidine and phenylpropano-
lamine had been discontinued, her psychosis remitted. Cau-
tion must clearly be used in prescribing phenylpropanola-
mine to former methamphetamine users with histories of
psychosis. It is possible that other stimulant decongestants
may be hazardous in these patients as well.
1. Sato M, Numachi Y, Hamamura T: Relapse of paranoid psy-
chotic state in methamphetamine model of schizophrenia.
Schizophr Bull 1992; 18:115–122
2. Sato M: A lasting vulnerability to psychosis in patients with pre-
vious methamphetamine psychosis. Ann NY Acad Sci 1992; 28:
3. Angrist B, Rotrosen J, Kleinberg D, Merriam V, Gershon S:
Domaminergic agonist properties of ephedrine: theoretical
implications. Psychopharmacology 1977; 55:115–120
ANGELIQUE GOODHUE, M.D.
ROXANNE L. BARTEL, M.D.
NANCY B. SMITH, PHARM.D.
Salt Lake City, Utah
Serotonin Syndrome From Addition of
Low-Dose Trazodone to Nefazodone
Serotonin syndrome, a potentially fatal condition of sero-
tonergic hyperstimulation, has been characterized by diag-
nostic criteria that include at least three of the following:
mental status changes (confusion or hypomania), restless-
ness, myoclonus, hyperreflexia, diaphoresis, shivering,
tremor, diarrhea, and incoordination (1). Nefazodone has
previously been implicated in two cases of serotonin syn-
drome, one with paroxetine (2) and one with valproic acid (3),
whereas trazodone has been implicated in combination with
buspirone (4), paroxetine (5), fluoxetine (6), and amitriptyline
with lithium (7). To our knowledge, this is the first report of se-
rotonin syndrome associated with the combination of nefaz-
odone and trazodone.
Ms. A was a 60-year-old woman with a long history of
depression and hypertension. After a relapse of depres-
sion due to noncompliance with her medication, nefaz-
odone therapy was initiated at 200 mg/day and increased
to 400 mg/day 6 weeks before she was seen in the emer-
gency room. Four days before admission she was evalu-
ated for an exacerbation of her depression. Her dose of
nefazodone was raised to 500 mg/day and trazodone, 25–
50 mg/day, was added as a hypnotic. Ms. A used traz-
odone for three nights. She was seen in the emergency
room after her blood pressure increased to 240/120 mm
Hg when she measured it at home.
Ms. A reported intermittent numbness of the right side
of her lips and nose and the fingers on her right hand,
which improved with time. She described an appearance
of flushed pruritic skin for a day. She noted nausea and
several loose stools. Her son found her to be confused,
and she reported concentration difficulties.
On examination, Ms. A was restless, hyperreflexic, and
diaphoretic (oral temperature: 36°C, pulse: 92 bpm, blood
pressure: 255/130 mm Hg, dilated pupils: equal at 4 mm).
Her creatinine kinase level was 180 U/liter (normal range:
30–170 U/liter), and her total cholesterol level was 249.8
mg/dl. All other laboratory values were within the normal
range. Nefazodone and trazodone therapy was immedi-
ately discontinued. Ms. A was treated with labetalol, cloni-
dine, amlodipine, and an increase in her usual dose of
irbesartan for high blood pressure. The confusion, rest-
lessness, hyperreflexia, nausea, diaphoresis, flushed pru-
ritic skin, and intermittent numbness all disappeared
within 12 hours, and her blood pressure was stabilized at
160 mm Hg systolic pressure within 48 hours.
This patient clearly had an episode of serotonin syndrome
meeting at least five of Sternbach’s criteria for diagnosis, in-
cluding confusion, restlessness, hyperreflexia, diaphoresis,
and diarrhea (1). Serotonin hyperstimulation can also ac-
count for her hypertension, nausea, and flushing (1). A tran-
sient ischemic attack was ruled out since this could account
only for the transient intermittent right-side numbness; a
computerized tomography scan of her head was normal. Fur-
thermore, nefazodone has been associated with paresthesias
(pp. 859–862, Physicians’ Desk Reference, 53rd ed.). An aller-
gic reaction to trazodone was considered, but this would ac-
count only for her pruritic flushed skin. Although it is com-
mon practice, the addition of low-dose trazodone as a
hypnotic to another serotonergic agent can lead to potentially
fatal serotonin syndrome.
1. Sternbach H: The serotonin syndrome. Am J Psychiatry 1991;
2. John L, Perreault MM, Tao T, Blew PG: Serotonin syndrome as-
sociated with nefazodone and paroxetine. Ann Emerg Med
3. Brazelton T, Blanc PD, Olson KR: Toxic effects of nefazodone
(letter). Ann Emerg Med 1997; 30:550–551
4. Goldberg RJ: Serotonin syndrome from trazodone and bus-
pirone. Psychosomatics 1992; 35:235–236
5. Reeves RR, Bullen JA: Serotonin syndrome produced by parox-
etine and low-dose trazodone. Psychosomatics 1995; 36:159–
6. George TP, Godleski LS: Possible serotonin syndrome with traz-
odone addition to fluoxetine. Biol Psychiatry 1996; 39:384–
7. Nisijima K, Shimizu M, Abe T, Ishiguro T: A case of serotonin
syndrome induced by concomitant treatment with low-dose
trazodone and amitriptyline and lithium. Int Clin Psychophar-
macol 1996; 11:289–290
HOWARD C. MARGOLESE, M.D., C.M.
GUY CHOUINARD, M.D., M.SC., F.R.C.P.(C.)
Montreal, Que., Canada
Acute Onset of Auditory Hallucinations After
Initiation of Celecoxib Therapy
Celecoxib is a nonsteroidal anti-inflammatory agent that is
selective for cyclooxygenase-2. This selectivity offers the abil-
ity to control inflammation and pain without some of the side
effects of traditional nonsteroidal anti-inflammatory drug
therapy, particularly gastrointestinal bleeding (1). The multi-
ple side effects of traditional nonsteroidal anti-inflammatory
drugs are well described; they include psychiatric manifesta-
tions such as psychosis, delirium, agitation, and depression
(2–5). We report a case in which a patient developed overt au-
ditory hallucinations after treatment with celecoxib.
Am J Psychiatry 157:6, June 2000
LETTERS TO THE EDITOR
Ms. A, a 78-year-old woman, reported a progressive on-
set of auditory hallucinations within 10 days of initiating
celecoxib therapy, 200 mg b.i.d., for osteoarthritis pain.
She reported initially hearing thumping sounds within the
first 2 days of therapy, and they progressively increased to
the sound of voices calling her name and repeating words
from the television and radio. She underwent evaluation
by an internist and consulted an otolaryngologist with
negative results. She was seen for psychiatric evaluation
because of fears that she was “going crazy.” Ms. A’s con-
current medications included 20 mg/day of quinapril, 240
mg/day of sustained-release verapamil for hypertension,
20 mg b.i.d. of isosorbide dinitrate for angina, 20 mg/day
of tamoxifen due to a history of breast cancer, and 500 mg
t.i.d. of calcium carbonate for osteoporosis. There had
been no changes in her concurrent medication regimen
over the past year until the addition of celecoxib.
Ms. A was alert and mildly anxious, with a clear senso-
rium and intact cognition. She described vivid hallucina-
tions and stated that voices were calling her name and re-
peating the words of radio and television announcers. She
was advised to discontinue the celecoxib; she then re-
ported gradual improvement over the next 3 days. By the
4th day her symptoms disappeared, but she reported in-
creasing pain and was advised by her internist to resume
taking the celecoxib at a lower dose—100 mg b.i.d. daily.
After 5 days, Ms. A again began hearing voices calling her
name and repeating phrases from the television. She was
again advised to discontinue the celecoxib, and her symp-
toms rapidly improved. Because of multiple gastrointesti-
nal side effects from other nonsteroidal anti-inflamma-
tory agents, she resumed taking celecoxib, 100 mg/day,
on an as-needed basis. She reports hearing occasional
thumping noises but feels able to tolerate them.
Cyclooxygenase-2 is present in brain tissue (1). The poten-
tial exists for psychiatric side effects with drugs selective for
this enzyme. This patient developed hallucinations that were
clearly associated with the initiation of celecoxib therapy and
recurred with retreatment. Despite the significant advantages
of celecoxib over traditional nonselective nonsteroidal anti-
inflammatory drugs in their lack of effects on the gastric mu-
cosa and platelets (1), practitioners must be aware of poten-
tial adverse psychiatric events with the use of celecoxib.
1. Hawkey CJ: COX-2 inhibitors. Lancet 1999; 353:307–314
2. Browning CH: Nonsteroidal anti-inflammatory drugs and se-
vere psychiatric side effects. Int J Psychiatry Med 1996; 26:25–
3. Hoppmann RA, Peden JG, Ober SK: Central nervous system side
effects of nonsteroidal anti-inflammatory drugs: aseptic men-
ingitis, psychosis, and cognitive dysfunction. Arch Intern Med
4. Griffith JD, Smith CH, Smith RC: Paranoid psychosis in a patient
receiving ibuprofen, a prostaglandin synthesis inhibitor: case
report. J Clin Psychiatry 1982; 43:499–500
5. Tollefson GD, Garvey MJ: Indomethacin and prostaglandins:
their behavioral relationships in an acute toxic psychosis. J Clin
Psychopharmacol 1982; 2:62–64
MELINDA S. LANTZ, M.D.
VINCENT GIAMBANCO, R.PH.
New York, N.Y.
for Control of Acute Mania
Valproate is a safe and effective antiepileptic and thymo-
leptic drug. An intravenous form has been available since
December 1996; it is approved for the treatment of complex
partial and absence seizures but not for bipolar disorder
(package insert, Abbott Laboratories). A recent case report
discusses using intravenous valproate for controlling agita-
tion in an 8-year-old autistic patient (1). We are not aware of
any literature on its use for mania.
Ms. A was an 81-year-old nursing home resident with a
long history of bipolar disorder, including approximately
10 hospitalizations for manic episodes; her condition was
maintained with divalproex and haloperidol therapy. She
had recently developed Alzheimer’s dementia.
The dementia manifested in growing apathy, with Ms. A
often sitting for long periods fully awake but with her eyes
closed. It was thought that she was becoming lethargic,
and her haloperidol dose was tapered off. This brought no
dramatic change, so her divalproex dose was tapered and
discontinued. Within a week, Ms. A was euphoric, loud, ar-
gumentative, grandiose (“I’m the greatest lion tamer in
the Ringling Brothers Circus!”), paranoid, and hallucinat-
ing. Soon she refused to take her medications, and a hasty
attempt to restart divalproex therapy failed. Finally, she
refused all food and fluids, necessitating hospitalization.
Ms. A was admitted to the psychiatric department. Her
manic symptoms continued. Attempts to persuade her to
take nutrition, fluids, or medications orally failed, and in-
travenous fluid replacement was begun. Because of her
declining nutritional status and persistent agitation, we
decided her condition constituted an emergency. With
family consent, we began intravenous haloperidol, 1 mg
b.i.d. This lessened her agitation but had no effect on her
mania or psychosis.
As Ms. A’s nutritional status continued to deteriorate,
we decided, again with family consent, to administer val-
proate intravenously. We began with 125 mg of valproate
in 100 cc of 5% dextrose in water, infused over 1 hour ev-
ery 6 hours, and increased the infusion to 200 mg after
the first two doses. After 2 days (nine doses), Ms. A’s mania
cleared. She began eating, drinking, and taking her medi-
cations, including divalproex, orally. In the few days until
discharge, Ms. A exhibited her baseline playful and pleas-
antly disoriented dementia (“Get me a beer!”), with no ev-
ident psychosis, grandiosity, or other indication of mania.
The nine intravenous doses brought her serum valproate
level to 53 µg/ml, which had historically been therapeutic
for her. She experienced no untoward effects. Ms. A was
discharged back to her nursing home with the manic epi-
When acute mood stabilization is required, but the oral
route is unavailable, parenteral valproate appears to be a safe
and effective alternative. Oral and intravenous total daily
doses are equivalent (package insert, Abbott Laboratories).
The rapid response in this case suggests that a loading effect
might be achieved expeditiously with intravenous adminis-
tration (presumably with diminished gastrointestinal side ef-
fects), although aggressive intravenous dosing for routine
control of mania would merit further study for safety.
Am J Psychiatry 157:6, June 2000
LETTERS TO THE EDITOR
1. Hilty DM, Rodriguez GD, Hales RE: Intravenous valproate for
rapid stabilization of agitation in neuropsychiatric disorders
(letter). J Neuropsychiatry Clin Neurosci 1998; 10:365–366
PAUL B. HERBERT, M.D.
J. CRAIG NELSON, M.D.
New Haven, Conn.
Weight Criteria for Diagnosis
of Anorexia Nervosa
The primary weight criterion for a diagnosis of anorexia
nervosa is a weight less than 85% of what is considered nor-
mal for that person’s age and height (DSM-IV and ICD-10).
According to DSM-IV, a body mass index less than or equal to
17.5 kg/m², which originated from the ICD-10 diagnostic cri-
teria for research, is an alternative and somewhat stricter
guideline. However, this alternative criterion is not adjusted
for age and sex. Because anorexia nervosa typically begins in
late childhood, adolescence, or early adulthood, it is crucial to
consider age in making a diagnosis because the relation of
weight to height changes substantially during this age span.
This is highlighted by the fact that with increasing age, the
proportion of individuals with a body mass index less than or
equal to 17.5 drops dramatically from 57% at 10 years to be-
low 1% at age 35 in the German female population (similar
percentages apply to the U.S. population) (1).
From a clinical perspective, the body mass index of a pa-
tient can only be interpreted appropriately when the age-
and sex-specific distribution of the body mass index is
known. Hence, the use of sex-specific age percentiles for
body mass index has been proposed in order to assess the de-
gree of underweight in acute anorexia nervosa, to determine
target weight, and to assess weight outcome (1, 2). Indepen-
dent of age and sex, the main DSM-IV weight criterion (less
than 85% of expected body weight) corresponds to a body
mass index between the fifth and 10th percentiles of the body
mass index in both the U.S. and German populations (1). To
address the issue of age- and sex-dependent distributions of
the body mass index and to introduce a convenient and epi-
demiologically based definition of the weight criterion for
anorexia nervosa, we suggest the use of the 10th percentile of
the body mass index as a cutoff for underweight in industri-
On the basis of these considerations, we warn against in-
discriminate use of the two weight criteria, both in clinical
practice and research, because misclassifications with poten-
tially serious sequelae can ensue. Thus, a body mass index of
17.5 in a 14-year-old girl is by no means indicative of anorexia
nervosa. It is obvious that epidemiological studies cannot
readily be compared if they are based on these different
weight criteria. It should be realized that a body mass index of
17.5 is a strict weight cutoff only for individuals over age 20.
For children, and to a lesser extent adolescents, the body
mass index cutoff is less strict than the primary DSM-IV
1. Hebebrand J, Himmelmann GW, Wewetzer C, Gutenbrunner C,
Heseker H, Schäfer H, Remschmidt H: Body weight in acute an-
orexia nervosa and at follow-up assessed with percentiles for
the body mass index: implications of a low body weight at re-
ferral. Int J Eat Disord 1996; 19:347–357
2. Hebebrand J, Himmelmann GW, Herzog W, Herpertz-Dahl-
mann BM, Steinhausen HC, Amstein M, Seidel R, Deter HC,
Remschmidt H, Schäfer H: Prediction of low body weight at
long-term follow-up in acute anorexia nervosa by low body
weight at referral. Am J Psychiatry 1997; 154:566–569
JOHANNES HEBEBRAND, M.D.
PETER M. WEHMEIER, M.D.
HELMUT REMSCHMIDT, M.D., PH.D
Schizophrenias in the
The Journal regularly highlights the nosology of the schizo-
phrenias from various perspectives (1, 2). Notably, the exten-
sion of the schizophrenic spectrum depends on classification
(1). We wish to elaborate on the Wernicke-Kleist-Leonhard
school within the traditional and current context.
Originally, Kraepelin’s dementia praecox and Bleuler’s
schizophrenias described heterogeneous psychoses with no
restitutio in integrum. They were unbiased against deficit
symptoms. The Wernicke-Kleist-Leonhard school sustained
this heuristic notion of the schizophrenias. It rejected noso-
logical hybridism, empirically separating developmentally
more conspicuous systematic schizophrenias of insidious on-
set or course from genetically higher loaded unsystematic
schizophrenias inclined to some bipolarity (i.e., periodic
catatonia, affective paraphrenia, and cataphasia). Thus, peri-
odic catatonia is but one of three unsystematic schizophre-
nias that were previously omitted from mention (2). Some
well-described conditions dominated by specific deficits
were grouped into hebephrenias, which exist only in system-
atic forms. In fact, various cases of heboidophrenia (Kahl-
baum), dementia simplex (Weygandt and Diem), schizoidia
(Bleuler and Kretschmer), schizotypes (Rado), latent schizo-
phrenia (Bleuler), and pseudoneurotic schizophrenia (Hoch
and Polatin) corresponded clinically to Kleist and Leonhard’s
hebephrenias. Moreover, Leonhard passionately differenti-
ated early childhood catatonias from mental retardation. De-
mentia infantilis (Heller) and early infantile autism (Kanner)
overlapped to some extent with Leonhard’s systematic child-
hood catatonias, whereas autistic psychopathy of childhood
(Asperger) denoted a nonschizophrenic condition.
Currently, DSM and ICD criteria strive for atheoretical con-
sentaneity but impose several inconsistencies. Productive
manifestations are appointed as primary gatekeepers for psy-
choses and schizophrenias that are subtyped but essentially
treated as a single entity. Positive symptoms dominate the “A”
criteria for DSM-IV schizophrenia. In anamnestic pervasive
developmental disorders, DSM-IV defines comorbid schizo-
phrenia as contingent on delusions or hallucinations. Simi-
larly, the inclusion criteria for ICD-10-schizophrenia are
largely neo-Schneiderian. Mere duration segregates schizo-
phrenia from schizophreniform disorder. For both, ICD-10 re-
fers to active duration but DSM-IV to total duration as well.
Schizotypes are conceptualized either as personality disor-
ders (DSM-IV) or as tightly related to schizophrenia (ICD-10).
However, ICD-10 disfavors schizotypes as hardly demarcated
from simple schizophrenia and schizoid and paranoid per-
sonality disorders. According to DSM and ICD consensus, the
Am J Psychiatry 157:6, June 2000
LETTERS TO THE EDITOR
schizoaffective disorders are essentially hybrids. Schizoaffec-
tive (Kasanin) and schizophreniform (Langfeldt) psychoses
preferentially describe cases of Leonhard’s unsystematic
schizophrenias or fast-cycling cycloid psychoses. Represent-
ing neither schizophrenic nor affective psychoses, cycloid
psychoses recur frequently in phases—unlike brief psychotic
disorders (DSM-IV)—and without progressing to residua.
According to the Wernicke-Kleist-Leonhard school, the
DSM and ICD criteria exclude some cases of schizophrenic
psychoses (namely hebephrenias and catatonias) from re-
search on schizophrenias. In other nonschizophrenic (partic-
ularly cycloid) psychoses, “schizophrenia” may be misas-
serted. Supported by genetic evidence (3), Leonhard’s
classification has just been retranslated (4). For technological
sophistication to eventually alleviate a centurial enigma (1),
psychopathological disputes (e.g., of primary negative symp-
toms and alternative dimensional descriptors) must consider
the history of the schizophrenias.
1. Andreasen NC: Understanding schizophrenia: a silent spring?
Am J Psychiatry 1998; 155:1657–1659
2. Carroll BT: Karl Leonhard, 1904–1988. Am J Psychiatry 1998;
3. Franzek E, Beckmann H: Different genetic background of
schizophrenia spectrum psychoses: a twin study. Am J Psychia-
try 1998; 155:76–83
4. Leonhard K, Cahn CH: Classification of Endogenous Psychoses
and Their Dfferentiated Etiology, 2nd rev ed. Edited by Beck-
mann H. Vienna, Springer Verlag, 1999
HELMUT BECKMANN, M.D.
ANDREAS J. BARTSCH, M.D.
KLAUS-JÜRGEN NEUMÄRKER, M.D.
BRUNO PFUHLMANN, M.D.
MARIA F. VERDAGUER, M.D.
ERNST FRANZEK, M.D.
Medication Treatment Versus Cognitive
I was disturbed by the mega-analysis of studies of severely
depressed outpatients by Robert J. DeRubeis, Ph.D., and col-
leagues (1), which concluded that cognitive behavior therapy
is a treatment equivalent to medication therapy. First, of the
four studies analyzed, three used imipramine and one
nortriptyline—the latter in modest doses at best. Neverthe-
less, the authors generalized these two drugs to represent all
antidepressant medications. Second, severity was deter-
mined by using summed scores on the Hamilton Rating Scale
for Depression or the Beck Depression Inventory, as if those
alone can tell us the factors most clinicians consider when
characterizing a patient’s depression severity. For example,
were most patients in these studies dysthymic or melan-
cholic? Most clinicians would consider the latter or even ma-
jor depression without melancholic features to be more se-
vere, whereas they would consider dysthymia more chronic
and resistant to medication. Other severity considerations
might include the degree of anxiety or agitation, suicidal feel-
ings, suspiciousness or outright delusions or perceptual dis-
turbances, and a history of bipolarity. If we are to take the
conclusions of Dr. DeRubeis et al. literally, then a depressed
man over 60 years of age with early-morning insomnia,
weight loss, mild cognitive problems, psychomotor retarda-
tion, and dysphoria is better off with cognitive behavior ther-
apy than with antidepressants and their potential side effects.
Does anyone really believe that?
1. DeRubeis RJ, Gelfand LA, Tang TZ, Simons AD: Medications ver-
sus cognitive behavior therapy for severely depressed outpa-
tients: mega-analysis of four randomized comparisons. Am J
Psychiatry 1999; 156:1007–1013
MICHAEL ALAN TAYLOR, M.D.
North Chicago, Ill.
Dr. DeRubeis and Ms. Gelfand Reply
Our report was more a reaction to treatment guidelines that
have been based on scant data than a definitive statement
about how specific patients with depression should be treated.
In his letter, Dr. Taylor emphasizes important issues about the
generalizability of conclusions that can be drawn from any re-
view of empirical research, including our mega-analysis. Spe-
cifically, one should not haphazardly generalize our findings
to antidepressant medications or doses that were not used in
the studies we reviewed, nor should one blithely generalize
our findings to all depressed patients. As clinicians, our job is
to combine the most relevant research findings with our clini-
cal experience and knowledge of available resources to make
the best decision we can about how to proceed in a given case.
Indeed, ongoing studies of cognitive behavior therapy versus
medication treatment will inform us in the future regarding
some of the more specific circumstances that Dr. Taylor de-
scribes (e.g., depression subtypes and patient ages).
Until such time, we hope that clinicians will bear in mind
the findings summarized in our report. In our mega-analysis,
based on four major studies that compared antidepressant
medications and cognitive behavioral therapy in the acute
treatment of severely depressed outpatients, we found equiv-
alent performance of these two treatment modalities. As for
Dr. Taylor’s specific concerns, first, especially with severe or
melancholic depression, the evidence suggests that tricyclic
antidepressants such as those used in the four studies re-
viewed are at least as potent as other classes of antidepres-
sants, including selective serotonin reuptake inhibitors (1).
Second, we understand that there are many ways to define se-
vere depression. As our aim was to present data germane to
existing treatment guidelines, we applied the criteria used in
those guidelines. Moreover, despite the original investigators’
attempts to find interactions of treatment and subtype in the
data sets of the four studies reviewed, there is virtually no em-
pirical evidence for the presumed superiority of medication
treatment over cognitive behavioral therapy in patients with
melancholic depression or in any subgroup of depressed pa-
tients for that matter. As for the 60-plus-year-old man with
classic melancholic symptoms, we have observed that many
such patients benefit greatly from well-delivered cognitive
behavioral therapy, just as they do from well-managed anti-
depressant medication regimes.
1. Perry PJ: Pharmacotherapy for major depression with melan-
cholic features: relative efficacy of tricyclic versus selective se-
Am J Psychiatry 157:6, June 2000
LETTERS TO THE EDITOR
rotonin reuptake inhibitor antidepressants. J Affect Disord
ROBERT J. DERUBEIS, PH.D.
LOIS A. GELFAND, M.A.
Short Screening Scale for Posttraumatic
I offer these comments about the short screening scale for
posttraumatic stress disorder (PTSD) described by Naomi
Breslau, Ph.D., and colleagues (1), both to help those who will
use this instrument and to promote better understanding of
the roles of sensitivity, specificity, and prevalence in test inter-
From the authors’ Table 3, one sees that when four or more
items on the seven-symptom scale are endorsed, sensitivity
equals 0.803 and specificity equals 0.973. In the population
evaluated, the proportion of individuals with PTSD, or disor-
der prevalence, was 142/1,830=0.078. Using this prevalence,
Dr. Breslau and colleagues correctly calculated the positive
predictive value as (sensitivity × prevalence) / (sensitivity ×
prevalence + [1 – prevalence] × [1 – specificity]) = 0.713 and
the negative predictive value as (specificity × [1 – prevalence])
/ (specificity × [1 – prevalence] + prevalence × [1 – sensitivity])
= 0.983. The authors should have pointed out, however, that
these values are correct only for the prevalence in their group.
The prevalence of PTSD has been reported to range from 1%
in community-based samples to 75% among rape victims (3).
If the cutoff of four or more symptoms is used in a population
where the prevalence is 2%, the positive predictive value is
0.378 and the negative predictive value is 0.996; if the same
cutoff is used in a population in which the prevalence is 50%,
the positive predictive value is 0.967 and the negative predic-
tive value is 0.832.
The authors say that with the cutoff of four or more symp-
toms, “less than 2% of ‘true’ cases of PTSD were missed,
whereas 29% of subjects without PTSD were falsely identified
as having PTSD” (p. 910). This is incorrect. A positive predic-
tive value of 0.713 means that about 29% of the subjects iden-
tified as having PTSD do not actually have it, and a negative
predictive value of 0.983 means that about 2% of the subjects
identified as not having PTSD actually have the disorder. The
scale’s sensitivity is the probability that it will detect PTSD if a
subject has the disorder; the scale’s specificity is the probabil-
ity that a subject without PTSD will be deemed not to have the
disorder. So if sensitivity is 0.803, the fraction of “true” cases
missed is (1 – sensitivity) = (1 – 0.803) = 0.197, or about 20% of
the subjects. If specificity is 0.973, the chance of falsely identi-
fying someone without PTSD as having the disorder is (1 –
specificity) = (1 – 0.973) = 0.027, or about 3%.
Dr. Breslau and colleagues say that the cutoff of four or
more symptoms is well suited “to maximize the number of
true cases of PTSD” detected initially when a subsequent
evaluation will be used “to reclassify those who were wrongly
classified as having the disorder” (p. 911). However, better
cutoffs for such a purpose might be two or more symptoms
(sensitivity=0.993) or three or more symptoms (sensitivity=
0.951); these cutoffs would miss only 1% or 5% of the actual
cases of PTSD, respectively.
1. Breslau N, Peterson EL, Kessler RC, Schultz LR: Short screening
scale for DSM-IV posttraumatic stress disorder. Am J Psychiatry
2. Mossman D, Somoza E: Neuropsychiatric decision making: the
role of prevalence in diagnostic testing. J Neuropsychiatry Clin
Neurosci 1991; 3:84–88
3. Davidson JRT, March JS: Traumatic stress disorders, in Psychia-
try. Edited by Tasman A, Kay J, Lieberman JA. Philadelphia, WB
Saunders, 1997, pp 1085–1099
DOUGLAS MOSSMAN, M.D.
Dr. Breslau and Colleagues Reply
Dr. Mossman correctly interprets the positive and negative
predictive values in Table 3 of our article—namely, the values
of those who scored four or more on our scale (defined as
cases). A total of 29% did not have PTSD according to the clin-
ical assessment, whereas of those who scored four or more
(defined as noncases), 2% did have PTSD according to the
clinical assessment. Despite our awareness of the common
tendency to confuse positive and negative predictive values
with sensitivity and specificity, we regret that we did not avoid
this pitfall and that we misstated the positive and negative
predictive values as if they were sensitivity and specificity
(i.e., using as denominators the number of cases and non-
cases defined by the clinical assessment instead of by a score
of four or more on the screening scale).
However, we disagree with Dr. Mossman in his emphasis on
sensitivity and specificity instead of on positive and negative
predictive values. The latter constructs are of considerable
utility for potential users, who wish to know what proportion
of persons identified as cases by the scale would be confirmed
by clinical examination.
We wish to emphasize that the prevalence of PTSD in our
study is typical of that in recent general population studies,
such as the National Comorbidity Survey. Therefore, esti-
mates of the performance of the scale based on this preva-
lence estimate are useful.
Our recommended cutoff of four or more symptoms was
based on the data for both positive and negative predictive
values in Table 3. We considered the fact that positive predic-
tive value diminishes rapidly with lower cutoffs (i.e., two or
three), whereas the accompanying improvement in negative
predictive value is slight. If we were to use a cutoff of two or
three instead of four, the number of identified cases that
would not be confirmed would double or triple.
NAOMI BRESLAU, PH.D.
EDWARD L. PETERSON, PH.D.
RONALD C. KESSLER, PH.D.
LONNI R. SCHULTZ, PH.D.
Reprints of Letters to the Editor are not available.