Rapid Responses to:

RESEARCH: Annalisa Rubino, Neil Roskell, Pat Tennis, Daniel Mines, Scott Weich, and Elizabeth Andrews Risk of suicide during treatment with venlafaxine, citalopram, fluoxetine, and dothiepin: retrospective cohort study BMJ 2006; 0: bmj.39041.445104.BEv1 [Abstract]

Rapid Responses: Submit a response to this article

Rapid Responses published:

Congratulations !

EUDIER Francis   (13 December 2006)

Risk of suicide during treatment with antidepressants: the importance of bipolarity.

Daniel J Smith, James T. Walters, Clinical Lecturer.   (10 January 2007)

A contemplation but does it change?

Shamuz Oowise   (14 January 2007)

Suicidal risk with Venlafaxine assessed with the use of GPRD: Some Thoughts!

Nitin Gupta, Nasrullah Khan-SHO Old Age Psychiatry, Rakhee Gupta-SHO Psychiatry   (1 February 2007)

Surviving antidepressant treatment

Rajeev Krishnadas   (5 February 2007)

is venlafaxine the same as others antidepressants?

nicholas Moore   (5 February 2007)

Prescribing Venlafaxine

sheeba ninan   (5 February 2007)

An Index of anti-depressant suicide risk

Michael D Beary   (6 February 2007)

Impact on patients?

Abhilasha sharma, Chitra Srinivasan, SpR   (7 February 2007)

Undetected bipolar depression

James Paul Pandarakalam   (13 February 2007)

Venlafaxine in depressed suicidal patients: What’s the message?

Nilamadhab Kar, Rajesh Dasi, Nicola Felton   (7 March 2007)

Risk of suicide during venlafaxine treatment: further considerations

Elizabeth B. Andrews, Annalisa Rubino, Neil Roskell, Pat Tennis, Daniel Mines, and Scott Weich   (9 March 2007)

Congratulations ! 13 December 2006
EUDIER Francis, MD Psychiatry and Psychiatry-Liaison University Hospital Rennes. Hop. Pontchaillou 35033 Rennes Cedex . France Send response to journal: Re: Congratulations !	Thanks for this beautiful study which, contrary to the others, takes account of the history of patients (previous suicide attempts, severity of depression, hospitalizations...) to explore the problem of suicide, and not only of pharmacology. Now, the question about links between suicide and antidepressants seems to me almost closed !(but it can be continued in the same way) Competing interests: None declared
Risk of suicide during treatment with antidepressants: the importance of bipolarity. 10 January 2007
Daniel J Smith, Clinical Lecturer Cardiff University, CF14 4XN, James T. Walters, Clinical Lecturer. Send response to journal: Re: Risk of suicide during treatment with antidepressants: the importance of bipolarity.	Rubino and colleagues have identified that most (if not all) of the excess risk of suicide in a venlafaxine-treated group of patients could be explained by a higher burden of risk factors for suicide [1]. While it may be true that this group of patients had more severe or ‘difficult-to- treat’ unipolar depression, it is also possible that bipolar features in this group may be responsible for the observed elevated rates of suicidality. Perhaps because of limitations of space, the authors do not discuss this as a possibility, despite an adjusted relative risk of completed suicide of 4.94 (95% CI 1.30-18.84) for ‘past history of bipolar disorder’ (Table 3) [1]. Recent work suggests that at least 50% of ‘difficult-to-treat’ unipolar depressed patients may have an undetected bipolar disorder [2] and it is now well documented that antidepressant monotherapy for bipolar depression runs a high risk of precipitating hypomanic and/or mixed affective states [3], which have been strongly associated with self-harm and completed suicide [4]. It is also the case that venlafaxine appears to be more likely than other antidepressants to precipitate a switch into hypomania or mania in bipolar depression [5]. Furthermore, many of the variables reported by Rubino and colleagues could be considered to point towards high levels of bipolarity in the venlafaxine-treated group [6], including: higher rates of a family history of psychiatric disorder; more frequent prescription of antipsychotics and mood stabilisers; a history of non-response to several different antidepressants; and more frequent lifetime depressive episodes. References 1. Rubino A, Roskell N, Tennis P, Mines D, Weich S, Andrews E. Risk of suicide during treatment with venlafaxine, citalopram, fluoxetine, and dothiepin: retrospective cohort study. BMJ 2006: bmj.39041.445104.BE. 2. Sharma V, Khan M, Smith A. A closer look at treatment resistant depression: is it due to a bipolar diathesis? Journal of Affective Disorders 2005;84(2-3):251-257. 3. Goldberg JF, Truman CJ. Antidepressant-induced mania: an overview of current controversies. Bipolar Disorders 2003;5:407-420. 4. Akiskal HS, Benazzi F, Perugi G, Rihmer Z. Agitated "unipolar" depression re-conceptualized as a depressive mixed state: implications for the antidepressant-suicide controversy. Journal of Affective Disorders 2005;85(3):245-258. 5. Post RM, Altshuler LL, Leverich GS, Frye MA, Nolen WA, Kupka RW, et al. Mood switch in bipolar depression: comparison of adjunctive venlafaxine, bupropion and sertraline. British Journal of Psychiatry 2006;189(2):124-131. 6. Angst J. Do many patients with depression suffer from bipolar disorder? Canadian Journal of Psychiatry 2006;51:3-5. Competing interests: None declared
A contemplation but does it change? 14 January 2007
Shamuz Oowise, Senior House Officer in Psychiatry Crisis Resolution and Home Treatment Team , Lime Trees, Calnwood Road, Luton, LU4 0FB Send response to journal: Re: A contemplation but does it change?	The study is a landmark, in contemplating thoughts for future research and therefore generated the following. The study does not denote the reasons for the higher suicide risk associated with Venlafaxine compared to the Citalopram group. It also mentions that 30% of the confounders remained uncontrollable, and most importantly impulsivity, hopelessness and suicidal ideation are among them. Following could be possible explanations. Venlafaxine enables the patient to suicide or attempt due to 1. Increase in impulsivity 2. Improvement in psychomotor retardation in patients with ongoing suicidal ideations and hopelessness 3. Worsening depressive symptoms leading to an attempt or suicide Firstly, If the suicide or attempt is due to an increase in impulsivity a careful evaluation is needed prior to the commencement of Venlafaxine, especially if a past history of impulsivity exists. Secondly, If its due to the improvement of severe psychomotor retardation, monitoring at higher levels of observation during the early stages would help to prevent a suicide or an attempt. Lastly, If the suicide or attempt is due to worsening symptoms due to the use of Venlafaxine, the Citalopram group should be considered in favour of Venlafaxine at every possible opportunity. Further RCTs are needed to prove all these associations. My only query is whether the study changes our current views and prescribing practices in anyway? Competing interests: None declared
Suicidal risk with Venlafaxine assessed with the use of GPRD: Some Thoughts! 1 February 2007
Nitin Gupta, Consultant Psychiatrist-South Staffordshire Healthcare NHS Foundation Trust Margaret Stanhope Centre, Belvedere Road, Burton upon Trent, DE13 0RB., Nasrullah Khan-SHO Old Age Psychiatry, Rakhee Gupta-SHO Psychiatry Send response to journal: Re: Suicidal risk with Venlafaxine assessed with the use of GPRD: Some Thoughts!	The authors draw conclusions that venlafaxine use is consistently associated with higher risk of suicide; but the presence of residual confounding factors does not make the link causal. The study is indeed a welcome addition to the ongoing debate on antidepressant use and suicide risk, more so from the fact that it is based on the UK GPRD. The authors mention that GPRD is a robust longitudinal database available to assess a rare outcome like suicide. They further acknowledge certain limitations of the data coding and utility of GPRD for assessing the confounding variables in the study. But, in the same vein, certain tacit assumptions related to the robustness and accuracy of the recorded data have been made which may not entirely be appropriate. We would like to mention a few such examples. In terms of overall number of participants, venlafaxine was prescribed the least. Is this a reflection of it being initiated less in primary care? This factor does need to be taken into account as prescribing can be influenced by a host of factors that are relatively independent of the illness for which the medication is being prescribed e.g. cost of medication, place of use of a medication on the drug formulary etc. These factors can operate equally in primary and secondary care, but nevertheless can be important confounding factors (and we strongly suspect that these would not be recorded on the GPRD database). Another possible reason could be that GPs were not using venlafaxine as routinely as the SSRIs due to the depression not being severe at initial presentation, or due to the side-effect profile (and monitoring required) with venlafaxine [1]. It needs to be kept in mind that SSRIs (since the introduction of fluoxetine) have generally been used as first-line treatment for depression, and the same would realistically hold true for citalopram. Hence, although both citalopram and venlafaxine were introduced in the same year, the pattern of prescribing may not be concordant for the above-mentioned reasons, generating another inherent (and probably robust) confounding factor. There is no information available to take into account these important issues as to where was the medication initiated- in primary or secondary care? The authors do mention that “venlafaxine use in this study population was associated with markers of severe and difficult to treat depression” (Page 3 of 6, under Discussion, para 2). This appears to be a tenable observation, based upon the data studied and also as suggested by NICE guidelines [1]. However, they further mention that this may be explained due to lack of desired therapeutic effect of previous antidepressant leading to a switch to venlafaxine (Page 3 of 6, under Discussion, para 2). We feel that certain other key variables contributing to switchover to another antidepressant (in this case-venlafaxine) have not been alluded to. Change of antidepressant can also be due to (a) intolerability and side-effects, (b) and/or if the patient desires a change of medication (which is agreed to by the clinician- willingly or otherwise) prior to an adequate therapeutic trial, (c) the clinicians in the hope of getting the patient better as soon as possible, (d) issues related to (dis)continuity of care and communication within/between primary/secondary care. These are probably inherent issues occurring in the context of clinical practice in the best interest of the patients for managing their depression, and hence may be largely unavoidable for the clinician. We feel that it is important to appreciate the relevance of these caveats as the GPRD database does not necessarily record these issues related to prescribing. Quality control criteria were applied in this study to the GPRD database, and it may be possible that some of the issues mentioned above may have been part of this quality control exercise. Nevertheless, we need to be equally mindful of the conclusions by Stevens et al [2]that- routine data are plentiful but of limited use in Health Technology Assessment, the data sets generally do not include the effects of treatment, and coding is inadequate. Our observations, per se, do not dilute the utility of the GPRD and this important study in any manner. REFERENCES [1] http://www.nice.org.uk/guidance/CG23 (last accessed 31 January 2007) [2] Stevens AJ, Raftery J, Roderick P. Can health technologies be assessed using routine data? International Journal of Technology Assessment in Health Care 2005; 21 (1): 96-103. Competing interests: Dr. Nitin Gupta has received speaker honoraria and educational support grants from Wyeth, Lundbeck, Eli-Lilly, and AstraZeneca.
Surviving antidepressant treatment 5 February 2007
Rajeev Krishnadas, SHO Tranwell Unit, QE Hospital, Gateshead. Send response to journal: Re: Surviving antidepressant treatment	I read with interest the article by Rubino et al (1). I congratulate the authors on the effort. The study adds to a host of other studies, including a metaanalysis on the subject (2). The authors suggest a causal association of suicide to venlafaxine, but rightly acknowledge the numerous limitations of the study and the need to interpret the data with caution. Although this was an epidemiological study, the authors did not take into consideration the mean dose of the medications the patients were on, and whether this varied between those who attempted suicide and those who "survived" the medication. This may be important because of the fact that Venlafaxine shows a dose response effect, especially considering the fact that it is an SNRI only at higher doses. So, if the dose was below optimum for a particular patient, it would have acted just like any other SSRI (Citalopram/Fluoxetine). Morover, the data may suggest that, treating severely depressed patients with venlafaxine at an SSRI dose was not adequate, especially with earlier data to support the evidence of a causal link between SSRIs and attempted suicide.(3) The period of study also saw changes in prescribing restrictions of venlafaxine in primary care. I wonder if that could have had an effect on the study findings. It would be a shame if a patient is denied venlafaxine on the basis of these findings. Reference 1. Annalisa Rubino, Neil Roskell, Pat Tennis, Daniel Mines, Scott Weich, and Elizabeth Andrews Risk of suicide during treatment with venlafaxine, citalopram, fluoxetine, and dothiepin: retrospective cohort study BMJ 2007; 334: 242. 2. Fergusson D, Doucette S, Glass KC, Shapiro S, Healy D, Hebert P et al. Association between suicide attempts and selective serotonin reuptake inhibitors: systematic review of randomised controlled trials. BMJ 2005; 330: 396. 3. Didham RC, McConnell DW, Blair HJ, Reith DM. Suicide and self- harm following prescription of SSRIs and other antidepressants: confounding by indication. Br J Clin Pharmacol 2005; 60: 519-525. Competing interests: None declared
is venlafaxine the same as others antidepressants? 5 February 2007
nicholas Moore, Pharmacology department head 33000 Bordeaux, France Send response to journal: Re: is venlafaxine the same as others antidepressants?	This is an extremely nice study, done by very competent pharmacoepidemiologists (and friends). They confirm what has been suspected for quite some time, that depression is a major risk factor for suicide, and this is severity of depresssion dependent. I seem to recall having been taught about 35 years ago during my medical studies that the risk of suicide was greatest during the first month of treatment, because antidepressants would improve the motor component of depression before improving the mood itself, giving the patient the courage and strength to attempt suicide (thank you, doctor, I feel much better with your treatment. Now I can at last do what I've been wanting to do for some time, said the patient as he(she) gets up and jumps out the window…) The more severe the depression initially, the greater the risk, and the more careful one should be. This risk was less with placebo which does not improve the patient thus (which justified the continued use of placebos in clinical trials of depression). At least that was what was taught, but that was before EBM. I suspect the situation has changed since, when I see all that has been said about antidepressants and suicide. So I'm happy to see it confirmed here. The second point is that venlafaxine does not quite have the same properties as the other drugs: the others are selective inhibitors of serotonine only, whereas venlafaxine like clomipramine inhibits the recapture of both serotonine and noradrenaline. In that, it is considered equivalent to the older tricyclics, without all the anticholinergic and cardiac side-effects. Whether or not this changes the antidepressant efficacy of the drug compared to pure SSRI may be debated but physicians consider venlafaxine as different from the SSRI, and use it in more severe patients. This we found some time ago (1,2). We're happy to see it confirmed here (even if we were not cited). Another minor point, that may also be related: we found that the patients seen by psychiatrists were more depressed and had had more previous episodes than those seen by GPs. GPRD probably does not contain this information, since it is only a GP-based ressource, but is there any indication of the initiator of the prescription: the GP or a psychiatrist? Congratulations again Nicholas Moore 1. Rambelomanana S, Depont F, Forest K, Hebert G, Blazejewski S, Fourrier-Reglat A, et al. Antidepressants: general practitioners' opinions and clinical practice. Acta Psychiatr Scand 2006;113(6):460-7. 2. Depont F, Rambelomanana S, Le Puil S, Begaud B, Verdoux H, Moore N. Antidepressants: psychiatrists' opinions and clinical practice. Acta Psychiatr Scand 2003;108(1):24-31.) Competing interests: I have been working with various drug companies over the years, including most of those marketing antidepressants, including on these antidepressants, but no competing interests in this specific response, apart from the opportunity of self-citation
Prescribing Venlafaxine 5 February 2007
sheeba ninan, sho hillingdon hospital,UB8 3NN Send response to journal: Re: Prescribing Venlafaxine	This study gives information to guide clinicians in practice while prescribing Venlafaxine .It is interesting to note that some of the patients respond very well to the drug.But the key point is to keep in mind history of risk when prescribing.It is vital to maintain a fine line between risks and benefits. Also presecence of confounding factors is a field which may benefit from exploration. Competing interests: None declared
An Index of anti-depressant suicide risk 6 February 2007
Michael D Beary, Consultant Psychiatrist Priory Hospital North London, London N14 6RA Send response to journal: Re: An Index of anti-depressant suicide risk	An index of anti-depressant suicide risk I have constructed an Index of anti-depressant suicide risk from the data presented by Rubino et al. (BMJ 2007; 334:242-5), which has two findings of clinical note. The first is the high risk of suicide with Venlafaxine. The second is the safety of the tricyclic Dothiepin (dosulepin). The NICE guidelines state that for treatment of moderate to severe depression only specialist mental health professionals should routinely initiate treatment with dosulepin. Rubino’s data suggest that concern that toxicity in overdose of tricyclics makes them too dangerous for general practitioners to prescribe is unfounded. The data suggests that they are indeed a safe first line treatment. Index of anti-depressant suicide risk Unadjusted incidence of completed suicide per 1000 person years of antidepressant Unadjusted incidence of completed suicide per 1000 person years of fluoxetine Fluoxetine (0.23) 1 Dothiepin (0.27) 1.17 Citalopram (0.26) 1.13 Venlafaxine (0.64) 2.78 Dr Michael Beary Consultant Psychiatrist The Priory Hospital North London, Southgate N14 6RA michaelbeary@prioryhealthcare.com Competing interests: None declared
Impact on patients? 7 February 2007
Abhilasha sharma, specialist registrar Wonford house hospital,Exeter,EX2 5BB, Chitra Srinivasan, SpR Send response to journal: Re: Impact on patients?	We have used venlafaxine for the treatment for depression over the past few years. The results have been varied but most of our experiences have been positive. Hence we were curious to know, what the impact would be on patients who are currently using venlafaxine? The study by Rubino is likely to be used by media as well as the competing drug companies to highlight the association between venlafaxine and suicide. One of the studies (1) has shown the effects of media. Our thoughts are- ‘Are we going to issue a media warning based on the study conclusion or have we already done it through the editorial?’’ It is somewhat reassuring to note that the authors (conclusions), editorial (paragraphs 4,5,6) and, detailed discussion in the rapid responses have discussed the limitations of the study. However, this is true mostly if professional or person with critical appraisal orientation reads the results. Given the impact of media and modern day anxieties (2), is it right to accept the results and call this study a landmark? We felt that this study is far from being a definitive and probably needs more research. 1. Effects of professional and media warning about the association between aspirin use in children and Reyes syndrome. S B Sumerai, The MillbBank Quarterly, Vol-170, No.1 (1992) 2. Modern worries, new technologies and medicine. Editorial, BMJ 2002,324 Competing interests: None declared
Undetected bipolar depression 13 February 2007
James Paul Pandarakalam, consultant psychiatrist, 5 Borough Partnership NHS Trust St Helens North CMHT, Peasley Cross Resource Centre, St Helens, Merseyside WA 9 3DA Send response to journal: Re: Undetected bipolar depression	S.N.R.I.s and S.S.R.I.-N.R.I. combination run the risk of hypomanic or manic switch. It is quite possible that some of the depressed patients treated with venlafaxine may be having undetected bipolarity which has a higher mortality rate. Bipolar depression differs from other forms of depression because of the high risk both of completed suicide and of psychotic features 1(Akiskal , Walker , Puzantian 1983). Bipolar depression is commonly misdiagnosed because of the similarities between its symptoms and those of major depression. Uniopolar depression needs to be readily differentiated from bipolar depression because the inappropriate use of antidepressants may precipitate a hypomanic or manic episode. Misdiagnosis can lead to unfocused treatment that may exacerbate the disease 2. The severity of the bipolar depression within the depressive episodes is variable, and this needs to be distinguished for treatment purposes. Atypical depressive features are more noticeable in bipolar depressives, as is psychomotor retardation 3. Patients also have a previous history of psychotic depression. Bipolar depression is associated with more mood lability and a relatively acute onset. When a young person presents with psychotic depression or psychomotor retardation, bipolar depression should be ruled out. Non-specific psychological symptoms and behavioural disturbances may be the precursor of bipolar disorder in young people 4. Any patient presenting with depressive symptomatology should be interrogated about their past history of mood elevation, and the family history of affective disorders should be explored also. Bipolar depression can occur with or without precipitating factors. Bipolar depression should be especially ruled out when depressions occur without an identifiable psychogenic stressor. A subset of patients diagnosed with unipolar depression in fact suffers from bipolar depression 3.When patients fail to respond to even one antidepressant, bipolarity should be screened if non adherence and suboptimal dosing are not the reasons for the antidepressant failure. Non response to antidepressant medication, comorbid anxiety, feelings of people being unfriendly, recent depression diagnosis, bipolar disorder family history and legal problems may prove useful indicators of bipolarity. It should be remembered that most patients with bipolar disorder seek treatment for depression, and not for mania or hypomania. It is estimated that one third of patients require 10 years from the onset to correct bipolar disorder diagnosis 5. It is a matter of common knowledge that depressed patients commit suicide when they are getting better and not always when they are in the nadir of depression and this observation could explain some the reported high suicidal risk with venlafaxine which is relatively more potent than S.S.R.Is. References 1.Akiskal HS, Walker P, Puzantian VR (1983) Bipolar outcomes in the course of depressive illness: phenomenological, familial, and pharmacological predictors. J. Affective disorders 5:115-28. 2.Hirschfild R M A, lewis L, Vornik LA(2003) Perceptions and impact of bipolar disorder, hoe far have we really come? Results of the National and Depression and Manic Depressive Association ,survey of Individuals with Bipolar Disorder. J.Clin Psychiatry. 64: 161-174. 3.Mitchell PB,Wilhelm K, Parker G, Austin MP, Rutgers P, Malhi GS,(2001)The Clinical features of bipolar depression: a comparison with matched major depressive disorder patients. J Clinical Psychiatry 62:212- 216. 4.Goodwin G.M (2003) Evidence-based guidelines for treating bipolar disorder: recommendations from the British Association for Psychopharmacology. Journal of Psychopharmacology 17:2149-173. 5.Suppes T.,Leverich GS., Keck PE (2001) The Stanley Foundation Bipolar.Treatment Outcome Network. Demographics and illness charecteristcis of the first 261 patients. J Affective Disorders .67: 45- 59. Competing interests: None declared
Venlafaxine in depressed suicidal patients: What’s the message? 7 March 2007
Nilamadhab Kar, Consultant Psychiatrist Wolverhampton City Primary Care Trust, Corner House, 300 Dunstall Road, Wolverhampton WV6 0NZ, Rajesh Dasi, Nicola Felton Send response to journal: Re: Venlafaxine in depressed suicidal patients: What’s the message?	Rubino et al [1] in a study involving 219 088 patients found that venlafaxine use was consistently associated with higher risk of suicide compared to citalopram, fluoxetine, and dothiepin. Venlafaxine users had higher burden of suicide risk; adjustment for measured confounders substantially reduced the excess risk but did not eliminate. Authors hoped that residual confounding could explain much or all of the remaining risk. So what’s the message regarding use of venlafaxine in depressed suicidal patients? How are the study conclusions going to affect the clinical practice? In spite of large population size as a major strength, the index study has several limitations which the authors have stated generously. With major limitations like probability of exposure and outcome misclassification, missing information on prescribing within secondary care, assumed compliance and time at risk, the reliability of the conclusions can be debated. In addition, there were indirect and partial measurements of many variables. Is it possible that these obvious limitations have biased the authors to believe that the remaining risk can be explained by the residual confounding? Anyway, with all these limitations the issue of venlafaxine and increased risk of suicidality appears to remain inconclusive. Considering the fact that three meta-analyses of clinical trials in adults had found no increased risk of suicide associated with antidepressants compared to placebo [2, 3, 4], the results of this study suggesting a probability of increased risk with venlafaxine does raise some concern. This issue needs to be addressed, probably using different methodologies. Considerable number of patients in the index study received venlafaxine as first line antidepressant. Probably these were not severely ill and had less burden of risk of suicide. It would be interesting to know the outcome in these patients compared to that in patients who received other antidepressants as first line. References 1. Rubino A, Roskell N, Tennis P, Mines D, Weich S, Andrews E. Risk of suicide during treatment with venlafaxine, citalopram, fluoxetine, and dothiepin: retrospective cohort study. BMJ. 2007 Feb 3;334(7587):242. Epub 2006 Dec 12. 2. Khan A, Khan S, Kolts R, Brown WA. Suicide rates in clinical trials of SSRIs, other antidepressants, and placebo: analysis of FDA reports. Am J Psychiatry 2003;160:790-792. 3. Storosum JG, van Zwieten BJ, van den Brink W, Gersons BPR, Broekmans AW. Suicide risk in placebo-controlled studies of major depression. Am J Psychiatry 2001;158:1271-1275. 4. 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;330:385. Competing interests: None declared
Risk of suicide during venlafaxine treatment: further considerations 9 March 2007
Elizabeth B. Andrews, Epidemiologist RTI Health Solutions, Annalisa Rubino, Neil Roskell, Pat Tennis, Daniel Mines, and Scott Weich Send response to journal: Re: Risk of suicide during venlafaxine treatment: further considerations	We would like to thank the authors of the editorial (1) and numerous rapid responses to our article (2), for generating an interesting debate. We feel some points are worth further consideration. In our study, the prevalence of bipolar diagnosis was higher among venlafaxine patients (1.2%) than among citalopram patients (0.5%), fluoxetine patients (0.3%), and dothiepin patients (0.3%). Our study, after adjusting for other risk factors, showed that bipolar disease is associated with a 1.5-fold% increased risk of attempted suicide and 4.9- fold increase in suicide. We agree with Smith and Walters that patients with undiagnosed bipolar depression could account for some of the remaining excess risk associated with venlafaxine (3, 4). However, because of the observational nature of our study, we could not evaluate the role of misdiagnosed psychiatric co-morbidities. We appreciate the general consensus on the methodological challenges and limitations of observational studies such as the one that we conducted in the GPRD. Prescribing is by definition a non-random process that involves patient selection and clinical judgment. This poses a remarkable methodological challenge to any comparisons between populations treated with different medications. We agree that controlling for prescribing behaviour and for possible discrepancies in prescribing patterns in primary care as opposed to specialist psychiatric care is not always feasible. However, in the UK prescribing is almost always transferred to the GP following initial prescription by specialists. Regarding the recommendation of venlafaxine as second choice treatment noted in several responses, it is noteworthy that NICE clinical guidelines for the management of depression (5) were dated 15 December 2004, while this analysis spanned from 1995 to 2005. We a priori planned an analysis that would adjust for confounding by severity of depression. Our study is noteworthy precisely because it tested the important hypothesis that venlafaxine use is an independent risk factor for suicide, after taking account as far as possible for indications for prescribing. We have concluded that the excess risk remaining after control for confounders could well have been due to residual confounding. Our view is that, based on the results of our study, the excess risk of suicide associated with venlafaxine treatment should not be seen as causal. This conclusion is consistent with a meta-analysis of randomized clinical trials evaluating antidepressants among adults published after the submission of our manuscript. The aforementioned meta-analysis (6) showed that adults randomized to venlafaxine were not at increased risk of suicide or suicidal behaviour, although it also showed that among adults less than 25 years of age use of any antidepressant compared with placebo was associated with increased risk of suicidal behaviour). Finally, Beary derived an “index of antidepressant suicide risk” from the unadjusted incidence rate of suicide during dothiepin, citalopram and venlafaxine treatment compared with that during fluoxetine treatment. Such an estimate of relative incidence rate was incorrectly used to make inferences on the toxicity in overdose of tricyclic antidepressants (7). Several studies have evaluated fatality rates among those who have taken an overdose with antidepressants (8,9). Our study measured the association of antidepressant treatment with suicide risk but did not evaluate the fatal toxicity index of antidepressants. The latter was beyond the scope of the present analysis and would have required an ad hoc study design and analytical approach. REFERENCES 1. Andrea Cipriani A, John R Geddes JR, and Corrado Barbui C. Venlafaxine for major depression. BMJ, Feb 2007; 334: 215 - 216; doi:10.1136/bmj.39098.457720.BE 2. Rubino A, Roskell N, Tennis P, Mines D, Weich S, Andrews E. Risk of suicide during treatment with venlafaxine, citalopram, fluoxetine, and dothiepin: retrospective cohort study. BMJ, Feb 2007; 334: 242; doi:10.1136/bmj.39041.445104.BE 3. Smith DJ, Walters JT. Bipolarity is important during treatment with antidepressants. BMJ 2007;334:327; doi:10.1136/bmj.39121.846667.1F 4. Pandarakalam JP. Undetected bipolar depression. BMJ 2007;334:327 (13 February), doi:10.1136/bmj.39121.846667.1F 5. http://www.nice.org.uk/guidance/CG23 (last accessed 14 February 2007). 6. http://www.fda.gov/ohrms/dockets/ac/06/briefing/2006-4272b1-01-FDA.pdf (last accessed 23 February 2007) 7. Beary MD. An Index of anti-depressant suicide risk. BMJ 2007;334:327 (6 February), doi:10.1136/bmj.39121.846667.1F 8. Buckley NA, McManus PR. Fatal toxicity of serotoninergic and other antidepressant drugs: BMJ 2004; 325:1332-1333 9. Morgan O, Griffiths C, Baker A, Majeed A. Fatal toxicity of antidepressants in England and Wales, 1993-2002. Health Statistics Quarterly, Office of National Statistics 2004; 23:18-24. Competing interests: This is a submission from all the authors of the original article.