The article you requested is

Suicidality as Rare Adverse Event of Antidepressant Medication: Report From the AMSP Multicenter Drug Safety Surveillance Project

J Clin Psychiatry 2010;71(10):1293-1307

Objective: Since the advent of antidepressant drug treatment, the question of whether these substances induce suicidal ideation and behavior has not been satisfactorily answered. The aim of this study is to contribute to this ongoing discussion by taking a heuristic case-based approach to the question.

Method: A large data set from a European drug surveillance program (Arzneimittelsicherheit in der Psychiatrie; AMSP) performed in 85 psychiatric hospitals from 1993 until 2008 was analyzed. A series of single cases were carefully assessed. The observed frequencies of adverse drug reactions (ADRs) in this sample were related to the total AMSP population, who used the imputed medication.

Results: A total of 142,090 adult patients taking antidepressant medication were observed. Thirty-three incidents of suicidality (12 cases of suicidal ideation, 18 attempts, and 3 completed suicides) were documented. Fourteen cases were assumed to be probably, and 19 to be possibly, related to the drug. Twenty-three cases judged as suicidal ADRs were associated with restlessness, 10 with ego-dystonia, 9 with impulsiveness, and 3 with psychosis. A higher incidence of suicidal ADRs was observed for selective serotonin reuptake inhibitors (0.034%; 95% CI, 0.020–0.054) and serotonin-norepinephrine reuptake inhibitors (0.034%; 95% CI, 0.015–0.068) compared to noradrenergic and specific serotonergic antidepressants (0.009%; 95% CI, 0.002–0.027) and tricyclic antidepressants (0.002%; 95% CI, 0.000–0.014).

Conclusion: Despite the methodological limitations of this study, the large AMSP data set supports the assumption that antidepressant drugs rarely trigger suicidality.

J Clin Psychiatry

Submitted: December 15, 2009; accepted March 31, 2010.

Online ahead of print: August 10, 2010 (doi:10.4088/JCP.09m05912blu).

Corresponding author: Susanne Stübner, PD Dr, Department of Psychiatry, Ludwig-Maximilian University, School of Medicine, Nussbaumstraße 7, D-80336 Munich, Germany (

Suicidality as Rare Adverse Event of Antidepressant Medication: Report From the AMSP Multicenter Drug Safety Surveillance Project

Objective: Since the advent of antidepressant drug treatment, the question of whether these substances induce suicidal ideation and behavior has not been satisfactorily answered. The aim of this study is to contribute to this ongoing discussion by taking a heuristic case-based approach to the question.

Method: A large data set from a European drug surveillance program (Arzneimittelsicherheit in der Psychiatrie; AMSP) performed in 85 psychiatric hospitals from 1993 until 2008 was analyzed. A series of single cases were carefully assessed. The observed frequencies of adverse drug reactions (ADRs) in this sample were related to the total AMSP population, who used the imputed medication.

Results: A total of 142,090 adult patients taking antidepressant medication were observed. Thirty-three incidents of suicidality (12 cases of suicidal ideation, 18 attempts, and 3 completed suicides) were documented. Fourteen cases were assumed to be probably, and 19 to be possibly, related to the drug. Twenty-three cases judged as suicidal ADRs were associated with restlessness, 10 with ego-dystonia, 9 with impulsiveness, and 3 with psychosis. A higher incidence of suicidal ADRs was observed for selective serotonin reuptake inhibitors (0.034%; 95% CI, 0.020–0.054) and serotonin-norepinephrine reuptake inhibitors (0.034%; 95% CI, 0.015–0.068) compared to noradrenergic and specific serotonergic antidepressants (0.009%; 95% CI, 0.002–0.027) and tricyclic antidepressants (0.002%; 95% CI, 0.000–0.014).

Conclusion: Despite the methodological limitations of this study, the large AMSP data set supports the assumption that antidepressant drugs rarely trigger suicidality.

J Clin Psychiatry 2010;71(10):1293–1307

Submitted: December 15, 2009; accepted March 31, 2010.

Online ahead of print: August 10, 2010 (doi:10.4088/JCP.09m05912blu).

Corresponding author: Susanne Stübner, MD, Department of Psychiatry, Ludwig-Maximilian University, School of Medicine, Nussbaumstraße 7, D-80336 Munich, Germany (

Pharmacoepidemiologic studies have shown that an increase in the prescription rate of antidepressant drugs is associated with a decreased risk of suicide. This holds true for large epidemiologic studies linking prescription data with suicide data, as well as for special intervention studies (eg, “Gotland Study,” “Nürnberger Modell,” “Zurich Follow-up Study”; for reference, see Möller1).

Nevertheless, some earlier observations suggested that there was a connection between the initiation of treatment with the tricyclic antidepressant imipramine2 or later on with the selective serotonin reuptake inhibitor (SSRI) fluoxetine3 and the emergence of suicidal ideation. Furthermore, Rouillon and colleagues4 reported significantly higher numbers of suicide attempts by patients on long-term treatment with the tetracyclic antidepressant maprotiline compared to placebo. Additionally, suicidal attempts were observed in the context of psychomotor activation in connection with less sedative antidepressants.5,6 Finally, suicidality was linked to some cases of antidepressant drug–induced psychosis.6

In the last few years, the discussion has focused more on suicidality in connection with SSRIs, mainly in children and adolescents7 but also in adults.7–9 The increase of suicidal behavior during antidepressant medication was attributed to a so-called drive-mood dissociation and to the disinhibition of impulsive behavior.10–13 Nevertheless, it is still difficult to decide whether suicidal behavior is an “adverse effect” of the medication or a “symptom” of the disease. Some experts fear that an overcritical attitude to antidepressants could increase the number of suicides due to undertreatment.11,14

The degree of suicidal risk reported so far depends on the method of the study,1 patient characteristics such as age, personality traits and diagnoses, the type of suicidality examined11,15 (ie, suicidal ideation, suicide attempt, or completed suicide), and the type of antidepressant investigated.15–17

One meta-analysis of randomized controlled trials (RCTs)18 revealed weak evidence that SSRIs increased the risk of suicidal self-harm. However, for SSRIs, neither a beneficial nor a harmful effect on suicide deaths could be stated because they occur so rarely.18 Another meta-analysis16 showed that users of SSRIs had a 2-fold increase of fatal and nonfatal suicidal attempts. A recent meta-analysis (comparing the SSRI sertraline to placebo)19 did not find any treatment effect on suicidal behavior, neither with regard to different age groups nor to different types of suicidal behavior. Likewise no differences in suicidal behavior were reported to be due to the various classes of antidepressants15,16 with the exception of one study20 that reported more deliberate self-harm with SSRIs than with tricyclic antidepressants (TCAs). A comprehensive analysis of the US Food and Drug Administration11 reported that antidepressants (including SSRIs) pose only a small risk of inducing suicidal thoughts and suicide attempts in groups under 25 years of age; this risk decreased as age increased. A recent study21 discussed that the antidepressants are even possibly protective for suicidal ideation in adults aged 25–64.

Meta-analyses of RCTs are assumed to be the gold standard for detecting causal relationships between the use of antidepressants and suicidal behavior. However, the included RCTs usually exclude suicidal patients, refer only to short periods of time, and are usually not designed to identify completed or attempted suicides.22 Because the absolute number of patients attempting or committing suicide is very low, the failure to report a few cases can have a high impact.23 Another disadvantage of meta-analyses is that data are pooled across different antidepressants, thus masking the risk profile of a single drug.

To obtain more precise information on the characteristics of the afflicted patients as well as on the development and precise nature of suicidality, it has been recommended that well-documented case reports be collected in addition to meta-analyses of RCTs and to pharmacoepidemiologic studies.18 Case reports from the German spontaneous adverse drug reaction (ADR) reporting system have already been published.24,25 In our study, a large data set from a European drug surveillance program (AMSP)26 carried out in psychiatric hospitals in Germany, Austria, and Switzerland was analyzed. By allowing careful assessment of a series of single cases, this approach may help to detect risk factors and early signs of emergent antidepressant-induced suicidality. It should be stressed that our study is a heuristic approach, not a confirmatory test.


The AMSP program for Drug Safety in Psychiatry (Arzneimittelsicherheit in der Psychiatrie) was established in 1993 as a continuous open-end study for the assessment of severe ADRs to all marketed psychotropic drugs during routine treatment of psychiatric inpatients. It currently includes 54 psychiatric institutions (15 university and 39 nonuniversity institutions, ie, 30 state hospitals and 9 psychiatric departments of general hospitals in Austria, Germany, and Switzerland). The AMSP replaces an earlier, similar drug surveillance program (Arzneimittelüberwachung in der Psychiatrie [AMÜP]27).

The AMSP methods were described in detail earlier.26,28 In brief, only clinically severe ADRs are assessed. The criteria of the program are based on the severity of the event itself and its potential danger to the patient’s health. Adverse drug reactions are documented by trained psychiatrists, who contact the ward psychiatrists at regular intervals (at least every 2 weeks) and document the ADR cases in standardized questionnaires. Age, gender, and psychiatric and somatic diagnosis are recorded along with a detailed description of the adverse event, potential risk factors, clinical measures taken because of the ADR, and the subsequent outcome of the ADR. Psychiatric and somatic drug data (dosage, time course, concurrent medication), previous exposure to the imputed drugs, treatment after the adverse event, and outcome in case of rechallenge are also documented.

Probability rating: all these factors are taken into account in order to rate the probability of a causal relationship between medication and adverse event. The probability ratings are as follows: grade 1 = possible (ADR not known, time course or dosage unusual for drug in question, or alternative explanation for adverse event more probable), grade 2 = probable (ADR known for drug in question, time course and dosage in accordance with previous experience, and alternative explanation less probable), grade 3 = definite (in addition to criteria necessary for a “probable” rating, reappearance of the ADR after rechallenge with drug in question).

The cases are forwarded to the AMSP center at the psychiatric department of the University of Munich, reviewed by a senior member of the team, and then discussed at central case conferences attended by drug monitors from all centers, representatives of the German Federal Institute for Drugs and Medical Products (Bundesinstitut für Arzneimittel und Medizinprodukte, BfArM) and the Drug Commission of the German Medical Association (Arzneimittelkommission der deutschen Ärzteschaft, AKdÄ) and drug safety experts from the pharmaceutical industry. After a consensus is reached and a probability rating is given to the adverse drug reaction, the completed case descriptions are sent to the various authorities as well as to the pharmaceutical companies involved. They are also stored in the central surveillance database for further analysis. Expert rating using the AMSP (and AMÜP) criteria was shown to be as reliable as the use of a formalized algorithm during the earlier AMÜP study.29

Data on drug use in all patients under AMSP surveillance at the participating centers are regularly documented by AMSP in order to estimate the frequency of an ADR occurring with a certain drug. On 2 reference days per year, all drugs and their dosages are recorded along with age, gender, and diagnoses. In addition, the participating centers provide the number of inpatients monitored per year as well as the mean duration of inpatient treatment for all monitored patients; both are broken down according to diagnostic groups.

AMSP assesses mainly ADRs occurring during inpatient treatment. In addition, ADRs leading to hospitalization are recorded. Both are explicitly documented.

Special Procedure of This Analysis

For this study, all cases of suicidal ideation or behavior judged to be related to antidepressants as an ADR were extracted from the database covering the period from 1993 to 2008 and were analyzed along with background information on demographic data and drug use in the patients under surveillance.

Criteria for assessing suicidality as an ADR: suicidality is a frequent illness-related feature of psychiatric inpatients. It is often difficult to differentiate between drug-related and illness-related events. The AMSP rates suicidality as an ADR only if it is connected with treatment-emergent symptoms that allow a judgment that the event is at least “possibly” drug related, eg, agitation, psychosis, unusual change in behavior, inner or motor restlessness. The unexpectedness of a suicidal action alone is not sufficient to rate it as drug-related. Special symptoms like emergent agitation have to be observed by the staff or reported by the patient. In particular, the assessment of ego-dystonic suicidal impulses, such as impulses experienced by the patients as strange to them, is relevant; it relies mostly on the explicit description by the patient concerned. For this reason, causality assessment is particularly difficult and often impossible in completed suicides, which are then rated as “unassessable.” Such cases are not included in the present study.

While assigning probability ratings, researchers considered the possibility of an illness related event as an alternative cause in all cases of suicidal ADRs; however, dependent on the clinical features this was judged as less likely in the “probable” cases and as more likely in the “possible” cases. All cases of suicidal ADRs due to antidepressants were reassessed by the AMSP leader team including the heads of the regional and national AMSP groups in February 2009. These were discussed until a consensus was reached.

It would have been desirable to assess all cases of suicidal behavior irrespective of judgments on causality. However, due to the limited means of the AMSP, this was not feasible; even concerning completed suicides, only a few hospitals systematically document all of them.

Classification of Substances

Antidepressant drugs were classified according to clinical convenience (for details, see Table 1).

Table 1

Click figure to enlarge

Data Evaluation

Data are presented as absolute numbers of ADRs and as incidence rates related to the number of patients exposed. The latter are estimated on the basis of data collected on the reference days per year from every hospital bed under surveillance and the absolute number of patients per year. Adverse drug reaction incidence rates are presented together with their confidence intervals. In view of the very low rate of serious side effects combined with the high number of patients exposed, confidence intervals are calculated according to the exact (asymmetric) method first described by Clopper and Pearson.30 This method avoids the bias of the commonly used approximation methods (for a discussion of methods, see, eg, Engel et al28). Statistical inference is used as an exploratory technique, as the focus of the AMSP project is the descriptive presentation of ADR data, not the proof of hypotheses.


Demographic Characteristics and Data on Drug Use of Patients

The AMSP project surveyed a total of 311,374 patients at 85 hospitals between 1993 and 2008 (139,509 men and 171,865 women). Of these, 142,090 patients were treated with antidepressant medication (63.2% were women). There was no relevant difference in age, gender, and psychiatric diagnoses between patients taking antidepressants who had been surveyed by AMSP and those taking antidepressants who had exhibited suicidal ADRs (see Table 2).

Table 2

Click figure to enlarge

A detailed overview of antidepressant subgroup drug use is given in Table 3. Approximately one-third (32%) of the patients with antidepressant medication were also treated with benzodiazepines.

Table 3

Click figure to enlarge

Table 4

Click figure to enlarge

Antidepressant use notably changed over time. Between 1993 and 2000, predominantly TCAs (51.2%) were given, whereas between 2001 and 2008, TCAs were administered less often (17.0%), and SSRIs (41.7%), noradrenergic and specific serotonergic antidepressants (NaSSAs) (30.2%), and serotonin norepinephrine reuptake inhibitors (SNRIs) (22.3%) predominated.

Details of Suicidal ADRs

From 1993 to 2008, 33 patient episodes of suicidality were observed, in which a relationship to antidepressant treatment was considered possible or probable. Fourteen of these cases occurred during the period 1993–2000 (53,042 antidepressant patients, 0.026%) and 19 during the period 2001–2008 (88,011 antidepressant patients, 0.022%).

They included 12 episodes of suicidal ideation, 18 suicide attempts, and 3 completed suicides (Table 4).

Fourteen events were assumed to have a probable, and 19 to have a possible, relationship to the drug (here: case-related, ie, 1 rating per case, Table 5).

Characteristics of Patients With Suicidal ADRs

The ages of the afflicted patients (male: n = 11; female: n = 22) ranged between 19 and 77 years. In the group under 26 years of age, there was 1 case with suicidal ADRs (ie, a relative frequency of 0.0011%). The group older than 25 years included the remaining 32 cases of suicidal ADRs (ie, a relative frequency of 0.024%). The majority of main diagnoses (22 patients) were affective disorders (ICD-10: F3; for an overview, see Table 2). Three patients had organic brain damage. Prior suicide attempts were known in 8 patients. Approximately half of the patients with suicidal ADRs experienced the first manifestation of psychiatric illness, and about two-thirds were hospitalized for the first time.

Table 5

Click figure to enlarge

Clinical Features During ADR

In the majority of patients (n = 23, 69.7%) suicidal ADRs were related to restlessness, 9 patients (27.2%) exhibited impulsiveness, and 3 (9.1%) developed a psychosis. Ten patients (30.3%) reported afterward that they had had an awkward ego-dystonia, ie, they had experienced suicidal impulses as strange to them (Table 4). Two of the patients who developed psychotic symptomatology had organic brain damage.

In comparison, within the cases of severe nonsuicidal psychic ADRs under antidepressant treatment (n = 215, toxic delirium not classified as psychic ADR), in 70 cases (ie, only in 32.6%), severe restlessness was recorded. For this symptom, the Fisher Exact Test comparing suicidal versus nonsuicidal ADRs was significant (P < .001). For psychosis (n = 42 in nonsuicidal psychic ADRs, 19.5%), no significant difference between suicidal and nonsuicidal psychic ADRs was found. Ego-dystonia and impulsiveness were not recorded as independent ADRs, ie, they were assessed only in the context of other psychic ADRs.


Selective serotonin reuptake inhibitors were involved in the majority of cases (n = 18); the causal relationship was considered to be probable in 10 episodes. Adverse drug reactions were observed during treatment with SNRIs in 8 cases, 3 of which were considered to have a probable causal relationship (see Tables 3 and 5).

The dosages of antidepressants at the onset of ADRs were rated as low, medium, and high according to product information. Adverse drug reactions occurred with SSRIs mostly at medium dosages (in 14 out of 18 cases). For all other antidepressants, medium dosages were found as frequently as higher dosages (for details, see Table 6).

Table 6

Click figure to enlarge

In one case, 2 antidepressants were considered to have caused the ADR (SI9, see Tables 7A, 7B, and 7C); in another case, 1 antidepressant and 1 antipsychotic drug (SA18); and in a further case, 1 antidepressant, 1 antipsychotic drug, and 1 benzodiazepine (S1). Monotherapy was administered in 8 cases (24%), and a probable causal relationship was considered in 4 of these cases (for details, see Tables 7A, 7B, and 7C). In the total population of antidepressant patients, only 8.2% were on monotherapy.

Table 7A a

Click figure to enlarge

Table 7A b

Click figure to enlarge

Table 7B a

Click figure to enlarge

Table 7B b

Click figure to enlarge

Table 7B c

Click figure to enlarge

Table 7B d

Click figure to enlarge

Table 7C

Click figure to enlarge

At least 2 drugs in combination were administered in 25 cases. In 15 of these cases, benzodiazepines were given (see Tables 3, 7A, 7B, and 7C). The numbers of cases of ADRs with the involved drugs and the corresponding data on drug use are shown in Table 3.

Most ADRs occurred shortly after onset of the imputed medication (12/33 cases within the first 7 days); in another 7 cases, the dosage of the imputed drug had been increased during the week before the ADR (for details, see Tables 7A, 7B, and 7C).

Figure 1 shows all incidences of suicidal ADRs (possible and probable) for the antidepressant drug groups together with their asymmetric exact confidence intervals (SSRIs: incidence 0.034% [95% CI, 0.020–0.054]; SNRIs: 0.034% [95% CI, 0.015–0.068]; NaSSAs: 0.009% [95% CI, 0.002–0.027]; TCAs: 0.002% [95% CI, 0.000–0.014]). Figure 2 gives the results only for relationships rated as probable (SSRIs: 0.019% [95% CI, 0.009–0.035]; SNRIs: 0.013% [95% CI, 0.003–0.038]; NaSSAs: 0.000% [95% CI, 0.000–0.011]; TCAs: 0.000% [95% CI, 0.000–0.009]). Both figures indicate a higher incidence of suicidal ADRs for SSRIs and SNRIs compared to NaSSAs and TCAs.

Figure 1

Click figure to enlarge

Figure 2

Click figure to enlarge

Only 1 patient used drugs from 2 antidepressive drug groups (SI9). For a pragmatic statistical treatment, we focused on those 32 patients who used only antidepressant substances categorized in 1 group of drugs and, accordingly, restricted the number of patients in the comparison group to those who used only antidepressive drugs from 1 of the groups. A simple 2-way table shows a significant imbalance of the drug groups, regardless of whether only the cases with a probable causal relationship (likelihood based P = .003) or all cases (likelihood based P = .0001) were considered. Selective serotonin reuptake inhibitors and SNRIs were more often imputed than TCAs and NaSSAs.

Tables 7A, 7B, and 7C gives detailed information on the individual cases, including clinical features, medication, causal relationship between ADRs and the drugs used, subsequent treatment, and outcome.


Thirty-three of the 142,090 patients taking antidepressant medication exhibited suicidal behavior as an ADR to antidepressants. Fourteen of these incidents were judged to be probably related to drug use. Restlessness was the most frequently observed clinical feature among the ADR symptoms in our cases. Furthermore, some patients explicitly reported impulsiveness and ego-dystonia. Only 8 of the 33 patients had a history of previous suicide attempts. Selective serotonin reuptake inhibitors and SNRIs were the most frequently imputed classes of antidepressant drugs.

In general, it is difficult to discriminate between suicidality as a symptom of the primary disorder and as a drug-induced event. Suicidality is known to be a complication of depression and is mentioned in the ICD-1031 as well as in the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition32 criteria. Thus, in general, one cannot assume a definite causal relationship between suicidal ADRs and drug treatment. In cases of completed suicide, it is even difficult to make a possible rating, since there is often a deficiency of information on the period immediately preceding the event, and it is no longer possible to ask the patient.

The analysis of ADRs in the international AMSP project is based on repeated and detailed assessments by experts, including the consideration of all clinical and differential diagnosis information.26 The separation of probable from possible events is somewhat debatable as it has to rely more on subjective information than other areas of ADR assessment. Therefore, it is worth noting that the observed trends in relative frequencies of different AD groups are similar for all cases and those rated as probable.

The majority of our sample of patients with suicidal ADRs was women. This corresponds to the gender distribution of all inpatients treated with antidepressants in the AMSP project (in the analyzed period of 1993–2008). The distribution of diagnoses was also comparable. We were unable in our sample to investigate the association between the occurrence of ADRs and younger age, as was reported in the literature,1,21,33 because the AMSP project includes only adult psychiatric inpatients (with a higher proportion of gerontopsychiatric patients). Our study observed only 1 case in the group under 26 years of age.

The notion that antidepressant drugs might facilitate suicidal events was discussed earlier, eg, within the AMÜP project, which preceded the AMSP project.6 Enhancement of drive before ameliorating mood and also triggering of psychosis were considered possible risk factors for suicidality during treatment with first-generation antidepressant drugs.6 These results referred mainly to activating TCAs. In the meantime predominantly sedative TCAs are used, due to the fact that SSRIs and SNRIs are preferred as drive-enhancing drugs.

Increased impulsiveness was first reported for fluoxetine,3,34 and subsequently also for other SSRIs (for review see Healy et al35). Because SSRIs are administered to treat disorders with deficits of impulse control, this raises the issue of a paradoxical effect. Similarly, an enhancement and disinhibition of libido have been observed in single patients taking SSRIs,36 whereas a loss of libido is considered a common adverse effect.

Motor restlessness, which resembles akathisia, was seen in several patients with suicidal ADRs in our study. SSRIs and other antidepressants have been reported to possibly trigger an akathisic state, thus increasing vulnerability to suicidality.35 Moreover, akathisia has been considered a risk factor for suicidality.3,24,37

Suicidality was judged as an ADR only, if it was connected with relevant treatment-emergent symptoms. Severe restlessness occurred more frequently in connection with suicidal than with nonsuicidal psychic ADRs, which was not the case for psychosis. Comparisons of the frequency of the symptoms ego-dystonia and impulsiveness between suicidal and nonsuicidal psychic ADR cases were not analyzed here, since these symptoms have been assessed only in the context of other ADRs. To confirm a possible relation between the symptoms described (restlessness, ego-dystonia, impulsiveness) and suicidal ADRs, further investigations are necessary.

As regards current drug use, we observed suicidal ADRs mainly in patients taking SSRIs and SNRIs in our study. To the best of our knowledge, SNRIs have not yet been focused on in this context. The main imputed serotonergic medication (SSRIs, SNRIs) and the clinical features described (impulsiveness, inner and motor restlessness) suggests an involvement of the serotonergic system in suicidal ADRs.38 However, other mechanisms may be also relevant; 1 case of suicidal ideation was observed with reboxetine (case number SI5).

The discrepancy between the results of our study and those of the RCTs or meta-analyses of RCTs is due to methodological differences. Random controlled trials on the one hand have in general the tendency to exclude patients with emergent suicidality. On the other hand, RCTs count all events of suicidality during antidepressant treatment irrespective of causality considerations whereas AMSP counts only cases with at least some evidence for the drug involvement. Furthermore, our study refers only to inpatients.

An inspection of the case reports revealed a larger proportion of monotherapy among patients with suicidal ADRs in comparison to the total population. In general, polypharmacy is a known risk factor for ADRs and was explicitly assumed, eg, in case SI9. It is striking that there was a higher proportion of co-medication with benzodiazepines in the suicidal ADR cases (15 of 33 cases, ie, 45% compared to 32% of the total population). Benzodiazepines are commonly prescribed with antidepressants (in almost one-third of all antidepressant patients) as can be seen in the data on drug use. It could be argued that benzodiazepines play a role in the disinhibition of impulses contributing to suicidal behavior. On the other hand, benzodiazepines may have been used in the treatment of more severely ill patients, who later developed suicidal ADRs. In contrast to the imputed antidepressants, benzodiazepines were successfully continued in most of the 15 cases and introduced for treatment of the ADR in other cases. In their assessment of single cases, the experts did not include benzodiazepines except in 1 case in which oxazepam was imputed (case number S1).

The results have to be interpreted with the AMSP method in mind. The outstanding advantage of this surveillance system is the detailed analysis of single cases. This allows precise assessment of all factors contributing to ADRs. Furthermore, the observed ADRs can be related to the rate of drug use in the participating centers. A general limitation of the method, however, is that ADRs are probably underreported, especially behavioral ADRs, which may go unrecognized. Another general problem in the analysis of ADRs is that the more frequent reports of ADRs in connection with newer substances may bias the findings. Furthermore, the increasing public debate on suicidality during treatment with SSRIs might influence awareness and lead to selective observation and reporting. However, in our study, no increase in the number of reports in the more recent observation period was found. We also registered suicidal ADRs during treatment with other drugs, eg, moclobemide and reboxetine.

The advantages and shortcomings of the AMSP approach have to be compared with those of other methods of ADR assessments such as controlled trials, cohort studies, or case control studies.28 The AMSP method assumes a distinct position within drug surveillance systems due to the following features. It allows a detailed analysis of single cases, the possibility of relating observed ADRs to data on actual drug use, and the comparison of ADR data for different compounds within the same population.


The large data set of AMSP supports the view that antidepressant-triggered suicidality (primarily by SSRIs and SNRIs) is rare; however, it is a very serious ADR. It is important to be aware of suicidal ideation and behavior as an at least possible adverse reaction to antidepressant treatment. The following clinical symptoms were frequently observed in patients who have been judged to exhibit suicidal ADRs: restlessness, ego-dystonia, and/or impulsiveness. Whether there is an actual connection between these characteristics and suicidal ADRs should be examined in further studies.

Drug names: amiloride (Midamor and others), atorvastatin (Lipitor), biperiden (Akineton), bisoprolol (Zebeta and others), bupropion (Aplenzin, Wellbutrin, and others), captopril (Capoten and others), carbamazepine (Carbatrol, Equetro and others), celecoxib (Celebrex), citalopram (Celexa and others), clomipramine (Anafranil and others), diazepam (Diastat, Valium, and others), diclofenac (Flector, Zipsor, and others), doxepin (Zonalon, Silenor, and others), duloxetine (Cymbalta), escitalopram (Lexapro and others), fluoxetine (Prozac, Sarafem, and others), flurazepam (Dalmane and others), fluvoxamine (Luvox and others), gabapentin (Neurontin and others), haloperidol (Haldol and others), hydrochlorothiazide (Microzide, Oretic, and others), ibuprofen (Caldolor, Ibu-Tab, and others), imipramine (Tofranil, Surmontil, and others), lamotrigine (Lamictal and others), lithium (Lithobid and others), lorazepam (Ativan and others), metformin (Riomet, Fortamet, and others), metoprolol (Toprol, Lopressor, and others), mirtazapine (Remeron and others), misoprostol (Cytotec and others), nortriptyline (Pamelor, Aventyl, and others), olanzapine (Zyprexa), paroxetine (Paxil, Pexeva, and others), promethazine (Promethegan and others), quetiapine (Seroquel), ramipril (Altace and others), risperidone (Risperdal and others), sertraline (Zoloft and others), simvastatin (Zocor and others), theophylline (Elixophyllin, Theochron, and others), tramadol (Ultram, Ryzolt, and others), tranylcypromine (Parnate and others), trazodone (Oleptro and others), trimipramine (Surmontil and others), valproate (Depacon and others), venlafaxine (Effexor and others), zolpidem (Ambien, Zolpimist, and others).

Author affiliations: Department of Psychiatry, Ludwig-Maximilian University, Munich (Drs Stübner, Grohmann, von Stralendorff, Rüther, Möller, Engel, and Greil); Drug Commission of the German Medical Association, Berlin/Cologne (Dr Müller-Oerlinghausen); Department of Psychiatry, Georg August University, Göttingen (Dr Rüther), Germany; and Psychiatric Hospital Kilchberg, Switzerland (Drs Horvath and Greil).

Potential conflicts of interest: Dr Grohmann is the project manager of AMSP (Arzneimittelsicherheit in der Psychiatrie). Dr Rüther has received research grants from Eli Lilly, Lundbeck, Bristol-Myers Squibb, AstraZeneca, and Bayer; and is a consultant for and on the speakership bureaus of Servier, Otsuka, Lundbeck, Wyeth, and Lilly. Dr Möller has received grants or is a consultant for and on the speakership bureaus of AstraZeneca, Bristol-Myers Squibb, Eisai, Eli Lilly, GlaxoSmithKline, Janssen Cilag, Lundbeck, Merck, Novartis, Organon, Pfizer, Sanofi-Aventis, Schering-Plough, Schwabe, Sepracor, Servier, and Wyeth. Dr Horvath is a consultant for and on the speakership bureaus of Eli Lilly, AstraZeneca, and Servier. Dr Greil has received an investigator-initiated research grant from Eli Lilly Switzerland, has received an unrestricted educational grant for CME Blended Learning Courses from Eli Lilly Germany and Janssen-Cilag Switzerland, and is member of speakers or advisory boards for AstraZeneca, Sanofi-Aventis, Eli Lilly, GlaxoSmithKline, Janssen-Cilag, Lundbeck, Organon, and Pfizer. Drs Stübner, von Stralendorff, Müller-Oerlinghausen, and Engel have no conflict of interest to be declared.

Funding/support: The AMSP Drug Safety Program is organized by nonprofit associations in Germany, Austria, and Switzerland. Almost all pharmaceutical companies involved in CNS research contribute financial support to the 3 associations. Educational and research grants since 1993: Austrian companies: AstraZeneca Österreich, Boehringer Ingelheim Austria, Bristol-Myers Squibb, CSC Pharmaceuticals, Eli Lilly, Germania Pharma, GlaxoSmithKline, Janssen-Cilag, Lundbeck, Novartis, Pfizer Med Inform, Wyeth Lederle; German companies: Abbott, AstraZeneca, Aventis Deutschland, Bayer Vital, Boehringer Mannheim, Bristol-Myers Squibb, Ciba Geigy, Desitin Arzneimittel, Duphar, Eisai, Esparma Arzneimittel, GlaxoSmithKline, Hoffmann-La Roche AG Medical Affairs, Janssen-Cilag, Janssen Research Foundation, Knoll Deutschland, Lilly Deutschland Niederlassung Bad Homburg, Lundbeck, Novartis, Nordmark Arzneimittel, Organon, Otsuka-Pharma Frankfurt, Pfizer, Pharmacia & Upjohn, Promonta Lundbeck Arzneimittel, Rhone-Poulenc Rohrer, Sanofi-Synthelabo, Sanofi-Aventis Deutschland, Schering, SmithKline Beecham, Solvay Arzneimittel, Synthelabo Arzneimittel, Dr Wilmar Schwabe, Thiemann Arzneimittel, Troponwerke, Upjohn, Wander, Wyeth; Switzerland companies: AHP (Schweiz), AstraZeneca, Bristol-Myers Squibb, Desitin, Eli Lilly (Suisse), Essex Chemie, GlaxoSmithKline, Janssen-Cilag, Lundbeck (Suisse), Organon, Pfizer, Pharmacia, Sanofi-Aventis (Suisse), Sanofi-Synthélabo, Servier, SmithKlineBeecham, Solvay, Wyeth AHP (Suisse), Wyeth Pharmaceuticals.

Acknowledgment: This publication is kindly dedicated to Professor Hanns Hippius on the occasion of his 85th birthday.


1. Möller H-J. Is there evidence for negative effects of antidepressants on suicidality in depressive patients? a systematic review. Eur Arch Psychiatry Clin Neurosci. 2006;256(8):476–496.PubMed doi:10.1007/s00406-006-0689-8

2. Damluji NF, Ferguson JM. Paradoxical worsening of depressive symptomatology caused by antidepressants. J Clin Psychopharmacol. 1988;8(5):347–349.PubMed doi:10.1097/00004714-198810000-00007

3. Teicher MH, Glod C, Cole JO. Emergence of intense suicidal preoccupation during fluoxetine treatment. Am J Psychiatry. 1990;147(2):207–210. PubMed

4. Rouillon F, Phillips R, Serrurier D, et al. Recurrence of unipolar depression and efficacy of maprotiline. Encephale. 1989;15(6):527–534. PubMed

5. Ströbel C, Grohmann R, Rüther E, et al. Stationäre UAW-Erfassung: Antidepressiva. In: Grohmann R, Rüther E, Schmidt LG, eds. Unerwünschte Wirkungen von Psychopharmaka: Ergebnisse der AMÜP-Studie. 1st ed. Berlin, Germany: Springer; 1994:134–204.

6. Grohmann R, Ströbel C, Rüther E, et al. Adverse psychic reactions to psychotropic drugs: a report from the AMÜP study. Pharmacopsychiatry. 1993;26(3):84–93.PubMed doi:10.1055/s-2007-1014348

7. Whittington CJ, Kendall T, Fonagy P, et al. Selective serotonin reuptake inhibitors in childhood depression: systematic review of published versus unpublished data. Lancet. 2004;363(9418):1341–1345.PubMed doi:10.1016/S0140-6736(04)16043-1

8. Whittington CJ, Kendall T, Pilling S. Are the SSRIs and atypical antidepressants safe and effective for children and adolescents? Curr Opin Psychiatry. 2005;18(1):21–25. PubMed

9. Healy D, Whitaker C. Antidepressants and suicide: risk-benefit conundrums. J Psychiatry Neurosci. 2003;28(5):331–337. PubMed

10. Culpepper L, Davidson JR, Dietrich AJ, et al. Suicidality as a possible side effect of antidepressant treatment. J Clin Psychiatry. 2004;65(6):742–749.PubMed doi:10.4088/JCP.v65n0603

11. Möller H-J, Baldwin DS, Goodwin G, et al. WPA Section on Pharmacopsychiatry. Do SSRIs or antidepressants in general increase suicidality? WPA section on pharmacopsychiatry: consensus statement. Eur Arch Psychiatry Clin Neurosci. 2008;258(suppl 3):3–23. PubMed

12. Soloff PH, George A, Nathan RS, et al. Paradoxical effects of amitriptyline on borderline patients. Am J Psychiatry. 1986;143(12):1603–1605. PubMed

13. Teicher MH, Glod CA, Cole JO. Dr Teicher and associates reply. Am J Psychiatry. 1990;147(12):1692–1693.

14. Hegerl U. Antidepressants and suicidality: risk-benefit analysis [in German]. Nervenarzt. 2007;78(1):7–14.PubMed doi:10.1007/s00115-006-2109-8

15. Martinez C, Rietbrock S, Wise L, et al. Antidepressant treatment and the risk of fatal and non-fatal self harm in first episode depression: nested case-control study. BMJ. 2005;330(7488):389.PubMed doi:10.1136/bmj.330.7488.389

16. Fergusson D, Doucette S, Glass KC, et al. Association between suicide attempts and selective serotonin reuptake inhibitors: systematic review of randomised controlled trials. BMJ. 2005;330(7488):396.PubMed doi:10.1136/bmj.330.7488.396

17. Khan A, Khan S, Kolts R, et al. Suicide rates in clinical trials of SSRIs, other antidepressants, and placebo: analysis of FDA reports. Am J Psychiatry. 2003;160(4):790–792.PubMed doi:10.1176/appi.ajp.160.4.790

18. Gunnell D, Saperia J, Ashby D. 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. 2005;330(7488):385.PubMed doi:10.1136/bmj.330.7488.385

19. Vanderburg DG, Batzar E, Fogel I, et al. A pooled analysis of suicidality in double-blind, placebo-controlled studies of sertraline in adults. J Clin Psychiatry. 2009;70(5):674–683.PubMed doi:10.4088/JCP.07m04004

20. Donovan S, Clayton A, Beeharry M, et al. Deliberate self-harm and antidepressant drugs: investigation of a possible link. Br J Psychiatry. 2000;177(6):551–556.PubMed doi:10.1192/bjp.177.6.551

21. Stone M, Laughren T, Jones ML, et al. Risk of suicidality in clinical trials of antidepressants in adults: analysis of proprietary data submitted to US Food and Drug Administration. BMJ. 2009;339:b2880.PubMed doi:10.1136/bmj.b2880

22. Cipriani A, Barbui C, Geddes JR. Suicide, depression, and antidepressants. BMJ. 2005;330(7488):373–374.PubMed doi:10.1136/bmj.330.7488.373

23. Sweeting MJ, Sutton AJ, Lambert PC. What to add to nothing? use and avoidance of continuity corrections in meta-analysis of sparse data. Stat Med. 2004;23(9):1351–1375.PubMed doi:10.1002/sim.1761

24. Arzneimittelkommission der deutschen Ärzteschaft. SSRI und Suizidalität? Dtsch Arztebl. 2004;101:A2642.

25. Göder R, Friege L, Treskov V, et al. Association of paroxetine with suicide attempt in obsessive-compulsive disorder. Pharmacopsychiatry. 2000;33(3):116–117.PubMed doi:10.1055/s-2000-7977

26. Grohmann R, Engel RR, Rüther E, et al. The AMSP drug safety program: methods and global results. Pharmacopsychiatry. 2004;37(suppl 1):S4–S11.PubMed doi:10.1055/s-2004-815505

27. Grohmann R, Hippius H, Helmchen H, et al. The AMÜP study for drug surveillance in psychiatry: a summary of inpatient data. Pharmacopsychiatry. 2004;37(suppl 1):S16–S26.PubMed doi:10.1055/s-2004-815507

28. Engel RR, Grohmann R, Rüther E, et al. Research methods in drug surveillance. Pharmacopsychiatry. 2004;37(suppl 1):S12–S15.PubMed doi:10.1055/s-2004-815506

29. Grohmann R, Dirschedl P, Scherer J, et al. Reliability of adverse drug reaction assessment in psychiatric inpatients. Eur Arch Psychiatry Neurol Sci. 1985;235(3):158–163.PubMed doi:10.1007/BF00380987

30. Clopper CJ, Pearson ES. The use of confidence or fiducial limits illustrated in the case of the binominal. Biometrika. 1934;26(4):404–413. doi:10.1093/biomet/26.4.404

31. Cooper JE. Pocket Guide to ICD-10 Classification of Mental and Behavioural Disorders: With Glossary and Diagnostic Criteria for Research DCR-10. London, United Kingdom: Churchill Livingstone; 1994.

32. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR). Washington, DC: American Psychiatric Association; 2000.

33. Gibbons RD, Brown CH, Hur K, et al. Early evidence on the effects of regulators’ suicidality warnings on SSRI prescriptions and suicide in children and adolescents. Am J Psychiatry. 2007;164(9):1356–1363.PubMed doi:10.1176/appi.ajp.2007.07030454

34. Gualtieri CT. Paradoxical effects of fluoxetine. J Clin Psychopharmacol. 1991;11(6):393–394.PubMed doi:10.1097/00004714-199112000-00022

35. Healy D, Langmaak C, Savage M. Suicide in the course of the treatment of depression. J Psychopharmacol. 1999;13(1):94–99.PubMed doi:10.1177/026988119901300110

36. Greil W, Horvath A, Sassim N, et al. Disinhibition of libido: an adverse effect of SSRI? J Affect Disord. 2001;62(3):225–228.PubMed doi:10.1016/S0165-0327(00)00150-6

37. Seemüller F, Riedel M, Obermeier M, et al. The controversial link between antidepressants and suicidality risks in adults: data from a naturalistic study on a large sample of in-patients with a major depressive episode. Int J Neuropsychopharmacol. 2009;12(2):181–189.PubMed doi:10.1017/S1461145708009139

38. Meyer JH, McMain S, Kennedy SH, et al. Dysfunctional attitudes and 5-HT2 receptors during depression and self-harm. Am J Psychiatry. 2003;160(1):90–99.PubMed doi:10.1176/appi.ajp.160.1.90