Psychotherapy Brown Bag

Treating female victims of domestic abuse: A test of dialectical behavior therapy (DBT)

by Michael D. Anestis, M.S.

Domestic abuse, which refers to physical, sexual, psychological, and verbal abuse between intimate partners, is a pervasive and distressing phenomenon.  Studies estimate that between 22% and 29% of women will be the victim of an abuse relationship at least once in their lifetime (Tjaden & Thoennes, 2000).  Sadly, a number of dangerous outcomes are associated with being the victim of domestic abuse, including post-traumatic stress disorder (PTSD), depression, general distress, difficulties in social adjustment, and elevated risk of suicide (Campbell, 2002). 

Given this startling information, the need to develop treatments capable of assisting victims of domestic abuse is obvious, and today I would like to discuss an article published in Professional Psychology: Research and Practice by Katherine Iverson, Chad Shenk, and Alan Fruzzetti (2009) in which they examined the potential utility of a brief, 12-week version of dialectical behavior therapy (DBT) for this population.  As you might recall from prior PBB articles, DBT is comprised of four components: mindfulness, interpersonal effectiveness, emotion regulation, and distress tolerance.  The treatment was originally designed for the treatment of individuals with borderline personality disorder (BPD) and each of these components addresses core symptoms of that disorder.  Recent research, however, has indicated that emotion regulation difficulties are at the core of numerous mental illnesses and, as such, DBT has been adapted for a number of other presentations and, in fact, is considered an empirically supported treatment for both bulimia nervosa and binge eating disorder (e.g., Telch, Agras, & Linehan, 2001). 

Marsha Linehan, the creator of DBT, has proposed a biosocial model (also referred to as a transactional model)  of BPD in which an individual's innate vulnerability to dysregulated emotions interacts with invalidating environments, resulting in the onset of BPD pathology.  Iverson, Shenk, and Fruzzetti (2009) explained that this same model might be successfully applied to female victims of domestic abuse, as an abusive intimate partner represents a severely invalidating environment.  As such, they believed that teaching such women skills to better manage their emotions and impulses and to better evaluate the value of particular interactions and assert their needs could potentially represent a successful method for improving the lives and psychological well-being of female abuse victims.

DBT in its truest form involves both group skills training (2-hours per group) and individual treatment and is intended to last two full years of weekly sessions, representing at least two complete cycles through the entire treatment protocol.  The authors sought to drastically reduce this time requirement by revising the protocol such that the entire treatment consisted of weekly group skills training and lasted a total of twelve weeks.  46 women were initially involved in the study, although only 31 completed the entire protocol, representing a 33% attrition rate.  I will address the attrition rate in more detail later, but obviously this represents one fairly substantial problem.

Participants ranged in age from 22 to 56, with an average age of 40.7.  54% of the participants had been in an abusive relationship for 1 to 5 years and 26% were still living in the same home as their abuser.  Each participant was given a measure of depression symptoms, hopelessness, social adjustment, and overall psychological well-being both an pre-treatment and post-treatment.  Each group consisted of 6-8 women.  The first hour of each group meeting involved the discussion of diary cards and the description of how skills were successfully or unsuccessfully implemented into group members' lives (behavioral chain analysis).  The second hour of each group meeting involved skills training, with the particular skills changing as the group advanced from one module to the next.  Each week, the DBT clinicians met amongst themselves for a consultation meeting.

So, what did they find?  For participants who remained in treatment, there were substantial (medium to large effect size) reductions in depression symptoms, hopelessness, suicidality, and general distress and a substantial increase in social adjustment.  At the beginning of treatment, the average depression score for the overall group was in the moderate range; however, at post-treatment, the majority of participants reported either a complete absence of mood symptoms or only mild symptoms.  At the beginning of treatment, 25% of the sample met criteria for severe suicide risk whereas at post-treatment, only 7% met that criteria.  Additionally, 93% of the participants reported being very satisfied with the treatment program (highest possible rating) and 7% reported being satisfied (second highest rating).

What about that 33% that dropped out of treatment though?  On average, those participants attended 1-3 sessions before stopping treatment.  Obviously, this is a substantial portion of the sample, which raises questions as to whether severely ill participants were simply dropping out of treatment and thus inflating the impact of treatment for those who remained.  As it turns out, however, this was not the case.  Individuals who dropped out of treatment were no more severe on any of the clinical measures than were the individuals who remained in treatment. The only significant difference between those who remained in treatment and those who dropped out was education level, with those who dropped out reporting fewer years of education.  The most common explanation for dropping out of the program was an inability to consistently attend sessions at the scheduled time and/or day.  In other words, severity of illness and satisfaction with treatment had nothing to do with the dropout rate.

This study had a lot of obvious positives.  First of all, it demonstrated positive effects on a range of outcome measures for a population in need of help.  Secondly, it found these effects for a brief version of DBT, meaning that treatment can be provided efficiently and at lower costs and which is received positively by clients.  Third, the study utilized a number of outcome measures representing a wide range of variables rather than focusing only on a limited set of outcomes.

Along with these positives, however, were a number of important limitations. Iverson and colleagues (2009) noted that this was a preliminary study aiming to establish initial evidence of utility and, as such, they were not aiming to conduct a large scale randomly controlled trial.  Unfortunately, this means that, in addition to lacking a comparison to another active treatment, the study lacked a control group.  As such, we can not tell whether or not the improvements experienced by participants were due to the treatment itself rather than common factors (e.g., therapeutic alliance) or the simple passage of time.  Additionally, although the authors did assess a number of different outcomes, they only used self-report questionnaires, which while useful also involve many important limitations.  The authors also provided no diagnostic information.  In other words, we do not know how many participants met criteria for mental illnesses either pre- or post-treatment, which thus prevents us from knowing whether the treatment actually impacted diagnostic status.  The authors only used two time points, so we do not know how quickly treatment gains occurred and whether or not all 12 sessions were even necessary.  Finally, the authors did not conduct any follow-up assessments, so it is entirely unclear to what degree treatment benefits were maintained following treatment and whether or not treatment resulted in participants avoiding involvement in future abusive relationships.

On the whole, this study provided solid preliminary evidence for the utility of brief DBT in treating female victims of domestic abuse.  Treating an outcome rather than a diagnosis is not the traditional approach of the empirically supported treatments paradigm, but that, in and of itself, is not a criticism.  It simply means that the authors took an alternative approach to addressing a problem.  Future studies that address the limitations mentioned above will be important, but readers who themselves are the victims of domestic abuse or who know others who are should feel encouraged by these findings and in the knowledge that there are researchers out there attempting to develop interventions aimed to precisely address this situation.

If you would like to learn more about dialectical behavior therapy, we recommend the following resources, all of which are available through our online store:

• Skills Training Manual for Treating Borderline Personality Disorder by Marsha Linehan

• Cognitive-Behavioral Treatment of Borderline Personality Disorder by Marsha Linehan

• Dialectical Behavior Therapy in Clinical Practice: Applications across Disorders and Settings by Linda Dimeff and Kelly Koerner

• The High Conflict Couple: A Dialectical Behavior Therapy Guide to Finding Peace, Intimacy, & Validation by Alan Fruzzetti

• Dialectical Behavior Therapy for Binge Eating and Bulimia by Debra Safer, Christy Telch, and Eunice Chen

Mike Anestis is a doctoral candidate in the clinical psychology department at Florida State University

Can video-based suicide prevention programs be harmful? A follow-up on yesterday's article

by Michael D. Anestis, M.S.

By far, one of the most enjoyable aspects of PBB has been that I have come into contact with a number of interesting and accomplished mental health professionals - whether they are clinicians, researchers, or both - that I might not otherwise have met.  This has led to a number of fascinating and entertaining conversations and has exposed me to a wealth of research I had not previously seen.  Yesterday, I had a timely email exchange with Captain Luther Dhillon-Davis of the United States Air Force during which he alerted me to an in press article he co-authored with Craig Bryan (author of the April 2009 Psychotherapy Brown Bag featured article) and Kieran Dhillon-Davis that touched upon a topic very similar to what we covered on the site yesterday. Capt. Dhillon-Davis made no request that I discuss the paper on PBB, but I thought the timing was too good to pass up on and the study was too interesting not to cover.

If you read yesterday's article, you will remember that Gould and colleagues (2005) found that, despite fears that assessing for suicide risk would increase suicidality in adolescents, the opposite occurred, with suicidal individuals experiencing significant decreases in distress and suicidal ideation after being asked questions about symptoms of suicidality.  That, quite clearly, is important news, as it means that discussing and systematically assessing suicide risk causes no harm and can help provide information essential for effective interventions.  In the article sent to me by Capt. Dhillon-Davis, the authors examined a similar phenomenon by testing the impact of the video-based Air Force Suicide Prevention Program (AFSPP).  My intention in covering this article today is to expand upon yesterday's discussion and to apply it to another population, as I am particularly interested in studying suicidality in the military and helping to develop effective approaches to diminishing risk and believe that discussing the topic in forums like this can be a useful enterprise.

The utility of studies like the one conducted by Bryan, Dhillon-Davis, and Dhillon-Davis (in press) rests in their ability to empirically test our fears that discussing or being exposed to information about suicide can increase suicide risk.  To be fair, in the past, there have been findings that indicate that such a connection could exist.  In fact, Madelyn Gould, one of the authors of the study discussed yesterday, published a study in 2003 that found exposure to information on suicide warning signs could have iatrogenic effects.  At the same time, such findings are far from universal and, in fact, represent a small minority of published research.  For instance, Rudd and colleagues (2006) found that individuals exposed to a list of suicide warning signs experienced no more distress than did individuals exposed to a list of warning signs for a heart attack.  Along those lines, we have to be careful not to become too attached to ideas that sound compelling or instill fear but lack a wealth of empirical support.  Sometimes, the most compelling narrative does not match up with reality and sometimes the most interesting sounding theory proposed by the most intelligent person turns out to be wrong.

Suicide prevention efforts are not always limited to lists and discussions of risk factors as the case was in the Rudd et al (2006) study mentioned above.  Some approaches use other forms of media in an effort to convey information.  The example focused on in the Bryan, Dhillon-Davis, and Dillon-Davis (in press) article was the AFSPP, which relies heavily upon video footage as a teaching tool.  The AFSPP was created more than 10 years ago in response to increasing rates of death by suicide among active duty personnel (Knox, Litts, Talcott, Feig, & Gaine, 2003) and it focuses on reducing malleable risk factors for suicide while bolstering protective factors.  The program seeks to increase awareness regarding the nature of suicide risk, to create prevention programs, and to change policies in an effort to reduce the stigma attached to help seeking behaviors related to mental health difficulties.  Prior research has demonstrated that the program has been highly efficacious in reducing suicide rates amongst active duty airmen (Knox et al., 2003).  The AFSPP consists of 11 initiative and a description of the program is available for no cost at the following URL: http://afspp.afms.mil

The video component of the AFSPP, which is shown during the mandatory public education and suicide awareness training for USAF community members, involves four emotionally charged vignettes focusing on identifying suicide warning signs, responding to suicidal crises, the impact of suicide on survivors, and obstacles that prevent individuals from seeking help for mental health difficulties.  The AFSPP utilizes video for a number of reasons, including the belief that it might enhance memory encoding (e.g., Bryan et al., in press).  The potential for enhanced learning is countered for some, however, by the fear that viewing such stimuli could be emotionally overwhelming and harmful, particularly for vulnerable individuals.  As such, Bryan, Dhillon-Davis, and Dhillon-Davis (in press) sought to empirically test the impact of the footage.

In this study, the participants were 286 active duty USAF personnel who were participating in a two-week introductory class required for all Airmen assigned to their first duty station after graduating from technical training school.  The sample was predominantly Caucasian and male.  11.2% of the sample was classified as currently suicidal based upon their scores on the Suicidal Behaviors Questionnaire - Revised.  Amongst those classified as suicidal, a significantly higher number were female.  Similarly, a significant number of participants who reported a history of past suicide attempts were female.  Participants were assessed on their levels of positive and negative affect both prior to and following exposure to the suicide related video footage.

The participants' mean level of positive affect decreased following exposure to the films; however this effect was quite small.  The participants' mean level of negative affect also decreased following exposure to the films and this effect was actually quite large.  In other words, looking at the entire sample, following the films individuals tended to feel slightly less positive emotions and substantially less negative emotions than they did before viewing the films.  Importantly, prior to the film, the sample reported a level of negative affect consistent with normative samples and an elevated level of positive affect relative to normative samples.  Consequently, it is certainly possible that the minor decrease in positive emotions actually represents regression to the mean.  Either way, this first finding did not indicate that there was any harmful impact on the overall sample as a result of viewing the films.

Still, the question remained whether suicidal individuals or individuals with a past history of suicide attempts might be impacted differently than individuals without such symptoms or histories.  Fortunately, the authors compiled data capable of answering that very question.  First, individuals with a past history of suicide attempts experienced no difference from individuals without such a history in their emotional responses to the films.  Second, there was a nearly significant interaction (p = .051) of sex and suicidality, such that suicidal females experienced the greatest decrease in negative affect, followed by non-suicidal males and non-suicidal females.  Suicidal males did not experience any change in negative affect in response to viewing the films.

So what do all of those findings in the previous paragraphs mean?  First of all, whether an individual has a past history of suicide or not, viewing suicide-related films in the context of a suicide prevention program does not cause any harm.  Obviously, the content of the film is an issue to consider, but these films emphasized educational material, not shocking or gruesome imagery.  A small but significant decrease in positive emotions is to be expected in response to this type of program, but so is a substantial decrease in negative affect.  Secondly, although males classified as currently suicidal experienced no change in their levels of negative emotions, suicidal females experienced a decrease in negative emotions and nobody experienced an increase in negative emotions.  As such, there was absolutely no evidence supporting the idea that viewing these films caused any harm and, in fact, there was substantial evidence that the films actually had a positive effect above and beyond their goals of conveying information.

One important consideration raised by Brian, Dhillon-Davis, and Dhillon-Davis (in press) is the idea that there might be an age effect that impacts the effects of this type of suicide prevention program.  Rudd et al (2006) proposed that, for adolescents, there could potentially be some danger and, given that the sample in this study was comprised of adults, that idea certainly can not be refuted by the data.  As such, future research that investigates this phenomenon in adolescents would be of great value.

Ultimately, the take home point of studies like the ones discussed today and yesterday is that we need to openly discuss the phenomenon of suicide in order to help prevent it.  Learning about suicide and being exposed to information regarding risk factors, whether on a paper list or through video format, does not appear to cause any harm and may, in fact, result in emotional benefits.  Obviously suicide is an uncomfortable and emotionally charged topic, even if an individual has no experience with it and does not know anyone who does, but avoiding the topic out of fear of its potential impact is a disservice, particularly given the data we have indicating that at best it is helpful and at worst it is a harmless source of important information.

If you or anyone you know is experiencing thoughts of suicide, please immediately dial 1-800-273-TALK to reach the national suicide prevention lifeline.  If you are not comfortable calling that number, please dial 911 or get somebody to take you to the emergency room.

If you would like to learn more about suicide or suicide prevention efforts, we recommend the following resources, all of which are available through our online store:

• Why People Die by Suicide by Thomas Joiner

• Treating Suicidal Behavior: An Effective, Time-Limited Approach by David Rudd, Thomas Joiner, and Hasan Rajab

• The Interpersonal Theory of Suicide: Guidance for Working With Suicidal Clients by Thomas Joiner, Kim Van Orden, Tracy Witte, and David Rudd

Mike Anestis is a doctoral candidate in the clinical psychology department at Florida State University

Debunking a harmful myth: Assessing for suicide risk does not cause people to become suicidal

by Michael D. Anestis, M.S.

Having been trained in a lab that specializes in the systematic investigation of suicide, I have been frequently exposed to the perspective that evaluating suicide risk is imperative and should be done routinely, even with patients for whom there is no reason to suspect high levels of risk.  I was taught to think of psychologists assessing suicidality as akin to physicians assessing resting heart rate: this might not be the primary reason for presenting and there might be small meaningless fluctuations from time to time, but we want to keep an eye on the situation so as to ensure that we can intervene when necessary.

All of that being said, I realize this is not the universal approach to things.  Many people fear that, if we ask somebody about suicide, it might increase the likelihood that they will think about and potentially even attempt to kill themselves.  After all, might we be putting such thoughts into people's heads?  As it turns out, the answer is no, but rather than asking you to take my word for it, I would like to describe a study published in the Journal of the American Medical Association (JAMA) in 2005 by Madelyn Gould of Columbia University and several of her colleagues (thanks to Ted Bender for calling my attention to this article).  In this study, the authors sought to investigate this question empirically, collecting data capable of informing us as to whether or not assessing for suicide risk can be legitimately considered iatrogenic.

To do this, the authors recruited a total of 2,342 students from six high schools in the northeastern United States.  Students ranged in age from 13 to 19, with an average age of 14.8.  Participants were randomized into two groups - the experiment group and the control group.  In both groups, students were assessed on two occasions separated by two days.  In the experiment group, both the first and second assessments consisted of measures of distress as well as measures of suicidal ideation and behavior.  In the control group, only the second assessment included the questions regarding suicidal ideation and behavior.  In other words, the experiment group was assessed for suicide risk at both time points to see if the first assessment caused an increase at the second time point and the control group was only assessed for suicide risk at Time 2 in order to compare their scores at the second assessment with those of the experiment group.  This design allowed the authors to determine if asking about suicide increased suicidal ideation or behavior or distress levels (e.g., depression).  Additionally, it allowed for a comparison across groups to see if the group that was screened for suicide risk at Time 1 as well as Time 2 would be more distressed and/or suicidal than the control group.

The results, quite fortunately, indicated that the suicide risk screening procedure was not at all harmful.  First of all, the experiment group and control group did not differ on distress levels immediately after the first survey.  In other words, even though the experiment group was asked about suicide, they did not experience more distress than did the control group.  Additionally, rates of depressive feelings in the two days between the assessments did not differ between the groups, meaning that asking about suicide did not cause an increase in depressive symptoms in the days following the assessment.

Taking this result a step further, the experiment group reported no more suicidality than did the control group after the initial survey and the same result held true for the two-day interim period between assessments.  So, in a direct test of the central research question, the results indicated that asking about suicide had no negative impact on levels of suicide risk.

An astute reader might look at these results and say "well, this was conducted in a high school, so maybe none of these kids were suicidal anyway and the results would not generalize to a clinic when people are actually struggling with a mental illness."  Fortunately - as might be expected of an article published in JAMA - the authors were careful to consider and investigate this question as well.  Step 1 in this process was to investigate individuals with high levels of depressive symptoms (higher than 15 on the Beck Depression Inventory).  Results indicated that being exposed to suicide questions did not increase distress or suicidal ideation amongst depressed students.  In fact, depressed students in the experiment group had slightly lower depression scores after suicide risk was assessed than the control group exhibited after their first assessment (which lacked any assessment of suicidality).  As such, it appears that, for depressed students, being asked about suicidality may have had significant beneficial effects.

That last finding was impressive, but the authors did not stop there.  The greatest predictor of future suicidal behavior is a past history of suicide attempts, so the authors examined the impact of suicide screening on students with a history of past attempts.  Much like with the depression analyses, the authors found that students with a past history of suicide attempts who were screened for suicide risk at Time 1 reported less suicidal ideation than did students with a past history of suicide attempts who were not assessed for suicide risk at Time 1.  So, for high risk students, assessing suicide risk appeared to have significant beneficial effects on suicidal ideation.

So, why is it that assessing for suicide risk is beneficial?  For one thing, it might simply be that, for individuals suffering from suicidal ideation, having somebody normalize the experience and express concern by asking about it actually feels good.  In other words, rather than carrying around a secret that makes them feel isolated and different, perhaps such individuals feel more understood and connected when a clinician systematically assesses whether or not he or she is at risk.  Regardless of the answer to the question, however, these results are highly meaningful.  They not only failed to support the belief that assessing for suicide risk is harmful - they actually supported the opposite conclusion, that suicide risk has beneficial effects, particularly for those at the greatest level of risk.  For those of you who read Dr.Jill Holm-Denoma's featured article on diagnostic feedback, these results might not be all that surprising.  Remember, Dr.Holm-Denoma reported that, by providing careful accurate diagnostic feedback (information regarding a client's diagnosis/diagnoses), clinicians can actually help their clients feel better.

It appears that, at times, our concern regarding labeling people or making them feel judged might actually get in the way of us providing effective care.  Of course, shouting labels or callously asking about personal topics without empathy would not be a good approach.  Just like in everything else we do, it is useful to maintain a reasonable level of social skills.  Assuming that the clinician is not boorish, however, and that he or she can follow the proper protocol, it appears that assessing for suicide risk is, in fact, an incredibly useful and beneficial option that should be performed and documented routinely.

If you would like to learn more about suicide and suicide risk assessment procedures, we recommend the following resources, all of which are available through our online store:

• Why People Die by Suicide by Thomas Joiner

• Treating Suicidal Behavior: An Effective, Time-Limited Approach by David Rudd, Thomas Joiner, and Hasan Rajab

• The Interpersonal Theory of Suicide: Guidance for Working With Suicidal Clients by Thomas Joiner, Kim Van Orden, Tracy Witte, and David Rudd

Mike Anestis is a doctoral candidate in the clinical psychology department at Florida State University

Treating pediatric obsessive-compulsive disorder

by Michael D. Anestis, M.S.

I have not spent a great deal of time discussing mental illness in children on PBB; however, today I came across an interesting article on the treatment of pediatric obsessive-compulsive disorder (OCD) and thought it might be useful to describe its summary of treatment outcomes.  In this article, published in Cognitive and Behavioral Practice by Wendi Marien of the Monroe Clinic and Eric Storch, Gary Geffken, and Tanya Murphy of the University of South Florida and the University of Florida (2009), the authors discussed the relative virtues of cognitive behavioral therapy (CBT) and antidepressant treatment for pediatric OCD and presented some preliminary data on an intensive family-based version of CBT for children who have had poor responses to pharmacological treatments in the past.

What is OCD?

As we have discussed in prior articles on this topic, OCD is an anxiety disorder marked by two distinct characteristics: obsessions and compulsions.  Obsessions are defined as intrusive thoughts and/or images that are experienced as unwanted by the individual and which cause significant anxiety or distress.  Compulsions are repetitive behaviors (e.g., hand-washing) or mental actions (e.g., counting to particular numbers) utilized in an effort to reduce distress or prevent unwanted outcomes.  In some cases, such as excessive hand washing in order to reduce the obsessive fear of contamination by germs, the compulsive behavior represents a logical but exaggerated response to the obsession.  In other cases, such as counting to a particular number in order to prevent a loved one from being injured, the compulsive behavior exhibits no logical connection to the obsession.

In order for a diagnosis of OCD to be warranted, the obsessions and compulsions need to cause significant distress and/or impairment and must take up a substantial amount of time (e.g., several hours per day).  In other words, normative childhood anxiety or ritualistic behaviors are not necessarily indicative of any form of pathology.  Additionally, avoidance behaviors (e.g., avoiding using public restrooms or touching objects in the restroom due to obsessive fear of contamination) should be considered as contributing to distress and/or impairment.

In this article, Marien and colleagues (2009) cite prior work indicating that the 6 to 12 month prevalence rates of OCD in children and adolescents is 1.3%-4.0%.  Those numbers might seem small, but they are essentially on par with the prevalence rates of most mental illnesses in adults.

Pharmacological treatments for pediatric OCD

Marien and colleagues (2009) provided a summary of research findings on the utility of prescribing pharmacological treatments - particularly serotonin reuptake inhibitors (SRI's) - for pediatric OCD as well as a summary of the findings comparing pharmacological treatments to CBT.  The authors noted that a substantial portion of clients who present at their clinic were first treated for pediatric OCD through the prescription of an SRI and/or the use of non-empirically supported forms of psychotherapy despite clear evidence that the best approach is to begin treatment with CBT, either alone or in combination with a selective serotonin reuptake inhibitor (SSRI; Pediatric OCD Treatment Study Team [POTS], 2004).  There are a number of reasons why this scenario might unfold and the authors mentioned several of them (e.g., inadequate training in the use of empirically supported treatments), but these reasons are not the focus of our article.

Although the use of SRI's is not the most effective form of treatment, there is substantial evidence that they are quite helpful in the treatment of pediatric OCD.  In several trials, SRI's have been shown to be superior to placebo in the treatment of pediatric OCD (e.g., Abramowitz, Whiteside, & Deacon, 2005).  This is obviously a promising finding; however, when SRI's are compared to CBT, the results are not nearly as favorable (e.g., POTS, 2004).  In several studies, although children on SRI's exhibited significant improvement in their symptoms of OCD, their symptoms remained in the moderate range of severity (e.g., Abramowitz et al., 2005; March et al., 1998).  In fact, of those who do respond positively to SRI's, only a small minority actually remit.  In the POTS (2004) study, only 21.4% of children treated with sertraline remitted.  Additionally, once medication is discontinued, relapse rates are extremely high, meaning that pharmacological treatments for pediatric OCD, in many cases, require chronic treatment (Leonard et al., 1991; Pato, Zohar-Kadouch, Zohar, & Murphy, 1988).

CBT for pediatric OCD

CBT for pediatric OCD is not much different than CBT for adults with OCD.  Treatment includes an initial psychoeducation component, in which the children are taught about the nature of OCD and CBT and told what to expect from treatment.  Additionally, cognitive components, including the challenging of distorted thoughts, the generation of helpful thoughts, and the use of supportive self-talk are used.  Clinicians differ in the degree to which they utilize cognitive skills with children, although there is no reason to believe that dropping this component is in any way essential.  Finally, treatment also involves exposure exercises.  For this component, the child generates a list of feared situations, ranked in order by the degree to which they are fear provoking, and is systematically exposed to those situations over time, beginning with the least frightening situations.  The child is asked to remain in the exposure situation until his or her anxiety either disappears or decreases to a manageable level and is told not to engage in compulsive behaviors while this is happening.  Exposure exercises can be conducted as imaginal exercises (e.g., imagine the situation) or in vivo, with the latter being preferable when possible.  In addition to engaging in these exercises in session, the child is encouraged to practice at home daily in order to ensure that treatment gains generalize to the child's home environment.  This entire process is often referred to as exposure and response prevention (EXRP).

Studies have shown that CBT results in greater symptom reduction than do SRI's or placebo (Abramowitz et al., 2005).  Additionally, children treated with OCD, on average, exhibit only mild OCD symptoms post-treatment as opposed to the moderate symptoms experienced by children treated with SRI's.  In the POTS (2004) trial, which compared CBT alone to sertraline alone and the combination of the two, combination treatment was found to be superior to the other two conditions with respect to reducing symptoms, although combo and CBT alone did not differ post-treatment on the proportion of participants who remitted.  Ashbahr et al., (2005) found that, while group CBT and sertraline both significant reduced OCD symptoms in children, children treated with group CBT has significantly lower relapse rates than did children treated with sertraline 9 months after treatment ended.

So are you saying that pharmacological treatments for OCD do not work?

Marien and colleagues (2009) were definitely not saying that SRI's or SSRI's are ineffective.  Rather, they were simply pointing out that, while such treatments are better than nothing, they do not produce effects that are as impressive as those produced by CBT, either in individual or group format.  As such, CBT is considered the front-line, empirically supported treatment for pediatric OCD.  There are plenty of disorders for which the opposite conclusion is true (e.g., schizophrenia or bipolar spectrum disorders in adults), so this statement is not a biased commentary by a clinical psychologist attempting to further his or her own cause, but rather a frank and open discussion of empirical evidence which, in this case, is more supportive of psychosocial than psychopharmacological approaches to treatment.

What about the families of children and adolescents with OCD?

The second half of Marien et al.'s (2009) article discussed prior research on the use of families in the treatment of pediatric OCD and a proposed intensive form of family based CBT for this population.  Summarizing all of this information is beyond the scope of today's post, but let me give you a quick sense of what they found and what they have proposed.

First of all, there is compelling evidence that incorporating families into CBT for pediatric OCD can be helpful (e.g., POTS, 2004).  Given that the child or adolescent with OCD is not the only one impacted by the symptoms, this seems like a promising line of research potentially capable of helping many people at once.

In addition to this basic finding - including families can be good - there is also a debate as to whether intensive treatment might be a better approach.  Typically, CBT involves weekly sessions spread out over the course of several months.  Intensive treatments on the other hand, often involve daily sessions spread out only across a few weeks.  Storch and colleagues (2007) compared intensive CBT to weekly CBT for pediatric OCD and found that, at post-treatment, the intensive treatment resulted in a higher rate of remittance (75%) than did the weekly treatment (50%), but that the two groups were essentially equal at 3-month follow-up (72% and 77% respectively).  This seems to indicate that intensive treatments result in quicker gains, but that, in the long run, the two approaches result in equal improvements.  As such, the relative value of intensive versus weekly treatment depends upon the degree to which fast improvement is needed and whether one approach is more convenient and/or financially plausible than the other.

In response to the positive findings for group and family-based CBT for pediatric OCD, the authors detailed their own proposal for an intensive form of family-based CBT.  The treatment includes daily 90-minute treatment sessions for 3 weeks.  At least one parent is required to be present at all sessions and a substantial amount of attention is spent on teaching parents to be informed consumers aware of the nature of OCD, the treatments that are available, cutting edge research, and ways to help their child implement the treatment protocol without reinforcing OCD behaviors.  Marien and colleagues (2009) actually provide a detailed description of the session-by session content and I encourage you to read the original article if you are interested in learning precisely how their approach to therapy is conducted.

As of now, the research into this particular approach is still ongoing.  That being said, the authors were able to provide some preliminary data on the efficacy of the approach.  Importantly, the sample used by the authors in their work is comprised of children and adolescents who have failed to fully respond to pharmacological treatments.  As such, their findings not only demonstrate that their treatment approach is useful, but that it works with children and adolescents who have not responded to prior attempts at treatment.  In an open trial of 5 youths ranging in age from 4 to 14, they found that all five responded to treatment, with symptoms reduced, on average, by nearly 70% after 3 weeks (Storch et al., 2007).  Obviously, this is too small a sample for any sort of conclusion, but it was simply the initial trial the authors used to ensure that the treatment was promising enough to pursue further.  In a subsequent trial using 30 youths ranging in age from 7 to 19 years of age, in which every participant had a history of two or more failed trials of pharmacological treatments (SRIs and/or SRI augmented with atypical antipsychotics), symptoms were reduced, on average, by 54% in 3 weeks and treatment gains were maintained at 3-month follow-up.  Again, the sample is too small to draw any conclusions, but a massive clinical trial of this treatment approach is now ongoing and results will be forthcoming. 

Conclusion

In conclusion, the authors presented a compelling case that, while SRI's can be useful in treating pediatric OCD, CBT is substantially better.  This statement is based upon the fact that it reduces symptoms more significantly, leads to greater rates of remittance, and exhibits lower rates of relapse.  Additionally, the authors provided preliminary evidence that an intensive form of family-based CBT for pediatric CBT could be highly useful, but more data are needed before we can draw any firm conclusions on that point.

If you would like to learn more about obsessive compulsive disorder, we recommend the following products, all of which are available through our online store:

• Stop Obsessing!: How to Overcome Your Obsessions and Compulsions by Edna Foa and Reid Wilson
  

• Mastery of Obsessive-Compulsive Disorder: A Cognitive-Behavioral Approach Client Workbook by Edna Foa and Michael Kozak

• Mastery of Obsessive-Compulsive Disorder: A Cognitive-Behavioral Approach Therapist Guide by Edna Foa and Michael Kozak

• Family Based Treatment for Young Children With OCD: Therapist Guide by Jennifer Freeman and Abbe Marrs Garcia

• Family-Based Treatment for Young Children with OCD Workbook by Jennifer Freeman and Abbe Marrs Garcia

Mike Anestis is a doctoral candidate in the clinical psychology department at Florida State University

Interesting YouTube video by Happier.com

I just came across this video, which discusses the nature of mindfulness and the idea that emotions are simply a source of information rather than experiences that need to be altered.  The speaker is Todd Kashdan, whose work we have discussed on PBB before.

Soon, we intend to launch a YouTube channel and PBB video podcasts.  I think this is a great way to provide folks with accurate and valuable information on mental health.  Enjoy the video!

Digit ratios and disordered eating in men: Using the ratio of the second finger to the fourth finger as a measure of prenatal testosterone exposure

by Michael D. Anestis, M.S.

We have devoted a fairly substantial amount of attention to eating disorders on PBB.  If you have read these articles or are familiar with eating disorders in general, you likely know that these diagnoses are significantly more common in women than they are in men.  In fact, in anorexia nervosa (AN), more than 90% of those diagnosed are female.  That being said, although they comprise only a small minority of individuals diagnosed with eating disorders, men nonetheless have been known to exhibit symptoms of disordered eating.

In the past, some researchers have focused primarily upon environmental causes of eating disorders and there is reason to believe that variables such as exposure to extremely thin celebrities in mass media can negatively influence eating behaviors (e.g., Iancu, Spivak, Ratzoni, Apter, & Weizman, 1994).  Other researchers have focused primarily upon biological risk factors, with a particular focus on gonadal hormones as a potential explanation of the sex differences in eating disorders (Klump et al., 2006).

Today, I would like to discuss a study conducted by April Smith, Sean Hawkeswood, and Thomas Joiner  of Florida State University that has been published in the International Journal of Eating Disorders but not yet been made available in print.  In this study, the authors sought to examine the degree to which prenatal testosterone exposure - the amount of testosterone a fetus is exposed to while in the womb - predicted various facets of disordered eating in men.  In this paper, the authors opened up with a discussion of the literature on the role of testosterone as an organizing hormone for men.  Testosterone has been shown to play an important role in the masculinization of the human body and plays a vital role in the organization and development of several body parts, including genitalia, the brain, and other organ systems.  Additionally, studies have demonstrated that higher levels of prenatal testosterone exposure predict cognitive abilities (e.g., mental rotation) and a variety of behavioral tendencies (Honekopp, bartholdt, Beier, & Liebert, 2007).

Importantly, prenatal testosterone exposure has also been directly linked to eating behavior.  One study found that mice who received prenatal testosterone exposure consumed greater amounts of food and weighed more than mice who were not exposed to testosterone in utero (Donohoe & Stevens, 1983).  In humans, studies have found that both female and male twins were less likely to exhibit eating disordered behaviors if they had been exposed to higher levels of prenatal testosterone (Culbert, Breedlove, Burt, & Klump, 2008).  In that study, males from male-male twin pairs had less disordered eating than did males from male-female twin pairs and females from male-female twin pairs had less disordered eating than did females from female-female twin pairs.  Because sharing a prenatal twin environment with a male twin is believed to increase prenatal testosterone exposure, the authors theorized that prenatal testosterone levels explained the findings.

Smith, Hawkeswood and Joiner (in press) pointed out, however, that despite these promising findings, measuring prenatal testosterone levels is difficult and potentially dangerous, so alternative methods for obtaining this information are necessary.  As it turns out, digit ratios - specifically, the ratio of the second finger to the fourth finger (2D:4D) - are strongly correlated with prenatal testosterone exposure, such that a lower ratio (longer fourth than second finger) is indicative of exposure to greater levels of prenatal testosterone.  Men tend to have smaller 2D:4D ratios than women and research indicates that testosterone promotes the growth of the fourth finger whereas estrogen promotes the growth of the second finger (Honekopp et al., 2007).  Now, if you are anything like me, you likely stopped reading halfway through this paragraph and looked to see what your ratio looks like.  First of all, remember that you can not tell just from glancing how big or small your ratio is (although you can tell, to some extent, whether or not there is a difference between your second and fourth fingers).  Additionally, remember that the research is not saying the individuals with a particular ratio are fated to particular outcomes, but rather that they are more vulnerable to those outcomes given the right environmental stress (or right environmental support, in the case of some of the more positive findings listed below).

Given all of these prior findings, Smith and colleagues (in press) wanted to determine whether a lower 2D:4D ratio - a sign of greater prenatal testosterone exposure - would predict lower levels of disordered eating in men.  Additionally, because research has demonstrated that greater levels of prenatal testosterone exposure predict male physical fitness and athletic prowess, even when controlling for age, experience, and type of sport (Bailey & Hurd, 2005; Manning & Taylor, 2001), the authors wanted to see if a lower 2D:4D ratio would be associated with a lower drive for leanness and a greater drive for muscularity.

The sample of this study consisted of 204 undergraduate males, 78% of whom were Caucasian and 37% of whom reported being gay.  Ages ranged from 17 to 24.  To measure disordered eating, the authors used the eating disorders examination questionnaire (EDEQ-4).  Importantly, to measure eating attitudes and drive for particular body type, the authors used a scale specifically designed for men - the male body attitudes scale (MBAS).  The authors used the digit ratio from participants' left hands, as they found a significant difference between the ratio on the left and the right, with the left hand proving to be more "masculinized."  Prior studies have reported similar findings (Luce & Crowther, 1999), leading to the assumption that the left hand provides the truer estimate of prenatal testosterone exposure (although this is not a universally accepted belief).

Results from this study were, in large part, consistent with the authors' hypotheses.  Drive for thinness was lower in men with a lower 2D:4D ratio and drive for muscularity was higher in men with a lower 2D:4D ratio.  In other words, the greater the level of prenatal testosterone exposure, the less likely a male was to feel driven to be lean and the more likely he was to feel driven to have a muscular physique.  Both of these relationships remained significant when gay men were included in the analyses.  Additionally, both relationships remained significant when the authors averaged the ratios of the left and right hand and the drive for muscularity finding remained significant when the authors used the right hand as the predictor.

A lower 2D:4D ratio also predicted lower levels of eating disordered behaviors in men.  The relationship only approached statistical significant (p = .09) when gay men were included in the sample and the results were not significant when the average of both hands or the right hand were used as the predictor.

So, what do these findings tell us?  First of all, it appears that, for men as with women, a greater level of prenatal testosterone is associated with less of a drive to be thin and lower levels of eating disordered behaviors.  Additionally, a greater level of prenatal testosterone exposure appears to predict a greater drive to develop a muscular physique.

Admittedly, when I first heard about the use of digit ratios in psychopathology research, I was skeptical.  In fact, several years ago I had an opportunity to review a paper that involved digit ratios, but declined because I had not heard of this phenomenon and was entirely surprised by it.  Having now had the opportunity to read more about this phenomenon and to collect data on it myself, I am a firm believer in its utility, even if it is an imperfect measure of prenatal testosterone exposure.  Certainly better methods exist, but the digit ratio offers an easy and useful tool for considering biological risks.

Importantly, keep in mind that the authors are not attempting to say that eating disorders are entirely biologically determined.  They are simply attempting to clarify one of the ways in which a particular biological marker might be used to predict vulnerability to eating disordered beliefs and behaviors.  Quite obviously, there is nothing we can do after the fact to adjust how much testosterone an individual was exposed to in utero, so this is not a target for therapeutic interventions; however, understanding this relationship can help clarify one of the ways in which our own body makes us more or less vulnerable to particular outcomes and can help us predict risk well before the onset of symptoms and potentially utilize preventative care for vulnerable individuals.

Future research using clinical samples and comparing both male and female participants simultaneously would be highly useful, but the authors did a great job of supplementing the current literature and providing a foundation for future examinations of biological predictors of eating disorders in males.

If you would like to learn more about eating disorders, we recommend the following resources, all of which are available through our online store:

• Overcoming Binge Eating by Christopher Fairburn

• Crave: Why You Binge Eat and How to Stop by Cynthia Bulik

• Dialectical Behavior Therapy for Binge Eating and Bulimia by Debra Safer, Cristy Telch, and Eunice Chen
  

• Overcoming Your Eating Disorder: A Cognitive-Behavioral Therapy Approach for Bulimia Nervosa and Binge-Eating Disorder, Guided Self Help Workbook by Stewart Agras and Robin Apple
  

• Treatment Manual for Anorexia Nervosa: A Family-Based Approach by James Lock, Daniel le Grange, and Stewart Agras
  

• Eating with Your Anorexic: How My Child Recovered Through Family-Based Treatment and Yours Can Too by Laura Collins

Mike Anestis is a doctoral candidate in the clinical psychology department at Florida State University

A dimensional model of personality disorders: Moving towards DSM-V

by Michael D. Anestis, M.S.

Yesterday, I wrote about Brown and Barlow's (2009) proposed dimensional model of mood and anxiety disorders for the DSM-V.  The author's original article was published in Psychological Assessment as part of a special issue on suggestions for DSM-V and reading over their thoughts left me fascinated with the rest of the journal issue.  That being said, today I would like to continue yesterday's discussion, but I will shift the focus to a dimensional model of personality disorders, proposed in a paper published in the same issue of Psychological Assessment by Thomas Widiger of the University of Kentucky, John Livesley of the University of British Columbia, and Lee Anna Clark of the University of Iowa (2009).

What is wrong with our current diagnostic system for personality disorders?

Personality disorders, for a variety of reasons, have always been a source of controversy, frustration, and misinformation.  The very nature of the idea that a fairly stable and genetically determined aspect of an individual can, in and of itself, constitute a mental illness is troublesome to some people, as they interpret that to mean that psychologists are arguing that some individuals are inherently flawed and need to be changed or, alternatively, that they are simply pathologizing people for being different.  This, of course, represents a misunderstanding of the diagnoses, which are intended to recognize long standing patterns of thought, feelings, and behaviors that cause clinically relevant levels of distress and/or impairment.  In other words, nobody is looking to criticize anyone for being different, but they are definitely interested in identifying aspects of an individual's life that are causing problems and require help.

Unfortunately, the structure of the personality disorder categories in the DSM are based on very little research and, in many cases, involve vague and/or overlapping symptoms across disorders.  In the DSM-IV-TR, personality disorders are divided into three clusters:

Cluster A - Odd/eccentric

Schizoid Personality Disorder
Schizotypal Personality Disorder
Paranoid Personality Disorder

Cluster B - Dramatic/emotional

Borderline personality disorder
Antisocial personality disorder
Narcissistic personality disorder
Histrionic personality disorder

Cluster C - Anxious/fearful

Obsessive-compulsive personality disorder
Dependent personality disorder
Avoidance personality disorder

As is stands, however, there is little research supporting the validity of these distinctions and some evidence that actually refutes it (Sheets & Craighead, 2007; Widiger & Mullins-Sweet, 2005).  In addition, there are some that argue that schizoid personality disorder does not cause enough distress or impairment to constitute a mental illness (I have even heard some prolific researchers state the case that this disorder does not truly exist), there is compelling evidence that schizotypal personality disorder should be moved to the schizophrenia spectrum section of the DSM, and some evidence that avoidant personality disorder is not fundamentally different from social anxiety disorder (SAD).

Widiger, Livesley, and Clark (2009) further emphasized the problematic structure of personality disorders in the DSM-IV-TR by pointing out that two individuals can meet criteria for antisocial personality disorder (ASPD) without sharing a single symptom.  The disorder includes seven possible symptoms and only requires that an individual exhibit three of them in order to meet criteria for a diagnosis.  Quite obviously, if two people can have the same diagnosis without sharing a single symptom, the validity of that diagnosis is open to serious question, even when we consider the research on different factors of ASPD and its relationship to psychopathy.

Yesterday, when I discussed the potential value of Brown and Barlow's dimensional model of mood and anxiety disorders, I mentioned my concern that we would no longer be able to apply our research and understanding of empirically supported treatments for those disorders.  In the case of personality disorders, this is much less of an issue.  Other than dialectical behavior therapy (DBT) for borderline personality disorder (BPD), there is a dearth of research on treating personality disorders and an empty list of empirically supported treatments.  In other words, the risk of shifting to a new system is substantially reduced in this case.

All of this brings me to the central point of Widiger, Livesley, and Clark's (2009) article and their proposal for a new system: there is substantial evidence that personality disorders are dimensional constructs rather than categorical entities (e.g., Clark, 2007) and that personality disorders represent extreme deviations of normal characteristics rather than completely foreign sets of symptoms and behaviors (e.g., Blashfield, 1984).  I realize that last sentence was a bit of a mouthful, so let me explain that more clearly.  It appears that personality disorders are dimensions rather than categories in that everybody has a certain amount of these symptoms/characteristics.  In other words, it's not a matter of simply having them or not, but rather a matter of how much of each characteristic you have.  Some people have more or less than others and that deviation in degree is at the root of the problematic outcomes associated with personality disorders.  Along those lines, personality disorders appear, at least to some degree, to represent extremely high or low amounts of personality characteristics present to some degree in all people (e.g., emotional reactivity).  As such, a personality disorder does not represent some foreign presence clouding an individual's character as a tumor alters the functions of an internal organ, but rather represents an amplified or dulled character trait - likely in large part genetically determined - that causes an individual to think and/or behave differently than most other people in such a manner that the difference is distressing and/or impairing.

So what do we do about this?

Widiger, Livesley, and Clark (2009) propose a system not altogether different from that proposed for mood and anxiety disorders by Brown and Barlow (2009).  The authors based their model on a combination of three popular measures of personality: the Dimensional Assessment of Personality Pathology - Basic Questionnaire (DAPP-BQ; Livesley & Jackson, in press), the Schedule for Nonadaptive and Adaptive Personality (SNAP; Clark, Simms, Wu, & Casillas, in press), and the NEO Personality Inventory - Revised (NEO PI-R; Costa & McCrae, 1992).  All three of these measures attempt to break personality into smaller components and to provide personality profiles for individuals based upon their scores on each subscale.

Using a statistical procedure known as factor analysis, the authors attempted to determine the degree to which these three models of personality could be integrated into a single model, thereby giving us a unified manner in which to understand and measure personality.  What they found was a hierarchical model that could potentially serve as a valid measure capable of being integrated into clinical practice in a useful manner.

Essentially, personality in this model is broken down into two levels.  The first level features four categories: 

• Emotional instability
• Introversion
• Antagonism
• Constraint

When a client first presents to a clinic, Widiger, Livesley, and Clark (2009) suggest that they be assessed on these four categories.  Clinicians should look to see if the client exhibits a deviation - either high or low - on any of them.  Remember, every individual has some degree of all four of these traits, but significantly high or low levels of any or them could result in problematic outcomes.

If a client does demonstrate a deviation from the norm on one or more of the four categories, the clinician should then follow-up on that particular category with an assessment of level two personality features - subcomponents within each of the four first level categories.  The following are examples of level two features, organized within their respective level one categories.

Emotional Instability:

• Anxiousness
• Affective lability
• Narcissism
• Self-harm

Introversion:

• Intimacy problems
• Restricted expression
• Social avoidance
• Stimulus seeking (for low introversion scores)
• Exhibitionism (for low introversion scores)

Antagonism:

• Aggression
• Entitlement
• Suspiciousness
• Callousness

Constraint:

• Compulsivity
• Workaholism
• Impulsivity (for low constraint scores)

So, for every client that comes in to a clinic, the clinician will assess their profile on the four first level categories.  In this sense, the first level serves as a screener.  The clinician administers a self-report questionnaire to see if further assessment is warranted and, in the meantime, develops a rough picture of both the assets and liabilities in the profile.  For any client that deviates with a high or low score on one or more of those categories, a further assessment will be conducted to examine the specific facets of the first level category that are causing the deviation. 

So what would happen to our diagnostic categories?

If this system were adopted in the DSM-V, the traditional personality disorder categories would no longer be used.  Instead, individuals would be assessed for their individual personality profile, which would detail both assets and liabilities.  The profile would consist of a number of non-overlapping concepts, each of which has been heavily researched and shown to correlate with distinct outcomes.  The specific, level two features (e.g., impulsivity, entitlement, emotional lability) would serve as targets for treatment.

Are these necessarily the best variables to use?

The authors freely admit that their model likely needs some modification.  Variables from measures other than the three mentioned above might be useful additions and some of the variables included above might not contribute much use.  In this sense, the authors are much more certain about the basic structure and purpose of the model than they are of the specific components.

This is a tricky situation, however, because many different people would have a stake in what actually become integrated into the final model.  For instance, as a researcher in the field of "impulsivity" I can tell you without hesitation that the word "impulsivity" is of almost no use, as it represents many different distinct ideas (Whiteside & Lynam, 2001).  Much of my research has focused on negative urgency and, as such, if that variable were integrated into the model, this would be great for my career personally and a decision with which I would agree wholeheartedly.  At the same time, there are countless other researchers out there with similar thoughts on different variables and no clear cut way to determine which of us makes the best case.  As a result, much like with the creation of the current DSM system, we would be left at the mercy of the committee to make that decision for us and, in all likelihood, biases will be involved in the selection of items. 

All of that being said, the variables mentioned above are all very well researched and represent strong choices for inclusion in a dimensional model of personality disorder.

What does this mean for treatment?

Treatment research for personality disorders other than DBT for BPD is limited.  Additionally, although DBT is highly successful in reducing many components of BPD, including suicidal behavior and non-suicidal self-injury, it does not fully address each symptom of the diagnosis.  As such, we are currently faced with a situation in which, regardless of whether or not the system is overhauled, a massive amount of research is needed.  Along those lines, implementing this system would require immense amounts of research indicating that individuals with particular personality profiles benefit in clearly identifiable ways from specific treatment protocols.  Such research would enable the creation of a new generation - or least an updated version of the current generation - of empirically supported treatments.

Conclusion

Here is what we know: the current diagnostic system for personality disorders is highly flawed in a number of ways.  In fact, the flaws inherent in this section of the DSM far outweigh those of just about any other part of the system.  For years, many individuals have pushed for a dimensional system for assessing personality and recognition of the fact that symptoms of personality disorder might represent extreme deviations of normal personality characteristics, but no consensus has been reached on a suitable model.  Widiger, Livesley, and Clark (2009) presented a thorough and compelling case for a new model based on empirical research, but acknowledge that their model is also imperfect.  That being said, there seems to be reason to believe that some form of dimensional model will be incorporated into DSM-V, at least for personality disorders, so it is important to discuss the pros and cons of such a move and to become informed on what this type of model would mean, both for clients and clinicians. Hopefully, you will share your thoughts on this model in particular or on changes needed for personality disorders in general.  If you are a clinician, how do you think this will impact your practice?  If you are an individual struggling with a personality disorder, to what degree does this model make sense to you?  If you are a researcher, what do you think needs to be integrated into this or other models of personality?

If you would like to learn more about the DSM or personality disorders, we recommend the following resources, all of which are available through our online store:

• Diagnostic and Statistical Manual of Mental Disorders DSM-IV-TR Fourth Edition (Text Revision)
  

• Cognitive-Behavioral Treatment of Borderline Personality Disorder by Marsha Linehan

• Personality Disorders: Toward the DSM-V by William O'Donohue, Katherine Fowler, and Scott Lilienfeld

Mike Anestis is a doctoral candidate in the clinical psychology department at Florida State University

A dimensional model of mood and anxiety disorders: Moving towards DSM-V

by Michael D. Anestis, M.S.

In May 2012, the fifth version of the Diagnostic and Statistical Manual of Mental Disorders (DSM-V) is due for publication.  As has been the case each time the DSM - our guideline for mental health diagnostic decisions - has approached revision, substantial controversy exists regarding the changes that should be made.  In the first two versions of the DSM, mental illnesses were described in vague paragraphs that made assumptions regarding causes (e.g., unconscious conflicts from childhood) often based upon the scientifically weak theories of psychoanalysts.  Beginning with DSM-III in 1980, a greater emphasis was placed upon reliability, with specific symptoms based in large part upon observable behaviors taking precedent in an atheoretical system.  The system was further refined in subsequent years, leading to our current version, the DSM-IV-TR.  Despite substantial improvements in the validity and reliability of diagnoses, the DSM still has many problems and, as a result, researchers and clinicians have been campaigning for various shifts, some more radical than others, in our approach to conceptualizing mental illnesses.

This article will by no means touch upon all of the controversies surrounding the creation of the DSM, nor will it effectively summarize all of the strengths and weaknesses of our current system.  Instead, the goal is to provide a description of an article recently published in Psychological Assessment by the eminent scholars Timothy Brown and David Barlow (2009) of Boston University that details their proposal for a new diagnostic system for anxiety and mood disorders.

The Problem - Weaknesses in the Current Diagnostic System

Brown and Barlow (2009) opened their article by explaining the various weaknesses of the DSM-IV-TR, with the aim being to emphasize the areas in which DSM-V needs to provide tangible improvements.  Along these lines, the authors noted that, while the reliability of the current DSM is a marked improvement over earlier versions, there are still several sources of problems in this area.  They cited a study by Brown, DiNardo, and colleagues (2001) in which participants were administered structured clinical interviews by two different clinicians in order to determine the degree to which the clinicians would agree on diagnoses.  The goal, of course, was to determine how reliable the DSM diagnoses actually were.  Although the results were fairly strong, they noted six distinct sources of error: 

1. Clients often reported different information to the different clinicians.  This reduces reliability because the clinicians are not left with the same information from which to work.
2. Clients reported identical information to each clinician, but the clinicians disagreed on whether or not the symptoms caused sufficient distress and/or impairment to justify a diagnosis
3. The clients' symptom levels actually changed between interviews, thereby leading to legitimate disagreements
4. The clinician improperly applied the DSM diagnostic rules and thereby arrived at improper diagnostic decisions
5. The clinicians disagreed on whether symptoms of one disorder were better accounted for by another diagnosis
6. The DSM itself was unclear in its description of symptoms, thereby hindering differential diagnoses

In addition to reliability problems, Brown and Barlow (2009) noted that the current DSM exhibits problems with comorbidity.  When an individual has comorbid diagnoses, this means that they meet criteria for more than one diagnosis at the same time.  As it turns out, comorbidity is the rule, not the exception.  In other words, clients meet criteria for multiple diagnoses more often than they meet criteria for only one.  At first glance, this might not seem like much of a problem.  After all, this might simply reflect the fact that different disorders often cluster together.  A closer examination of the phenomenon, however, reveals several problems that emerge from this situation.  For instance, Brown and Barlow (2009) noted that the comorbidity of social anxiety disorder (SAD) and depression is, in large part, accounted for by the fact that both disorders are characterized by low trait levels of positive affect.  In other words, it might not be that these two distinct phenomena co-occur but rather that both the symptoms of depression and SAD are, in large part, accounted for by the general tendency to experience chronically low levels of positive emotions and that making the distinction between the two different disorders is less important than emphasizing the shared characteristic. 

Brown and Barlow (2009) also noted that the high levels of comorbidity in the current DSM might actually cause clinicians to understate the presence of certain symptoms.  For instance, Brown, Campbell, and colleagues (2001) found that the presence of panic disorder was predictive of decreased risk for SAD and specific phobias.  Given that all three disorders share several important characteristics, including avoiding feared situations, this was surprising.  The explanation, however, is quite simple,  The DSM diagnostic rules often lead clinicians to assume that symptoms of diagnoses such as SAD and specific phobias are better accounted for by other diagnoses such as panic disorder and, as a result, the client's list of diagnoses does not make any mention of clinically impairing symptoms that are associated with the other disorders.

Brown and Barlow (2009) saved their strongest criticism of the current DSM for last.  As it currently stands, the DSM calls for categorical diagnoses.  In other words, you either have a diagnosis or you do not.  With the exception of depression, which allows for specifiers such "severe" or "mild," there is no description of severity in any DSM diagnoses.  Importantly, the decision to use a categorical system and the number of symptoms required to meet the cut point for each diagnosis are not based on empirical research.  In other words, an arbitrary line was drawn by a committee to determine whether or not an individual meets criteria for a mental illness.  Now, to be fair, the DSM also includes "not otherwise specified" diagnoses, which allow clinicians to still provide a recognized diagnosis to an individual who is obviously distressed or impaired but who does not meet criteria for a mental illness, but the arbitrary nature of the cut points is an obvious issue nonetheless.

The Proposed Solution - Brown and Barlow's (2009) Idea for a New Approach

Having established that the current version of the DSM is imperfect, Brown and Barlow were left to offer up something better.  After all, it is easy to be a critic, but the real value rests in the ability to provide incrementally valuable alternatives capable of improving the situation.  Simply tearing apart imperfections does little to help anyone and, remember, this entire field exists to provide effective care to individuals suffering from very real problems.  Fortunately, neither Brown nor Barlow were interested in being mere critics and, while we can certainly debate the value of their proposed system, it is worth taking a look at what they propose in order to figure out where the field might be heading and what the best possible solution might be.

Quite frankly, the system proposed by Brown and Barlow (2009) is complex and the article in which they described the system was fairly dense, so my efforts to explain their thoughts here come with the disclaimer that, in order to fully understand the concepts, your best bet is to read the actual article.  That being said, their system essentially boils down to ditching our current system of diagnostic categories for mood and anxiety and instead measuring the degree to which individuals exhibit elevations on particular variables known to relate to mood and anxiety disorders.  Specifically, they propose considering:

• Anxiety/neuroticism/behavioral inhibition (A/N). In other words, the degree to which the individual experiences chronically elevated levels of negative emotions and demonstrates a tendency to inhibit prepotent responses.
• Behavioral activation/positive affectivity (BA/P).  In other words, the degree to which an individual tends to engage in approach behavior and to experience chronically elevated levels of positive emotions.

Research has shown that individuals experiencing anxiety disorders as well as depression tend to exhibit highly elevated levels of A/N.  Additionally, depression and SAD tend to involve low levels of BA/P and mania involves high levels of BA/P.  In other words, all of these disorders are related to particular combinations of these variables and, as such, developing a profile of the degree to which an individual exhibits high or low A/N and BA/P could provide a more comprehensive picture of what is going on with a client than we can get from a simple yes/no decision regarding whether or not the client meets criteria for one or more diagnoses.

Simply looking at these two variables, however, does not give us enough information.  After all, if SAD and depression are characterized by similar patterns of A/N and BA/P but are also characterized by different symptoms and outcomes (e.g., suicide risk profile), then we need to know more about what is going on with the client.  Brown and Barlow (2009) thus propose that we further assess the presence, severity, and frequency of avoidance behaviors and examine the functional relationships of A/N with the characteristics traditionally associated with particular diagnoses. 

So, Brown and Barlow (2009) see mood and anxiety disorders unfolding as follows: an individual has genetically predetermined levels of A/N and BA/P.  In the presence of life stress, specific features of classic diagnoses are triggered and become evident, resulting in clinical distress and impairment.  Importantly, although specific factors (e.g., intrusive thoughts) are often associated with one particular diagnosis (e.g., OCD), they often appear in other disorders as well, so assessing for their presence instead of just assessing for diagnoses would allow clinicians to better understand everything going on with each client.  Brown and Barlow (2009) listed the following as specific features that should be assessed:

• Fight or flight response: when triggered inappropriately, this results in panic attacks.  Panic attacks themselves often become the central focus of A/N.
• Somatic symptoms: the degree to which the individual is focused on unexplained physical symptoms and/or sensations of unreality associated with dissociation.
• Ego-dystonic intrusive thoughts: the degree to which an individual experiences unwanted and aversive thoughts (e.g., thoughts of harming one's own children). 
• Social evaluation: the degree to which an individual exhibits a fear of being negatively evaluated by others
• Past trauma: the degree to which an individual focuses on or responds strongly to (e.g., flashbacks) cues of a past traumatic event

Pulling it All Together

Brown and Barlow (2009) thus propose that, when a client presents at a clinic, he or she should be assessed for levels of A/N and BA/P.  This will provide a basic profile of their vulnerability to traditional mood and anxiety disorders.  Furthermore, the clinician should assess the presence, severity, and frequency of avoidance behavior and the degree to which the client is exhibiting the specific features listed above.  So, in the end, the client does not receive a diagnosis, but rather a description of the degree to which he or she maintains particular levels of negative and positive emotions as well as the specific manner(s) in which his or her emotions manifest in response to stress.

What is the Purpose of All This?

The idea here is to ensure that our assessment provides us with the optimal amount of information about the client and his or her struggles rather than focusing too narrowly on arbitrarily designed categorical diagnoses that might leave out important information relevant to treatment approaches.  This system would remove the need to make differential diagnoses that cause one diagnosis to be subsumed within another and would no longer overlook the clinical implications of subthreshold symptoms.

Are there potential pitfalls to this system?

Here is where my article diverges a bit with the Brown and Barlow (2009) one.  The authors noted that clinicians might be hesitant to utilize this system because it is relatively complex and unfamiliar.  They also noted that, as of now, there is no single empirically tested assessment measure capable of performing the tasks they recommend and, as such, we have no research upon which to base the cut points that would be needed in order to inform insurance companies regarding the need for services.  I agree with all of these critiques.

What they did not spend much time discussing - and to be fair, they could not discuss everything in depth in a single article - was the impact this would have on empirically supported treatments (ESTs).  ESTs have been developed for several disorders.  Although various forms of cognitive behavioral therapy (CBT) work for essentially all of the disorders involved in this new system, other forms of therapy such as interpersonal psychotherapy (IPT) have not been validated on all of the diagnoses.  As such, it becomes unclear when each treatment would be most useful.  The authors proposed that we develop transdiagnostic treatments and noted that they had developed one based upon CBT, but in doing so, we will be left without the mountains of prior work implicating particular treatments for specific diagnoses.  We would have to essentially start over, which can take a very long time.  The end result might be better, but we would be assuming substantial risk and would be left with a transition period certain to cause a significant amount of confusion and unrest both within and without the professional community.

So is this the answer?

Well, Timothy Brown and David Barlow are both substantially more informed on these matters than I am, but I find myself undecided on the best course.  Data indicate that many disorders are dimensional in nature, so I am very much in favor of a system that measures severity of illness and assesses all symptoms.  At the same time, I am a huge proponent of ESTs and am skeptical of any approach that would leave us without the benefit of all the work that has led us to this moment.  Additionally, I am not convinced that a dimensional system will result in cut points that are any more valid that those that have currently been established and I can not help but fear that insurance companies would utilize their own cut points that minimize the likelihood of clients receiving effective mental health care.

The bottom line is, there will be no perfect answer.  The question is how to best ensure incremental improvements in DSM-V that allow for more valid and reliable diagnoses and an easier path toward empirically supported care.

Your thoughts?

Where do you see the DSM-V going?  Do you like Brown and Barlow's model?  What do you think are its pros and cons?  Even if you think it is the best system, do you think it could be effectively implemented in every day practice?

If you would like to learn more about the DSM and current conceptualizations of mental health and psychotherapy, we recommend the following resources, all of which are available through our online store:

• Diagnostic and Statistical Manual of Mental Disorders DSM-IV-TR Fourth Edition (Text Revision)
  

• Clinical Handbook of Psychological Disorders, Fourth Edition: A Step-by-Step Treatment Manual by David Barlow

• A Guide to Treatments that Work by Peter Nathan and Jack Gorman

Mike Anestis is a doctoral candidate in the clinical psychology department at Florida State University

Guest Article - From "Verbal Behavior" to Relational Frame Theory

by Brian Thompson, Ph.D.

In a cognitive psychology class in graduate school, I was taught an account of the fall of behaviorism and rise of cognitive psychology that has achieved near mythical status. According to this narrative, Skinner’s (1957) Verbal Behavior, an attempt to describe in behaviorist terms how humans develop language, was criticized by Noam Chomsky in his review as too simplistic and mechanistic to account for the complexity of human language. This mortal wound weakened the hegemony of behaviorism and its hold on academic psychology, allowing for cognitive approaches to seize power and advance the field.

Like many students, I accepted this story without question, including its characterization of Skinner and behaviorism, which supported many of prejudices against behaviorism I had picked up during my education. However, as I began to learn more about behaviorism a few years later, I found that there was another side to this story. Among behaviorists, the view is that Skinner did not respond to Chomsky’s critique because Chomsky misunderstood his book and had attacked a caricatured version of behaviorism different from Skinner’s brand of radical behaviorism. Skinner did not appear to feel the attack was worth responding to, although MacCorquodale (1970) eventually published a reply. 

Skinner’s disinterest in responding to Chomsky’s critique may in fact have weakened the behaviorist position in the eyes of many psychologists and led to it’s gradual disappearance among many graduate programs in following decades. Moreover, some behaviorists also believe that Skinner’s account of language was flawed—just not for the reasons Chomsky outlined. Friman, Hayes, and Wilson (1998) argue that Skinner developed a valid theoretical foundation for studying private events (e.g., thoughts, emotions), but that he simultaneously indicated that studying private events was “theoretically and practically unnecessary” (p. 140). In this view, Skinner overemphasized the study of direct contingencies of emotional experiences rather than the emotional experiences themselves.

Within the past 20 years, behaviorists have steadfastly developed a theoretical and research literature tackling language from a behaviorist perspective called Relational Frame Theory (RFT). This has yielded an edited book, Relational Frame Theory: A Post-Skinnerian Account of Human Language and Cognition. RFT is “post-Skinnerian” in the sense that it is indebted to Skinner’s work but moves beyond it with a re-working of Skinnerian ideas and expanded nomenclature. RFT stands on its own, but Barnes-Holmes, Barnes-Holmes, and Cullinan (2000) discuss how RFT and Skinner’s work may be bridged.

RFT is dense, even for seasoned behaviorists, so only a cursory discussion of the main principles will be offered here. It is rooted in functional contextualism, a philosophy of science with a focus on the study of an act or event within a particular context (See Hayes, 1993). Functional contextualists believe that there are no unshakable, universal truths; instead, they value what is useful and practical. The basic unit of RFT is the relational frame, which is the “action of framing events relationally” (Hayes, Fox, et al., 2001, p. 43). The idea of a “frame” is like a picture frame in that relational responding can involve any sort of events, just as a frame can hold any picture.

Relational frames have three key properties. The first is called mutual entailment, which simply means that relations between stimuli are bidirectional. If I told you that my cell phone was called a “shug wuggle,” you’d instantly understand that a shug wuggle is also my cell phone: A = B, B = A. A nonhuman organism can be trained to choose a picture of my cell phone after being shown a blue square, but it would need additional training to pick a blue square after being shown my cell phone. A major feature of relational frames is that they are arbitrarily applied; that is, they don’t depend on any formal or physical properties of the environment. There’s nothing particularly “shug wuggle-like” about my cell phone—the relationship we established is completely arbitrary. In addition to mutual entailment, relational frames involve what is called combinatorial entailment. If I told you that a shug wuggle costs $50, you may also surmise that my cell phone costs $50: A = C, C = A, B = C, C = B. Lastly, relational frames involve transformation of stimulus functions. This refers to a modification of functions in the stimulus relations.  If prior to telling you that a shug wuggle is my cell phone I told you that a shug wuggle is “a pain,” you might initially think that a shug wuggle referred to something physically painful. After you found out a shug wuggle is my cell phone, you may decide that what I meant was that a shug wuggle is difficult or inconvenient to operate. 

I’ve provided just a sketch here, but the relatively simple idea of how relational framing forms the basis of human language unfolds into many insights about how language and cognition work. These include applications of RFT in the realms of development, education, and even spirituality. Perhaps the most widely studied application of RFT is in understanding the processes of psychopathology and of psychotherapeutic change, which has helped to inform psychological interventions such as Acceptance and Commitment Therapy (ACT) across hundreds of studies.

RFT is the foundation for ACT, a form of applied behavioral analysis and one of the “third wave” treatments within the cognitive behavioral tradition (Hayes, 2004). Although one does not need to be familiar with RFT to be a competent ACT therapist, an understanding of RFT can deepen one’s understanding of ACT. For example, RFT and ACT have informed theories about the processes of mindfulness (*Editor's note: see our prior articles on mindfulness for a summary of this concept - M.A.*).  In addition to supporting an understanding of the processes involved in ACT, RFT has been applied with very promising results to children with development disabilities such as autism. This recent application of RFT has resulted in the book, Derived Relational Responding Applications for Learners with Autism and Other Developmental Disabilities: A Progressive Guide to Change.

As you might have gathered from my brief example, RFT is not an easy read. The edited RFT book is often referred to within the ACT community as a “cure for insomnia.” During a recent ACT training I attended with a major contributor to the ACT and RFT literature, he referred to the RFT book as something for “super nerds,” which earned me amused glances from colleagues who knew I was studying it as part of my postdoc at Portland Psychotherapy. Yet the book is far from impenetrable. In a comparison to literature, it’s closer to Joyce’s Ulysses than Finnegans Wake: sometimes difficult but not baffling so, and rewarding with additional re-readings.

The main website for both ACT and RFT is the Association for Contextual Behavioral Science. For paid members—and you only pay what you feel the membership is worth, based on your values (at minimum $1)—you have access to publications, treatment protocols, and listervs for ACT and RFT. For anyone interested in learning more about RFT and/or ACT, the access to PDF’s of related publications is worth joining for in itself.

If you would like to learn more about Acceptance and Commitment Therapy, Psychotherapy Brown Bag recommends the following resources, which are also available through the online store:

• Get Out of Your Mind and Into Your Life: The New Acceptance and Commitment Therapy by Steven Hayes

• Learning ACT: An Acceptance & Commitment Therapy Skills-Training Manual for Therapists by Jason Luoma, Steven Hayes, and Robyn Walser

• Mindfulness for Two: An Acceptance and Commitment Therapy Approach to Mindfulness in Psychotherapy by Kelly Wilson

• A Practical Guide to Acceptance and Commitment Therapy by Steven Hayes and Kirk Strosahl

• Acceptance and Commitment Therapy: An Experiential Approach to Behavior Change by Steven Hayes, Kirk Strosahl, and Kelly Wilson

• The Mindfulness and Acceptance Workbook for Anxiety: A Guide to Breaking Free from Anxiety, Phobias, and Worry Using Acceptance and Commitment Therapy by John Forsyth and Georg Eifert

Brian Thompson is a psychologist resident in Portland, Oregon

Psychological wounds: Testing the "scar hypothesis" of depression

by Michael D. Anestis, M.S.

Depression is a cruel and greedy intruder, not only for those who battle its symptoms, but also for their loved ones.  It brings seemingly unending waves of negative emotions, it robs people of their ability to experience joy, it disrupts sleep and eating, and it steals hope.  For some, it tragically culminates in loss of life due to suicide.  For all of these reasons and more, many individuals have devoted their lives to studying and treating depression with the hope of helping people avoid such pain.

One of the many things we have learned through research is that depression is episodic.  In other words, it will show up, hang around for a while, disappear, come back, and repeat this pattern again and again if left to its own devices.  There are several empirically supported treatments like cognitive-behavioral therapy (CBT) and interpersonal psychotherapy (IPT) that have been shown to successfully treat symptoms and, in large part, prevent them from coming back, but the episodic nature of depression remains an important consideration.  For one thing, past episodes of depression are one of the strongest predictors of future episodes of depression.  Adolescents with a past history of depression, for example, are more likely to develop future depressive episodes than are adolescents with no such history (Garber, Kriss, Koch, & Lindholm, 1988).  Because of findings like this, Lewinsohn, Steinmetz, Larson, and Franklin (1981) developed the "scar hypothesis," which states that depressive episodes cause individuals to develop characteristics that linger even after an episode has ended, thus leaving them vulnerable to future depression.

The idea of psychological scars is compelling.  It gives us a poignant and familiar image - physical scars - as a guide to help us conceptualize an abstract concept like mental illness.  It allows us to compare emotional struggles with familiar physical sensations and provides us with a way to envision the impact of depression on those who have experienced its binding grip.  The question, however, is whether this compelling hypothesis is supported by empirical evidence.  Along these lines, several researchers have attempted to study whether "scars" indeed are formed in response to depressive episodes and, if so, what those scars actually represent.

In one study of 352 grade school children, Nolen-Hoeksema, Girgus, and Seligman (1992) assessed depressive symptoms and attributional style repeatedly over a five year period and found that, in children who became depressed during the study, pessimistic attributional style increased in severity and remained elevated after depression symptoms subsided.  As such, they concluded that a pessimistic attributional style may represent a "scar" of depression.  In a separate study, Rohde, Lewinsohn, and Seeley (1994) examined depression and a variety of other variables in 45 adolescents at two time points separated by one year.  These researchers found that depressive symptoms, internalizing behavior problems, stressful life events, and excessive emotional reliance on others might represent "scars" of depression. 

Today, I would like to discuss a newer study on this same phenomenon, published in 2007 in the Journal of Consulting and Clinical Psychology by Christopher Beevers, Paul Rohde, Eric Stice, and Susan Nolen-Hoeksema.  The authors of this newer study - several of whom were also authors on the two studies cited above - mentioned that, while those earlier works were impressive in a number of ways, they also had important limitations that prevent a full understanding of whether or not any of the variables above actually represent "scars."  For one thing, the Nolen-Hoeksema et al (1992) study relied on self-report measures and, as such, no diagnostic information was available.  Additionally, the Rohde et al (1994) study only had two time points and a period of one-year, which prevented any assessment of whether or not the "scars" were actually present before depression onset and whether they faded with time after depression remitted.

In this newer study, Beevers and colleagues (2007) followed 496 female adolescents over a 7-year period.  Each year, participants took part in structured diagnostic interviews and filled out a series of questionnaires.  The authors identified 49 participants from this group who experienced their first ever depressive episode during the study and, in their analyses, they compared these individuals to 98 individuals with no history of depression prior to or during the experiment.  Importantly, the authors only included depressed individuals who did not meet criteria for depression one year, experienced their first depressive episode the next year, and did not meet criteria for depression the year after that.  The central aim of the study was to determine if the onset of an individual's first ever depressive episode resulted in an increase of particular variables that would remain elevated even after depression remitted, thereby serving as "scars" of depression.

The authors' findings were quite interesting.  When comparing the depressed group to the non-depressed group, each participant had three years of data.  For individuals in the depressed group, this meant they were looking at data from the year before depression, the year of depression, and the year after depression.  For the non-depressed group, they were simply looking at three consecutive years.  Along these lines, Beevers and colleagues (2007) found that the depressed group was more likely to have sub-threshold depressive symptoms in years 1 and 3 than were the non-depressed group; however, only 7% of the individuals in the depressed group had sub-threshold symptoms during those years.  In other words, although a small minority of individuals in the depressed group had symptoms of depression in the years preceding and following their depressive episode, most individuals did not follow that pattern.

Beevers and his co-authors (2007) next looked at a series of 13 other variables to see if they might serve as "scars."  The results indicated that none of these variables fit the pattern of emerging in response to the initial depressive episode and lingering in the year after depression had remitted.  That being said, they did find compelling evidence that these 13 variables serve a different types of risk factors for depression.  

Negative emotionality, rumination, and social maladjustment were all elevated prior to the onset of the initial depressive episode, became more severe during the depressive episode, then returned to their Year 1 levels after depression remitted.  As such, the authors referred to these variables as "exacerbated risk factors."  In other words, they were not "scars" of depression because they were there before depression.  Instead, they were risk factors for the development of depression and were, in turn, impacted by the depression while it was ongoing.  Depressive symptoms followed a similar pattern.  For those who experienced lingering depressive symptoms after their actual depressive episode, those same symptoms tended to be present prior to the depressive episode as well.  Most individuals who experienced sub-threshold depression, however, did not experience a later depressive episode.

Low self-esteem, low parental support, low levels of peer support, substance use, antisocial behavior, bulimic symptoms, and negative life events all served as "chronic risk factors."  In other words, individuals in the depressed group had higher levels of each of these risk factors prior to, during, and after depressive episodes and those who were depressed exhibited elevations of each of these factors prior to their depressive episodes, not just after them.  As such, it appears that while these variables predict vulnerability to depression, they do not serve as "scars" in that they do not emerge in response to depression.

Looking at all of these findings, the evidence does not seem to support the idea that depression leaves "scars" that account for the episodic nature of the illness.  Does this mean that depression does not have an impact on the lives of individuals after the symptoms fade?  Absolutely not.  But it does mean that many of those impacts were already established prior to the depressive episode as well and that, while individuals are more likely to reflect back upon their depressed days as a starting point for these variables, this might not accurately reflect how things unfolded.

There are some important limitations to this study that must be considered.  The authors noted that their sample consisted entirely of adolescent girls.  As such, the findings can not necessarily be generalized to boys or populations in different age groups.  Additionally, the authors could not possibly measure every possible variable that could serve as a "scar" of depression.  As such, although they did not find any variables that fit that pattern, this does not mean that there are not other variables that would.  The authors also only looked at individuals for whom depression lasted a year or less, so it is possible that "scars" only exist for individuals that experience a longer course of illness.  Finally, although the authors used impressive longitudinal data, they could not experimentally manipulate many of the variables and, as such, they could not conclusively determine whether any of the risk factors actually caused depression.  Correlation, after all, does not equal causation.

Despite these limitations, this study offered us a lot of important information.  Primarily, despite the philosophical appeal of the "scar hypothesis," there is no evidence supporting that the theory is valid.  Instead, there appears to be a wide variety of risk factors for depression, some of which become worse during depressive episodes and then return to their initial levels once depression remits.  Along these lines, there are many different paths to depression for adolescent girls, but no evidence that, once they experience their first depressive episode, they are likely to develop "scars" from the experience that, in turn, leave them responsible to future depressive episodes.  Their vulnerability to future depression, which is a very real and well-established phenomenon, is instead better accounted for by variables that were already present before they ever experienced depression.

As I said in the opening of this article, depression is a dark and disheartening experience for everyone involved.  Our intense and entirely understandable impulse to grasp the meaning of depression can leave us latching onto ideas that make intuitive sense, regardless of whether or not there is any evidence supporting the validity of those ideas.  This one study by no means indicates that the "scar hypothesis" is incorrect; however, to this point, there is more evidence contradicting than supporting the idea of "scars."  The better way to think about the phenomenon appears to be that depression is the result of many different vulnerabilities and that treatments must target these vulnerabilities in order to reduce risk because, left without intervention, they are likely to linger and continue to lead to problematic outcomes.

If you would like to learn more about depression, we recommend the following resources, all of which are available through our online store:

• Handbook of Depression, Second Edition by Constance Hammen and Ian Gotlib

• Skills Training Manual for Diagnosing and Treating Chronic Depression: Cognitive Behavioral Analysis System of Psychotherapy by James McCollough

• Cognitive Therapy of Depression by Aaron Beck, John Rush, Brian Shaw, and Gary Emery

• Feeling Good: The New Mood Therapy Revised and Updated by David Burns

Mike Anestis is a doctoral candidate in the clinical psychology department at Florida State University