The field of psychiatry is awash with systematic reviews, meta-analyses and other published articles proclaiming various discoveries. But can this research be trusted? Is it of good quality? Is it ethical? Let us examine one such article and see.
In his article “Suicide prevention strategies revisited: 10-year-old review,” Zalsman (2016, p. 1)1 shares with us his discoveries:
“The anti-suicidal effects of clozapine and lithium have been substantiated, but might be less specific than previously thought.”
Clozapine and lithium have “anti-suicidal effects”?
As I found this statement shocking, I decided to check the sources.
This has led me into a quite extensive maze of variable quality meta-analysis and research. Although this analysis is not exhaustive, it proposes to study the first articles cited by Zalsman, to see if he correctly reports them and if they are good quality.
Insofar as we are not going to find very good science in this journey, it is an encouragement for the public to systematically check the questionable sources of publications.
Clozapine according to Zalsman
Clozapine is the only drug indicated in the USA for reduction of suicide risk in psychosis. A meta-analysis (Asenjo Lobos, 2010) of the effects of clozapine in comparison with other dopamine and serotonin-receptor antagonists (eg, olanzapine and risperidone) demonstrated anti-suicidal effect in schizophrenia. (Asenjo Lobos, 2010)
(Zalsman, 2016, p. 3)
By design, this meta-analysis can not demonstrate that clozapine has an anti-suicide effect. This meta-analysis compares neuroleptics with each other: it is quite possible that all neuroleptics increase the risk of suicide, but that clozapine does it a little less than the others. But more importantly:
Do the authors of the review tell the truth?
Clozapine according to the sources
Asenjo Lobos, 20102
Here are all the citations of Asenjo Lobos (2010) about clozapine and suicide, in the abstract and in the body text. Citations from the study cards and from the tables were omitted because they are difficult to present and they are essentially the same content. This publication is open access.
I highlight some interesting passages:
Other findings that should be replicated were: clozapine improved social functioning less than risperidone and fewer participants in the clozapine group had to be hospitalised to avoid suicide attempts compared to olanzapine.
1. Comparison 1. CLOZAPINE versus OLANZAPINE
Only two studies reported the mortality during the trials (Conley 2003; Meltzer 2003). Deaths from any reason (1 RCT, n=980, RR 1.50 CI 0.62 to 3.64), natural causes (2 RCTs, n=993, RR 1.40 CI 0.45 to 4.38) and suicide (2 RCTs, n=993, RR 1.67 CI 0.40 to 6.94) were all similarly likely whether allocated to clozapine or olanzapine.
[Note: the Conley’s study reports all discontinuations and none of them is the death. (Kelly, 2003 p. 183 pdf p. 3)]
1.19 Service use: Hospital re-admission
Most of studies did not provide data on service use. Only one long term study (Meltzer 2003) reported the hospitalisation for imminent risk of suicide as a rescue intervention. Significantly fewer people taking clozapine (20%) were hospitalised compared to those taking olanzapine (26%)(1 RCT, n=980, RR 0.78 CI 0.62 to 0.98, NNT 18 CI 9 to 230).
5. Comparison 3. CLOZAPINE versus RISPERIDONE
Only two short term studies reported on death. There were no reports of suicides (Wahlbeck 2000). Regarding death due to natural causes, no difference was found between clozapine and risperidone (2 RCTs, n=293, RR 0.98 CI 0.06 to 15.48).
Other findings that should be replicated were that clozapine improved social functioning less than risperidone and fewer participants in the clozapine group had to be hospitalised to avoid suicide attempts in comparison with those from olanzapine group.
1. Comparison 1: CLOZAPINE versus OLANZAPINE
Only two studies assessed this outcome. Data on deaths for any reason as well as for natural causes and suicide were available. A tendency in favour of olanzapine was observed, but these differences were not statistically significant. Hence, it can be concluded that the incidence of death was generally low and similar between treatment groups.
1.19 Service use
A long term study, Meltzer 2003, showed a significant difference, where more participants in the olanzapine group had to be hospitalised to avoid suicide attempts. The relevance of this finding is limited due to the small effect size and the CI observed, which showed that it is possible that the true RR value is likely to be very close to 1.
5. Comparison 3: CLOZAPINE versus RISPERIDONE
Death is an outcome not usually evaluated in studies. There was no evidence of any difference between short term treatment groups with respect of the incidence of death due to natural causes. In the case of death by suicide, a small study reported that it did not occur, therefore the treatment effect was not estimable.
3. For managers/policy makers […] Service use was reported by only one large study [clozapine vs olanzapine] which showed fewer hospitalisations to avoid suicide attempts. This evidence base is too limited for making recommendations.
That’s it, that’s all.
Subsequently, I checked the only study according to which clozapine might have a slight advantage over olanzapine.
The principal publication of this study (Meltzer, 2003) is open access.
This study initially gathered 980 suicidal subjects for two years. There were more suicides in the clozapine group than in the olanzapine group (1.0% vs 0.6%, table 4), but this difference is not significant.
To prevent suicides, it was planned to hospitalize subjects in suicidal crisis, drug them and submit them to electroshocks (Alphs, 2004, p. 583, pdf p. 7)4. “All other appropriate means were allowed.” Even psychotherapy.
Almost all subjects were subjected to massive polypharmacy (Glick, 2004, abstract)5. 90% of subjects took at least two psychotropic drugs. On average, the clozapine group took 4.8 drugs and the olanzapine group 5.2. The olanzapine group took higher doses than the clozapine group.
The olanzapine group took more antidepressants (55.1% vs 49.1%) and more axiolytics/soporifics (69.4% vs 60.2% ) than the clozapine group (Meltzer, 2003, table 4).
38.7% of the subjects left the study along the way: 39.2% in the clozapine group and 38.2% in the olanzapine group (ibid., table 3). 28.7% of the subjects were considered “much worsening from baseline,” 24.5% in the clozapine group and 32.9% in the olanzapine group (ibid., table 4).
1.3% died: 1.6% in the clozapine group, and 1.0% in the olanzapine group (ibid., table 3), the majority by suicide in both groups (ibid., table 4).
This study was sponsored by Novartis Pharmaceuticals Corp.
* * *
In summary, Zalsman cites a single meta-analysis in support of his claim that clozapine has “anti-suicide effects.”
In this meta-analysis, no studies compare clozapine to placebo or to no medical treatment. Only one study shows that clozapine is associated with a slightly lower rate of rehospitalization due to suicide crisis compared with olanzapine. The authors of the meta-analysis report that no one should rely on it.
Finally, Zalsman attributes to his assertion “an anti-suicidal effect of clozapine in psychosis has been demonstrated” the level of evidence 1a according to the Oxford criteria — the highest level (table 2, p. 5).
Let’s move to lithium.
Lithium according to Zalsman
“Reasonably strong evidence from randomised controlled trials (RCTs) shows that lithium is effective in reducing the risk of suicidal behaviour in people with mood disorders, (Cipriani, 2013; Baldessarini, 2006; Kessing, 2005) which was supported by the findings of a large-scale naturalistic cohort study (Goodwin, 2003) comparing lithium with valproate. A specific anti-suicidal effect of lithium was suggested in a population of people who had attempted suicide and who were treated with lithium, compared with people who had attempted suicide and were treated with placebo, (Lauterbach, 2008) although the number of deaths was very small (three deaths vs no deaths on lithium). A large-scale naturalistic study (Gibbons, 2009) suggested that anticonvulsant mood stabilisers might also have protective effects against suicide attempts.”
(Zalsman, 2016, p. 3)
If there is less suicidal behavior in a lithium group than in a valproate group, it does not prove that lithium has anti-suicide effects.
But at least here, it seems that there are studies vs placebo. We will focus on studies comparing lithium to placebo.
Lithium according to the sources
This is a meta-analysis in open access.
Let’s go straight to the part that talks about placebo studies.
Lithium versus placebo
Lithium was more effective than placebo in reducing the number of suicides (Peto odds ratio 0.13, 95% confidence interval 0.03 to 0.66; fig 2⇓)
Bauer, 2000, lithium: 0/14, placebo 1/15
Lauterbach, 2008, lithium: 0/84, placebo 3/83
Prien, 1973a, lithium: 0/45, placebo: 1/39
Prien, 1973b, lithium: 0/101, placebo: 1/104
Suicide rate in placebo groups is absolutely insane. Let’s check the sources.
Source? Source? The bibliography is not available online: you have to ask the author for it. But thanks to the internet, we can find these studies by another way.
This is the abstract:
Use of lithium to augment antidepressant medication has been shown to be beneficial in the acute treatment of depression. The authors examined the efficacy of lithium augmentation in the continuation treatment of unipolar major depressive disorder.
Thirty patients with a refractory major depressive episode who had responded to acute lithium augmentation during an open 6-week study participated in a randomized, parallel-group, double-blind, placebo-controlled trial of lithium augmentation during continuation treatment. After a 2-4-week stabilization period following remission, patients were randomly assigned to receive either lithium or placebo for a 4-month period (?!). Antidepressant medication was continued throughout the study.
Relapses (including one suicide (!!)) occurred in seven (47%) of the 15 patients who received placebo in addition to antidepressants. None (0%) of the 14 patients who received lithium augmentation with antidepressants suffered a relapse during the double-blind phase of the study. Five of the seven relapsing patients in the placebo group developed a depressive episode, and the other two experienced a manic episode.
Lithium augmentation in the continuation phase of treatment of unipolar major depressive disorder effectively protects patients against a relapse. Patients who respond to lithium augmentation should be maintained on lithium augmentation for a minimum of 6 months or even longer.
Withdrawal! Bauer gradually increases the dose of lithium with antidepressants, then… he removes it! With antidepressants! Result… a dead person!
The design of this study had some limitations that could confound interpretation of the data. First, the 1-week duration of lithium withdrawal was relatively short. There is a controversy in the literature about the existence of a rebound phenomenon caused by rapid withdrawal of lithium.
(p. 1433, pdf p. 5)
Ho yes! A “controversy”! Antidepressants increase the risk of mania and lifted inhibitions while lithium decreases it. By acute augmentation and fast withdrawal of lithium, Bauer maximizes the risk of suicide caused by antidepressants and lithium withdrawal.
Is this how science is done? Of course! Read more:
Of the seven relapses in the placebo group, five were depressive relapses and two were manic relapses. The two patients with a manic relapse had never before experienced a manic episode, and both patients subsequently received a diagnosis of bipolar disorder.
(p. 1431, pdf p. 3)
Is he completely unconscious? It’s crazy.
167 suicidal people followed for one year.
This is not a true placebo-controlled study: 60.5% of the cohort took antidepressants at the same time as lithium (66.7%) or placebo (52.2%) (Table 1, p. 474, pdf p. 6).
In our sample, about 60% of all recruited patients and more than 80% of suicide attempters and completers received antidepressant co-medication.
(p. 476, pdf p. 8)
In addition, there is no significant difference in the number of suicidal acts between the two groups, the number of suicidal acts is very high in both groups (lithium: 8.4% and control: 12.0%) (Table 4, p. 475, pdf p. 7). 68.9% of participants left the study before the end (fig. 1 p. 473 pdf p. 5). The annual suicide rate in control group was 6.5% and in the experimental group 0% (Table 4).
Because the dropout rate is huge, and the antidepressants are a confounding factor, no conclusions can be drawn from this study, only hypotheses.
One of these hypotheses is that lithium could delay the suicidogenic effects of antidepressants, but not for a very long time. Indeed, the number of suicide attempts under lithium + antidepressants or other drugs increases sharply after 300 days, while there are no more new suicide attempts in the antidepressants or other group after 150 days. If the study had lasted longer, it is possible that the lithium + antidepressants or other group would have more suicide attempts than the antidepressants or other group (fig. 2 p. 475 pdf p. 7).
205 people at the beginning, 122 at the end, two years of follow-up.
This is not a placebo study but a brutal withdrawal study:
Following remission of acute manic episode and prior to discharge patients were stabilized on maintenance doses of lithium carbonate which yielded serum lithium level of 0.5 to 1.4 mEq/liter. Medication was administered in unmarked 250mg capsules. At discharge, the patients were randomly assigned to lithium carbonate or placebo.
(p. 337, pdf p. 1)
Moreover, it is not a double-blind trial:
Major treatment decisions such as hospitalization were made by the treatment physician who knew the identity of the patient’s medication.
(p. 340, pdf p. 4)
It was found that only 16% had their dosage raised because of poor clinical response. Another 31% required dosage adjustment for low serum lithium levels while 12% had their dosage increased for unspecified reasons. The remaining 41% had no upward adjustment of dosage.
(p. 340, pdf p. 4)
Twenty-four percent of the patients on lithium carbonate had reactions severe enough to require dosage reduction or temporary drug withdrawal.
(p. 340, pdf p. 4)
In this study, investigators were instructed to increase placebo dosage as well as lithium dosage at the first sign of symptomatology. In practice, placebo dosage was rarely manipulated.
(p. 341, pdf p. 5)
Early Terminations. – Early terminations are summarized in Table 3. There were significantly more early terminations on placebo than lithium carbonate ( χ² = 13.4, P<0.001). This was due mainly to the high proportion of terminations for poor clinical response in the placebo group (40% compared to 11% in the lithium group). A patient was dropped for poor clinical response when the treatment physician considered that further participation in the project would seriously jeopardize the patient’s mental or physical health.
(p. 339, pdf p. 3)
In addition, Table 1 (p. 338, pdf p. 2) shows that the treatment physicians ordered the hospitalization of 67% of the weaning group and of 31% of the lithium group. According to Prien, hospitalization = “severe relapse” (p. 337, pdf p. 1 “Assessment of clinical change”).
There is no detail on the person who committed suicide in the weaning group (p. 339, pdf p. 3).
In summary, subjects in the control group were brutally weaned, treatment physicians knew the identity of the patient’s medication, precisely adjusted the dose for a large majority of subjects in the lithium group, but almost never in the weaning group, even when the subjects felt terribly bad; they excluded subjects in the weaning group four times more often than those in the lithium group, and hospitalized them twice as often.
Psychiatric “science” in action. It is not for nothing that there are double-blind trials. Prien may defend himself that:
[…] several investigators were highly skeptical about the prophylactic efficacy of lithium carbonate at the start of the study
(p. 341, pdf p. 5)
The results of his study were literally forged by the experimenters. They did not “discover” anything.
It’s the same kind of junk science.
122 unipolar and bipolar subjects at the beginning, 67 at the end (table 3, p. 422, pdf p. 3), two years of follow-up. Brutal weaning. Electroshocks. No blindness of treatment physicians (pdf p. 2). The physicians expelled 20% of the subjects in the lithium group and 54% in the weaning group (table 3).
In less than four months, 67% of the weaning group and 27% of the lithium group relapsed (table 2, pdf p. 3). ~80% of patients who relapse need to be readmitted (pdf p. 3), i.e., ~53% in the weaning group and ~21% in the lithium group.
Five people died during the study (4% of all groups), including one suicide in the weaning group (3% of this group) (pdf p. 3). No details on this suicide.
Above are the first placebo-controlled studies from the meta-analyses cited directly by Zalsman. You can read the others’ sources in the meta-analyses of Tondo (2001)11, Baldessarini (2006)12 and in the “narrative review” of Lewitzka (2015)13. In addition, Kessing (2005)14 is not a RCT, but an observational study.
Zalsman cites also Gibbons (2009)15, a naturalistic study, which compares anti-epileptic drug treatments to the total absence of pharmacological treatment, and he does not notice that this study also speaks about lithium.
In this study:
- All patients were diagnosed bipolar (abstract)
- 2285 patient-years took only lithium and 40 made a suicide attempt, i.e., 18 suicide attempts per 1000 patient-years (table 1)
- 11,207 patient-years took absolutely no medication and 170 made a suicide attempt, i.e., 15 suicide attempts per 1,000 patient-years (table 1).
Although the difference is significant, it does not prove that lithium increases the risk of suicide attempts (there are many confounding factors and no homogeneity). However, this study does not raise any concerns about not prescribing lithium to bipolar people.
On the other hand, the suicide attempt rate among this class of patients is much lower in real life than in the experiments cited by Zalsman. The abrupt weaning, the combination of weaning with antidepressants, the emergency hospitalizations in the so-called “placebo group” are certainly what has contributed the most to the insane suicide rate in Prien’s and Bauer’s studies.
You can not trust psychiatrists. Neither in the clinic nor in science. Many published articles are full of lies, many systematic reviews are a complete failure. I think that all psychiatric articles should be considered false until proven otherwise; that is, after careful verification of the original data and confrontation with other contradictory sources. We need, in particular, researchers who focus on the verification and criticism of poorly conducted studies.
Some studies are specifically designed to defeat the “placebo” group, which is actually a weaning/antidepressants/electroshocks group. These studies cause more deaths than the conventional clinic. Is this how psychiatrists prove their treatments are effective? Some researchers should be subject to a criminal investigation, not only a scientific investigation.
- Zalsman G, Hawton K, Wasserman D, (2016). Suicide prevention strategies revisited: 10-year systematic review. The Lancet Psychiatry DOI: https://doi.org/10.1016/S2215-0366(16)30030-X https://www.sciencedirect.com/science/article/pii/S221503661630030X ↩
- Asenjo Lobos C, Komossa K, Rummel-Kluge C, et al. (2010) Clozapine versus other atypical antipsychotics for schizophrenia. Cochrane Database Syst Rev 2010; 11: CD006633. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4169186/ ↩
- Meltzer HY, Alphs L, Green AI, et al. (2003a). Clozapine Treatment for Suicidality in Schizophrenia: International Suicide Prevention Trial (InterSePT). Arch Gen Psychiatry. 2003;60(1):82–91. doi:10.1001/archpsyc.60.1.82 https://jamanetwork.com/journals/jamapsychiatry/fullarticle/207092 ↩
- Alphs L, Anand R, Islam MZ, et al. (2004). The International Suicide Prevention Trial (InterSePT): Rationale and Design of a Trial Comparing the Relative Ability of Clozapine and Olanzapine To Reduce Suicidal Behavior in Schizophrenia and Schizoaffective Patients. Schizophrenia Bulletin, 30(3):577-586, 2004. https://www.ncbi.nlm.nih.gov/pubmed/15631247 ↩
- Glick ID, Zaninelli R, Hsu C, et al. (2004). Patterns of concomitant psychotropic medication use during a 2-year study comparing clozapine and olanzapine for the prevention of suicidal behavior. J Clin Psychiatry. 2004 May;65(5):679-85. https://www.ncbi.nlm.nih.gov/pubmed/15163255 ↩
- Cipriani A, Hawton K, Stockton S, Geddes JR. Lithium in the prevention of suicide in mood disorders: updated systematic review and meta-analysis. BMJ 2013; 346: f3646.https://www.bmj.com/content/346/bmj.f3646 ↩
- Bauer M, Bschor T, Kunz D, Berghöfer A, Ströhle A, Müller-Oerlinghausen B. Double-blind, placebo-controlled trial of the use of lithium to augment antidepressant medication in continuation treatment of unipolar major depression. Am J Psychiatry. 2000;157:1429–35. doi: 10.1176/appi.ajp.157.9.1429. ↩
- Lauterbach E, Felber W, Müller-Oerlinghausen B, et al. Adjunctive lithium treatment in the prevention of suicidal behaviour in depressive disorders: a randomised, placebo-controlled, 1-year trial. Acta Psychiatr Scand. 2008;118:469–79. doi: 10.1111/j.1600-0447.2008.01266.x. https://www.ncbi.nlm.nih.gov/pubmed/18808400 ↩
- Prien RF, Caffey EM, Jr, Klett CJ (1973b). Prophylactic efficacy of lithium carbonate in manic-depressive illness. Report of the Veterans Administration and National Institute of Mental Health collaborative study group. Arch Gen Psychiatry. 1973;28:337–41(b). doi: 10.1001/archpsyc.1973.01750330035006. ↩
- Prien RF, Klett CJ, Caffey EM., Jr (1973a). Lithium carbonate and imipramine in prevention of affective episodes. A comparison in recurrent affective illness. Arch Gen Psychiatry. 1973;29:420–5(a). doi: 10.1001/archpsyc.1973.04200030104017. ↩
- Tondo, L., Hennen, J., & Baldessarini, R. J. (2001). Lower suicide risk with long‐term lithium treatment in major affective illness: a meta‐analysis. Acta Psychiatrica Scandinavica, 104(3), 163-172. https://www.ncbi.nlm.nih.gov/pubmed/11531653 ↩
- Baldessarini, R. J., Tondo, L., Davis, P., Pompili, M., Goodwin, F. K., & Hennen, J. (2006). Decreased risk of suicides and attempts during long‐term lithium treatment: a meta‐analytic review. Bipolar disorders, 8(5p2), 625-639. https://onlinelibrary.wiley.com/doi/full/10.1111/j.1399-5618.2006.00344.x ↩
- Lewitzka, U., Severus, E., Bauer, R., Ritter, P., Müller-Oerlinghausen, B., & Bauer, M. (2015). The suicide prevention effect of lithium: more than 20 years of evidence—a narrative review. International Journal of Bipolar Disorders, 3, 15. http://doi.org/10.1186/s40345-015-0032-2 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4504869/ ↩
- Kessing, L. V., Søndergård, L., Kvist, K., & Andersen, P. K. (2005). Suicide risk in patients treated with lithium. Archives of general psychiatry, 62(8), 860-866. https://www.ncbi.nlm.nih.gov/pubmed/16061763 ↩
- Gibbons RD, Hur K, Brown CH, Mann JJ. Relationship Between Antiepileptic Drugs and Suicide Attempts in Patients With Bipolar Disorder. Archives of general psychiatry. 2009;66(12):1354-1360. doi:10.1001/archgenpsychiatry.2009.159. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3926811/ ↩
Mad in America hosts blogs by a diverse group of writers. These posts are designed to serve as a public forum for a discussion—broadly speaking—of psychiatry and its treatments. The opinions expressed are the writers’ own.