GME Logo

User login

User menu

Issue 43, Nov 2015
Share This
Does psychotherapy for major depression work as well as claimed?

Driessen et al. Does Publication Bias Inflate the Apparent Efficacy of Psychological Treatment for Major Depressive Disorder? A Systematic Review and Meta-Analysis of US National Institutes of Health-Funded Trials. PLoS One. 2015 Sep 30;10(9):e0137864. PubMed PMID: 26422604

  • It is obviously important to get an accurate estimate of the efficacy of different treatments so that the best treatments can be recommended.
  • There was a big scandal in the recent past when data was published showing that the efficacy of antidepressant medications had been exaggerated, in part due to negative studies not getting published (e.g., Turner et al., 2008).
  • Now, a study found that the same thing has happened for psychotherapy. This has led to a lot of attention for this study in the press. Your patients may ask you about it too!
  • This is the first study to directly compare the effects of psychotherapy for major depression in published studies versus after including unpublished studies. 
  • The authors identified grants from the United States National Institutes of Health to fund randomized clinical trials comparing a psychological treatment to control conditions or to other treatments in patients diagnosed with major depressive disorder.
  • The period covered was 1972 to 2008.
  • They then found out whether or not those grants led to publications.
  • For studies that were not published, they requested the data from the investigators of those unpublished studies.
  • Then, they did their own meta-analysis of the data including both published and unpublished studies.
  • Of 55 research grants awarded that did start the clinical trial, 23.6% did not result in publications.
  • When the unpublished studies were added to the analysis, the efficacy of psychological treatments for major depression was reduced by 25% from what it was thought to be before the unpublished studies were added to the analysis.
  • The effect size for efficacy of psychotherapy for major depression based on published studies only was 0.52. When the unpublished findings were added to the published findings, the effect size was reduced to 0.39. 0.39 is 75% of 0.52. That’s why the authors say that the efficacy decreased by 25% when unpublished studies were included. 
  • By ignoring the unpublished studies, previous meta-analyses have in the past overestimated the efficacy of psychotherapy for major depression by 25%.
Clinical Commentary
  • The efficacy of psychological interventions for major depression has been overestimated in the published literature. This is the same problem and for the same reasons that has been shown for antidepressant medications. For example, Turner et al. (2008) had found that omission of unpublished studies had led to overestimation of the efficacy of antidepressant medications by 24%, almost exactly the same as was found for psychotherapy in this study.
  • This study was not done by proponents of antidepressant medications “getting back” for the previous report showing publication bias with regard to antidepressant medications. No conspiracy theories, please! The authors are themselves psychologists who conduct research on psychotherapy.
  • It is very important to note that the findings of this study do not mean that psychotherapy does not work for major depression. They only mean that it did not work as well as we thought it had. The degree of the difference was not such that it would greatly affect our decision to use or not use psychotherapy for major depression.
  • It should also be noted that this meta-analysis only refers to treatment of major depression. It does not have any bearing on treatment of other forms of depressive disorders (e.g., dysthymia) with psychotherapy.
  • Clinicians, guidelines developers, and decision makers should be aware that the published literature overestimates the effects of the both antidepressants and psychotherapy for major depression.
  • The authors noted that in this study they could not assess whether there was also another kind of bias. Could it be that not only were the studies that did not find psychotherapy to be as efficacious remain unpublished, but that even in the published studies, those outcome measurements that did not find psychotherapy to be as efficacious may have been left out of the papers?
  • For that, the authors suggested that funding agencies and journals should archive both the original study protocols and raw data from treatment trials. By comparing these to the published data, any bias in reporting outcomes could be detected and corrected. 
Expand Content
Share this Section:
Long-acting aripiprazole lauroxil for schizophrenia: Clinical trial

Meltzer et al. A Randomized, Double-Blind, Placebo-Controlled Trial of Aripiprazole Lauroxil in Acute Exacerbation of Schizophrenia. J Clin Psychiatry. 2015 Aug;76(8):1085-90. PubMed PMID: 26114240

  • Aripiprazole lauroxil (Aristada), an injectable, long-acting formulation of aripiprazole was approved by the United States Food and Drug Administration on October 5, 2015 for the treatment of schizophrenia.
  • Long-acting (or depot) antipsychotics are important due to the widespread problem of non-adherence.
  • We already have a long-acting version of aripiprazole (Abilify Maintena) but aripiprazole lauroxil was not approved by the FDA on the basis of research with Abilify Maintena. A specific study was done with this particular formulation. We are looking here at that study.  
  • Aim: to evaluate the efficacy, safety, and tolerability of aripiprazole lauroxil for the treatment of schizophrenia.
  • This was an international multicenter, randomized, double- blind, placebo-controlled trial.
  • The study included 623 patients, 18 to 70 years old, with a diagnosis of schizophrenia with an acute exacerbation.
  • The patients were randomized to receive one of the following:

- Aripiprazole lauroxil 441 mg

- Aripiprazole lauroxil 882 mg

- Placebo


  • The aripiprazole lauroxil was given as an intramuscular injection in the gluteal muscle. (Note: aripiprazole lauroxil is approved for administration in the deltoid for the 441 mg dose only. All three approved doses – 441 mg, 662 mg, and 882 mg – can be administered in the gluteal muscle.)
  • It was given once a month for 12 weeks. 
  • On the primary outcome measure, the change in the total score on the Positive and Negative Syndrome Scale (PANSS), there was a statistically significantly greater improvement in the groups that received aripiprazole lauroxil than in those who received placebo injections.
  • The benefit of the drug over placebo was apparent by treatment day 8.
  • On the secondary outcome measure, the Clinical Global Impression Scale (CGI), there was also statistically significantly greater improvement in patients who receiving aripiprazole lauroxil.
  • The most common treatment-emergent adverse events were insomnia, akathisia, headache, and anxiety.
  • The incidence of injection site reactions was low. Mainly it involved injection site pain after the first injection.
  • This study found that aripiprazole lauroxil is efficacious for the treatment of acute exacerbation of schizophrenia. 
Clinical Commentary
  • Please note: approval of Aristada will have no effect at all on the patent life of Abilify Maintena. So, introduction of this medication is not an attempt to extend the patent life of an existing medication as some people have commented on social media.
  • Of course, from this study, we cannot say whether either Abilify Maintena or Aristada is better than the other medication.


Aristada may have some advantage in terms of convenience of use though. See 

Expand Content
Share this Section:
Can antidepressants lead to intracranial hemorrhage?

Shin et al. Risk of intracranial haemorrhage in antidepressant users with concurrent use of non-steroidal anti-inflammatory drugs: nationwide propensity score matched study. BMJ. 2015 Jul 14;351:h3517. PubMed PMID: 26173947.

  • Both antidepressants and non-steroidal anti-inflammatory drugs (NSAIDs) are commonly used.
  • If it is true that combining antidepressants with NSAIDs increases risk not only of gastrointestinal bleeding but of intracranial bleeding as well, this is a serious matter!
  • Aim: To compare the risk of intracranial hemorrhage among patients treated with an antidepressant and an NSAID with the risk among those treated with an antidepressant without an NSAID.
  • This was a nationwide study: all patients in the Korean nationwide health insurance database (during a specified period, of course) were included. This reduces potential bias due to patients with a particular insurance or from a particular source being included in a study. Here all patients in the country were included.
  • The study design used in this study was a retrospective cohort design. What that means is that even though the study was done retrospectively, patients were selected on the basis of having or not having the risk factor (being on an antidepressant plus an NSAID). If you don’t understand that, please make a mental note to learn about types of study design when you have a chance.
  • Patients were included if they had started on an antidepressant for the first time. “First time” was based on not having received a prescription for an antidepressant during the preceding year. That is, they included only patients who started on an antidepressant but had not been on an antidepressant in the previous one year.
  • The study then focused only on the 30 days after starting the antidepressant. Patients who received an NSAID during these 30 days were included in the study. No minimum duration of NSAID treatment was specified.
  • Patients who had been diagnosed as having a cerebrovascular disease as their primary or secondary diagnosis within a year before their index date were excluded.
  • Now, the reason all observational studies, whether population-based or not, are very problematic is that the outcome could have occurred not simply due to the risk factor but due to various other factors called Confounders. In this study, the risk of having an intracranial hemorrhage could have also been related to having one of a variety of diseases (diabetes, hypertension, atrial fibrillation, etc.) or to being on one of various other medications (aspirin, warfarin, platelet aggregation inhibitors, direct thrombin inhibitors, etc.). These authors attempted to deal with this problem by matching the two groups on as many of these other potential risk factors as possible by using a statistical method called  “propensity score.”
  • The main outcome measure was time to first hospital admission with intracranial hemorrhage but ONLY when such admissions occurred within 30 days of starting an antidepressant.
  • This outcome measure was compared between those who were or were not also taking an NSAID along with the antidepressant.
  • About 4.1 million patients taking an antidepressant were used in the analyses – about 2.7 million taking an antidepressant without an NSAID and about 2.4 million taking an antidepressant with an NSAID. This is the power of using a nationwide database!
  • The risk of intracranial hemorrhage was statistically significantly higher for those who were taking both an antidepressant and an NSAID compared to those taking an antidepressant alone.
  • By how much was the risk higher? By 60%. The “hazard ratio” was 1.6. (Hazard ratio is a ratio of the risk in the two groups but over time rather than only at one point of time.)
  • Importantly, the hazard ratio was much more in males (2.6) than in females (1.2). That is, the combination of an antidepressant and an NSAID increased the risk of intracranial hemorrhage mainly in males.
  • The actual risk of intracranial hemorrhage, even in those taking both an antidepressant plus an NSAID is, thankfully, low: 5.7 per 1000 person-years. What is a person-year? It is one person for one year, two persons for 6 months each, 12 persons for one month each, and so on.
  • Was any particular type of antidepressant more likely to be associated with risk of intracranial hemorrhage? No difference was identified in this study.
  • Use of an antidepressant and an NSAID was associated with an increased risk of intracranial hemorrhage within 30 days of being on the combination.
Clinical Commentary
  • That taking an antidepressant and an NSAID together increases risk of bleeding is not really in doubt.
  • However, some but not all previous studies found that taking an antidepressant and an NSAID together increases the risk of intracranial hemorrhage.
  • This study had many strengths compared to most previous studies:

    1. Nationwide study

    2. Started with patients who took an antidepressant plus NSAID or an antidepressant alone (cohort study design). Most previous studies started with patients who did or did not have an intracranial hemorrhage (case-control design).

    3. Focused only on the 30-day period after starting the antidepressant.
Expand Content
Share this Section:
Anger/ hostility/ irritability in major depressive disorder affects severity of illness and response to treatment

Fisher LB, Fava M, Doros GD, Alpert JE, Henry M, Huz I, Freeman MP. The Role of Anger/Hostility in Treatment-Resistant Depression: A Secondary Analysis From the ADAPT-A Study. J Nerv Ment Dis. 2015 Oct;203(10):762-8. PubMed PMID: 26348584.

  • There has been limited success in using subtypes or clinical features of major depressive disorder to predict response to treatment.
  • While anger/ hostility/ irritability are common in patients with major depressive disorder, their predictive value is not clearly known.
  • About 40% of outpatients enrolled in the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) study reported feeling irritable more than half of the time.
  • Also, about one third of outpatients with major depressive disorder present with anger, hostility, and irritability in the form of “anger attacks.” These are sudden spells of anger accompanied by rapid heartbeat, hot flashes, sweating, and tightness of the chest.
  • Nevertheless, these symptoms are not considered diagnostic features of major depressive disorder and are often not included in ratings scales used in major depressive disorder. 
  • This was a secondary analysis of data from a previously published study of using low-dose (2 mg/day) aripiprazole as an adjunct to antidepressants in patients with major depressive disorder who had had an inadequate response to an antidepressant alone (Fava et al., 2012).
  • Patients with high levels of anger/hostility (n = 135) and patients with low levels of anger/hostility (n = 86) were compared.
  • Note: these patients had been systematically evaluated to rule out a history of bipolar disorder. 
  • There were more Hispanic patients in the high anger/ hostility group than in the low anger/hostility group.
  • Compared to patients in the low anger/hostility group, patients in the high anger/hostility group had:

    1. More lifetime depressive episodes

    2. Greater severity of depression at baseline
  • Number of suicide attempts and age of onset of first depressive episode were not different between the two groups.
  • Placebo response rates were much lower in patients with high anger/ hostility (11%) than in patients with low anger/hostility (30%).
  • Overall, response rate to adjunctive aripiprazole was lower among patients with high anger/hostility (10%) than in those with low anger/hostility (25%). 
  • Depressed patients with high anger/hostility had more severe major depressive disorder and lower rates of response to treatment.
Clinical Commentary
  • The findings of this study are very interesting because we have relatively ignored the issue of anger/ hostility/ irritability in patients with major depressive disorder. 


  • It is fairly common to use a low dose of antipsychotic for patients who are irritable, so it was surprising to me that these patients did not benefit as much from addition of low dose of aripiprazole.
Expand Content
Share this Section:
What effect do risperidone and SSRIs have on bone development in boys?

Calarge et al. Longitudinal examination of the skeletal effects of selective serotonin reuptake inhibitors and risperidone in boys. J Clin Psychiatry. 2015 May;76(5):607-13. PubMed PMID: 26035190.

  • Psychotropic medications are frequently prescribed to children and adolescents.
  • Bone mass (density) increases in growing children but it is then less amenable to change later on.
  • Age-related bone loss later in life is directly related to peak bone mineral density achieved by young adulthood.
  • Therefore, anything that impairs formation of bone mass during childhood and adolescence is likely to have long-term effects and is a serious matter. 
  • In a previous cross-sectional study (Calarge et al., 2010), these authors had reported lower bone mass in boys being treated with SSRIs or risperidone.
  • In that study, hyperprolactinemia was present in 49% of boys treated with risperidone. That is a very high percentage!
  • But, cross-sectional studies don’t provide good data on cause and effect. That is, just because A and B occur together doesn’t mean that A causes B.
  • Therefore, the study we are discussing here evaluated the effects of risperidone and of SSRIs on bone formation over time, i.e., during follow up rather than cross-sectionally.
  • Medically healthy 7- to 17-year-old boys who had been treated with risperidone for at least 6 months were enrolled.
  • After baseline evaluation, they returned 1.5 years later for a second evaluation.
  • Adherence to risperidone was verified by measurement of serum risperidone and 9-hydroxyrisperidone levels.
  • 94 boys were evaluated.
  • The most common diagnoses were attention-deficit/ hyperactivity disorder (88%) and disruptive behavior disorders (87%).  Risperidone was prescribed primarily (78%) for irritability and aggression.
  • The subjects had received risperidone for an average of 2.5 years or an SSRI for an average of 1.6 years.
  • 26% of the boys stopped taking the risperidone.
  • Those who continued to take the risperidone were compared to those who stopped the risperidone.
  • At the follow up visit, none of the subjects who had stopped taking risperidone had hyperprolactinemia, while 48% of those who were still taking the risperidone did.  Please note this: hyperprolactinemia does not   usually resolve with continued use.
  • Analyses controlled for other factors like age, gender, height, and race that may affect bone density.
  • Those who continued to take the risperidone:

    a) Had lower bone mineral density (measured in the lumbar spine)

    b) Had lesser increase in bone mineral density (measured in the radius bone)
  • Those who were on an SSRI:

    a) Had lower bone density in both the lumbar spine and the radius

    b) But, did not show further worsening over one year.
  • Longer-term administration of risperidone to boys is associated with reduction in the normal increase in bone mineral density.
  • Longer-term SSRI treatment in boys is associated with reduced bone mass for age but this did not progress over one year of follow up.
Clinical Commentary
  • It is important for us to know that hyperprolactinemia is not an occasional adverse effect of treatment with risperidone; it occurs in about half of the patients taking it.
  • This study is important but not definitive. Firstly, the sample size was small. Secondly, even though the authors tried to control for other factors that affect bone mineral density, it was not possible for them to control for differences in the severity of the illness itself. Therefore, further studies should be done.
  • However, the findings of this study are consistent with other literature and so are quite believable.
  • In interpreting any study, in addition to statistical association, it is also important to think about biological plausibility.
  •  Given that risperidone causes hyperprolactinemia in a very high proportion of patients, and that hyperprolactinemia has a negative effect on bone density, it should come as no surprise that treatment with risperidone was associated with reduced gain in bone mineral density.
Expand Content
Share this Section:

Rajnish Mago, MD
Associate Professor of Psychiatry, Thomas Jefferson University

GME Research Review is a monthly newsletter edited by Rajnish Mago, MD, who is Associate Professor of Psychiatry at Thomas Jefferson University and is author of "The Latest Antidepressants and Side Effects of Psychiatric Medications: Prevention, Assessment, and Management." Dr. Mago selects, summarizes, and provides a clinical commentary on the latest published research in psychiatry.

We are always carefully evaluating which research papers to discuss in GME Research Review. Have you come across a research paper published in the last 6 months that you think is clinically relevant? If you would like to ask Dr. Mago to consider analyzing it, please email him the citation at: [email protected]

To contact GME, email us at [email protected]

GME does not provide medical advice. The website and articles are intended for informational purposes only. They are not a substitute for professional medical advice, diagnosis or treatment. Never ignore professional medical advice in seeking treatment because of something you have read on the GME Website. If you think you may have a medical emergency, immediately call your doctor or dial 911.

Log In

Join GME For Free

Become a GME subscriber and gain full access to our extensive library of 700+ psychiatric medical education videos, free CME webcasts, latest research updates, and more. To sample our content, watch the featured videos below.
Join Today!