Escolar Documentos
Profissional Documentos
Cultura Documentos
456
INTERNE T TREAT MENT OF DEP RES SI ON
(h) prepared to work with the self-help were automatically sent to the therapist, who regardless of how many treatment modules
programme several hours each week; in turn gave e-mail feedback on the answers they had completed. This could also be
(i) no obstacle to participation (e.g. travel
and gave the participant access to the next referred to as ‘intention to treat’, as we
abroad during the treatment, major treatment module within 24 h. Each module included all those who provided post-
surgery); was available on the website in hypertext treatment data. However, for the main
markup language (HTML) for-mat. The outcome measure we also calculated results
(j) completion of the pre-treatment website was built by JavaServer Pages (JSP) on a last observation carried forward basis,
assessment.
programming and MySQL databases. The replacing missing values post-treatment
participant could also print each module by with pre-treatment values. Since this pro-
Participants were randomised by an first downloading them as rich text format or cedure assumes that values remain frozen
independent person (not involved in the as portable document format documents in time, we did not report this for all
study or recruitment), who drew the (PDFs). The amount of time advised for measures. Confidence intervals, analysis of
numbers of the (consecutively numbered) completion of all five mod-ules was 8 weeks. variance and t-tests were used for outcome
applicants from a bowl and placed them However, the mean time for completion was analyses. Significance was set at 0.05 and
alternately into one of two separate envel- 10 weeks. The time spent on each participant all tests were two-tailed. Power was esti-
opes, which were handed to the researchers for completion of treatment was estimated to mated by assuming an effect size (Cohen’s
later. Once allocated to treatment or the be 2 h in total, including screening, d, defined as the standardised difference
control condition, each participant was sent responding to e-mails and monitoring the between groups obtained by calculating the
an e-mail with a log-in user name. On log- discussion group. In total 506 messages were mean difference and dividing by their
ging in for the first time, the participants sent to the par-ticipants, which included a few pooled standard deviation) of 0.80, which
were required to fill out the pre-treatment e-mails sent to the control group. would require 52 participants to obtain a
questionnaires. However, participants were power of 80% with a conventional alpha
not informed about their group status until level of 0.05. The actual power for the main
The discussion groups were separate
they had completed the questionnaires. outcome measure with 85 participants was
and differed in their content, since the
groups had different topics to discuss. For over 95%.
Outcome measures example, the treatment group could discuss
The principal outcome measure of de- the contents of the self-help material,
pression was the 21-item Beck Depression whereas the control group was more likely RESULTS
Inventory (BDI; Beck et al, 1961), and the to bring up topics such as sick leave and the
results are based upon this instrument. We experience of being depressed. All activity Of the 343 persons who completed the
also included MADRS–S (9 items), the 21- in the discussion groups was closely inclusion forms, 117 (34%) were included
item Beck Anxiety Inventory (BAI; Beck et monitored, with the possibility of deleting (Fig. 1); 226 persons were excluded from
al, 1988) and the Quality of Life Inventory inappropriate postings. However, this never the study. The most common reason for
(QoLI; Frisch et al, 1992). The QoLI occurred. In addition, the thera-pists in the exclusion was risk of suicide (n¼77).
includes 16 dimensions of life (e.g. health, study answered some of the questions posed This was measured by item 9 on the
economy); for each dimension a rating is by members of the discus-sion groups when MADRS–S, and/or reported previous sui-
made regarding importance (scored 0 to 2) appropriate, for example questions cide attempts. Since the aim was to target
and of how pleased the person is with that regarding the website. people with mild-to-moderate depression,
dimension (scored 73 to +3, but with no 0 Each time a participant in either group 67 persons were excluded after reporting
alternative). The QoLI has been reported to logged on to the website, the MADRS–S severe depression (a score of more than 30
have satisfactory reliability and validity was automatically administered, with the on the MADRS–S) and 36 after reporting
(Frisch et al, 1992). All outcome measures restriction that at least 7 days had to have minor depression (a score below 15 on the
were administered using the internet. passed since the previous form was com- MADRS–S). The CIDI–SF was used to
pleted. This was done in order to monitor obtain a probability of diagnosis according
Treatment conditions depression levels – and in particular zest to the full CIDI, and 32 persons were
for life – on a regular weekly basis. excluded after reporting a probability
The cognitive–behavioural self-help treat-
below 0.55 (see above). To control for
ment was based on Beck’s cognitive
Follow-up effects of recent medication (e.g. initial side-
therapy, as presented in numerous sources
effects), 65 persons who had started
(e.g. Burns, 1999), and on behavioural For ethical reasons the control group mem-
antidepressant medication, or altered its
activation (Lewinsohn et al, 1986; Martell bers were given access to the treatment
modules after the intervention group had dosage within the last month, were
et al, 2001). The material (presented in
finished their treatment. Participants were excluded. Use of antipsychotic medication
Swedish) consisted of 89 pages of text,
contacted by e-mail and asked to fill in the led to the exclusion of 2 persons. People
divided into five modules: introduction;
questionnaires again on the internet 6 meeting the DSM criteria for bipolar dis-
behavioural activation; cognitive restruc-
turing; sleep and physical health; and months after the treatment had ended. order (American Psychiatric Association,
relapse prevention and future goals. The 1994) were excluded (n¼28). Other rea-
sleep module was based on a programme sons for exclusion were not filling out the
Analysis
for insomnia (Stro¨m et al, 2004). Each pre-treatment measures (n¼11), receiving
All randomised participants with follow-up cognitive–behavioural therapy before the
module ended with a quiz, with questions
data were included in the analyses trial start (n¼6), being under 18 years old
on the content of the module. Responses
4 57
ANDERS SON E T AL
458
INTERNE T TREAT MENT OF DEP RES SI ON
Adherence and separately. At this stage the control group in an internet support group (Houston et
modules completed also had received the treatment. al, 2002). Finally, being engaged in self-help
Participants in the intervention group nor- Analysis of the difference between the seems to affect the tendency to participate
groups at follow-up revealed no statisti- in a discussion group, with less activity in
mally reached at least the fourth module,
cally significant difference. Improvements the group who received the active
with 65% completing all modules. The
between pre-treatment and follow-up were, treatment immediately.
average number of modules completed was
however, found for both groups on the BDI,
3.7 (s.d.¼1.9). The number of modules Adherence
MADRS–S, BAI and the QoLI (see Table 2
completed was weakly correlated with post- for change scores and confidence intervals). Although self-administered treatments have
treatment BDI scores (Spearman’s r¼70.33, As the control group had received shown promising results in many studies, a
P50.05). treatment, we expected changes between crucial problem is how well participants
post-treatment and follow-up for this adhere to the treatment. For example, in a
Activity in discussion groups
group, but no difference for the treatment recent study only 41 out of 139 randomised
Activity in the discussion group was not group. This assumption was confirmed by participants were available for assessment at
correlated with improvement in the treat- means of paired t-tests for the BDI, the 3-month follow-up (Richards et al,
ment group. However, there was a marked MADRS–S, BAI and QoLI (all P values 2003), which makes it likely that some
difference in activity between the discussion were less than 0.05 in the control group and failed to complete the self-help material.
groups, with a total of 233 postings in the more than 0.05 in the treatment group). Internet-based self-help facilitates monitor-
treatment discussion group and 842 post- ing of adherence to treatment, because
ings in the control discussion group, which modules are provided only when the pre-
was also reflected in the mean difference vious module has been completed. There
DISCUSSION
be-tween the groups of 711.0 (95% CI) was, however, a differential rate of with-
721.5 to 70.6. Overall, the form of activity drawal between the two groups, and
This randomised controlled trial of internet-
differed between the groups, as the control judging from the comments we received,
delivered self-help based on cognitive–
group tended to discuss their own problems some perceived the text and the exercise as
behavioural therapy yielded three major
more, whereas the treatment group leaned too demanding. A solution to this is to
results. First, the active treatment, which
more towards discussing the treatment. adjust the text, and to allow a longer
included standard cognitive–behavioural
approaches and behavioural activation, treatment period.
Follow-up resulted in decreased depressive symptoms
Implications for cognitive^
At the 6-month follow-up, 71 participants immediately after treatment and at the 6-
(all in the treatment group and 35 in the month follow-up. Benefits were also observed behavioural therapy
control group) completed the question- regarding anxi-ety symptoms and quality of Self-help treatment of depression is an
naires again, yielding a 16% rate of life. Second, participation in a web-based attractive treatment option, as practitioners
withdrawal from post-treatment to follow- discussion group only had no effect on often wish to offer their clients effective
up (0% in the treatment group and 29% in depressive symptoms, which is in contrast to a psychosocial interventions, but hesitate to
the control group). Table 2 shows the study showing some benefits from do so because of lengthy waiting lists
outcomes at follow-up for each group participation (Williams & Whitfield, 2001). Indeed,
4 59
ANDERS SON E T AL
developing self-help approaches has been cognitive–behavioural therapy does not diagnoses before initiation of treatment; in
recommended several times (Hollon et al, yield incremental improvements, this our study, we did not use a clinician-
2002). Our study was preceded by other cannot be directly inferred, given that a administered interview. However, one of the
applications of internet-based self-help therapy-only group was not included. A potential benefits of internet-delivered
treatments (Carlbring et al, 2001), and differs plausible explanation for the lack of an treatments is that geographical distances are
from other applications of internet-based effect in the waiting-list discussion group immaterial. Requiring participants to come in
treatments of depression (Clarke et al, 2002). could be that the patients were aware of for a clinical assessment would therefore
First, we divided the material into modules to being placed on a waiting list, and hence introduce a limitation. It is poss-ible that the
be provided on a consecu-tive basis dependent were not expecting any change from internet could be used for diagnoses in the
on progress. Second, individualised feedback participation in their group. future, perhaps comple-mented with web-
was given by a therapist who was clearly camera technology or video conferencing. The
identified with a name and a photograph on Limitations validity of such procedures has yet to be
the website. In a recent review it was Although self-report was used to obtain a assessed. Internet technology might also be
concluded that self-help results in effect sizes likely diagnosis using DSM criteria, no for- used in the future for preventing relapse,
roughly equivalent to the average effect size mal diagnosis was made in an interview. perhaps in combina-tion with medication. All
obtained in psychotherapy studies Hence, it is possible that people with these suggestions point to the importance of
(McKendree-Smith et al, 2003). In common depression were excluded and people with- evaluating the cost-effectiveness of internet
with our study, most self-help studies on out depression were included. However, this treatments. No attempt was made here to do
depression would be better described as is not very likely, particularly the latter this, as a proper assessment of costs would
testing minimal therapist contact treat-ments, possibility of including people who would include the costs of programming and
as it is common to have either meetings or not fulfil DSM depression criteria in a computer equipment, as well as therapist time
telephone calls to monitor progress and structured interview. Internet administra- de-voted to writing the self-help material and
adherence. Internet-based self-help does not tion of both interviews and questionnaires processing the participants’ responses to the
therefore exclude clini-cian input and can be is a research area on its own that needs modules. Finally, effective mechanisms are yet
demanding for the therapist. However, given further investigation. Independent ratings to be disclosed, as most studies of cognitive–
that responses are not given directly in ‘real by clinicians would have strengthened the behavioural therapy include packages of
time’, collea-gues can be consulted and self-reported findings, but was not done, treatment ingredients. Our study was no
specific questions can be directed to the given that participants were not requested exception in this respect.
specialist, all being done within 24 h. For to attend a research clinic.
example, in our study the psychiatrist was Confounding with respect to medi-cation
consulted about some of the participants’ status cannot be ignored. First, self-report was ACKNOWLEDGEMENTS
questions, whereas the psychologists handled used to ascertain medication use. Second,
other questions deal-ing with the contents of those with ongoing but stabilised medication The L.J. Boethius Foundation and Swedish
the programme. It is, however, interesting to regimens were not excluded. Although no Research Council are acknowledged for funding
compare our findings with the results of effect of medication status was found, in line this study.We also thank Olafur Jakobsson for
providing web host-ing and technical support at
Proudfoot et al (2003), who used a stand- with other research (Oei & Yeoh, 1999), a
Uppsala University Hospital, and Daniel Gidlo«f for
alone computer in a general practice setting, better approach would have been to control
JSP and database programming.
and Christen-sen et al (2004), who used an for medication status in the first place in order
open web page, both finding promising to enable investigation of drug–therapy
results. interactions. REFERENCES
The study period was relatively short,
American Psychiatric Association (1994) Diagnostic
Interestingly, participation in the dis- and it would have been preferable to have and Statistical Manual of Mental Disorders
cussion group only did not confer any had a control group that had not received (4th edn) (DSM^IV).Washington, DC: APA.
immediate benefits; this is in contrast to an any self-help or psychotherapy at the 6- Beck, A. T.,Ward,C. H., Mendelson, M., et al (1961)
observational study in which benefits were month follow-up, or at least data from a An inventory for measuring depression.
found (Houston et al, 2002). Findings in the less specific attention control treatment. Archives of General Psychiatry, 4, 561^571.
latter study were attributed to the effects of This was not possible in the study protocol Beck, A. T., Epstein, N., Brown, G., et al (1988) An
breaking down social isolation by for ethical reasons. inventory for measuring clinical anxiety.
Psychometric properties. Journal of Consulting
participation in the discussion group. As we Limited access to the internet is often
and Clinical Psychology, 56, 893^897.
did not include any measure of social put forward as an argument against
Burns, D. D. (1999) Feeling Good. The New
isolation, we cannot exclude the possibility internet treatments, and although a large
Mood Therapy. NewYork: Avon Books.
that members of our study group were less proportion of the Swedish population does
socially isolated. On the other hand, have access to the internet (about 70%), Carlbring, P.,Westling, B. E., Ljungstrand, P., et al
(2001) Treatment of panic disorder via the Internet
participants in the waiting-list discussion there is still a significant minority who do ^ a randomized trial of a self-help program.
group spent more time with the discussion not, and this is even more the case in Behavior Therapy, 32, 751^764.
group compared with the therapy group, countries outside northern Europe.
Christensen, H. & Griffiths, K. M. (2002) The
most probably because the therapy group prevention of depression using the Internet.
members were occupied with the treatment. Further directions Medical Journal of Australia, 177, S122^S125.
Although our study indicates that adding Research on internet-based self-help for Christensen, H., Griffiths, K. M. & Jorm, A.
discussion group activity to depression would benefit from clear-cut (2004) Delivering interventions for depression by
460
INTERNE T TREAT MENT OF DEP RES SI ON
Frisch, M. B.,Cornell, J.,Villanueva, M., et al (1992) & Outcome with internet-based therapy resembles that in
Clinical validation of the Quality of Life Inventory: a measure of life
controlled studies of clinician-delivered therapy.
satisfaction for use in treatment planning and outcome
assessment. Psychological Assessment, 4,
& Participation in a discussion group within a controlled study
92^101.
does not seem to confer any benefits.
Hollon, S. D., Munoz, R. F., Barlow, D. H., et al (2002)
Psychological intervention development for the LIMITATIONS
prevention and treatment of depression:
promoting innovation and increasing access.
Biological Psychiatry, 52, 610^630. & No structured clinical interview was conducted with a
Houston,T. K.,Cooper, L. A. & Ford, D. E. (2002)
clinician to secure diagnoses.
Internet support groups for depression: a 1-
year prospective cohort study. American & Medication status was not controlled.
Journal of Psychiatry, 159, 2062^2068.
& Lack of access to the internet might be a problem in future
Kessler, R. C., Andrews, G., Mroczek, D., et al
(1998) The World Health Organization composite
dissemination of the therapy.
international diagnostic interview short-form (CIDI).
International Journal of Methods in
Psychiatric Research, 7, 171^185.
«
Lewinsohn, P. M., Munoz, R. F.,Youngren, M. A., et al GERHARD ANDERSSON,PhD,Department of Behavioural Sciences,Linko«ping University; JAN BERGSTROM,
(1986) Control Your Depression. NewYork: Prentice-Hall. MSc, Department of Clinical Neuroscience, Section of Psychiatry,Karolinska Institute, Stockholm, FREDRIK
«
HOLLANDARE, MSc, PER CARLBRING, PhD,VIKTOR KALDO, MSc, LISA EKSELIUS, MD, PhD, Department of
Marks, I. M., Mataix-Cols, D., Kenwright, M., et al
Psychology,Uppsala University, Department of Neuroscience,Uppsala University Hospital,Uppsala, Sweden
(2003) Pragmatic evaluation of computer-aided
self-help for anxiety and depression. British
Correspondence: Professor Gerhard Andersson,Department of Behavioural Sciences,Linko«ping
Journal of Psychiatry, 183, 57^65.
University, SE-58183 Linko«ping, Sweden. Fax: +46 (0)13 28 2145; e-mail: Gerhard.Andersson@ibv.liu.se
Martell,C. R., Addis, M. E. & Jacobson, N. S. (2001)
Depression in Context. Strategies for Guided (First received 27 May 2004, final revision 9 December 2004, accepted 21December 2004)
Action. New York: W.W. Norton.
Mattila-Evenden, M., Svanborg, P.,Gustavsson, P., et
Oei,T. P. S. & Yeoh, A. E. O. (1999) Pre-existing self-help cognitive behavioural therapy in primary care.
al (1996) Determinants of self-rating and expert
antidepressant medication and the outcome of British Journal of General Practice, 53, 764^770.
rating concordance in psychiatric out-patients,
group cognitive^behavioraltherapy. Australia and
using the affective subscales of CPRS. Acta
New Zealand Journal of Psychiatry, 33, 70^76.
Psychiatrica Scandinavica, 94, 386^396. Stro«m, L., Pettersson, R. & Andersson, G. (2004)
Internet-based treatment for insomnia: a
McKendree-Smith, N. L., Floyd, M. & Scogin, F. R. Proudfoot, J., Goldberg, D., Everitt, B., et al (2003)
Computerized, interactive, multimedia cognitive^ behavioural controlled evaluation. Journal of Consulting
(2003) Self-administered treatments for depression: a
program for anxiety and depression in general practice.
and Clinical Psychology, 72,113^120.
review. Journal of Clinical Psychology, 59, 275^288.
Psychological Medicine, 33, 217^227.
Montgomery, S. A. & —sberg, M. (1979) A new Williams,C. & Whitfield, G. (2001) Written and
depression scale designed to be sensitive to change. Richards, A., Barkham, M.,Cahill, J., et al (2003) computer-based self-help treatments for
British Journal of Psychiatry, 134, 382^389. PHASE: a randomised, controlled trial of supervised depression. British Medical Bulletin, 57,133^144.
4 61