Antidepressant use and mortality in Finland: a register-linkage study from a nationwide cohort.

Abstract

It is generally acknowledged that depressed patients need specific attention during the first weeks after initiation of antidepressant (AD) treatment because of the increased risk of suicide.
The study population consisted of all individuals residing in Finland from 1999 to 2003 who had purchased a prescribed antidepressant at least once but had no preceding antidepressant prescription. Data sources were the National Prescription Register, the Causes of Death Register, Census Data of Statistics Finland, and the National Care Register. Follow-up started at the first purchase and ended at the end of 2003 or death. Data on prescriptions were used to construct contiguous treatment periods of follow-up time. Life-table analysis with Poisson regression was used to estimate risk ratios (RR) of antidepressant use with respect to all-cause mortality and to deaths from suicide.
Current AD use was associated with a lowered all-cause mortality (RR = 0.18, 95% CI = 0.18-0.19) compared with those who filled one previous prescription only. There was no difference in suicide mortality when any current antidepressant usage was compared to the one-prescription group. Current SSRI usage was associated with lower risk of suicide compared to the one-prescription or other antidepressant groups (RR 0.47, 0.38-0.59).
Current AD treatment is associated with decreased all-cause mortality rates in patients who have ever had AD treatment.

Full text

Page 1

PHARMACOEPIDEMIOLOGY AND PRESCRIPTION
Antidepressant use and mortality in Finland:
a register-linkage study from a nationwide cohort
Jari Haukka & Martti Arffman & Timo Partonen &
Sinikka Sihvo & Marko Elovainio & Jari Tiihonen &
Jouko Lönnqvist & Ilmo Keskimäki
Received: 8 September 2008 /Accepted: 8 January 2009 /Published online: 5 March 2009
# Springer-Verlag 2009
Abstract
Background It is generally acknowledged that depressed
patients need specific attention during the first weeks after
initiation of antidepressant (AD) treatment because of the
increased risk of suicide.
Methods The study population consisted of all individuals
residing in Finland from 1999 to 2003 who had purchased a
prescribed antidepressant at least once but had no preceding
antidepressant prescription. Data sources were the National
Prescription Register, the Causes of Death Register, Census
Data of Statistics Finland, and the National Care Register.
Follow-up started at the first purchase and ended at the end
of 2003 or death. Data on prescriptions were used to
construct contiguous treatment periods of follow-up time.
Life-table analysis with Poisson regression was used to
estimate risk ratios (RR) of antidepressant use with respect
to all-cause mortality and to deaths from suicide.
Findings Current AD use was associated with a lowered
all-cause mortality (RR=0.18, 95% CI=0.18–0.19) com-
pared with those who filled one previous prescription only.
There was no difference in suicide mortality when any
current antidepressant usage was compared to the one-
prescription group. Current SSRI usage was associated with
lower risk of suicide compared to the one-prescription or
other antidepressant groups (RR 0.47, 0.38–0.59).
Interpretation Current AD treatment is associated with
decreased all-cause mortality rates in patients who have
ever had AD treatment.
Keywords Antidepressant.Suicide.Mortality.SSRI
Introduction
It is acknowledged that depressed patients should be
monitored intensively during the first weeks after initiation
of antidepressant (AD) treatment because of temporarily
increased risk of suicide [1]. However, the temporal
evolution of mortality and suicide risk after discontinuation
of AD treatment remains unknown. Based on ecological
evidence, the increased use of AD and especially of
selective serotonin re-uptake inhibitors (SSRI) is associated
with decreased rate of suicides [2–6].
Still, concerns have been raised about the potential
connection between antidepressant use and suicidal behav-
ior. This has been studied using meta-analysis, case-control
studies, ecological studies, and observational cohort stud-
ies. A meta-analysis based on drug company data from
placebo-controlled, randomized trials indicated that in-
creased risk of suicide and self-harm caused by SSRIs
cannot be ruled out [7], but on the other hand analysis of
FDA reports showed no difference in suicide risk between
Eur J Clin Pharmacol (2009) 65:715–720
DOI 10.1007/s00228-009-0616-9
J. Haukka (*):M. Arffman:T. Partonen:S. Sihvo:
M. Elovainio:J. Lönnqvist:I. Keskimäki
National Institute for Health and Welfare,
Helsinki, Finland
e-mail: jari.haukka@thl.fi
M. Arffman
e-mail: martti.arffman@thl.fi
J. Tiihonen
Department of Forensic Psychiatry,
University of Kuopio, Niuvanniemi Hospital,
Kuopio, Finland
J. Lönnqvist
Department of Psychiatry, University of Helsinki
and Helsinki University Central Hospital,
Helsinki, Finland

Page 2

antidepressant-treated and placebo-treated depressed sub-
jects in controlled trials [8].
A retrospective cohort study with a total of 57,361
patients and follow-up time of 120 days concluded that the
observed connection between suicidal behavior and pre-
scription of SSRIs is due to confounding by indication [9].
UK General Practice Research Database (GPRD) was used
in a matched case-control study that compared risk of
suicidal behavior among users of four antidepressant drugs
[10]. In a nested case-control study, the adjusted odds ratio
of suicide was 0.57 [95% confidence interval (95% CI)
0.26–1.25] in people prescribed SSRIs compared with those
prescribed tricyclic antidepressants [11]. In a case-control
study of an elderly population (66 years and older) in
Canada, SSRI therapy was associated with increased risk of
suicide during the month following its initiation [12]. In an
ecological study of association between use of antidepres-
sants and suicide rate in Finland, increased use of
antidepressants was associated with a decline in suicide
rate [13]. Similarly, the fall in suicide rates in Norway was
associated with the increased sales of non-TCA antidepres-
sants [14]. An aggregated data analysis of National Vital
Statistics from the U.S. Centers for Disease Control and
Prevention showed that a higher number of tricyclic
antidepressants (TCA) was associated with higher suicide
rates compared with SSRI use [15].
An observational, individual-level register linkage study
of 15,390 patients with previous attempted suicide showed
that treatment of any antidepressant was associated with
increased risk of attempted suicide and, at the same time,
decreased risk of completed suicide and death [16]. A recent
study based on the GPRD with 219,088 patients concluded
that there is excess risk of suicide associated with use of
venlafaxine, which is probably due to the higher burden of
suicide risk among venlafaxine users [17]. A register-based
study from Denmark showed lower rate of suicides for those
who continued antidepressant use compared to those who
purchased once (rate ratio 0.31, 95% CI 0.26–0.36) [18]. On
the other hand, a longitudinal register study from Finland
demonstrated that lower socio-economic groups received the
lowest levels of treatment despite evidence of their increased
prevalence of depression and suicide [19].
The aim of this study was to investigate, in a nation-wide
cohort of antidepressant users, how the risk of suicide and
all-cause mortality are associated with AD treatment.
Material and methods
Study population
In Finland, information on causes of death is recorded by
Statistics Finland. All reimbursed medication prescriptions
purchased from a pharmacy are registered by the Social
Insurance Institution. The linking of these registers is
possible because every individual residing in Finland has
a unique personal identification number.
Construction of the study population began by identify-
ing all people who had reimbursed antidepressant prescrip-
tions from 1999 to 2003, using the database of the National
Social Insurance Institution. The study population consists
of new antidepressant users and hence patients with
antidepressant purchases in the preceding 5 years were
excluded. Prescription data contained the personal identifi-
cation number, date of purchase, anatomic therapeutic code
(ATC), and amount purchased as defined daily doses
(DDD) [20]. Prescription data also included information
on all antipsychotic drug (ATC code N05) prescriptions,
while data on prescriptions during the 5 previous years
were used as background variable.
Statistics Finland provided socioeconomic background
variables for the study population by record linkage using
the unique personal identification numbers. The following
background variables were used in analyses: sex, age at
the beginning of follow-up (10–14, 15–19, 20–29, 30–39,
40–49, 50–59, 60–69, 70–79, and over 80 years), socio-
economic status (non-manual employee, manual employee,
farmer, other entrepreneur, student, other, or unknown),
marital status (married, other), employment status
(employed, 0–6 months unemployed, 7–12 months unem-
ployed, retired, or other outside of workforce), education
level (basic level, secondary or higher), family net income
quintile, living alone (yes, no), time since start of follow-up
(first antidepressant prescription in days), and calendar year
(1999–2003).
In addition, information on hospitalizations was obtained
from the Care Register for Health. The following dichot-
omous variables for each individual were constructed,
based on hospitalizations during the previous 5 years:
hospital-treated attempted suicide, alcohol-dependence di-
agnosis (ICD-10 starting with F1), schizophrenia, schizo-
typal and delusional disorder (F2), mood disorder (F3), and
personality disorder (F6). Causes and date of death were
obtained from the Causes of Death Register of Statistics
Finland. Cause of death was classified as suicide (ICD-10
X60*-X84) or all other causes. Follow-up of the individual
was started at the first purchase of antidepressant medica-
tion and finished at death or the end of 2003. Thus, the
longest possible length of follow-up was 5 years.
Data analysis
The duration of antidepressant treatment was calculated
according to the purchased DDD [21]. Contiguous pre-
scription periods were combined if the gap between them
was shorter than 30 days. The 10 most frequently used
716 Eur J Clin Pharmacol (2009) 65:715–720

Page 3

Table 1 Mortality and use of antidepressants
Current use of antidepressant Person yearsAll deaths (n) All cause rate per 100,000RR unadjustedRR adjusted
RR 95% CIRR 95% CI
SSRI
TCA
Other
None
139,710
7,710
81,880
340,730
1,358
130
716
11,704
972
1,686
874
3,435
0.25
0.43
0.22
0.88
0.24
0.36
0.21
0.86
0.27
0.52
0.24
0.91
0.18
0.23
0.19
0.85
0.17
0.20
0.18
0.82
0.19
0.28
0.20
0.88
Citalopram
Fluoxetine
Sertraline
Mirtazapine
Paroxetine
Amitriptyline
Venlafaxine
Fluvoxamine
Moclobemide
Mianserin
Other
None
68,041
27,502
14,624
14,270
12,839
5,223
3,592
3,176
2,469
1,870
75,695
340,730
937
187
86
211
61
93
28
14
11
32
544
11,704
1,377
680
588
1,479
475
1,781
780
441
446
1,711
719
3,435
0.35
0.17
0.15
0.38
0.12
0.46
0.20
0.11
0.11
0.44
0.18
0.88
0.33
0.15
0.12
0.33
0.09
0.37
0.14
0.07
0.06
0.31
0.17
0.86
0.38
0.20
0.19
0.44
0.16
0.56
0.29
0.19
0.21
0.62
0.20
0.91
0.17
0.20
0.16
0.18
0.22
0.26
0.23
0.16
0.13
0.14
0.20
0.85
0.16
0.17
0.13
0.15
0.17
0.21
0.16
0.09
0.07
0.10
0.18
0.82
0.19
0.23
0.20
0.20
0.29
0.32
0.33
0.27
0.23
0.21
0.22
0.88
All antidepressants
One prescription
229,299
175,708
2,204
6,830
961
3,887
0.25
Reference
0.240.260.18
Reference
0.18 0.19
Rate ratios (RR with 95% confidence interval) based on Poisson regression model with sex, age, socio-economic and medical history variables,
calendar year, and follow-up time as background variables. Current antidepressant usage compared to one-prescription group
Table 2 Suicide mortality and use of antidepressants
Current use of antidepressantPerson years SuicideSuicide rate per 100,000RR unadjusted RR adjusted
RR 95% CIRR95% CI
SSRI
TCA
Other
None
139,710
7,710
81,880
340,730
123
10
167
355
88
130
204
104
0.67
0.99
1.55
0.79
0.54
0.52
1.27
0.67
0.83
1.86
1.89
0.94
0.47
0.57
0.84
0.73
0.38
0.30
0.69
0.61
0.59
1.07
1.03
0.86
Citalopram
Fluoxetine
Sertraline
Mirtazapine
Paroxetine
Amitriptyline
Venlafaxine
Fluvoxamine
Moclobemide
Mianserin
Other
None
68,041
27,502
14,624
14,270
12,839
5,223
3,592
3,176
2,469
1,870
75,695
340,730
53
28
10
36
13
6
6
5
4
3
136
355
78
102
68
252
101
115
167
157
162
160
180
104
0.59
0.77
0.52
1.92
0.77
0.87
1.27
1.20
1.23
1.22
1.37
0.79
0.44
0.52
0.28
1.35
0.44
0.39
0.56
0.49
0.46
0.39
1.11
0.67
0.80
1.15
0.98
2.73
1.35
1.97
2.86
2.90
3.31
3.81
1.69
0.94
0.43
0.56
0.33
0.80
0.52
0.57
0.58
0.70
0.82
0.47
0.83
0.72
0.32
0.38
0.17
0.56
0.29
0.25
0.26
0.29
0.31
0.15
0.67
0.61
0.58
0.83
0.61
1.14
0.90
1.29
1.31
1.69
2.22
1.46
1.03
0.86
All antidepressants
None
One prescription
229,299
340,730
175,708
300
355
231
131
104
131
1.00
0.79
Reference
0.84
0.67
1.18
0.94
0.63
0.72
Reference
0.53
0.60
0.75
0.85
Rate ratios (RR with 95% confidence interval) based on Poisson regression model with sex, age, socio-economic and medical history variables,
calendar year, and follow-up time as background variables. Current antidepressant usage compared to one-prescription group
Eur J Clin Pharmacol (2009) 65:715–720717

Page 4

substances were determined, and individuals who used only
one drug category were assigned to that respective group.
We used the group that had filled only one prescription
(“one-prescription”) of any AD purchased during follow-up
period as reference group. Mixed use (i.e., several antide-
pressant medications) or rarely used drugs were assigned to
a separate group (“other”), and “no current use of
medications” composed a separate group. We also applied
another grouping of drugs: “one-prescription”, “SSRI”,
“TCA”, other, or no current use of medication. All-cause
mortality and mortality due to suicide during the use of
antidepressants were compared with the hazard during no
antidepressant use by taking into account the effect of the
background variables of each individual. For each individ-
ual, the follow-up was divided into periods in which the
value of time-dependent variables (i.e., antidepressant
medication) was constant. Thus, the follow-up of each
individual consisted of several contiguous periods, each
defined with specific entry and exit times.
We constructed life tables using two time dimensions
(calendar year and follow-up time), medication, and
background variables. Follow-up time was divided into
five intervals (0–90, 91–180, 181–270, 271–360, over
360 days), and calendar year into five groups (1999–
2003). This resulted in a table with number of events and
sum of person-years in each cell. Number of events was
used as the response variable and logarithm of person-years
as the offset term in the generalized linear model with
Poisson distribution and logarithm link function [22]. The
discrete marginal variables of life tables were used as
explanatory variables in the models. Results are presented
as rate ratios (RR) with 95% confidence intervals.
Results
The study population consisted of 258,417 individuals who
were 10 years or older and contributed 745,737 person-
years. The average length of follow-up was 2.9 years.
During the follow-up, 20,738 deaths were observed
(including 886 suicides).
All-cause mortality during antidepressant treatment was
lower than with no medication, and mortality was lower in
both the group taking any antidepressant and the no
antidepressant group than in the one-prescription group
(Table 1). There were no clear differences among antide-
pressant groups or among specific substances in this regard.
Adjustment with background variables increased the RR for
paroxetine from 0.12 (95% confidence interval 0.09–0.16)
to 0.22 (0.17–0.29), whereas it decreased the RR of
mianserin from 0.44 (0.31–0.62) to 0.14 (0.10–0.21).
Current use of any antidepressant was associated with a
lower suicide RR of 0.63 (0.53–0.75) compared to the one-
prescription group (Table 2). Current SSRI usage was
associated with lower suicide mortality compared to the one-
prescription (RR 0.47, 0.38–0.59). Current SSRI usage was
also associated with lower suicide mortality compared to the
no-antidepressant group (Table 2). The same general differ-
ence between medication groups was apparent when specific
substances were compared, however no remarkable differ-
Table 3 Suicide mortality in youth (10–20 years of age) and use of antidepressants
Current use of antidepressantPerson years Suicide deaths (n) Suicide rate per 100,000RR unadjustedRR adjusted
RR95% CI RR95% CI
SSRI
Other
None
One prescription
9,313
3,906
17,759
7,478
7
6
15
9
75
154
84
120
0.62
1.28
0.70
Reference
0.23
0.45
0.31
1.68
3.59
1.60
0.66
0.94
0.58
Reference
0.24
0.33
0.24
1.82
2.69
1.38
Rate ratios (RR with 95% confidence interval) based on Poisson regression model with sex, age, socio-economic and medical history variables,
calendar year, and follow-up time as background variables. Current antidepressant usage compared to one-prescription group
Time (d)
RR
1
2
4
8
16
0-30 31-6061-90 91-180181-360 >361
Fig. 1 Rate ratio (RR) for suicide as function of time since start of
follow-up. Reference category is 0–30 days in one-purchase group
(solid line). The no-current-medication group is represented by long
dashes, SSRI by short dashes, and other antidepressants by short and
long dashes. The 95% confidence intervals of all groups overlap in all
time periods
718 Eur J Clin Pharmacol (2009) 65:715–720

Page 5

ence between substances could be detected. We also analyzed
suicide mortality among youth from 10 to 20 years of age
separately (Table 3). There was no significant difference
between one-prescription and any of the AD groups (no-
current-use, SSRI, or other AD). The usage of TCA was
combined with other groups because of very low usage.
We detected significant interaction between follow-up
period and medication group with respect to suicide. The
risk was highest between 31 to 60 days after purchase of
antidepressant for the one-purchase group. After that, risk
in all groups was quite similar (Fig. 1.)
Men had a risk of all-cause mortality that was more than
twice as high as women, and nonmarried people had 19%
(95% CI 14–23%) higher all-cause mortality risk higher
risk compared to married. Being in the lowest income
quintile was associated with 21% (14–28%) higher risk
than being in the highest quintile. A history of attempted
suicide, use of antipsychotics, or hospitalization was
associated with higher risk with respect to both end-points.
Discussion
Current use of antidepressants was associated with the
lowered risk of all-cause mortality. There were no differ-
ences between substances or groups of substances. This is
in line with observations from a general population-based
study [23]. The study reported that antidepressant exposure
was associated with increased risk, whereas withdrawal
from treatment doubled the risk. In our study, every person
received at least one prescription of an antidepressant,
which means that everyone with no treatment during the
follow-up time had discontinued treatment.
In the follow-up of this study population, having only one
prescription for an antidepressant was associated with the
highest risk of suicide. As shown in Fig. 1, this is probably
because the individuals with highest risk committed suicide
after a short follow-up. When current use of any antidepres-
sant was compared to the one-prescription group, no
significant difference was detected, but current SSRI
antidepressant use was associated with lower risk of suicide.
This study, like all observational studies, has a number
of limitations. Our data included quite detailed information
on socio-economic characteristics, medication, and hospi-
talizations. On the other hand, the lack of information about
certain clinically important characteristics, such as indica-
tion of antidepressant treatment, could hinder model-based
adjustments. Another limitation is that adherence rates for
prescribed medications could be low [24], and thus the
effect observed could be attenuated. Another shortcoming
is that the duration of AD treatment was estimated only
using purchased amount and DDD, without any informa-
tion on actual prescribed daily dose. This reduces the
accuracy of estimation of duration of treatment. It is also
possible that accuracy differs across substances. Because
our data consisted only of reimbursed medications, part of
the medication periods are missing. We did not have
information on hospitalization periods during follow-up.
Thus, some periods without medication are due to hospital-
izations. However, because the one-prescription group was
used as reference category, this will not distort other
comparisons. In observational studies everything can thus
be confounded by indication. It is possible that close
monitoring may reduce all-cause mortality.
In conclusion, our results show that the current usage of
SSRI antidepressants in particular is associated with lower
all-cause mortality in a population of antidepressant users.
However, because of the observed increased risk of suicide
at the early stages of antidepressant treatment, special
attention should be paid to care immediately after starting
treatment.
Acknowledgement
for the preparation of the data for research purposes.
The authors wish to thank Tiina Sevon, MSc.,
Conflict of interest
tion with Janssen-Cilag. Jari Tiihonen has been a faculty member of
the Lundbeck International Neuroscience Foundation 1997–2007, a
consultant/expert member of the Advisory Board of Janssen-Cilag
since 2006, a member of the Eli Lilly Advisory Board since October
2006, and a member of the Bristol Myers-Squibb Advisory Board
(2002–2004). He has done research collaboration with Organon (since
2007); has received fees for giving expert opinion from Bristol Myers-
Squibb and GlaxoSmithKline; and has received lecture fees from
Janssen-Cilag, Bristol Myers-Squibb, Eli Lilly, Pfizer, Lundbeck,
GlaxoSmithKline, and Astra Zeneca. Jouko Lönnqvist has done
research collaboration with Orion Corporation (2004–2007) and has
received fees for giving an expert opinion from Eli Lilly (2007) and
for serving as the chairman of an international meeting from Janssen-
Cilag (2008). All other authors declare no competing interest.
Jari Haukka has been done research collabora-
References
1. Sadock BJ, Sadock VA (eds) (2000) Kaplan and Sadock’s
comprehensive textbook of psychiatry. Lippincott Williams and
Wilkins, Philadelphia
2. Isacsson G (2000) Suicide prevention—a medical breakthrough.
Acta Psychiatr Scand 102(2):113–117
3. Isacsson G, Holmgren P, Ahlner J (2005) Selective serotonin
reuptake inhibitor antidepressants and the risk of suicide: a
controlled forensic database study of 14,857 suicides. Acta
Psychiatr Scand 111(4):286–290
4. Ludwig J, Marcotte DE (2005) Anti-depressants, suicide, and
drug regulation. J Policy Anal Manage 24(2):249–272
5. Rihmer Z, Akiskal H (2006) Do antidepressants t(h)reat(en)
depressives? Toward a clinically judicious formulation of the
antidepressant-suicidality FDA advisory in light of declining
national suicide statistics from many countries. J Affect Disord
94(1–3):3–13
6. Milane MS, Suchard MA, Wong M, Licinio J (2006) Modeling of
the temporal patterns of fluoxetine prescriptions and suicide rates
in the United States. PLoS Med 3(6):e190
Eur J Clin Pharmacol (2009) 65:715–720 719

Page 6

7. Gunnell D, Saperia J, Ashby D (2005) Selective serotonin
reuptake inhibitors (SSRIs) and suicide in adults: meta-analysis
of drug company data from placebo controlled, randomised
controlled trials submitted to the MHRA’s safety review. BMJ
330(7488):385
8. Khan A, Khan S, Kolts R, Brown WA (2003) Suicide rates in
clinical trials of SSRIs, other antidepressants, and placebo:
analysis of FDA reports. Am J Psychiatry 160(4):790–792
9. Didham RC, McConnell DW, Blair HJ, Reith DM (2005) Suicide
and self-harm following prescription of SSRIs and other anti-
depressants: confounding by indication. Br J Clin Pharmacol 60
(5):519–525
10. Jick H, Kaye JA, Jick SS (2004) Antidepressants and the risk of
suicidal behaviors. JAMA 292(3):338–343
11. Martinez C, Rietbrock S, Wise L, Ashby D, Chick J, Moseley J
et al (2005) Antidepressant treatment and the risk of fatal and non-
fatal self harm in first episode depression: nested case-control
study. BMJ 330(7488):389
12. Juurlink DN, Mamdani MM, Kopp A, Redelmeier DA (2006) The
risk of suicide with selective serotonin reuptake inhibitors in the
elderly. Am J Psychiatry 163(5):813–821
13. Korkeila J, Salminen JK, Hiekkanen H, Salokangas RKR (2007)
Use of antidepressants and suicide rate in Finland: an ecological
study. J Clin Psychiatry 68(4):505–511
14. Bramness JG, Walby FA, Tverdal A (2007) The sales of
antidepressants and suicide rates in Norway and its counties
1980–2004. J Affect Disord 102(1–3):1–9
15. Gibbons RD, Hur K, Bhaumik DK, Mann JJ (2005) The
relationship between antidepressant medication use and rate of
suicide. Arch Gen Psychiatry 62(2):165–172
16. Tiihonen J, Lönnqvist J, Wahlbeck K, Klaukka T, Tanskanen A,
Jari Haukka (2006) Antidepressants and the risk of suicide,
attempted suicide and overall mortality in a nation-wide cohort.
Arch Gen Psychiatry 63:1358–1367
17. Rubino A, Roskell N, Tennis P, Mines D, Weich S, Andrews E
(2006) Risk of suicide during treatment with venlafaxine,
citalopram, fluoxetine, and dothiepin: retrospective cohort study.
BMJ. doi:10.1136/bmj.39041.445104.BE
18. Søndergård L, Lopez AG, Andersen PK, Kessing LV (2007)
Continued antidepressant treatment and suicide in patients with
depressive disorder. Arch Suicide Res 11(2):163–175
19. Kivimäki M, Gunnell D, Lawlor DA, Smith GD, Pentti J,
Virtanen M et al (2006) Social inequalities in antidepressant
treatment and mortality: a longitudinal register study. Psychol
Med 37:373–382
20. WHO Collaborating Centre for Drug Statistics Methodolgy (2009)
ATC/DDD index. Available from: http://www.whocc.no/atcddd/.
Accessed 19 January 2009
21. Benson K, Hartz AJ (2000) A comparison of observational studies
and randomized, controlled trials. N Engl J Med 342(25):1878–
1886
22. McCullagh P, Nelder J (1994) Generalized linear models, 2nd ed.
Chapman and Hall, London
23. Tata LJ, West J, Smith C, Farrington P, Card T, Smeeth L et al
(2005) General population based study of the impact of tricyclic
and selective serotonin reuptake inhibitor antidepressants on the
risk of acute myocardial infarction. Heart 91(4):465–471
24. Haynes RB, McKibbon KA, Kanani R (1996) Systematic review
of randomised trials of interventions to assist patients to follow
prescriptions for medications. Lancet 348(9024):383–386
720 Eur J Clin Pharmacol (2009) 65:715–720