Background

Venlafaxine is hydroxycycloalkylphenylethylamine-derivative bicyclic antidepressant which inhibits the reuptake of both serotonin and noradrenalin. In the early days of its clinical use venlafaxine was thought to have a toxicity profile similar to that reported for selective serotonin reuptake inhibitors (SSRIs) and was sometimes grouped with the SSRIs in self-poisoning studies [1, 2].

More recently, a number of large epidemiological reviews in the UK and Finland have suggested that venlafaxine is associated with more deaths per written prescriptions than other SSRIs suggesting a greater inherent toxicity [3–6]. It is difficult to determine from these large studies whether this is a true toxic effect or one related to the possibility that venlafaxine is prescribed to patients at higher suicidal risk than those prescribed SSRIs [7].

This study aims to compare the demographic details, historical features and clinical sequelae of venlafaxine versus SSRI overdoses presenting to the emergency department of a large tertiary referral hospital over a nine year period.

Methods

This retrospective cohort study was conducted in a large tertiary referral hospital in the western suburbs of Sydney. The hospital’s emergency department is serviced by a consultation-liaison psychiatry and toxicology, both of which have maintained a patient database since 1997. All presentations of actual or suspected self-poisoning to the emergency department are referred to consultation-liaison psychiatry and/or toxicology for assessment. The data from the psychiatric assessment done by a psychiatrist, trainee or clinical nurse consultant are entered into the consultation-liaison database; a toxicology trainee collects and enters the toxicology data obtained by him/herself or by the toxicologist. The study included all patients who had taken an overdose of venlafaxine between 1997 and 2006 as identified from the databases. A comparison group of 44 SSRI self-poisoning cases was randomly selected from a pool of 257. The outcome variable for the power calculation was ‘expressed intent’ (present versus absent). A power analysis determined a sample size of 44 patients who ingested SSRIs would have 80% power to detect a 25% difference in intent compared to venlafaxine with a two-tailed alpha of 0.05. Approval for the study was obtained from the hospital’s human research ethics committee.

All variables were extracted by one author (ANC) who was not blinded to the overdose type. The following descriptive variables were extracted from the medical record of each patient: age, sex, relationship status, length of stay, reason for psychiatric referral and referral service, psychiatric diagnoses made, degree of suicide intent, past suicide attempts, psychiatric follow up, amount of antidepressant taken [expressed as defined daily dose (DDD)], concomitantly ingested medications, alcohol or illicit substances, and medical and surgical history. In addition, the following outcome variables were extracted: length of stay in a medical ward; length of stay in the intensive care unit; need for intubation; Glasgow Coma Score on admission; occurrence of seizures or arrhythmia; longest QRS, QT and QTc; highest BP, pulse and temperature; signs of serotonin toxicity according to the Hunter Serotonin Toxicity Criteria [8]; and death.

Degree of suicidal intent was judged from the information presented in the progress notes and recorded according to a five-point scale: no intent or accidental overdose; denied intent, but ambivalent about survival; admitted intent, but ambivalent about dying; admitted intent and made plans to escape detection.

The amount of venlafaxine (DDD) ingested was calculated from amounts recorded in the medical records. The DDD is based on the assumed average daily dose of the drug when used for its main indication by adults. It is the unit approved by the World Health Organization for drug use studies, and allows for comparisons independent of differences in price, preparation and quantity per prescription [9].

The Hunter Serotonin Toxicity Criteria group features of serotonin abnormalities under three broad areas of mental state (agitation, restlessness, confusion, hypomania), the motor system (clonus/myoclonus, tremor/shivering/hyperreflexia/hypertonia/rigidity) and the autonomic nervous system (diaphoresis, tachycardia, flushing, mydriasis). Three or more of these features suggest serotonin toxicity [8].

Data were analysed with SPSS for Windows Version 14. Two-tailed tests with a significance level of 5% were used throughout. No adjustment was made for multiple comparisons. Chi-squared or Fisher’s exact tests were used as appropriate to look for differences between patient groups in the distribution of categorical variables. Since many of the continuous variables had highly skewed distribution, Mann-Whitney tests were used for differences between the patient groups.

Results

Thirty seven cases of venlafaxine deliberate self-poisoning were identified but one of these could not be included as the patient’s file was lost. Of the 36 medical records reviewed, one patient coingested venlafaxine and sertraline and was included in the venlafaxine group for the purpose of this analysis. Forty four patients who took an overdose of SSRIs during the study period were randomly identified for inclusion in the study. The SSRIs ingested included sertraline, fluoxetine, citalopram, paroxetine and fluvoxamine.

Descriptive and outcome variables are shown below in Tables 1 and 2, respectively. Continuous, non-normally distributed values are expressed as median values, interquartile ranges and Mann-Whitney P values are used. Continuous, normally distributed values are expressed as mean values, standard deviations and t test P values are used. Categorical values are expressed as nominal values and percentage of the number in the group.

Table 1 Descriptive variables in patients who had ingested venlafaxine or an SSRI
Full size table
Table 2 Outcome variables in patients who had ingested venlafaxine or an SSRI
Full size table

Descriptive Variables

Table 1 shows the descriptive variables in patients who had ingested venlafaxine or an SSRI. The patients who ingested venlafaxine had a mean age of 37, compared with 29 for the SSRIs patients (p < 0.001).

Four cases of venlafaxine ingestion were omitted from the calculation because either the number or strength of the tablets ingested were unknown and could not be reasonably estimated. The median venlafaxine overdose was 35 DDDs, whereas the median SSRI overdose was 19.4 DDDs.

Among the venlafaxine group 22.2% (eight of 36) had expressed intent to suicide with plans to escape detection versus 4.7% (two of 44) of SSRI poisoning. In contrast, 25.6% (11 of 43) of SSRI poisoning group denied intent but were ambivalent about the suicide attempt compared with 5.6% (two of 36) of those who took venlafaxine. Overall, the patients who took venlafaxine exhibited a higher degree of intent to die (p ≤ 0.039).

Outcome Variables

Table 2 shows the outcomes of the patients who ingested venlafaxine and SSRIs. The outcome variables to reach a statistically significant difference between the two groups were recorded confusion, mydriasis and longest QT interval.

SSRIs were associated with a significantly longer QT on electrocardiography than venlafaxine. The median longest recorded QT in the SSRI group was 372 ms, while it was only 346 ms in the venlafaxine group (p ≤ 0.030). However, for further analysis, QT was plotted against the heart rate on a QT interval nomogram (Fig. 1) [10]. This showed that neither group demonstrated QT/HR results in the ‘at risk’ range.

Fig. 1
figure 1

QT Nomogram [10] for venlafaxine poisoning versus SSRI poisoning. The bold line is the ‘at risk’ line. If the QT/HR point is above the line, the combination is regarded as ‘at risk’ of developing torsades de pointes

Full size image

One of the 80 patients included in the study died. This patient had ingested venlafaxine XR 12,600 mg in an apparently impulsive overdose of ambivalent intent. She had also taken propranolol, alcohol and amphetamine. Her Glasgow coma scale on admission was 14 out of 15. She subsequently developed seizures, hypotension and ventricular tachycardia. Confusion, mydriasis and diaphoresis were also noted. She died on day 4. A coronial inquest found the cause of death to be related to pericarditis with evidence of superficial myocardial cell necrosis.

Discussion

Previous studies have suggested that the apparent greater toxicity of venlafaxine may be due to it being prescribed for patients at higher risk of overdose by virtue of a greater severity of depression [11] and to its inherently greater toxicity [12, 13].

There was a higher degree of suicidal intent in venlafaxine self-poisoning than SSRI self-poisoning. The most common diagnosis amongst the venlafaxine self-poisoning group was a major depressive disorder.

In our study, patients who ingested venlafaxine were more likely to be confused and have mydriasis than those who took SSRIs. Howell and co-workers also found a high prevalence of mydriasis in patients who took venlafaxine in overdose [14]. Venlafaxine is known to be more toxic in overdose than other SSRIs due to its pro-convulsant activity as well as cardiac sodium channel blocking capacity [15, 16].

This is the first comparison study between venlafaxine and the SSRIs to have investigated QT intervals. In spite of the statistical significance in the higher median QT in SSRIs than venlafaxine, we do not feel we can attribute clinical significance to this after further analysis using the QT Interval Nomogram [10]. As illustrated in Fig. 1, most of the QT intervals were below the ‘at risk’ line for torsades de pointes. There were three QT intervals for venlafaxine and three for SSRIs which were above the ‘at risk’ line. It is unclear as to whether clinical significance can be attributed to these values as moderate increase in QT prolongation is not necessarily associated with a substantially increased risk of torsades de pointes [17]. The reason for interpreting the QT and not the QTc in our study is that Bazett’s formula (QT corrected for heart rate) significantly overcorrects for fast pulse rates and significantly undercorrects for slow pulse rates [18, 19]. This is a problem in our comparison as venlafaxine overdose is associated with tachycardia; and SSRIs with bradycardia, hence rendering comparison difficult. A retrospective study of venlafaxine poisoning found a dose-dependent relationship with prolonged QTc [14].

The biggest limitation of this study is the small sample size. This may explain why some of the findings from previous studies into the comparative toxicity of venlafaxine and SSRIs, such as the occurrence of seizures, were not replicated. Another significant limitation is co-ingestions. As is often the case in the real world, many of the drug ingestions were polydrug overdoses further confounding any simple comparison of the inherent toxicity of the two types of agent. With the exception of the primary outcome variable, ‘expressed intent’, all analyses were purely exploratory. Since a large number of comparisons were made, false positive findings are possible and care must be taken in interpreting these results which require confirmation in future independent studies. The study has other limitations. Only one person (ANC) extracted the data from the files and she was not blinded to the type of overdose involved. The study’s retrospective nature and the use of routinely collected medical records raise the possibility that salient negative findings might have been overlooked, or assessed but not recorded. Both venlafaxine and SSRI overdose were based on historical evidence alone since serum confirmatory testing is not done routinely.

Conclusion

This is the first study to look in detail at the psychiatric profile of venlafaxine self-poisoning. We found that patients who deliberately ingested venlafaxine were more likely to exhibit suicide intent. They were also more likely to have confusion and mydriasis.

Our findings lend support to an evolving hypothesis in the literature that venlafaxine may be more toxic than SSRIs in overdose. However, venlafaxine’s greater association with completed suicide is partly explained by the fact that it tends to be prescribed to patients at higher risk of a serious suicide attempt.