Follow-up assessment of medication-treated dysthymia

ELSEVIER

Prq. h’euro_Psychophmmacol. &Bid. Psychiat. 1996. Vol. 20. pp. 427-442 Copyright C’ 1996 Elsevier Science Inc.

All rights reserved

PII SO278-5846(!36)00007-3 0278 - 5646/96 $32.00

FOLLOW-UP ASSESSMENT OF MEDICATION-TREATED DYSTHYMIA

DAVID J. HELLERSTEIN ls4 LISA WALLNER SAMSTAG’, MARC CANTILLON3, MARTIN MAURER’.‘, JESSE ROS’ENTHALzs4, PHILIP YANOWITCH3,4, ARNOLD WINSTON3,4

‘Psychiatric Outpatient Services and *Mood Disorders Research Unit, 3Beth Israel Medical Center, New York, N.Y., U.S.A.

4Department of Psychiatry, Albert Einstein College of Medicine, Bronx, N.Y., U.S.A.

(Final form, February, 1996)

Hellerstein, David J., Lisa Wallner Samstag, Marc Cantillon, Martin Maurer, Jesse Rosenthal, Philip Yanowitch and Arnold Winston: Follow-Up Assessment of Medication- Treated Dysthymia. Prog. Neuro-Psychopharmacol. & Biol. Psychiat. 1996, 20

1. The objective was to assess long-term efficacy of antidepressant medications in dysthymia.

2. In a naturalistic study, patients with DSMIII-R dysthymia who had participated in previous antidepressant trials with fluoxetine and trazodone were evaluated at a mean of 40.0 weeks of follow-up to assess whether medication response persisted over time. A multivariate analysis was performed for patients on vs. off medicatioin. Relapse rates (with relapse defined as HDRS score > 13) were also compared for these two groups.

3. Of 40 patients, the 24 still on medication showed significantly lower scores on most rating scales (HDRS, Cornell Dysthymia Rating Scale, and CGI, but not on the SCL- 58) than the heterogeneous group of 16 patients not taking medication. Relapse was low (17.4%) among patients remaining on medication.

4. These preliminary findings suggest that dysthymia patients who remain on medication maintain improvement over time.

Kaywords: chronic depression, clinical trial, dysthymia, fluoxetine, medication treatment

. Abbrevlatlons: Clinical Global Impressions (CGI), Cornell Dysthymia Rating Scale (CDRS), Diagnostic and Statistical Manual, III Edition, Revised (DSM-III-R), Hamilton Depression Rating Scale (HDRS), Hopkins Symptom Checklist (58 item) (SCL-581, Monamine Oxidase Inhibitor antidepressants (MAOIs), Structured Clinical Interview for DSM-III-R, Patient Version (SCID-PI, Selective Serotonin Reuptake Inhibitor antidepressants (SSRls), Tricycfic Antidepressants (TCAs).

427

428 D.J. Hellerstein et cd.

Dysthymia (American Psychiatric Association, 1987) or chronic depression, is a

psychiatric disorder affecting about 3% of the population. Recent studies have clarified

the phenomenology of dysthymia (Keller et al., 1991; Keller et al., 19951, and indicate

that dysthymics suffer significant medical and psychiatric comorbidity (Markowitz,

19931, with increased utilization of health services (Howland, 1993) and significant

psychosocial dysfunction (Friedman, 1993).

Medication studies demonstrate that dysthymia and other chronic depressive

disorders often respond to acute trials with medications, including TCAs (Kocsis et al.,

1988) and MAOls (Quitkin et al., 1989). Our studies have demonstrated positive

results with SSRls. In our recent 8-week double-blind study of fluoxetine vs. placebo

(Hellerstein et al., 19931, 62.5% of fluoxetine subjects (IO/161 were responders [with

response defined by a 50% or greater decrease in Hamilton-Depression Rating Scale

(HDRS) score (Hamilton, 1960) and a score of 1 or 2 “very much improved” or “much

improved” on the Clinical Global Impressions Scale (CGI) (Guy, 197611, compared to

only 18.8% (3/16) of placebo subjects. These findings confirmed the results of our

earlier open-label studies of fluoxetine vs. trazodone (Rosenthal et al., 1992; Hellerstein

et al., 19941, in which 61.3% of dysthymic patients completing 3-month medication

trials responded to fluoxetine (14/23 subjects; 60.9%) or trazodone (5/8 subjects;

62.5%).

However, few studies have assessed continued efficacy of antidepressant treatment

in dysthymia. While there is some evidence (Keller et al., 1983) for poor results with

naturalistically-treated chronic depression, other studies suggest that patients receiving

adequate antidepressant trials have good outcome, and that initial medication response

may be maintained over long periods of time (Howland, 1991; Harrison et al., 1986).

Patients who are initially medication responsive may have particularly good outcome.

In imipramine-treated patients, Kocsis et al. (1991) found 89% percent of initial

imipramine responders met the criteria for recovery at follow-up of one to five years,

compared with 31% of nonresponders and noncompleters.

Given the paucity of systematic follow-up studies, even preliminary naturalistic

Follow up assessment of medication-treated dysthymia 429

studies can shed light on the question whether dysthymic symptoms are improved after

longterm medication treatment. The authors report results from a naturalistic follow-up

of patients from their prior efficacy studies. Our hypothesis is that dysthymics taking

antidepressant medications will continue to have relief of depressive symptoms at

follow-up, while patients who have discontinued medication will tend to relapse.

Participants in previous dysthymia studies (Hellerstein et al., 1993; Rosenthal et al.,

1992; Hellerstein et al., 1994) at Beth Israel Medical Center’s Mood Disorders Research

Unit were contacted for follow-up assessment. Subjects had been enrolled in the

original studies by the following procedure:

After providing informed consent, a semistructured clinical interview was conducted,

and prospective subjects were systematically questioned using the Structured Clinical

Interview for DSM-III-R, Patient Version (SCID-P) (Spitzer et al., 1988) mood disorder

section. Subjects between the age of 21 and 65 were included if their diagnosis of

dysthymia was what DSM-III-R terms primary, that is, not related to a pre-existing,

chronic, nonmood Axis I or Axis Ill disorder. The SCID-P sections on major depression

and dysthymia were used to ensure that subjects met diagnostic criteria for dysthymia.

Exclusion criteria for the original studies were as follows: Axis I conditions such as

DSM-III-R psychoactive substance use disorders (within the past six months), panic

disorder, schizophrenia, anorexia or bulimia; a history of mania or hypomania; severe

stress, whether acute or chronic (a score of 4 or greater on DSM-III-R’s Axis IV); a

history of suicide attempts and self-mutilation; significant medical illness; and use of

any psychotropic medication within one month prior to the study. For a more detailed

description of the original studies’ inclusion criteria, the reader is directed to those

respective papers (Hellerstein et al., 1993; Rosenthal et al., 1992; Hellerstein et al.,

1994).

A total of 50 patients, of a possible 63 who had completed treatment protocols, were

430 D.J. Hellerstein et al

reached either by telephone or mail, and 40 patients (63.5% of the total sample) agreed

to a follow-up assessment. Patients provided informed consent for participation in a

follow-up study. The 40 subjects who made up the study sample of the present

research consisted of 35 who had met criteria for a DSM-III-R diagnosis of dysthymia

(American Psychiatric Association, 1987), without concurrent major depression, and

five who had met criteria for double depression (Keller et al., 1983). Patients had been

randomly assigned to either fluoxetine, trazodone, or placebo in the original studies.

The study sample of 40 patients was comprised of 21 (52%) women and 19 (48%)

men, with a mean age of 38.OA9.4 years. The mean number of weeks from the time

patients completed the treatment protocol to the date of the follow-up assessment was

40.0&29.1. Most patients were currently employed (85% of the sample); 27 (68%)

were single, 10 (25%) were married, and one (2%) was divorced (information regarding

marital status was unavailable for two patients). Thirty-two (80%) patients had

completed a college degree, with 14 (35%) having also completed a graduate degree.

The 40 patients agreeing to follow-up interviews were compared with the 23 who could

not be contacted or who declined reassessment, using chi-square (with exact p values

where sample sizes are small) or t-test analyses; they did not differ significantly on any

of the demographic variables (Table 1).

At the time of the follow-up assessment, 24 (60%) patients were taking medication

(23 (58%) taking fluoxetine, and one (2%) taking trazodone); the mean dosage of the

subjects on fluoxetine was 30.5 mgkl8.7 mg, with a mean of 35.9L25.3 weeks of

use. One patient was taking trazodone at a dose of 150 mg/day. Blood levels for

fluoxetine and metabolites have not consistently been related to response (Kelly, et al.,

1989); in this study, levels were drawn primarily to assess compliance. The mean&SD

blood level for fluoxetine was 214.4~195.4 ng/ml and norfluoxetine was

249.3h146.5 ng/ml. Only 7 patients reported entering individual or group

psychotherapy during the follow-up period; 4 were members of the on-medication

group, and 3 were members of the off-medication group.


Table 1

Demographic Characteristics of Patients Assessed for Follow-UP (N=W and Not Assessed for Follow-Up (N = 23)

Assessed Not Assessed Chi Sq df P (N=40) (N = 23) or t-test

Gender (female) Marital Status* Married Single Divorced

Education HS or less Some college College graduate Graduate school

Employed Yes No

Age (M.+SD)

21 17 2.78 1 .I15

10 3 27 13

1 7 2.41 2 .331

0 1 8 5

18 14 14 3 5.09 3 .I30

34 20 3 3 .38 1 .662

38.Ok9.4 37.318.9 .29 60 .775

* Marital status was unavailable for 2 patients in the Assessed group.

The remaining 16 patients were not currently taking medication at the time of the

assessment. Ten patients had discontinued medications after initial trials [five (31 %I

reported insufficient improvement or unpleasant side effects and had chosen to

discontinue medication; three (19%) had discontinued medication after entering

prolonged remission; one had stopped for reasons other than medication effects (i.e.,

reporting that it was becoming inconvenient to make appointments); and information

was not available on one patient]; and six patients initially assigned to placebo (38%)

never began medication. None of the patients had received courses of other types of

antidepressant medications following the original trials. There was no significant

difference between the groups (on vs. off medication) on any demographic variable.

Of the 26 patients who had completed an initial trial of active medication, 18 (69%)

were previously defined as medication responders, and 8 (31 %I as medication

nonresponders. Of the 14 patients intially treated with placebo, 8 went on to

medication treatment with fluoxetine.

432 D.J. Hellerstein et cd.

Subjects who provided informed consent to the follow-up assessment participated in

a structured clinical interview conducted by teams of clinicians and staff researchers.

Interviewers were blind to initial response status, but not to current medication use

(i.e, if patients were currently on or off medication). Patients completed the self-

report version of the Hopkins Symptom Checklist (SCL-58; Derogatis et al., 19741,

and were rated for current severity of depression with the 21-item Hamilton

Depression Rating Scale (HDRS; Hamilton, 1960) and the Cornell Dysthymia Rating

Scale (CDRS; Mason et al., 1993). These measures were completed in order to

assess current level of mood and general symptomatology, as well as to compare

ratings to post-treatment levels in the initial studies. Results of reliability assessment

with the HDRS and the CDRS are reported elsewhere (Hellerstein et al., 1993;

Rosenthal et al., 1992; Hellerstein et al., 1994). The interviewers provided a

Clinician’s Global Impression rating (CGI; Guy, 1976) and assessed the patient with

respect to DSM-III-R criteria for dysthymia and major depression. Fluoxetine and

norfluoxetine blood levels were also collected.

The sample of 40 patients was divided in order to compare those patients who

continued with medication treatment to those who discontinued medication or did not

receive medication trials (Fig 1 J. The authors thus compared samples of patients on

medication at the time of the follow-up interview (N = 24) and those off medication

(N = 16). A repeated measures multivariate analysis of variance (MANOVA) was used

to compare these two groups on the outcome measures (Procedure and Rating Scales)

over treatment. Relapse (operationally defined as HDRS score > 13 at follow-up) was

compared across these groups, analyzed by chi-square using exact p values where

appropriate. Demographic and descriptive variables were compared between groups

with chi-square or independent t-test analyses.


In this naturalistic study, the duration of the follow-up period for the 24 patients

remaining on medication varied greatly (M =37.0 weeks, SD~25.7 weeks). In order

to determine whether there was any effect of treatment duration on follow-up scores,

the authors divided the sample into those patients assessed at less than 40 weeks

(N = 16) vs. those assessed at greater than 40 weeks (N = 8). Follow-up ratings were

compared using a series of t-tests. HDRS scores (lO.lk3.9 vs. 7.91L4.9; t = 1 .I 9,

df=21, NS), CDRS scores (19.OA7.2 vs. 19.9f9.4; t=.24, df=21, NS), SCL-58

scores (I .8&.32 vs. 1.9k.4; t = .38, df =21, NS) and CGI scores (2.Ok.8 vs.

1.9+1 .l; t = .35, df =21, NS) did not differ significantly between groups. Thus the

two groups were collapsed for subsequent analyses,

The authors compared posttreatment and follow-up assessment scores of patients

who were on or off medication at the time of the follow-up interview (Table 2). A

2X2X4 repeated measures multivariate analysis of variance (MANOVA), with the

HDRS, CDRS, SCL-58 and CGI as dependent measures, revealed a significant

interaction of these two groups from post-treatment to follow-up (F(1,38) = 11.56,

p < .Ol). A significant interaction of the different indices over time was found

(F(3,36) = 10.34, p < .OOl), as well as a three-way interaction of the assessment

indices over time between the on- and off-medication groups (F(3,36) = 3.75, p < .Od).

While there was no significant difference between these groups at posttreatment, at!

follow-up the overall scores of the group off medication were significantly higher thdn

those of the group on medication (F(1,38) = 12.18, p < .OOl). Univariate F tests

revealed a significant group by time interaction with the HDRS (F(1,38) =6.13,

p<.O5), the CDRS (F(1,38)=11.72, p<.OOl), and the CGI (F(1,38)=5.63, pC.051,

with the off-medication group evidencing significantly higher HDRS (F(1,38) = 6.58,

p<.O5), CDRS (F(1,38)= 14.51, p<.OOl), and CGI (F(1,38) =16.50, p<.OOl) scores

than the on-medication group at follow-up. The SCL-58 scores of the two groups

were not significantly different either at post-treatment or follow-up. There was no

significant difference between groups with respect to the number of weeks from

Tab

le

2.

A C

om

par

iso

n

of

Ass

essm

ent

Sco

res

of

Pat

ien

ts

Cu

rren

tly

on

(N

= 2

4)

and

off

(N

= 1

6)

Med

icat

ion

at

Po

st-T

reat

men

t an

d

Fo

llow

-Up

.

Po

sttr

eatm

ent

Fo

llow

-up

a,b

,c,d

O

n M

edic

atio

n

Off

M

edic

atio

n

On

Med

icat

ion

O

ff

Med

icat

ion

(N

= 2

4)

(N =

16)

(N

= 2

4)

(N=1

61

Mea

n

SD

M

ean

S

D

Mea

n

SD

M

ean

S

D

HD

RS

C

DR

S

SC

L-5

8 C

GI

7.54

4.

10

7.56

4.

98

9.42

4.

14

13.4

4 5.

79e.

h

20.2

1 11

.23

17.0

0 13

.18

19.3

3 7.

53

29.7

5 9.

74f.

i 1.

57

.37

1.45

.2

4 1.

84

.34

1.90

.3

7 2.

33

.96

2.69

1.

08

1.96

.8

6 3.

12

l.O5g

,j tl

t

No

te.

HD

RS

= H

amilt

on

D

epre

ssio

n

Rat

ing

S

cale

(H

amilt

on

, 19

60);

C

DR

S =

Co

rnel

l D

ysth

ymia

R

atin

g

Sca

le

(Mas

on

et

al.,

19

93);

S

CL

-58

= H

op

kin

s S

ymp

tom

C

hec

klis

t (D

ero

gat

is

et a

l.,

1974

);

CG

I =

Clin

ical

G

lob

al

Imp

ress

ion

s (G

uy,

19

76).

Mu

ltiv

aria

te

F v

alu

es:

a F

(1,3

8)

= 11

.56,

p

< .

Ol;

tr

eatm

ent

gro

up

b

y ti

me

inte

ract

ion

b

F

(3,3

6)

= 10

.34,

p

< .

OO

l; as

sess

men

t in

dex

b

y ti

me

inte

ract

ion

c

F(3

,36)

=

3.75

, p

< .0

5;

trea

tmen

t g

rou

p

by

asse

ssm

ent

ind

ex

by

tim

e in

tera

ctio

n

d

F(1

,38)

=

12.1

8,

p <

,00

1;

off

-med

icat

ion

g

rou

p

> o

n-m

edic

atio

n

gro

up

at

follo

w-u

p

Un

ivar

iate

F

val

ues

: e

F(1

,38)

=

6.13

, p

< .

05;

trea

tmen

t g

rou

p

by

tim

e in

tera

ctio

n

f F

(1.3

8)

= 11

.72,

p

C .

OO

l; tr

eatm

ent

gro

up

b

y ti

me

inte

ract

ion

g

F

(1,3

8)

= 5.

63,

p <

.05

; tr

eatm

ent

gro

up

b

y ti

me

inte

ract

ion

h

F

(1,3

8)

=6.5

8,

p<

.05;

o

ff-m

edic

atio

n

gro

up

>

on

-med

icat

ion

g

rou

p

at f

ollo

w-u

p

i F

( 1,3

8)

= 14

.51,

p

< .

OO

l ;

off

-med

icat

ion

g

rou

p

> o

n-m

edic

atio

n

gro

up

at

follo

w-u

p

j F

(1,3

8)

= 16

.50,

p

< .

05;

off

-med

icat

ion

g

rou

p

> o

n-m

edic

atio

n

gro

up

at

fo

llow

-up


posttreatment to follow-up (N = 24:37.2&_9.0; N = 16:38.9&10.5).

The authors did an additional analysis of relapse at follow-up, between patients on

and off medication, where relapse was operationally defined as an HDRS score > 13

at follow-up. Of patients on medication at follow-up, only 4/23 (17.4K) had relapsed,

versus 7/14 (50%) of those who were off medication (3 cases deleted for missing

data). While this was a significant difference (Chi sq with Fisher’s Correction

factor =4.43, df = 1, p = .0424), this finding must be qualified by the fact that the

patients off medication were a heterogeneous sample (including medication

responders, medication nonresponders, and patients never treated with medication).

In addition to continuation of medication use, it has been suggested (Kocsis, et al.,

1991) that initial medication response may predict good long-term outcome. Of the

26 patients who had completed a medication trial (as opposed to placebo) in the

original studies, 18 were previously defined as medication responders and eight as

medication nonresponders, based on the two-dimensional criteria described earlier (Fig

I). Comparative scores for these groups on post-treatment and follow-up HDRS,

CDRS, SCL-58 and CGI are shown in Table 3. Because of the small size of these

samples, statistical analysis was not performed. The mean scores of the responder

group were lower at follow-up than those of the nonresponder group, and it is possible

that these differences would be statistically significant with a larger sample size. A

similar comparison was performed for the 26 patients who had completed a

medication trial (as opposed to placebo) in the original studies, of whom 18 were

previously defined as medication responders and eight as medication nonresponders

(Fig 1 and Table 4). Again, small sample size precludes statistical analysis, but scores

on several scales (HDRS, CDRS, CGI, but not SCL-58) suggest the possibility that

medication responders who discontinue medication may have recurrence of

symptoms, whereas patients remaining on medication may have persistent remission;

436 D.J. Hellerstein et al.

I. First analysis: Comparison of patients on and off medication at the follow-up interview.

N=40

J\

N=24 N=16 On medication Offmedication

II. Second analysis: Comparison of initial medication responders and nonresponders at follow-up.

N=40

N=26 N=14 Medication treated Placebo treated

N=18 N=8 Initial medication Initial medication

responders nonresponders

III. Third analysis: Comparison of medication responders who had continued or discontinued medication at follow-up.

N=26 Medicatio treated

/L

N=18 N=8 Initial medication Initial medication

respon ers nonresponders

/\ N=12 N=6

Medication responders Medication responders

on medication at off medication at follow-up follow-up

Fig 1. A description of the patient subsamples used in the three analyses


Table 3.

A Comparison of Follow-Up Assessment Scores for the Responder (N = 18) and Nonresponder (N = 8) Groups.

Scales Responder Nonresponder

Mean SD Mean SD

HDRS 9.22 4.63 11.38 4.34

CDRS 21.06 10.79 25.50 6.57

SCL-58 1.82 .32 1.96 .37

CGI 2.00 1.08 3.00 .93

Note. HDRS = Hamilton Depression Rating Scale (Hamilton, 1960); CDRS = Cornell Dysthymia Rating Scale (Mason et al., 1993); SCL-58 = Hopkins Symptom Checklist

(Derogatis et al., 1974); CGI = Clinical Global Impressions (Guy, 1976).

Now that there is an increasing body of literature indicating the short-term

efficacy of antidepressants in dysthymia, one major question is whether relapse

occurs when medication is continued. Given the scarcity of data on longterm

treatment of dysthymics, there is a value even to preliminary data suggesting that

patients continuing medication show significant alleviation of symptoms and less

relapse than patients off medication.

In earlier studies (Hellerstein et al., 1993; Rosenthal et al., 1992; Hellerstein et

al., 19941, the authors demonstrated the response of dysthymic symptoms to

acute treatment with fluoxetine and trazodone. The results of this preliminary

naturalistic follow-up study confirm the hypothesis that dysthymia patients who

remain on medication at follow-up generally maintain improvement over time: Of

the overall sample of 40 patients, the 24 patients who remained on medication for

an average of 35.9A25.3 weeks, HDRS-24 scores were 9.42+4.14 at follow-up,

not significantly different from the scores at termination (7.54+4.10). Furthermore

the rate of relapse is low (17.4%; 4/23) in patients remaining on medication. We

also hypothesize that initial medication response may predict response at follow-up,

but current sample sizes are too small to conclusively demonstrate this. This issue

Tab

le

4.

A C

om

par

iso

n

of

Ass

essm

ent

Sco

res

of

Med

icat

ion

R

esp

on

der

s O

n

(N =

12)

an

d O

ff

(N =

6)

Med

icat

ion

at

Pre

trea

tmen

t,

Po

sttr

eatm

ent

and

F

ollo

w-U

p.

Pre

trea

tmen

t P

ost

trea

tmen

t N

=12

N=6

N

=12

N=6

M

ean

S

D

Mea

n

SD

M

ean

S

D

Mea

n

SD

Fo

llow

-up

N

=12

N=6

M

ean

S

D

Mea

n

SD

HD

RS

18

.50

7.49

16

.17

4.62

5.

25

2.60

3.

83

1.72

7.

67

4.25

12

.33

3.98

CD

RS

39

.92

8.60

38

.00

8.02

12

.42

7.05

7.

83

5.42

17

.08

7.99

29

.00

11.9

0 u

t

SC

L-5

8 2.

03

.37

1.92

.2

7 1.

58

.45

1.30

.1

7 1.

83

.34

1.80

.3

3 g

K

C

GI

1.58

.5

2 1.

67

.52

1.67

.7

8 2.

67

1.34

1 8 5

No

te.

HD

RS

= H

amilt

on

D

epre

ssio

n

Rat

ing

S

cale

(H

amilt

on

, 19

60);

C

DR

S =

Co

rnel

l D

ysth

ymia

R

atin

g

Sca

le

(Mas

on

et

al.,

19

93);

S

CL

-58

= H

op

kin

s S

ymp

tom

C

hec

klis

t (D

ero

gat

is

et a

l.,

1974

);

CG

I =C

linic

al

Glo

bal

Im

pre

ssio

ns

(Gu

y,

1976

).

$ r


requires further study.

The findings in this study are in many ways similar to those of Kocsis’s

naturalistic follow-up study of 25 patients (Kocsis et al., 1991). In both studies,

patients by self-report state that their depressive symptoms average 20 or more

years in duration, and prior treatment consisted primarily of psychotherapy, and in

both studies patients continuing medication generally have persistent relief. One

significant difference is that whereas 96% (24/25) of Kocsis et al. (1988)‘s

patients suffered from “double depression, ” in our sample only 12.5% were doubly

depressed, and 87.5% (35/40) met criteria for dysthymia only. Thus our study

suggests that dysthymia uncomplicated by major depression responds on a long-

term basis to medication.

. . . LlmlmmsofIbisstuIiy

The present findings must be qualified by a number of factors. Follow-up

assessment was done cross-sectionally rather than longitudinally, and was done at

variable time intervals. Sample sizes were small and heterogeneous. The group off

medication at follow-up included patients who discontinued medications for a

variety of reasons, and some patients who never received medication, and

therefore there is no definitive evidence that discontinuation leads to relapse,

though our preliminary findings suggest that it may. Controlled discontinuation

studies are needed to assess this issue. In addition, small sample size makes it

impossible to rule out possible Type II errors, and replication is obviously required.

Patients were drawn from two studies, one with open-label and the other with

placebo-controlled design; however, differences are minimized by the continuity

both of study staff and assessment procedures between studies. In addition,

patients continuing medication also may have received therapeutic benefit from

contact with physicians, whereas patients who discontinued medication generally

ended ongoing treatment contact. The authors assessed other possibly important

factors (psychotherapy in follow-up period, variable duration of medication

treatment, and characteristics of patients agreeing to or refusing follow-up

assessment), and found no significant differences.


The lack of a significant between-group finding for the SCL-58, a patient-rated

scale, in comparison to the other (clinician-rated) inventories, raises the possibility

of rater bias. However, in the double-blind study (Hellerstein et al., 19931, the

SCL-58 also did not differ significantly between placebo and fluoxetine groups.

Studies of major depression (Greenberg, et al., 1992) also show consistently less

medication effect as assessed by patient-rated than clinician-rated scales, the

reasons for which are unclear. Additional patient-rated and observer-rated

inventories should be used in dysthymia outcome studies to clarify this issue.

Finally, there is the question of generalizability. The present sample consists of

patients with primary dysthymia, most without concurrent major depression.

Epidemiological studies (Klein et al., 1988; Markowitz 1993; Horwath et al., 1992)

suggest that about 75% of dysthymics have other comorbid medical or psychiatric

disorders, whereas our patients were screened to exclude such disorders. It is

important to determine whether similar findings apply to primary dysthymics with

comorbid disorders (e.g. anxiety disorders, etc.), and whether they apply to

patients with secondary dysthymia (e.g. dysthymia secondary to other psychiatric

or medical disorders).

The preliminary findings thus suggest that continued antidepressant treatment

leads to long-term symptom improvement in many dysthymics. Discontinuation of

medication may be associated with recurrence of symptoms. These results require

replication in prospective, controlled studies with blinded raters, and with a variety

of patient and observer-rated instruments.

This study was supported by a grant from Eli Lilly and Company.


AMERICAN PSYCHIATRIC ASSOCIATION (1987) Diagnostic and Statistical Manual of Mental Disorders, 3rd ed, revised. Washington DC.

DEROGATIS, L.R. LIPMAN, R.S., RICKELS, K. UHLENHUTH, E.H. and COVI, L. (1974) The Hopkins Symptom Checklist (HSCL): A Self-Report Symptom Inventory. Behav Science I.9 l-l 5.

FRIEDMAN, R.A. (1993) Social Impairment in Dysthymia. Psychiat Annals 23, 632- 637.

GREENBERG, R.P., BORNSTEIN, R.F., GREENBERG, M.D. and FISHER, S. (1992) A Meta-Analysis of Antidepressant Outcome Under “Blinder” Conditions. J Consulting and Clin Psycho1 6.0, 664-669.

GUY, W. (1976) ECDEU Assessment Manual for Psychopharmacology. DHEW Pub No (ADM)76-338. Rockville, MD: National Institute of Mental Health.

HAMILTON, M. (1960) A Rating Scale for Depression. J Neurosurg Psychiat 25, 56-62.

HARRISON, W., RABKIN, J., STEWART, J.W., MCGRATH, P.J., TRICAMO, E. and QUITKIN, F. (1986) Phenelzine for Chronic Depressions: A Study of Continuation Treatment. J Clin Psychiat 42, 346-9.

HELLERSTEIN, D.J., YANOWITCH, P., ROSENTHAL, J., SAMSTAG, L.W., MAURER, M., KASCH, K., BURROWS, L., POSTER, M., CANTILLON, M. and WINSTON, A. (1993) A Randomized Double-Blind Study of Fluoxetine Versus Placebo in Treatment of Dysthymia. Am J Psychiat l5.6, 1169-l 175.

HELLERSTEIN, D.J., YANOWITCH, P., ROSENTHAL, J., HEMLOCK, C., KASCH, K., SAMSTAG, L.W. and WINSTON, A. (1994) Long-Term Treatment of Double Depression: a Preliminary Study with Serotonergic Antidepressants. Prog Neuropsychopharmacol and Biol Psychiat I.8 139-l 47.

HORWATH, E., JOHNSON, J., KLERMAN, G.L. and WEISSMAN, M. (1992) Depressive Symptoms as Relative and Attributable Risk Factors for First-Onset Major Depression. Arch Gen Psychiat 49, 817-823.

HOWLAND, R.H. (1991) Pharmacotherapy of Dysthymia: a Review. J Clin Psychopharmacol 11, 83-92.

HOWLAND, R.H. (1993) Chronic Depression. Hospital and Comm Psychiat 44, 633- 639.

KELLER, M.B., KLEIN, D.N., HIRSCHFELD, R.M.A., KOCSIS, J.H., MCCULLOUGH, J.P., MILLER, I., FIRST, M.B., HOLZER C.P.3rd, KEITNER, G.I., MARIN, D.B., and SHEA, T. (1995) Results of the DSM-IV Mood Disorders Field Trial. Am J Psychiat l!i2, 843-849.

KELLER, M.B., LAVORI, P.W., ENDICOTT, J., CORYELL, W. and KLERMAN, G.L. (1983) “Double Depression”: Two Year Follow-Up. Am J Psychiat I4.9, 689694.

KELLER, M.B. and RUSSELL, C.W. (1991) Refining the Concept of Dysthymia. Hosp and Commun Psychiat 42, 892-893.


KELLY, NW., PERRY, P.J., HOLSTAD, S.G. and GARVEY, M.J. (1989) Serum Fluoxetine and Norfluoxetine Concentrations and Antidepressant Response. Therapeutic Drug Monitoring Il, 165-I 70.

KLEIN, D.N., CLARK, D.C., DANSKY, L. and MARGOLIS, E.T. (1988) Dysthymia in the Offspring of Parents with Primary Unipolar Affective Disorder. J of Abnorm Psycho1 9.2, 265-274.

KLEIN, D.N., TAYLOR, E.B., DICKSTEIN, S. and HARDING, K. (1988) Primary Early- Onset Dysthymia: Comparison with Primary Nonbipolar Nonchronic Major Depression on Demographic, Clinical, Familial, Personality, and Socioenvironmental Characteristics and Short-Term Outcome. J of Abnorm Psycho1 92, 387-398.

KOCSIS, J.H., FRANCES, A.J., VOSS, C.B., MANN, J.J., MASON, B.J. and SWEENEY, J. (1988) lmipramine for Treatment of Chronic Depression. Arch Gen Psychiat 45, 253-257.

KOCSIS, J.H., SUTTON, B.M. and FRANCES, A.J. (1991) Long-Term Follow-Up of Chronic Depression Treated with Imipramine. J Clin Psychiat 52, 56-59.

MARKOWITZ, J.C. (1993) Comorbidity of Dysthymia. Psychiatr Annals 23, 617-

624.

MASON, B.J., KOCSIS, J.H., LEON, A.C., THOMPSON, S., FRANCES, A.J., MORGAN, R.O. and PARIDES, M.K. (1993) Measurement of Severity and Treatment Response in Dysthymia. Psychiatr Annals 23, 625-637.

QUITKIN, F.M., MCGRATH, P.J., STEWART, J.W., HARRISON, W., WAGER, S.G., NlJNES, E., RABKIN, J.G., TRICAMO, E., MARKOWITZ, J. and KLEIN, D.F. (1989) Phenelzine and lmipramine in Mood-Reactive Depressives. Arch Gen

Psychiat 46, 787-793.

ROSENTHAL, J., HEMLOCK, C., HELLERSTEIN, D.J., YANOWITCH, P., KASCH, K., SCHUPAK, C., SAMSTAG, L.W. and WINSTON, A. (1992) A Preliminary Study of Serotonergic Antidepressants in Treatment of Dysthymia. Prog Neuropsychopharmacol and Biol Psychiat IL6 933941.

SPITZER, R.L., WILLIAMS, J.B.W., GIBBON, M. and FIRST, M.D. (1988) Structured Clinical Interview for DSM-Ill-R--Patient Version (SCID-P). New York, New York State Psychiatric Institute, Biometrics Research.

Inquiries and reprint requests should be addressed to:

David J. Hellerstein, MD 212-420-2800; FAX 212-420-3936 Psychiatric Outpatient Services Beth Israel Medical Center First Avenue and 16th St. New York, NY 10003 USA

Follow-up assessment of medication-treated dysthymia

Documents

Transcript of Follow-up assessment of medication-treated dysthymia