- Hide menu


I finished my PhD candidacy examinations in the fall of 2013. I read the latest volume (9) of the Annual Review of Clinical PsychologyAs I read the articles, I wrote about them & shared my thoughts here.

I should mention that these entries represent quick sketchpad-like thoughts about these papers aimed to help me keep them distinct in my mind. What these entries are not, is well refined masterpieces of intellectual synthesis and critical analysis. Feel free to get in touch via email if you have some perspectives on the issues discussed here. I’d very much like to hear your opinion.



‘Reagcovery’, or ‘Trickle-Down Mental Health’

Braslow, J. T. (2013). The manufacture of recovery. Annual Review of Clinical Psychology, 9, 781-809. doi: 10.1146/annurev-clinpsy-050212-185642

In brief, I’m not quite certain what to make about Braslow’s arguments about the nature of ‘recovery’. If I’ve understood the arguments correctly, we are to understand that: 1) recovery is poorly or nebulously defined; 2) recovery is not such a radical break from the policies of the past as we might think. I get the distinct sense that I’m missing some kind of political context to these arguments. I’m not sure exactly what Braslow is trying to convince me of, in part because I’m less sure about what he’d like me to do about it had he managed to convince me of it. What is the agenda here?

Let’s start with a solid definition of recovery. Well this is exactly Braslow’s first point. No such consensus definition really exists. He discusses three distinct conceptualizations of the concept, as outcomes for patients, as experiences of patients, but also as a kind of mission statement for health care facilities. Ok fair – but how does Braslow introduce us to the topic? He describes the ‘recovery movement’ coalescing from disparate intellectual and social movements including the antipsychiatry movement of the 1960’s and 1970’s, the psychiatric survivors movement of the 1970’s, and the consumer rights movement of the last 50 years or so. He states “These various strands coalesced in the 1990’s as the hegemonic guiding principle of public mental health policy. Mirroring neo-liberal discourse on welfare reform, recovery advocates argue that poor mental health outcomes have resulted from a system of care that iatrogenically creates helplessness and dependency. Further they argue, a biologically based, Neo-Kraepelinian vision of psychiatric disease has produced a system of care that breeds self-fulfilling pessimism, resulting in patients dependent on the very system meant to cure them” (p. 783). Wow. That’s going to take some unpacking.

Braslow appears to be arguing against these ‘recovery advocates’ by anecdotally and impressionistically softening our conceptualizations of asylums and mental hospitals of yesteryear. He argues that contrary to what the recovery advocates would have you believe, these were places of healing, care, and therapy, and that patients were not passive recipients of care, or helpless against the questionable whims of the attending physicians. Even in cases of lobotomy, electroshock therapy and sterilization, doctors often “though not always” (p. 791) sought patient assent. Furthermore, in the case of drug therapies, doctors considered these as only adjunctive to psychosocial therapies.

Again, against those recovery advocates that would have you believe that state hospitals of the past bred their own need via creating institutional dependency, Braslow provides statistics from the 1950’s for the proportion of first admission patients remaining hospitalized in state hospitals after 12 months – the numbers are rather surprisingly low e.g., 34% for New York hospitals. Ok fair, but these numbers don’t tell us much about the quality of care received, or the effectiveness of treatment received, not to mention treatment durability. They’re also quite anecdotal, covering only a handful of states, and providing no explanation for why state 12 month retention rates would be so variable – compare New York’s 34% to that of Arkansas at 12.7%. I would have thought that large groups of patients subjected to similar therapies should evince similar levels of recovery. This is where I start to get confused about Braslow’s argument. Is he advocating that treatment provided in the past are superior to today’s? I wouldn’t think so… So, then is he arguing more simply that the recovery model is not so different from the past? I think so, but to what end? What’s are we supposed to do with this information?

From here the argument takes a turn for the weirder. Braslow directly implicates the swearing in of Ronald Regan (on January 2, 1967 no less) in setting the stage for the recovery movement by creating the ‘cultural legitimacy’ for deinstitutionalization and all but the disbanding of the welfare state… Though he later states that “no single factor accounts for deinstitutionalization” (p. 794). Right, so why implicate one man? I know Reagan is a rather polarizing figure to say the least, but I think we need to recognize here that Reagan could be nothing more than a mere figurehead for a pattern of cultural ideas and initiatives. From here Braslow makes again makes an impressionistic case, but this time against the therapeutic atmosphere of the community-based treatment centres that sprang up as state hospitals shut down: “evidence suggests that community mental health centres provided scant care for those with severe mental illness and instead provided care for a different, less impaired population.” (p. 795). So what I’m getting so far is that psychiatric hospitals of the past weren’t as bad as we think, community based treatment hasn’t been as good as we’ve thought, and Ronald Reagan single handedly moderated the change from the former to the later (maybe).

Well not really. Braslow also implicates the psychopharmaceutical industry as a secondary cause of deinstitutionalization by shifting perceptions of medications as adjunctive treatments of severe mental illness, to virtual panaceas thereby reducing complex mental illnesses to simple constellations of symptoms readily treatable by the same very drugs. The proof? Two drug adverts from 1958 and 1975 implicating this shift in thinking. Hmm. Maybe I’m too young to appreciate the cultural milieus that these adds apparently typify, but I’m certainly too old to rely on such anecdotal evidence to sway my thinking. Maybe a different pair of adverts would tell a different story. Another issue I had with this piece of persuasive writing was the way in which Braslow admonishes the drug industry for how it “…transformed a myriad of psychological ills into a vast multi-billion dollar market” (p. 795). Not that I’m a staunch advocate of big pharma, but this argument while holding the potential to be true, may be trivially so. Every ‘market’ that exists in the modern economy exists because it has taken on some inefficiency or ill of the human condition, innovated solutions, and sold them for profit. Automobiles, medical devices, computer technology. Not that these industries are always ultimately benevolent (e.g., weapons etc.) but each fills a need for profit, and simply to say that it does is not a strong argument for its malevolence.

In keeping with this view, Braslow then links the shifting values in mental health care with the concurrent shifting of political values in the United States away from government intervention programs for the poor such as state hospitals and healthcare. Reagan’s argument was essentially to reduce government intervention (thereby saving tax dollars) and let the free market do its work; rearranging incentives and redistributing wealth accordingly. Under this conceptualization, the poor were being kept poor by the very government programs designed to help them; welfare itself was not an antidote to poverty, but a poverty trap! In this framework, Braslow argues that the recovery movement “nests neatly within the broader context of neo-liberalism and state policies of the past 25 years.” (p. 800). For Reagan, welfare begets poverty; for Recovery, mental health care begets psychopathology – or, “Dependency begets dependency in both cases” (p. 801).

Again, I’m not quite sure what’s being argued for here. What are we to do with this information? At a practical level, what does it mean to say that the recovery movement mirrors some abstract ideology from decades ago? This is exactly the question Braslow leaves us with in his conclusion, adding that “this history says nothing about whether recovery works” (p.805). Rather, he seems to be suggesting that there are viable alternatives to the recovery model. Hmm.

Some thoughts:

1) Ok, fair – from this historical and retrospective level of analysis there are viable alternatives to the recovery model. Consequently, under a different set of circumstances, I can see how sate run hospitals funded by bigger government may serve as a better model of care.

2) More importantly however, I think the emphasis needs to be on what actually works. For this reason I find Braslow’s argument to deliver all but the final blow against the recovery model. Neo-liberalism be damned if the recovery orientation is working for people.

3) Ultimately, I think Braslow hit the nail on the head when he discussed “cycles of hope and despair” in mental health. Analogously, in politics this cycle might be seen as more of a pendulum swinging back and forth between liberal and conservative ideology. Conservatives shrink government, liberals expand it – and all social services hang in the balance. Liberals will say welfare will raise the lot of the poor, while conservatives will retort that stimulating the economy will have a trickle-down effect thereby raising the lot of everyone. In this sense, recovery may viewed historically as a result of a particularly large dose of conservatively minded government-shrinking, concurrent with the rise of effective pharmacotherapy. Importantly though, if the opposite situation were to arise in the future – huge increases in government spending and institutionalization coupled with breakthroughs in psychosocial therapies – wouldn’t that re-energize another cycle of hope? Couldn’t Braslow then make the converse argument that this ‘neo-institutionalization’ is not so different than we might think, and that it is actually based on values of fiscal liberalism? Protection begets protection in both cases… In this hypothetical case, as it is now, what is most important is not the theoretical and philosophical links between politics and mental health policy, but rather the empirical links between that health policy and beneficial outcomes for patients, consumers, recipients, dependents, protectees, or whatever other name we might chose to assign to those who need our help.



Clinical Bullying?

Olweus, D. (2013). School bullying: Development and some important challenges. Annual Review of Clinical Psychology, 9, 751-780. doi: 10.1146/annurev-clinpsy-050212-185516

I must admit, I have relatively little to contribute for this review.

First, I can see that the topic of bullying is somewhat related to clinical psychology, but I’m not convinced that it is more related that other tangential topics such as maybe ‘hazing’ or even ‘street racing’. Problematic behaviors to be sure, but seemingly not squarely within the realm of clinical psychology. In support of that view, almost a full page of the reference section is filled with references to Olweus’ own work – so much so that a disclosure statement may have been considered prudent. Or, maybe he’s one of only a few researchers doing work on bullying, and this is a representative overview of the field. I just don’t know. Additionally, the review reads more like a meandering academic autobiography than a targeted review of key concepts.

Regardless, there are a few interesting points in the review, so let’s take a look. So, while the history of the construct of bullying may be interesting in its own right, it doesn’t seem particularly helpful in this context. After reviewing the concept of ‘mobbing’ for several pages, Olweus concludes “…the terms mob and mobbing are not very useful in describing the phenomenon of bullying” (p. 755). Let’s hop right to a definition that might turn out to be somewhat helpful then: “bullying is a subset of aggressive behaviour…intended to inflict injury or discomfort upon another individual” (p. 756). In addition to this, the aggressive behaviour needs to meet the following three criteria: 1) intentionality; 2) some repetitiveness; and 3) imbalance of power.

Does this definition of bullying apply to cyber bullying too? you may be asking. Olweus says yes. Additionally, despite sensationalized media reports of rampant cyber bullying, Olweus points to prevalence statistics “data from all over the United States” (p. 766-767) indicating that cyber bullying is much less prevalent than traditional bullying. In addition, his own research indicates that while traditional cyber bullying has negative effects on psychosocial adjustment and well-being, the added effect of cyber bullying evinced negligible effects in addition to traditional bullying. This points to an overall negative effect of being bullied, regardless of modality.

Is Olweus’ Olweus Bullying Questionaire (OBQ) reliable and valid? Very little research has been published on the matter. Additionally, Olweus offers several reasons why we might expect discrepancies between answers on the self-reported OBQ and the peer nomination method. First, many forms of bullying are subtle or secretive in nature, and thus would only be picked up on self-report measures such as the OBQ. Secondly, the OBQ asks students to comment on the frequency of bullying behaviours, whereas the peer nomination method does not. Third, because bullying is often perpetrated by older students toward younger students, the peer nomination method (which apparently has only ever been used within a single classroom) would not pick up on this, while the self-report OBQ would. Similarly, peer nominations are often (inexplicably) limited to same gender nominees thus being insensitive to instances cross-gender bullying. Hmm. While these problems all seem readily fixable, Olweus points out that for these reasons, the OBQ will generally return higher levels of bullying than the peer-nomination method, though neither appears to be regarded as the gold-standard in the assessment of bullying.

Olweus then discusses the longitudinal fate of bullies and their victims. Bullies as assessed by their teachers and peers were overrepresented in crime registers up to the age of 24, and even more so for violent crimes. Victims on the other hand, are approximately twice as likely to exhibit depressive symptoms later in life. Both bullies and their victims have been identified as “overconsumers” of the health and social support systems.

So knowing all this, how do we stop bullying? Dunno. Only one large meta-analysis has been completed to date which included 30 separate programs with “scientific evaluations of reasonable quality” (p. 771). The review concluded that these programs were able to reduve bullying by 20-23%, though Olweus points out the considerable variation in results leading to doubt about any consolidated statement on the efficacy of bullying reduction programs as a whole. All of the programs in the review, with the exception of Olweus’ own Olweus Bullying Prevention Program (OBPP) were first time non-replicated trials. The OBPP has been replicated only five times. Thus there is a “desperate need for replication” (p. 771).

Finally, Olweus reviews the academic/political context of the United States in relation to why research on this topic was late to arrive in that country. I’ll skip that here.

In concluding, Olweus writes that he is particularly happy with the fact that bullying is now regarded not as simply a part of growing up, but rather a social and public health issue that can be solved or ameliorated with research-based intervention programs.

Despite my questioning the relevance of bullying to the field of clinical psychology, my most hopeful reflection on this review is in relation to Olweus’ comments on reasons to continue to research the topic. He states “A focus on bully/victim problems leads naturally to an examination of larger units or contexts (a multilevel or ecological approach): classes, schools, and communities and related aspects including teacher behavior, classroom norms, social psychological mechanisms, and school policy” (p. 775). Though the research to date is rather patchy at best, in my opinion wherever the tools and techniques of clinical psychology might be applied to understand and reduce human suffering, they ought to be.



Health and SES: Arbitrary and Non-Arbitrary Associations

Chen, E., & Miller, G. E. (2013). Socioeconomic status and health: Mediating and moderating factors. Annual Review of Clinical Psychology, 9, 723-749. doi: 10.1146/annurev-clinpsy-050212-185634

Low socioeconomic status (SES) has classically been associated with all types of negative psychological and health outcomes, and serves as the stereotypical example of the ‘third variable’ in psyc 101 courses everywhere in helping to differentiate true causation from mere correlation. One memorable example from my psyc 101 days: Drinking a glass of wine per day is associated with better health outcomes including longevity. Contrary to sensationalized media reports that might have you believe that it’s the wine causing the health benefits, another important factor to consider is the SES of the one who can afford that glass of wine each day. What other health promoting benefits might that high-SES wine-drinking person have access to or participate in that may be causing some of the observed effects?

In the same way that SES serves to boost outcomes on multiple levels at its higher end, it likewise serves to reduce outcomes at its lower levels. Low SES is not simply a lack of financial means, but rather a lack of ability to obtain all the things that money can buy – including health. Chen and Miller examine the effects of SES on health, and importantly go beyond simply listing the now-familiar associations between low SES and poorer overall health, and discuss several purported mechanisms of action. They also discuss several protective factors which have been found to be protective for low SES individuals, shielding them from the otherwise negative effects.

I want to start by looking at some of the basic associations between low SES and poor outcome discussed by Chen and Miller at the beginning of their review. Relative to people in higher SES groups, people from lower SES groups are:

2.5 times more likely to have repeat emergency department visits over a one year period

2.7 times more likely to have repeat hospitalizations over a 1 year period

3.5 times more likely to suffer activity limitations due to disease

More likely to live under worse physical conditions (greater exposure to toxins etc.)

More likely to compromise their own health via smoking or inactivity

The authors note that access to health insurance (in the US context) is certainly a contributing factor to these health discrepancies. However, they also point out that the pattern of discrepancy remains very much stable in countries with universal health coverage. Though I have no hard reference to support this claim, I have to add here that I suspect that while the pattern of health inequity between high and low SES groups may be stable across health care systems, the relative degree of disparity is likely exaggerated in the absence of universal health coverage, and dampened in its presence.

With this background in mind, Chen and Miller set out to review the literature regarding the relevant psychosocial factors affecting this discrepancy at a variety of levels: neighborhood, family, and individual. At the neighborhood level they identify exposure to violence, and subsequent declines in social capital (overall cohesion in a community) as factors related to poorer overall health.

At this point I want to comment briefly on an issue I’ve been having in reading several of these reviews. When reviewing associations made between two factors or variables from original research, I think it’s important to list all the relevant empirical/statistical context about that association. Otherwise the reader is left to wonder where on the scale between obvious and earth-shattering that empirical fact lies. Let’s take an example from the current review: “…exposure to violence is associated with increased morbidity from a variety of health problems” (p. 726). The information that we intuitively want from such a statement is a determination of causality about the source of poor health. However, in this case we are left to wonder what exactly that causal factor might be, as we are not given any information about what might have been controlled for in making that association. Upon conservative interpretation then, we are left with a vague association between exposure to violence and poor health that really didn’t need empirical expounding at all. It’s obvious. We could have observed it, or guessed it. I’m now calling these types of statements ambiguous associations. The first section of Chen and Miller’s review is full of them. To be fair though, in many cases, this is simply where the literature is at. Future studies will undoubtedly push forward toward closer approximations of causal statements about the relevant factors involved.

Ok, back on track. At the family level, they identify parenting, family conflict and routines as associated to poorer health outcomes. Specifically, lower SES parents are more likely to exhibit inconsistent controlling and restrictive parenting styles with their children, with a focus on obedience vs. independent thinking found to be more associated with higher SES parents. Higher stress due to low SES can also lead to irritability, and overall poorer quality of time spent between family members. Here’s another ambiguous association: “family conflict during childhood (retrospectively reported) predicts adult illness and mortality 13 years later” (p. 729). Seems reasonable, but again, what’s causing the poorer health outcomes?

In terms of individual factors, the authors list psychological characteristics and health behaviours as relevant factors affecting health. Steps toward a more causal understanding were empirically afforded in this section by statistically controlling for certain factors: “…the association between low SES and cardiovascular mortality as well as all-cause mortality was substantially reduced when psychological risk factors (including depression, hopelessness, and social support) were statistically controlled” (p. 730). Better.

I’ll end my rant on ambiguous associations and move to what I think are the more interesting parts of this review – potential biological pathways, and protective factors. Specifically, the authors use the example of asthma, and review evidence suggesting that low SES children with asthma (controlling for several other demographic variables!) are ‘primed’ to respond more aggressively to some allergic stimuli, and are less responsive to asthma medications, thereby providing a putative path to worse asthma related impairment. Furthermore, these biological pathways were found to be mediated by several psychosocial factors such as perceptions of the social world, and family support.

Finally, the authors outline several characteristics of particularly SES-resilient people: shifting and persisting. ‘Shifting’ involves a psychological reframing of environmental stressors in less threatening ways. The emphasis here is on secondary coping, or changing one’s own perceptions rather than the trying to change the state of the world. ‘Persisting’ refers to “enduring adversity by finding meaning in difficult situations and maintaining optimism about the future.” (p.736). Skimming over the details, Chen and Miller sum up: “In sum, low-SES individuals who accept and adapt to life stressors by shifting—i.e., engaging in cognitive reappraisal and emotion regulation—and persisting—i.e., finding meaning and retaining optimism—even in the face of obstacles—appear to have better health outcomes.” (p. 739).

More than simply psychological factors associated with resilience, it seems clear that ‘shifting and persisting’ embodies many virtues essential to successfully navigating tumultuous times and life events regardless of SES. Perhaps it is the case that shifting and persisting also helps those at higher levels of SES as well. Chen and Miller mention that shifting and persisting does not continue to provide protective effects at in groups of high SES individuals (p. 471) but do not provide much further discussion on the idea. Perhaps we should be looking not for protective effects, but facilitative effects in these groups. They describe a “toughening effect” associated with the shifting and persisting stance that “facilitates subsequent task performance and mitigates emotional reactivity” (p. 740). Intuitively, it is exactly this type of mental toughening that might serve to benefit people in all kinds of ways (I’m thinking of post #24 re: self-efficacy being highly related to quitting tobacco in the long-term). Moving forward, I’ll be interested to see how findings like this are followed up in the research literature, and how the findings can be translated into clinical practice.

One final thought here. I will also be interested to see how the interactions amongst low SES and psychosocial factors change over time in relation to health and psychological outcomes. Levels of absolute wealth have risen over the past several hundred years, bringing within the reach of the middle class what was once not even available to kings (refrigerators, automobiles, cellular phones, basic health care etc). This brings up an interesting question: As low SES groups continue to prosper (in absolute terms), will some of the psychosocial risk factors observed today be resolved? Or is it the case that it’s the relative status of low SES groups that perpetuates psychosocial risk factors? Might we quantitatively compare the high SES neighborhoods, families, and individuals of yesterday to their low SES counterparts today? Hopefully time and hard-won non-arbitrary associations will provide the answers to these questions.




Alcohol Use in Adolescence: Preexisting Differences vs. Toxicity

Jacobus, J., & Tapert, S. F. (2013). Neurotoxic effects of alcohol use in adolescence. Annual Review of Clinical Psychology, 9, 703-721. doi: 10.1146/annurev-clinpsy-050212-185610

Jacobus and Tapert review the literature on the neurotoxic effects of alcohol in adolescents. Overall, they report that binge drinking behaviors in youth are associated with poorer neurocognitive performance across domains, alterations in grey and white matter, as well as aberrant patterns of neural activation as assessed via fMRI techniques.

In order to properly disentangle the sometimes-messy contributions of causal factors from merely correlated factors, the authors were careful to only include research that properly conducted randomly controlled trials in which adolescents were randomly assigned to either a binge drinking group or a non-binge drinking group. Wait what? Really? No. In fact all of the research reviewed relied on samples of adolescents who had non-randomly assigned themselves to either binge drinking or not, for thousands of different reasons, I’m sure. And fair enough – I’m certainly not suggesting that such studies should have been carried out, but rather that any evidence of group differences in the absence of such studies is tentative at best. As you might have guessed, this is my main problem with the review. Let’s look at some of the associations:

Teens with an alcohol use disorder have smaller hippocampal volumes, with total hippocampal volume being positively correlated with age at onset, and negatively correlated with duration of disorder

Binge-by-gender interactions have been found in terms of cortical thickness. Female bingers had thicker cortices than female controls, while male bingers had thinner cortices that male controls

Teens with alcohol use disorders have smaller overall white matter volumes, and poorer white matter integrity compared to healthy controls

Teen females with alcohol dependence showed decreased BOLD response during a spatial working memory task with a dose-dependent effect for both alcohol use and hangover symptoms

Alcohol use in teens is associated with poorer neuropsychological functioning across several domains including attention and information processing, memory, visuospatial functioning, language abilities, and executive functioning

Etc., etc.

Where possible, the authors of the review point out facts from the data that would suggest that the findings are more in line with true effects of alcohol on neural development rather than mere selection factors from non-randomized samples of teens. E.g., dose-dependent effects, correlations with age at onset and duration of illness, as well as preclinical animal studies that likely did include random assignment. Also, the authors mention that several studies demonstrating significant effects of alcohol use employ ‘matched controls’ as a comparison group, but do not elaborate on how the participants were matched, leaving the reader to wonder whether all possible sources of variance between groups might have been controlled via their matching processes.

So, having read this review, I’m relatively certain that alcohol consumption and especially alcohol use disorders have *some* deleterious effects on the developing human brain, I’m less sure about each of the specific findings discussed. For example, what other factors might account for the differences between the groups that was not controlled for by the individual studies? Home environment? Abusive/neglectful parenting? Depression, anxiety, other psychopathology? Genetic risk? General intelligence, or academic performance? And what about the effects of the activities that the binge drinking teens are not partaking in? Could some of the observed differences between the groups be caused by the beneficial effects of the activities of the non-bingers?

Ultimately, despite an interesting review of the literature I think that Jacobus and Talpert are dead on when they say that future studies are needed to “…disentangle preexisting differences from toxicity…” (p. 716).



Therapy Tolerability vs. Self-Efficacy in Quitting Smoking

Schlam, T. R., & Baker, T. B. (2013). Interventions for tobacco smoking. Annual Review of Clinical Psychology, 9, 675-702. doi: 10.1146/annurev-clinpsy-050212-185602

Well, I suppose it’s a good thing that McKee & Weinberger didn’t specifically delve into methods of reducing tobacco use/abuse as Schlam & Baker cover that in their review. The review itself is quite straightforward, so I won’t spend a whole lot of time reviewing the nuts and bolts. Essentially, smoking is quite dangerous and is estimated to cause approximately 20% of all deaths in the US. The authors discuss several interventions for smoking that fall neatly into one of four stages from the ‘stages of change’ model (i.e. motivation / precessation / cessation / maintenance). Therapies discussed included motivational interviewing, smoking reduction, precessation interventions, brief advice, smoking cessation counseling, quitlines, behavioral (non-counseling) approaches, and smoking medications.

Each of these interventions are shown to have *some* effectiveness at their assigned level of the stages of change model. This is an exceedingly vague statement, I realize, but unfortunately it is consistent with the types of conclusions stated by the review itself. Additionally, too little research has been conducted to date to accurately identify potential mechanisms of action for these therapies.

What grabbed my interest most in this review was the section related to interventions designed to maintain smoking cessation over time. First, the review mentions that extended counseling (up to one year post cessation) can have beneficial effects, but that this process may not be transferrable to general practice due to the unwillingness of many smokers to commit so much time to the process. Now, I understand that tolerability is an important factor in all forms of therapy, however, something about this seems rather ‘defeatist’ to me. To my thinking, if there’s an effective way of reducing psychopathology or otherwise augmenting positive outcomes it needs to be developed to its maximum therapeutic potential. Stated differently, it seems that finding an effective treatment, but discarding it on the basis of low tolerability, or convenience in real-world samples is tantamount to giving up on the search for effective psychotherapies altogether. Psychotherapy is difficult because it attempts to ameliorate multifaceted and deep-rooted psychological problems. Both in theory and practice, I think constraining our empirical search to easy solutions for hard problems is a step in the wrong direction.

In fact, I was interested to see that a sense of self-efficacy was most strongly related to long-term abstinence versus other variables like perceived social support (discussed on p. 691-692). In light of this, it could be the case that extremely unpalatable psychotherapies might actually work to foster a sense of self-efficacy which could then subsequently have a beneficial effect on helping to curb addiction. In this sense, providing people with relatively easy therapies like a nicotine replacement patch might be at cross-purposes to fostering a sense of inner strength which ought to be more strongly correlated with quitting over the long term. This is entirely speculative, but it seems reasonable, and certainly untested as of yet. What other interventions might serve to foster peoples’ sense of self-efficacy in such a way as to have beneficial effects on quitting behaviour.



‘Nudging’ People from Alcohol Use

McKee, S. A., & Weinberger, A. H. (2013). How can we use our knowledge of alcohol-tobacco interactions to reduce alcohol use? Annual Review of Clinical Psychology, 9, 649-674. doi: 10.1146/annurev-clinpsy-050212-185549

The first thing that struck me about McKee & Weinberger’s review of Alcohol-Tobacco interactions was its scope. It remains unclear to me why the focus of such a review would be squarely on reduction of alcohol use, rather than the reduction of both alcohol and tobacco use. Although the review doesn’t reveal the answer to this mystery, it has some pretty interesting things to say about the interaction between alcohol and tobacco, and how that information might be used in curbing alcohol use in the general population.

First, the authors discuss exactly is at stake. Estimates indicate that approximately 8.5% of the population meets criteria for alcohol abuse of dependence, and that excessive alcohol use is the third leading cause of death in the US, and is additionally associated with a host of adverse consequences such as hypertension, gastrointestinal bleeding, sleep disorders, major depression, hemorrhagic stroke, cirrhosis of the liver, cancer, unintentional injuries and violence. With this background in mind, the authors lay out their central thesis: “We contend that potentiated reinforcement is at the core of alcohol and tobacco interactions and it is possible to use this knowledge to reduce alcohol use” (p. 650).

In support of this thesis, McKee & Weinberger discuss a variety of associations between alcohol and tobacco use plucked from the literature:

Current smoking is associated with a greater number of alcoholic drinks per day, and consumption days per month, greater severity of alcohol dependence, and greater alcohol withdrawal symptoms.

Smokers are slower to mature out of patterns of heavy drinking

Higher rates of smoking occur in populations of past-year drinkers

Comorbid smoking and alcohol dependence is more probable in treatment samples than epidemiological samples

Health risks of comorbid alcohol and tobacco use are ‘supermultiplicative’ in some cases increasing risk of cancer many-fold for combined users.

Daily smoking increases the risk of meeting criteria for hazardous drinking and alcohol use disorders

Needless to say, there is an ‘association’ between drinking and smoking, and also between not drinking and not smoking. But why? The authors suggest that nicotine decreases subjective intoxication via the nicotinic acetylcholine receptor system via excitatory cholinergic input into the mesolimbic dopamine system. This attenuates the subjective properties of alcohol, allowing more alcohol to be consumed, and also to be experienced as more reinforcing.

However, the authors state that empirical findings are rather mixed with regard to the attenuation effect of nicotine on subjective intoxication by from alcohol… which seems to be a central tenet of their thesis. Some studies have even found increasing subjective intoxication in the context of nicotine.

Despite this, the authors move to discuss 1) whether or not smoking status might be used as a clinical marker for potential alcohol misuse; and 2) whether tobacco policies might be used to reduce alcohol use; and 3) whether nicotinic acetylcholine based medications might be used to treat alcohol use disorders. In brief, their answers are 1) yes; 2) yes; and 3) yes.

Some thoughts:

1) The authors suggest that smoking status be used as a “red flag to help identify primary care patients at higher risk for alcohol misuse and as a helpful mnemonic for alcohol screening in general”(p. 657). While I generally agree that this might be the case, I’m not certain that there is much practical relevance in this suggestion. We know that smoking and heavy drinking commonly occur together, but the absence of smoking behaviour does not (I think) give a clinician clear rationale to not ask about potential patterns of problem drinking. Also, if the ultimate goal is to curtail drinking behaviours (as per the focus of the article), why not simply prescribe that clinicians ask about problem drinking behaviours directly, rather than smoking behaviours as a probabilistic proxy for drinking behaviour? My impression is that screening for substance use/abuse of all kinds is generally lacking in clinical practice, and that routine screening across the board would perhaps be more beneficial that providing if-then rules for screening that could only serve to miss certain cases.

2) Both smoking bans and tobacco taxes have been empirically associated with decreased drinking behaviors. Both seem reasonable in their own right, and any reductions in drinking behaviours would seem to be an added bonus. I do however question the degree to which the field of clinical psychology has influence over either of these variables in wider society. That’s not to say that I don’t think these avenues should be pursued. To the contrary, I think much of what clinical psychology does is reactive at a personal level, where much greater effects on mental health could be achieved via proactive population based interventions such as this. Once we know something about the variables that affect psychopathology, it behooves us to at least try to ‘nudge’ people towards making better informed choices about their own health and wellness. In fact, one of my more ethereal interests in clinical psychology is how we might become more like ‘choice architects’ or behavioral economists (or at least collaborate in multi-disciplinary teams) to effect the kinds of mental health outcomes we typically seek on a much, much smaller scale.



Biomarkers of Alzheimer’s Disease, and Social Markers of Stigma

Risacher, S. L., & Saykin, A. J. (2013). Neuroimaging and other biomarkers for Alzheimer’s disease: The changing landscape of early detection. Annual Review of Clinical Psychology, 9, 621-648. doi: 10.1146/annurev-clinpsy-050212-185535

In reviewing the quickly changing landscape of reliable biomarkers for Alzheimer’s disease (AD), Risacher and Saykin begin with an overview of the relevant neuropsthological processes known to be involved in AD. The two ‘hallmark’ biomarkers for AD are amyloid plaques (extracellular aggregations of amyloid-beta peptides), and neurofibrillary tangles (insoluble  filamentous structures resulting from hyperphosporylated tau protein). While the underlying causes of these pathological processes are largely unknown, the temporal links between these processes and effected brain regions are now beginning to be better understood. Amyloid accumulation occurs first, followed by plaque deposition and neurofibrillary tangles, which leads to neuronal injury and death. In terms of brain regions affected, the medial temporal lobe including the entorhinal cortex, hippocampus, amygdala, and parahippocampal cortex are affected first, usually followed by regions of the cortex itself across the parietal, temporal, and frontal lobes. Importantly, the occipital lobe is largely sparred in the process of neural degeneration in AD, as are the primary sensory-motor regions.

Perhaps not surprisingly, these effects on underlying neurobiology manifest in clinical symptoms such as memory impairments in episodic and sematic domains, language deficits, as well as deficits in executive functioning. Mild cognitive impairment or MCI is thought to be a precursor to AD. Making its official debut in the DSM-5, MCI (officially termed Minor Neurocognitive Disorder) is similar to AD, but less severe in associated deficits, and does not impact social or occupational functioning, which the diagnosis of AD (now officially termed Major Neurocognitive Disorder) requires. All those meeting requirements for AD very likely met criteria for MCI at some point or other, however, not all those with MCI will progress to AD or other forms of dementia. Risacher & Saykin point out that conversion rates (to AD) have been estimated at 10-15% for those with MCI, compared to 1-2% for those in the general population.

Importantly, the authors point out here that AD is still diagnosed clinically, and that the availability and/or reliability of biomarker-informed diagnosis of AD is not yet a reality. Despite this, the field appears to hold promise in bringing new technologies and techniques to market. These technologies are especially important given the enormously large time gap between the onset of biomarkers for AD, and the eventual clinical presentation (up to three decades!). In this sense, reliable identification of biomarkers should massively aid early detection and prevention of AD.

Other biomarkers discussed by Risacher and Saykin include widespread brain atrophy with occipital and sensory-motor sparing on via structural MRI; decreased brain activation in response to various neuropsychological tasks in brain regions associated with AD via fMRI; decreased connectivity between task-related and resting-state networks via measures of functional connectivity; and impaired integrity of white matter tracts via investigations using diffusion tensor imaging. Interestingly, but again, not surprisingly similar investigations with MCI populations have revealed intermediate markers of impairment between healthy controls and AD patients in terms of general atrophy, glucose metabolism, amyloid accumulation, and functional brain activation.

From this landscape of empirical findings, the authors then move on to discuss possible routes for prevention, and identify several particularly high-risk populations: 1) clinically normal adults who have abnormal amyloid accumulation; 2) genetically high risk individuals; and 3) older adults with cognitive complaints. Finally, a nice hypothetical model of biomarker change across AD progression is presented (figure 5), and several factors are discussed that are thought to speed up the progression the overall process (e.g., being a APOE e4 carrier), or slow down the overall process (e.g., high cognitive reserve).

Overall, I think the authors present a nice review of the literature here. I have only two additional thoughts, one related to the content of the article, and one not. First, in discussing the impact of various factors that might serve to shift the curves of biomarker accumulation across disease progression, the authors discuss genetic risk (APOE e4 carrier) and high cognitive reserve as having hypothetically equal by opposite effects on biomarker progression. I can see how being at genetic high risk might predispose someone to, for example, develop higher concentrations of amyloid proteins at an earlier stage of life, however I have a harder time imagining how high cognitive reserve might counteract this accumulation. My understanding of cognitive reserve was that of being better able to cope with neural and cognitive loss, rather than being a neural or cognitive protective factor per se. Perhaps this is what the authors meant to convey.

Finally, I wanted to comment on the shift in terminology in the DSM-5 away from the term ‘dementia’ to the new term ‘neurocognitive disorder’. The DSM-5 itself does not discuss the reasoning behind this shift in terminology, but reports I’ve read elsewhere seem to suggest that the change was made to reduce the enormous stigma attached to the term ‘dementia’. I find that I’m resistant to changes like this on the basis that the title of a particular disorder ought not to be based on its palatability in clinical or common usage. ‘Dementia’ has developed a stigma not because of its particular constituent phonemes, but because of the very human emotions stirred by the psychological trauma and loss it stands to represent. Replacing the term with a new one does not serve to eradicate the stigmatization process, but rather renew the process with a new term, all the while causing conceptual confusion about the comparability of the terms. We’ve seen this process happen with other unpalatable terms, such as retardation à handicapped à disabled à differently abled and so on, each taking on the stigma from its predecessor as quickly as the common vernacular could catch up. My point is not that we should simply embrace the stigma, but rather that we should spend ourselves in fighting for that worthy cause head on, rather than continually flee from it.



Eating Disorders: Genes are Involved

Trace, S. E., Baker, J. H., Peñas-Lledó, E., & Bulik, C. M. (2013). The genetics of eating disorders. Annual Review of Clinical Psychology, 9, 589-620. doi: 10.1146/annurev-clinpsy-050212-185546

Trace et al. review the research on genetics and eating disorders. Much of it is rather straightforward and unsurprising. And, unlike almost all of the other articles that have discussed expected changes in the DSM-5, those discussed by these authors were actually implemented in the new manual! Criterion D (amenorrhea) for anorexia nervosa (AN) was removed due to lack of utility, and binge eating disorder (BED) was finally added as an official diagnostic category.

The authors discuss each of the three main eating disorders in turn, and review literature on the genetic contributions to the serotonergic system, the dopaminergic system, the opioidergic system, appetite regulation, food intake, weight regulation, as well as ‘other’ genes. Despite the implication of genetic components of each of these systems to the distal expression of formal eating disorder symptoms, the conclusions drawn are rather disappointing:

Anorexia Nervosa (AN):

“…it is difficult to make definitive conclusions about the specific genes that influence risk for AN, and GWAS with much larger sample sizes and homogenous definitions must be conducted in order to fully elucidate the genetic architecture of AN” (p. 597).

Bulemia Nervosa (BN):

“Although twin studies suggest consistently that BN is influenced by genetic factors, molecular genetic studies have not yet been adequate in scope or design to identify susceptibility loci” (p. 599).

Binge Eating Disorder (BED):

“Although family and twin studies suggest the role of genetic factors in BED, candidate gene studies have not clearly confirmed the involvement of any one gene or genetic pathway” (p. 602).

Huh. So genes are involved, but we’re not quite sure how. Ok. Well what else is happening in this article? One thing I found interesting was the way in which the review was organized. Despite highlighting the usefulness of focusing on the symptoms of eating disorders rather than the diagnostic categories themselves (p. 602-603), the authors chose to organize the preceding review by diagnostic category rather than genetic contributor (e.g., serotonin, dopamine etc.). Although it’s very likely that I’m just reading too much into this, organizing the review in this way seems to make more intuitive sense, and thus seems to underscore the psychological utility of diagnostic categories. As a discipline, we seem hesitant to leave the relative comfort of our airy DSM categories for the brave new biologically reductive  world of trans-diagnostic symptoms and endophenotypes. And fair enough, it’s a scary world down there.


For the sake of brevity, I’ll bypass the quite interesting topics of gene-by-environment correlations, and epigenetics. I want to talk about some of the ideas presented in the clinical implications section – namely answering what I’ll call the ‘So what?’ question often associated with genetic study of clinical disorders. I’ve heard a number of people throughout my student career decry the utility of genetic implications in clinical work. So what? they ask. How can knowing that you are particularly predisposed (or not) to a particular type of psychopathology help you in treatment? Well, Trace et al. tackle that question head on in this section. They offer the following suggestions spanning both prevention and treatment:

1)      Models of genetic risk might be used to target high risk populations

2)      Models of genetic risk might be used in a psychoeducational manner in helping individuals or families understand the “causes” of their illness.

3)      This kind of genetic reframing might then help to reduce stigma associated with the disorder in wider society

4)      By developing family trees showing how the disorder aggregates in families, individuals may gain a better understanding of their sensitivities to certain environments, help them understand why recovery is so challenging, and help them develop a long term maintenance plan

Now, I’m not sure that I agree with all these points, I just find it interesting that they are enumerated here – and think that each definitely warrants further investigation. I’ll be interested to see empirical validation of each of these tenets over time.



On the Spectrum…

de Lacy, N., & King, B. H. (2013). Revisiting the relationship between autism and schizophrenia: Toward an integrated neurobiology. Annual Review of Clinical Psychology, 9, 555-587. doi: 10.1146/annurev-clinpsy-050212-185627

De Lacy and King set out to provide sufficient evidence to convince the reader that far from being disparate categorical disorders, autism (ASD) and schizophrenia (SCZ) are actually simply different expressions of the same disease process. They begin by discussing overt similarities between the disorders: both share similar prevalence rates, gender distributions though differing in modal time of onset. Relatively few studies have looked into the association between child onset ASD and later SCZ, the extant literature points to much higher rates of SCZ in ASD samples than would be expected from the general population – somewhere between 12% and 50%. Perhaps not surprisingly, the same association was found in samples of SCZ with 52-60% of having the additional diagnosis of ASD. Other endophenotypes are common to both ASD and SCZ such as deficits in theory of mind, smooth pursuit and saccadic eye movements. Additionally, both disorders are found to be quite heritable ~80% for SCZ, and ~39-95% for ASD.

The authors then move to cover the specific genetic conditions linked to both ASD and SCZ, and provide estimates of the percentage of the disordered population who also have a positive ASD or SCZ diagnosis. For example, from table 4, we can see that in samples of patients with Prader-Willi syndrome 50% might be expected to have ASD, and 85% might be expected to have SCZ. I’ll be honest, much of the discussion of the genetic factors involved was methodologically over my head. The following phrases simply are not in my vocabulary: ‘transcription silencing’, ‘promoter hypermethylation’, ‘permutation carriers’ etc. etc. I gather I’m not alone here, because the authors provide a nice summary in plain English: “There appears to be no question that a phenotypic continuum links the SCZ and autism spectra; moreover it incorporates neuropsychiatric deficits associated with all of the classic neurodevelopmental disorders” (p. 562-563)

Unfortunately in discussing genome wide association studies (GWAS) and particular candidate genes, the language drifts back the incomprehensible: ‘copy number variants’ (CNV), ‘haploinsufficiency’, ‘stochiometric effects’. Hmm. Let’s find another summary. “Taken together, GWAS and CNV studies increasingly undermine the traditional psychiatric diagnostic specificity of ASD and SCZ” (p. 564). Better.

Following the genetic (and epigenetic) linkages between ASD and SCZ, de Lacy and King discuss the surprisingly similar effect of each disorder on neurodevelopment, specifically head and brain size. Essentially, head circumference is mostly normal or slightly smaller than normal at birth, followed by a period of ‘postnatal overgrowth’ peaking at one year. In later childhood and adolescence an opposite process occurs involving neuron loss, degeneration, and cortical thinning. Thus, brain size normalizes in both ASD and SCZ, but leaves abnormalities in cortical surface morphology resulting in “…altered whole brain parcellation, which by its nature causes local connectivity deficits and dyscoherence.” (p. 569).

And that’s about it. Well, that’s all I can make sense of anyway. The take home points appear to be that multiple lines of fairly reductionistic (in a good way) biologically based evidence support the notion that schizophrenia ought to be considered a part of the autism spectrum – or more formally, “Evidence supports the conclusion that ASD and SCZ are both disorders of cerebral specialization originating in the embryonic period” (p. 576). The implications here appear to be 1) identification of a plethora of biological targets at the molecular, synapse and systems levels; and 2) strong support for the cross-diagnostic/biomarker/endophenotypic/non-categorical view of psychiatric disorders, at least for these two categories.

It will be interesting to see how the field deals with developments like this. The safer and easier option might be to simply write these new discoveries into the traditional categorical criteria for the disorders. At several conferences I’ve been to, I’ve heard rumblings about exactly this – adding cognitive impairment to the SCZ diagnostic criteria. A bolder option would be ditching the criteria altogether.



On Moving Personality Disorders to Axis I or Axis None

Links, P. S., & Eynan, R. (2013). The relationship between personality disorders and axis I psychopathology: Deconstructing comorbidity. Annual Review of Clinical Psychology, 9, 529-554. doi: 10.1146/annurev-clinpsy-050212-185624

In reviewing comorbidity between axis I and II disorders, Links & Eynan debate the merits of collapsing across these axes to better reflect high rates of comorbidity in real world settings. Again starting with the conclusion, they decide that there is not currently sufficient evidence to collapse across axes, or delete select disorders to reduce comorbidity, but that borderline personality disorder may evince sufficient evidence to be moved to axis I. Alas, looking through the DSM-5 reveals that the axes have been removed altogether making their suggestions rather moot in practice. However, the new DSM appears to have some trouble letting them go. Page 16 of the DSM-5: “The DSM-5 has moved to a nonaxial documentation of diagnosis (formerly Axes I, II, and III), with separate notations for important psychosocial and contextual factors (formerly Axis IV) and disability (formerly Axis V)” Interestingly, we can get the authors’ opinion on the removal of the multiaxial system from the DSM-5 even before it was published! “One could argue that given our limited progress with understanding comorbidity between clinical disorders and PDs, a major conceptual change such as removing the separate axes will only serve to divert our attention from a more substantive and robust understanding” (p. 547). Oops.

However, the review raises some interesting points. First, the authors ‘deconstruct’ the term comorbidity, reviewing several deconstruction schemes whereby comorbidity is divided into more specific types indicating something in addition to simple co-occurrence (chronicity, course, causal status etc.). There are literally too many comorbidity-related terms introduced in this section to review here. In general, the thrust of thinking in this area seems to be increased attention given to the clinical and psychopathological relevance of the co-occurrence of disorders rather than simply their co-occurrence. They cite the opinions of several others in proposing that “overlap among disorders should not be referred to as comorbidity when the overlap is purely at a descriptive level”(p. 533).

After deconstructing comorbidity in this way, the authors move on to discussing whether or not several personality disorders ought to be moved to the now non-existent axis I. Due to the removal of multiaxial system in the DSM-5, the authors suggestions here are now of little import, though the review of the literature raise some interesting points about each of the personality disorders discussed.

One issue I had with this review was the somewhat inconsistent application of a critical lense to the concept of comorbidity. For instance, after thoroughly deconstructing the construct over several pages, and pointing out the need for a greater degree of specificity they state: “Comorbidity results in far more significant functional impairment than the individual effects of the disorder” (p. 534). Well, which type? I realize that it is this exact kind of ambiguity the authors are aiming to excise from common clinical usage, however it seems that here they are perpetuating the ambiguity. There needs to be some kind of latin term for this. Deconstructing a construct so useful or ubiquitous that you mistakenly use it subsequently without qualification, thereby creating more ambiguity than would have originally resulted from its usage. Hmm, I’ll have to think about that one for a while.

Similarly, if the term ‘comorbidity’ ought not to be used without qualification, neither should terms like homotypic continuity or heterotypic continuity (as the authors do on p. 542), as they bring with them all the same ambiguities as comorbidity. Simply to say that one disorder remains the same, or somehow transforms into another disorder is not saying much.

Finally, another issue I had was with the rather politically charged reasons to move borderline personality disorder to axis I. Links & Eynan state: “moving BPD to Axis I will decrease the underutilization of the diagnosis, encouraging the development of training programs, help establish reimbursement policies, and hopefully stimulate more funded research into the disorder” (p. 545). While all worthy considerations, these strike me as all the wrong motivations to carve nature at any particular joint. I understand that the DSM is in the awkward position of walking the line between being a collection of objectively hard-won empirical truths, and a manual for insurance and liability purposes, but strategically planning its organization for the later masquerading as the former is special kind of unhelpful scientific deception. Insofar as it is possible, the DSM should represent the facts, and politically oriented efforts to increase the lot of any particular disorder should take place outside of its pages.



Personality Disorders: Stability and Change in DSMs

Morey, L. C., & Hopwood, C. J. (2013). Stability and change in personality disorders. Annual Review of Clinical Psychology, 9, 499-528. doi: 10.1146/annurev-clinpsy-050212-185637

Personality is relatively stable over time. We know this not only via research findings, but also intuitively – we expect the people we know to remain much the same over the course of their adult lives. So, almost by definition, a disorder of personality should also be relatively stable over the lifetime, thereby causing a variety of philosophical and practical problems for treatment. If personality disorders are stable by definition, what hope could there be of ‘treating’ them? If on the other hand, a personality disorder was treated successfully, might it be the case that the presenting issue(s) was/were never personality based to begin with? This is starting to sound silly already. To make sure we’re on track here, let’s start with the DSM-IV definition of a personality disorder (PD): “an enduring pattern of inner experience and behavior that deviates markedly from the expectations of the individuals culture, is pervasive and inflexible, has an onset in adolescence or early adulthood, is stable over time, and leads to distress or impairment” (p. 500)

Given this definition, Morey and colleagues set out to cast doubt on the assertion that stability over time is an accurate diagnostic criterion for PDs. For maximum clarity here, let’s start with their conclusion and then work backwards through their reasoning: “It is concluded that no single answer can be given to the question ‘How stable are PDs?’” (p. 499). Huh. Well that’s disappointing. However disappointing their ultimate answer to this question might be, I think their approach to answering it is quite illuminating and casts some light on some oft overlooked issues in the assessment and treatment of PDs.

The authors posit five separate sources of variance in the PD literature that serve to obscure any consensus on the ‘true’ stability of PDs. These include: 1) the definition of the construct itself (i.e., categorical vs. dimensional); 2) the type of stability considered (they list five types. Yes, five types); 3) the particular assessment measures used (self-report vs. interview etc.); 4) the reliability of the assessment measures; and 5) the sampling methods used.

While reading through descriptions of each of these, I came to think of them like lenses. Each one individually obscures the truth only slightly, though jointly make an evaluation of some absolute truth all by completely impossible. I’ll comment briefly on each of these lenses.

First lense – the definition of the construct itself. The authors point out that PD’s have been considered categorical in nature since the DSM-III, but that greater stability is achieved when they are considered in a dimensional model, as much information is lost when PDs are scaled as categories. However, defining them as categories has several advantages including ease of communication, familiarity to the mental health field, and links to a large body of research based on the same category. Morey et al. mention that the DSM-5 proposal advocated a more dimensional view of PD’s, however, a quick skim through the final published version reveals that very little has changed from DSM-IV. While we’re on the topic, I have two thoughts on the categorical versus dimensional debate. The first is that I think we as a field need to realize that considering disorders as categorical or dimensional is more about us and our own psychology than it is about the actual nature of the world ‘out there’. Mental disorders are both categorical and dimensional and I think we should move to adopting whichever epistemological position provides the greatest utility for the task or issue at hand, and discard the ostensible inconsistency created by doing so. Second, I think we also need to have an appreciation for the possibility that we are all incorrigible category-mongers. Think about it – we’ve even turned the issue of category versus dimension into a categorical issue. We simply seem ill at ease with dimensions.

Second lense – type of stability. Morey et al. describe five separate types of stability: differential, absolute, interindividual, structural, and ipsative. I’ll define each here for my own benefit. Differential stability is the rank ordering of individuals in a particular sample. Thus, high retest coefficients would indicate that individuals the rank ordering of individuals has remained constant over time – but says nothing about the consistency of the group as a whole. For that, researchers have used the metric of absolute stability, or simply the average change observed in a sample over time. Interindividual stability captures the degree of variability around the average course over time. Structural stability captures how well the measured traits hang together over time, and ipsative stability captures the degree to which a constellation of traits hang together in within an individual over time. A few thoughts on this. In light of the other obscuring lenses here, such fine-grained detail-oriented look at these different types of stability almost seems like a little bit of overkill. Knowing the exact type of stability that a PD exhibits over time says little in the absence of confidence in sampling and assessment methods, their associated reliabilities, and even the definition of the construct itself! Stated differently, it seems odd to present graphs of conceptual possibilities for stability over time (figure 1) without presenting graphs of actual data. However, I think that these types of stability would actually be quite helpful in characterizing the nature of substance use disorders and whether or not people ‘mature out’ of addiction as discussed by Heyman (see post #2). These types of stability are probably good thinking tools for the problem I was grappling with in that post – the smoothness of remission statistics at a population level versus the cyclical course of remission and relapse at the individual level. In these terms, what’s missing from Heyman’s analysis is data on the ipsative stability of addiction. It’s not whether addiction is absolutely unstable over time that is most relevant to the ‘maturing out’ question, but rather ipsative stability.

Third and fourth lenses – instrumentation and reliability. I’ll skip these ones except to say that in extended study designs often employed in this type of research (up to 10 years in some cases!), the reliability of a measure would seem to take on several more degrees of complication. It’s one thing to say that a measure such as the WAIS-IV has a certain degree of reliability over weeks months or years, however a decade later there will be problems due to cultural factors such as the Flynn effect. The same must follow for measures and interviews for PDs. Stated differently, the reliability of a measure can always be calculated, but the concept of reliability would seem to have a kind of psychometric best before date.

Fifth lense – sampling. As if the above issues weren’t enough to shake you confidence in a solid answer to the question of stability of PDs over time, Morey et al. point out that the literature has used a wide variety of samples which have undoubtedly contributed to the variance in findings regarding stability. Sample size, sampling intervals, sample age and clinical status will all impact estimates of stability over time.

Ultimately, I think these lenses are more interesting that the conclusions they suggest – “stability is not an especially compelling differentiator of PDs and clinical disorders” (p. 520). No matter though, the DSM has remained stable in a variety of ways, one being the definition of a PD. Ctrl-C Ctrl-V’d right from the DSM-IV: “A personality disorder is an enduring pattern of inner experience and behavior that deviates markedly from the expectations of the individuals culture, is pervasive and inflexible, has an onset in adolescence or early adulthood, is stable over time, and leads to distress or impairment” (DSM-5, p. 645)



When Treatment Informs Etiology

Mueser, K. T., Deavers, F., Penn, D. L., & Cassisi, J. E. (2013). Psychosocial treatments for schizophrenia. Annual Review of Clinical Psychology, 9, 465-497. doi: 10.1146/annurev-clinpsy-050212-185620

The sheer number of approaches to the treatment of schizophrenia always amazes me. The way in which a disorder is treated, implicitly speaks to a proposed etiological theory of that disorder, and by this logic, there appears to be no consensus in sight for schizophrenia. Different types of treatments for schizophrenia also seem to attract strong adherents who would have you believe that looking at the disorder any other way would in fact be foolish, or misguided at best. What’s more amazing is that adherents on both sides of such a divide can often produce decent evidence in support of their point of view, which in turn is disregarded or criticized by the other side.

One prominent example here is the use of CBT for psychosis. Some claim that prevalence rates for positive psychotic symptoms (i.e., delusions & hallucinations) are actually quite prevalent in the general population. This view lends itself nicely to the use of CBT in psychosis in order to help people suffering from full blown psychosis to simply adjust their thinking, feelings and behaviours related to these symptoms in order to achieve both symptom reduction as well as a sense of subjective wellbeing. In contrast, others have pointed out the rather dismal effect sizes for CBT for the treatment of schizophrenia in the most rigorously controlled trials, suggesting a kind of pervasive bias in which only rather sloppily conducted  trials achieve clinically significant effects. For these people, anti-psychotic medications might constitute the first line of treatment, casting the etiology of schizophrenia in a completely different light. Still others, more recently have stressed the importance of cognitive and social cognitive deficits, again shifting the etiological essence of the disorder to something more inherently neurocognitive and/or social-cognitive in origin. Certainly, schizophrenia is all of these things, but it remains interesting that such a wide variety of treatment modalities cast in such vastly different lights.

All of this necessitates a review of psychosocial treatments for schizophrenia, and Mueser et al. take up the challenge well, dividing their summarization efforts into evidence based practices, and promising practices. In my own review of this topic, I looked at social skills training, cognitive behavioral therapy, cognitive remediation, and social skills training. Mueser et al. cover quite a bit more than this, though I’ll stop short of recapitulating their summaries here.

The one issue I did want to address here is the idea of using outcomes that resemble those tasks used in training – or essentially teaching to the test. Using a training task as an outcome measure (or vice versa) is a practical shortcut that essentially serves to undermine the validity of any proposed enhancements in functioning. Intuitively, what really matters to the betterment of the individual sufferer is the remediation or elimination of those factors most distal to the domain of training or remediation. In other words the treatment needs to generalize to other areas of the patient’s life, not directly addressed in any given psychosocial therapy. This perspective is certainly not new – but certainly one that should always be kept in mind in designing and testing potentially beneficial treatments. In light of this, I was surprised to see ‘supported education’ and ‘supported housing’ being included in this review. Each was associated with improvements in, you guessed it – educational, and housing related outcomes respectively. Not that these outcomes are not important in their contributions to easing the burdens of the disorder, just that I feel they fall slightly outside the domain of ‘treatment’, perhaps sharing more in common with the idea of ‘support’. However, these kinds of support, intuitively have the ability to affect (if not generalize to) quite distal areas of functioning for the individual, thereby capturing the essence of what researchers are trying to find in effective treatments. Also intuitive though, is the notion that simply providing housing and education for patients with schizophrenia in the absence of other treatments would likely not result in much beneficial effect.

I think I need to do some more thinking on the commonalities and differences between treatments and supports. Might all forms of treatment be considered special cases of support? And what might this mean for the perceived etiologies of schizophrenia?



Targeting Full Disclosure in Cognitive Interventions for Schizophrenia

Fisher, M., Loewy, R., Hardy, K., Schlosser, D., & Vinogradov, S. (2013). Cognitive interventions targeting brain plasticity in the prodromal and early phases of schizophrenia. Annual Review of Clinical Psychology, 9, 435-463. doi: 10.1146/annurev-clinpsy-032511-143134

I’ll start here with what this article leaves until the very end. Sophia Vinogradov “is a paid consultant of Brain Plasticity Inc. She is also a paid consultant to Amgen, Genentech, and Hoffman-LaRoche” (p. 457). How might these ties colour or otherwise influence the interpretation of findings and suggestions in the publication? I have no idea. Perhaps not at all, or perhaps completely. What I do know is that having such a disclosure statement has the immediate effect of making me less confident in the content of the article, if only because it is the conservative position to take while not fully comprehending the importance of these potential conflicts of interest. A quick check through the remainder of vol. 9 of the Annual Review of Clinical Psychology did not reveal any other such disclosure statements. At least none that the authors were “aware of”.

Ok so with this mind, what is the actual content of the article? Fisher and colleagues discuss the recent paradigm shift in the treatment of schizophrenia away from reactive therapies to early detection and intervention. They then distinguish between the term “prodrome” (which by definition assumes transition to full psychosis) from the terms clinical high risk (CHR), and ultra high risk, which make no such assumption. Thus while it would be most ideal to conduct research on prodromal individuals, uncertainty about conversion to full psychosis necessarily limits study to these CHR participants who have greatly increased likelihood of developing the disorder.

Several lines of research point to neuro-developmental abnormalities in the CHR population including aberrant (overly aggressive) patterns of synaptic pruning, possibly leading to grey matter reductions, and less than age-appropriate white matter increases possibly leading to reduced functional connectivity between key brain regions, and inefficiencies in general information processing. Additionally, stress can play a key role in the development of psychotic symptoms throughout the prodrome, via dysregulation of the hypothalamic-pituitary axis (HPA). Ordinary functioning of this system involves cells in the hypothalamus secreting corticotropin-releasing hormone, which then leads to adrenocorticotropic hormone secretion by the pituitary gland, which then stimulates the release of stress hormones like cortisol. In a negative feedback loop, cortisol then signals to the hypothalamus (via glucocorticoid receptors in the hippocampus) to cease production of the original corticotropin releasing hormone. Got that? Chronic stress can actually have the effect of mimicking psychosis, and can even enhance fear through its effects on the amygdala. Additionally, chronic stress can damage the negative feedback loop of the HPA leading to a failure to effectively shut down the stress response. A corpus of literature focused on MRI data support this notion, showing structural differences between HPA-associated regions in CHR youth versus controls. Similarly, fully psychotic patients who reported childhood trauma were shown to have increased negative affect and psychotic symptoms in response to daily stress compared to psychotic patients without a history of trauma. For these reasons, stress reduction and coping have become critical targets for intervention in CHR youth.

With this rationale, Fisher and colleagues review several types of therapies for CHR youth, including cognitive behavior therapy (referred to as just cognitive therapy in the article), and cognitive training. A few thoughts: Fisher et al. reviewed Wykes et al.’s (2008) meta-analysis regarding the effects of CBT in the treatment of schizophrenia. They summarize by saying that CBT for psychosis “has a beneficial effect on positive symptoms, negative symptoms, mood, and functioning” (p. 443). However, having read and reviewed the Wykes et al. meta-analysis for my own review, I think this summary obscures many important findings from that analysis. Wykes et al. investigated several moderators of treatment effect via a scale of study rigor. Sloppily run studies (no control group, raters not blinded etc.) showed large effects, while more rigorous studies showed effect sizes which included zero in their 95% confidence intervals. This should certainly be noted in any review discussing these results. Either way, the findings from the use fo CBT in the CHR population specifically seem promising, but inconclusive. I also found it rather odd that the subsequent sections discussed a ‘clinical staging model’ which prescribes cognitive behavioral treatment based on the stage at which an individual is at (i.e. CHR vs. first episode). It would seem that true effects of therapies ought to be established first, before therapies are combined (albeit, in an intuitive manner) in ostensibly comprehensive treatment plans.

Finally, the authors move to discuss what would appear to be their main interest – cognitive training. Relative to controls, CHR youth show impairments in general intelligence, executive functioning, verbal and visual memory, verbal fluency, attention, working memory, social cognition and processing speed. A variety of evidence also points to the fact that early deficits in cognition are risk factors for conversion to psychosis later on. For this reason, the authors underline the need for early detection and targeted intervention of these cognitive deficits in CHR youth, and in recent-onset schizophrenia. In this respect, the authors review several therapy modalities including, therapist guided paper and pencil tasks, Cognitive Enhancement Therapy, Cognitive Adaptation Training, and the Brain Fitness Program (an auditory processing / verbal learning training program). All except the last listed therapy here were shown to suffer from poor methodologies or null results. In studies (conducted by these same authors) the Brain Fitness Program was shown to have significant impacts on verbal working memory, verbal learning and memory, and global cognition with large and durable effect sizes in the mid-high .8 range in samples with chronic schizophrenia, and also samples with recent onset schizophrenia.

These findings are impressive, but I can’t help but feel that we’re missing some crucial information here on study rigor (i.e., who was blind to what?), and political transparency (i.e., what does the research team stand to gain from the success of the program). From what I’ve read elsewhere, overall I agree with the author’s conclusions that cognitive training in CHR and early onset populations are promising areas of future research but I’m not sure that I have enough confidence in it just yet to agree that the field is “facing the very real possibility that we may be able to move beyond preemption of schizophrenia, to inoculation” (p. 456).



Emotion Deficits in Schizophrenia

Kring, A. M., & Elis, O. (2013). Emotion defecits in people with schizophrenia. Annual Review of Clinical Psychology, 9, 409-433. doi: 10.1146/annurev-clinpsy-050212-185538

After elucidating the evolution of the definition of schizophrenia as a diagnostic concept, and some basics about what constitutes ‘emotion’, Kring & Elis move on to explain emotion deficits in schizophrenia. Drawing on the work of two relatively recent meta-analyses that I had actually read as part of my master’s thesis literature review (Kohler et al., 2010; Chan et al. 2010), the authors point out that individuals with schizophrenia do in fact demonstrate deficits in both emotion identification and emotion discrimination tasks. This should not come as a huge surprise, people with schizophrenia experience generalized deficits in neurocognitive and social cognitive functioning that should affect almost all tasks (on average). However, what is less clear from the 40+ years of research on emotion deficits in schizophrenia is the degree to which the deficits are caused by lower-level sensory processes involved in perception, or higher level cognitive factors responsible for integrating perceptions with surrounding context, or some combination thereof.

My masters work suggested that emotion perception deficits in schizophrenia may be caused in part by low-level perceptual processes as evidenced by aberrant patterns of visual facial information usage. The degree to which patients used visual facial information differently than controls was even found to be tentatively linked to poorer functioning in the community.

In contrast, Kring and Elis argue for the second option – that emotion deficits arise not from lower level perceptual deficits, but rather from a deficit in the ability to integrate such perceptual information with wider contextual cues from the environment. In support of their view, the authors point to evidence that suggests that people with schizophrenia experience similar levels of emotion as controls, but are simply less expressive of their internal states than controls are. Some interesting findings include: 1) people with schizophrenia have been shown to exhibit attenuated facial muscle activations to emotionally congruent stimuli; 2) using the experience sampling method, people with schizophrenia report a similar amount of subjective positive and negative emotion as controls; 3) in retrospective reports, people with schizophrenia are more accurate in reporting the nature of their emotional experiences; 4) people with schizophrenia show greater skin conductance reactivity that people without schizophrenia in response to emotionally evocative stimuli. Findings from fMRI and PET investigations of emotion in schizophrenia are more mixed, evincing both similarities and differences between schizophrenia and control populations.

So, up to this point the Kring and Elis have made the case that people with schizophrenia essentially feel emotions more or less the way that healthy controls might, but exhibit deficits in outwardly expressing those feelings. In order to explain the inconsistency between this finding, and the fact that up to 75% of people with schizophrenia have anhedonia (the “inability” to experience pleasure), the authors evoke a distinction between anticipatory and in-the-moment emotional responses. Compared to in-the-moment pleasure, anticipatory pleasure presumably relies on a host of cognitive skills like imagination, reflection, memory of past experiences, and active maintenance of an emotional state – cognitive skills that people with schizophrenia less often have. Thus while viewing emotionally evocative pictures cognitive neuroscientific indicators of neural activity are quite similar between those with schizophrenia and without. However, differences begin to emerge after the stimulus is removed, and new evidence suggests that the degree to which patients and controls differ on such markers of neural activation is actually predictive of clinical ratings of anhedonia. Kring and Elis sum up: “Taken together, these results are consistent with the notion that in-the-moment emotional responding in relatively intact, whereas anticipation of emotion and active maintenance of emotional information are disrupted in schizophrenia”

Finally, the authors discuss emotion deficits in the prodrome, in those at high-risk of developing the disorder, but are not yet evincing symptoms. Findings from the North American Prodrome Longitudinal Study (NAPLS) have indicated that the clinical high risk group does in fact exhibit deficits in both emotion identification and discrimination tasks relative to controls. Interestingly, the high-risk group also evinced less functional connectivity between amygdala and prefrontal cortex indicating potential lack of ‘communication’ between emotive and cognitive areas of the brain even before the onset of symptoms.

Ok, as a schizophrenia researcher dealing with emotion perception in the disorder, I actually have relatively little to add here, not even a catchy title apparently. I think the authors presented the facts regarding emotion deficits in an organized and coherent manner. However, I’m not sure that the evidence provided leads to the conclusion that the authors state at the outset – that emotion deficits arise not from lower level perceptual deficits, but rather from a deficit in the ability to integrate such perceptual information with wider contextual cues. If people with schizophrenia are experiencing emotions like controls, would this not preclude the possibility that those with schizophrenia are ‘failing to integrate’ context? If one had failed to integrate context into an emotionally-laden assessment, how could one then evince similar fMRI pattern of activation, or facial muscle responding as someone who had? Or perhaps the authors meant only to apply this lack of integration idea to anticipatory emotions (vs. in-the-moment emotions) which would presumably require greater application of cognitive skills to environmental factors?

Either way, I agree with the spirit of Kring & Elis’ calls for more research in order to resolve issues related to the heterogeneity of emotion deficits within the schizophrenia population, including individual differences in cognitive control, ambivalence and accessibility to beliefs. Moreover, I find their contrasting opinion to my own work (i.e., higher-level integrative deficits vs. lower-level perceptual deficits) almost completely non-confrontational, but rater complimentary. That is, I wouldn’t be the least bit surprised to find out that facilitating the integration of context into emotional processes would likely have aided participants in my study, and conversely that helping patients along as best as possible via low-level perceptual processes would similarly help in emotion based tasks and situations more generally.



Postpartum Depression, or Just Depression Postpartum?

O’Hara, M. W., & McCabe, J. E. (2013). Postpartum depression: Current status and future directions. Annual Review of Clinical Psychology, 9, 379-407. doi: 10.1146/annurev-clinpsy-050212-185612

Having read O’Hara & McCabe’s review of postpartum depression (PPD), I have more questions than answers. I’ll get to these, but first let’s cover some of the basics of PPD. Distinct from the postpartum blues, and postpartum psychosis PPD is defined as an episode of major depression (or sometimes including minor depression) that occurs in the postpartum period. What’s the postpartum period? Accounts differ. While the research literature includes postpartum timeframes up to twelve months, the DSM-IV defines the postpartum period as four weeks. The DSM-5 extends the postpartum period to six months based on scant evidence, though the authors note that there wasn’t great evidence for setting it at four weeks in the first place. ”In sum, there is no consensus as to what constitutes the postpartum period for the purposes of research on PPD, and it is likely that different time frames will be used for different purposes” (p. 382).

Ok, not a good start. The next roadblock to validity for the diagnostic category of PPD is verifying that the postpartum period (whatever that is) does actually represent a period of increased risk for the development of depressive symptoms. The authors discuss three different studies that provide some empirical insight into this question, and conclude that “the evidentiary base for concluding that depression is more common in the postpartum period [whatever that is] is still relatively weak” (p. 383). Additionally, physiological changes that occur during pregnancy can complicate the standard assessment and diagnosis of depression. Fatigue, appetite disturbance, and sleep disturbance have each been associated with the postpartum period in nondepressed women, raising the issue of whether or not PPD differs in a clinically significant manner from MDD. At least one study was reviewed which indicated that the psychometric structure of PPD and MDD are in fact the same. The authors take this to be a kind of validation of the standard MDD criteria during the postpartum period (whatever that is). However, my initial reaction to this was in support of an invalidation of the PPD concept altogether. Given that its not well defined, and there’s little evidence of increased risk, is it not more parsimonious to do away with the PPD descriptor anyway?

One of my complaints about this review is that it seems to start with a shaky definitional foundation of postpartum depression, and then continues to build on the concept as if it were rock-solid. The whole concept of PPD as a distinct diagnostic category is thrown into doubt in the first several pages, but the authors keep using the term PPD as if it carried none of the empirical and validity baggage described earlier. To avoid falling into this same pattern, from here on, I’ll use the term PPD to describe the diagnosis and the term ‘depression postpartum’ to describe the symptom of depression following childbirth, regardless of timeframe used or consequences for diagnosis.

The models of PPD reviewed by the authors also serve to enhance this doubt about the validity of the PPD specifier in their similarity to depression in general. O’Hara’s own cognitive behavioural model of PPD suggests “…that a woman’s psychological vulnerabilities prior to and during pregnancy would predict increases in depressive symptoms following a stressful life event such as childbirth.” (p. 386). How is this different from the process of MDD in a woman who does not have a child? Similarly, Beck proposed an interpersonal model in which “…the degree of mismatch between a woman’s desired versus received social support following childbirth influences her depressive symptoms.”(p. 386). Same question here: how is this different from the process of MDD in a woman who does not have a child? My understanding is that low social support is always a predictor of depression, regardless of childbirth. As far as I can tell, these are not models of PPD, but post hoc descriptions of ‘depression postpartum’. Succinctly stated, my objection is this: the instantiation of a specifier implies a clinically relevant distinction. If there is no such distinction, there should be no specifier.

However, O’Hara and McCabe discuss some data regarding the potential hormonal causes of depression postpartum which might indicate some kind of differentiating factor from MDD. Essentially there may be a subset of women who are particularly vulnerable to fluctuations in estradiol and progesterone in pregnancy, leading to a destabilization of the serotonin and dopamine systems, and thereby a greater risk of depressive symptoms.

Interestingly, the authors also discuss the role of depression postpartum on the developing child. Reviews generally suggest, not surprisingly, that maternal depression is associated with a variety of negative consequences for the child – including behavioural, cognitive, and health-related. This makes sense, but I worry again here that the effect of genetics has been completely forgotten. As an example, consider the claim that chronically elevated symptoms of depression in the mother is associated with poorer cardiovascular outcomes in the children. Given what we know of the association between depression and cardiovascular disease from the Whooley and Jang review (see post #12), it might make more sense to consider both the expression of depressive symptoms and poorer cardiovascular outcome at least partially mediated by the genetic contribution of such traits from parent to child. Fully appreciating these genetic contributions would inform clinical practice in disbursing clinicians of the notion that the negative effects of maternal depression on the developing child are mediated through her behaviour alone.

The final sections of the review move to discussing treatment and prevention options. I hasten to spend too much time on these as they are the standard treatments for depression in general (general counseling, interpersonal psychotherapy, CBT, and psychodynamic therapy), and there is no solid evidence for using one method over another: “…there is a dearth of information on the active ingredients of therapy. Moreover, there is no evidence that one approach to treatment is better than any other. Nor is there evidence regarding the optimal length of treatment.” (p.394-395). Similarly, pharmacotherapy does not appear superior to placebo. In terms of prevention, O’Hara and McCabe suggest that depression postpartum represents a unique opportunity in that women are in regular contact with the medical system during a very structured and predictable length of time. Early intervention during this time could prevent the further development of depressive symptoms, and full depressive episodes.

Checking to confirm that the DSM-5 made the changes that O’Hara and McCabe refer to (namely changing the timeframe of postpartum from four weeks to six months), it appears that is not the case. DSM-5 now actually makes no mention of the postpartum specifier, and instead includes a “with peripartum onset” specifier which retains the four week timeframe. Apparently the change was made in recognition of the fact that depressive symptoms can begin during pregnancy. However I find the new category linguistically confusing, as now all women who experience depression during or after childbirth only have the option of a peripartum specifier which will be inaccurate for most. Ugh, I’m dissociating again.



A Moratorium on Models in Interpersonal Processes in Depression

Hames, J. L., Hagan, C. R., & Joiner, T. E. (2013). Interpersonal processes in depression. Annual Review of Clinical Psychology, 9, 355-377. doi: 10.1146/annurev-clinpsy-050212-185553

Hames et al. begin their review of interpersonal processes in depression by first describing the well-known symptoms of the disorder as well as some lesser known associated features. In addition to the nine core criteria for depression (depressed mood, anhedonia, change in appetite and sleep, psychomotor agitation/retardation, low energy, feelings of worthlessness or guilt, concentration difficulties or indecisiveness, and suicidal ideation), the disorder is also characterized by its persistence and recurrence. The authors point out “the most frequent course…was recurrent, with antecedent dysthymia, without full interepisode recovery” (p. 357). People with depression have also been observed to express fewer facial expressions, engage in less eye contact, hold their head down, use fewer nonverbal gestures, and speak more slowly and quietly relative to non-depressed people. Deficits have also been noted in terms of social skills in depressed people.

Central to their review, the authors then move on to discuss the opposing interpersonal processes of excessive reassurance seeking (ERS), and negative feedback seeking (NFS). ERS is thought to be the behavioural means through which depressed individuals elicit rejection from other people. It’s measured via a four-item self-report questionnaire called the depressive interpersonal relationships inventory reassurance seeking subscale (DIRI-RS), and has been positively related to depressive symptoms and negative reactions from others. In contrast NFS is thought to stem from the depressed individuals desire to elicit feedback from the environment which is consistent with their self-concept. It’s measured via the feedback seeking questionnaire which assesses a person’s interest in receiving negative feedback across five domains: intellectual, social, musical/artistic, athletic abilities, and physical attractiveness. NFS has been positively linked to depressive symptoms and peer rejection. The literature is somewhat mixed on whether or not ERS and NFS are specific to depression.

Ok, so far we know that depressed people tend to seek both positive and negative feedback about themselves, which then seem to precipitate rejection from peers, thus contributing to the worsening of depression. While these two interpersonal processes seem to be at odds with each other, Hames et al. present a series of models which attempt to describe how these two seemingly opposite processes are expressed by the same depressed individual. I’ll run through each of these quickly before turning to a few thoughts about the implications for treatment.

First up is the cognitive-affective cross fire model which posits that ERS and NFS both produce either cognitive or affective discomfort, causing additional reassurance in the opposite domain. For example, ERS while affectively pleasing is supposedly cognitively discomforting due to the mismatch between the feedback and self-concept. For this reason, a depressed person would then engage in NFS which ought to be cognitively pleasing, but affectively discomforting, and so on, and so on, ad infinitum. It seems to me that a relatively simple way to test this hypothesis would be to empirically create a vivid description of the time course of ERS vs. NFS in a depressed sample (perhaps via some of the ambulatory assessment methods discussed by Trull and Ebner-Premier? See post #6). If the time course reliably showed alteration of ERS and NSF, the cognitive-affective crossfire theory might garner some support.

Next is the cognitive processing model, which posits that the type of feedback depressed people choose to elicit from peers depends on the cognitive resources the person has available at the time. Few cognitive resources available predicts that people will choose self-enhancing ERS over more cognitively demanding NFS. The particular reasons why negative feedback is considered more cognitively demanding are not well explained, but I agree with this evaluation on an intuitive level. Amazingly, this theory has never been tested in a sample of depressed people – though a quick check might be provided by comparing samples of depressed people high and low in cognitive resources to begin with (i.e., high vs. low intelligence) and see if there is a systematic difference in the type of feedback sought throughout the course of a given time period. Under this conceptualization, smarter depressed people would be expected to elicit more negative feedback that their cognitively slower counterparts. Seems rather unlikely to me…

Next, the authors discuss the integrative interpersonal framework, which in my reading doesn’t add much to the theories already discussed. The framework “argues that several depression-related mechanisms produce a variety of interpersonal problems and stressors; these problems in turn are strong predictors of future depressive symptoms and/or lengthened current episodes of depression” (p. 365). Hmm. Rather general, no? I might downgrade this one to a framework for a framework – that is, any more specific theory of interpersonal processes in depression that may be empirically validated in future, could almost certainly also be explained in these vague, dare I say vacuous, terms.

Finally, global enhancement and specific verification theory, which argues that depressed individuals seek self-enhancing feedback about their global traits and negative feedback about their specific traits. The extent to which this theory is true is “unclear”.

So which of these theories is correct? Or at least most correct? Certainly we should be able to discard at least one, as the theories make contradictoty predictions. Unfortunately the science possesses no answer on this issue. An apt metaphor here might be ships passing in the night piloted by armchair philosophers. I’m left scratching my head as to why some of these theories warrant inclusion in a review if they are simply speculation at odds with other speculation. At the very least, I’m going to suggest a moratorium on models in interpersonal processes in depression. Models are like toys. If you don’t use the ones you have, you don’t get new ones.

Following this, Hames et al. discuss three additional interpersonal processes in depression: interpersonal inhibition, interpersonal dependency, and attachment style. In order to avoid another rant about attachment styles and the complete neglect of all genetic contributions to behaviour (see post #10), I’m going to skip right to treatment implications.

In light of these interpersonal processes in depression, the authors discuss the several of the leading treatments including behavioral activation, cognitive behavioural analysis system of psychotherapy, and perhaps not surprisingly, interpersonal psychotherapy. Evidence is provided for the effectiveness of each of these therapies, however unfortunately the authors do not compare the therapies to each other or make specific recommendations about which might be best under which conditions.

In sum, this review leaves me wanting more. All we know is that individuals with depression have a tendency to seek both negative and positive feedback, though we’re not sure how (nor do we even seem particularly ready to empirically test how) these two opposing processes play out within the same depressed individual. We’re not exactly sure that seeking of positive and negative feedback is exclusive to depression, and neither are measured particularly well in the current literature anyway. How this corpus of knowledge could be used to inform treatment is beyond me, which is just as well, as the therapies outlined by the authors are the same ones that might be outlined in a review of depression that never mentions interpersonal processes at all. All of this leaves me wanting both positive and negative feedback about this post (i.e., feeling a little depressed).



Train Your Heart to Be Happy

Whooley, M. A., & Wong, J. M. (2013). Depression and cardiovascular disorders. Annual Review of Clinical Psychology, 9, 327-354. doi: 10.1146/annurev-clinpsy-050212-185526

Whooley and Jong begin their review of depression and cardiovascular disorders by laying out the logic of the bidirectional relationship between these two maladies, and the resulting negative self-perpetuating cycle of poor mental and physical health outcomes. It should come as no surprise that depression itself is a factor in lower levels of cardiovascular exercise, and similarly, that low levels of cardiovascular exercise can contribute to maintaining depression. Specifically, the authors examine the literature regarding depression in relation to several specific types of cardiovascular illness: coronary artery disease, cerebrovascular disease, peripheral artery disease, and heart failure.

One criticism I feel compelled to make about this first section of the review is that the four specific cardiovascular disorders discussed are not particularly well differentiated from each other. Without taking to Wikipedia, I wouldn’t have known that coronary artery disease is distinguishable from heart failure via the buildup of arterial plaque. In both cases, the heart stops. The differences between these conditions are somewhat beyond the realm of clinical psychology, but to the extent that there are important differences, they should be included in reviews like this one to aid in the process of hypothesis generation for future studies. Relatedly, I felt that for each of these disorders, the take home message was the same, and not terribly informative above and beyond common sense: the relationship between cardiovascular disease X and depression is complex and bidirectional. Let’s take a look at the introductory sentences for each section.

Coronary artery disease: “…[major depressive disorder] is an independent risk factor for mortality following [myocardial infarction]” (p. 329)

Cerebrovascular disease: “Depression is also common after stroke and predictive of adverse outcomes among patients with existing cerebrovascular disease” (p. 330)

Peripheral artery disease: “A growing body of evidence indicates that depression is a risk factor for the development of PAD and for adverse outcomes among patients with existing PAD” (p. 331)

Heart failure: “Depression is associated with an increased risk of developing HF and with adverse outcomes among patients with existing HF” (p. 331)

Hmm. As common sense would have advised – complex and bidirectional. Simply stating that two things are linked in this way is not particularly helpful. See here for a spirited wag-of-the-finger at this kind of discussion. However, this is perhaps being a little too harsh on Whooley and Jang, as they make good on their review of this link by delving into the mechanisms by which that link is perpetuated.

Again, these proposed mechanisms are not surprising, but warrant listing here. Behavioral factors include physical inactivity, medication nonadherence, smoking, dietary indiscretion, and social isolation. Biological factors include autonomic nervous system activation, systematic inflammation, alteration of the HPA axis, mental stress-induced ischemia, platelet activation, endothelial dysfunction, and genetic vulnerability. Each of these factors is discussed in turn, and evidence is provided for their *links* to both depression and cardiovascular disease.

Finally, the authors discuss implications for clinical practice. They recommend a collaborative care model of treatment in which patients are managed by a primary care provider, depression case manager, and consulting psychiatrist, and routing screening for depression amongst patients who have access to such care. Self-management, psychotherapy, and pharmacotherapy are the recommended front line treatments for MDD, which not-surprisingly, has shown positive effects on cardiovascular outcomes.

So, in light of this review, it’s clear that 1) depression and cardiovascular disease are *linked* in a complex and bidirectional manner; 2) the link is sustained by some likely suspects; and 3) standard treatment of depression is associated with better cardiovascular outcomes.

A few thoughts: First, is it only depression that’s linked to poorer cardiovascular outcomes? Intuitively, most of the mental disorders listed in the DSM (IV or 5, take your pick) individually or jointly could potentially cause some or all of the behavioural factors listed by Whooley and Jang. Following their model, these behavioural factors should then activate or aggravate the biological factors that lead to cardiovascular disorders, perpetuating the same cycle of mental and physical health deterioration. It would be interesting to see the prevalence of all mental disorders in the population of all those with cardiovascular disease. In this sense, using something like the PHQ-2 to assess for only depression (in the most rudimentary way) might be failing to detect a whole range of mental illnesses.

Second, looking at this issue through the network perspective discussed by Boorsboom et al. (see post #4) perhaps it is more beneficial to view the symptoms of MDD (or any other disorder) and cardiovascular disease as simply ‘hanging together’. Or, stated differently, should poor cardiovascular outcome simply be considered part of the criteria for MDD? Similarly, rather than attempting to determine the exact placement and direction of the causal arrow(s) between MDD and cardiovascular disease, it might be more beneficial to look at such associations between individual symptoms of the disorders. For example, if it were established that feelings of worthlessness (one of the 9 criteria for MDD) were more highly associated with cardiovascular disease, then assessment and treatment efforts could be employed more efficiently.

Third, I wonder about the association between mental illness and cardiovascular health in the positive direction. We expect depression and cardio illness to hang together, but should we also expect the same association to carry through the positive range of cardio functioning? Should we expect athletes to be less prone to depression because of their high cardio abilities? What would a graph of cardio ability versus depressive symptomatology look like for the entire population?

Ultimately, mental health (and life in general) is all about balance. Imbalance in one domain tends to beget imbalance in others, while balance in one domain seems to at least provide the opportunity for balance in others. Simple. While much of the material covered in this review is relatively commonsense, I think it reflects a trend in the literature toward integrating our understanding of mental and physical suffering, and most importantly, using what we know about balance in one domain to battle imbalance in another. Although I would certainly consider this discussion of links to be in its preliminary stages, I’m genuinely excited to see the progression of the literature in this area.



Dissociating from DSM-5

Spiegel, D., Lewis-Fernández, R., Lanius, R., Vermetten, E., Simeon, D., & Friedman, M. (2013). Dissociative disorders in DSM-5. Annual Review of Clinical Psychology, 9, 299-326. doi: 10.1146/annurev-clinpsy-050212-185531

“There are three fundamental dissociative ways to respond to a traumatic experience: detach from it, forget it, or separate the memory of the experience from one’s present identity.” (p.308) Spiegel et al. review the disorders that result from the pathological application of each of these methods: depersonalization disorder, dissociative amnesia, and dissociative identity disorder (DID) respectively, and their ostensibly unlikely links to post-traumatic stress disorder and conversion disorder. Where the latter two disorders are conceived of as quite unique from dissociative disorders in their DSM-IV formulations (though placed contiguously in the manual), the authors make the case that they are in fact quite similar in many ways, and may in fact be alternative psychological methods of dealing with traumatic experiences. Corresponding changes to the diagnostic criteria are suggested for DSM-5.

The first several sections of the review focus on proposed modifications to the criteria for the dissociative disorders, as well as a few other issues related to nomenclature. They propose that ‘pathological possession’ be added to the criteria for DID, that dissociative fugue be added as a subtype of dissociative amnesia, and that derealization be given equal prominence as depersonalization to create a merged depersonalization/derealization disorder category in the DSM-5.

While the authors present a good case for the alterations to dissociative amnesia, and depersonalization/derealization disorders, the proposed addition of pathological possession to DID criteria seems more problematic to me. Something about broadening the definition of a pathological process to suit the needs of wider cultural applicability seems backwards to me. The DSM has always been an enterprise aimed at carving the world of mental illness at its natural joints, and its success or failure in this regard is a matter of considerable controversy. Efforts to increase the accuracy of this endeavor over time have generally driven research to increasingly reductionistic explanations of the causes of disease processes, rather than the broader downstream expression of these causes in various cultures. Of course, this is not to say that the formal criteria for disorders ought not to apply to all cultures, rather exactly the opposite – the criteria should capture what is common about the disease process in all human beings rather than enumerate the ways in which it may be experienced differently across cultures. In this sense, the addition of things like pathological possession to the formal criteria will serve to push the criteria further away from the root causes of disorder sought by the endophenotype literature.

Secondly, adding the term ‘possession’ seems to give some legitimacy to a subjective experience that clearly has no correspondence with reality. This is not meant to invalidate the clinical relevance of the experience of possession, but rather that the experience of possession itself is certainly secondary to the general process of dissociation. Analogously, depression and anxiety are experienced in markedly different ways across cultures, and clinicians need to be keenly aware of these differences. However, to add these differences to the formal criteria for the disorder would seem to subvert the goal of carving nature at its joints.

In the second half of the review, Spiegel and colleagues make the case for the instantiation of a dissociative subtype for PTSD. Recent evidence points to some kind of joint in nature here evinced by a high degree dissociative symptoms in PTSD patients, similar etiologies based in trauma, and differential response to treatment. The authors point out that while only a subset of those with PTSD experience dissociative symptoms, most of those who qualify for a dissociative disorder would also meet criteria for PTSD. One study found that 32% of a sample of military veterans with confirmed PTSD belonged to this ‘dissociative taxon’. Interestingly, new evidence has revealed a pattern linking emotional undermodulation of limbic areas by frontal areas to symptoms of re-experiencing trauma (as classically captured by DSM-IV criteria for PTSD), and emotional overmodulation of limbic areas by frontal areas to symptoms of dissociation. This conceptualization is consistent with the cortico-limbic inhibition model of dissociation in which “…dissociation, including states of depersonalization and derealization, is an emotion regulatory strategy during conscious processing of threat that is employed to cope with extreme arousal in PTSD through hyperinhibition of limbic regions.” (p. 313). Not surprisingly, this has obvious implications for treatment wherein hyperinhibition leading to dissociative states during exposure-type therapy for PTSD should be expected to greatly reduce the effect of the treatment.

Finally, the authors discuss conversion disorder and its connections to dissociative symptoms. Simply stated, it has been proposed that “conversion is a somatoform manifestation of dissociation” (p. 317) – a claim that has been supported by dysfunction of brain regions implicated in the integration of affect and sensation in both conversion and dissociation. The authors conclude by laying out specific changes that might be made regarding conversion disorder in the DSM-5, though they state that the most likely outcome will be no change at all.

Now that the DSM-5 has been released, we don’t have to guess any more. So how do Spiegel et al.’s predictions match up with the actual text of the DSM-5? The new DSM now includes reference to possession in criteria A for DID, and drops all reference to ‘multiple personality disorder’. Dissociative amnesia now includes a specifier for dissociative fugue, though retains the same coding as in DSM-IV. Depersonalization disorder is now renamed depersonalization/derealization disorder and retains the same coding. PTSD was moved out of the ‘Anxiety Disorders’ section in DSM-IV to ‘Trauma- and Stressor-Related Disorders’ in DSM-5. It does now include specifiers for depersonalization and dearealization. The PTSD criteria now also include separate criteria for children 6 years and younger making the criteria span across four separate pages of the new manual. And finally, what of dissociative disorder not otherwise specified? DSM-5 replaces this terminology with two new disorders: other specified dissociative disorder, and unspecified dissociative disorder. Rationale? Straight from the DSM-5: “The other specified dissociative disorder category is used in situations in which the clinician chooses to communicate the specific reason that the presentation does not meet the criteria for any specific dissociative disorder.” (p. 306 of DSM-5) “The unspecified dissociative disorder category is used in situations in which the clinician chooses not to communicate the specific reason that the criteria are not met for a specific dissociative disorder, and includes presentations for which there is insufficient information to make a more specific diagnosis (e.g., in emergency room settings)” (p. 307 of DSM-5)

These new categories literally make the disorder that someone is labeled with contingent on what the clinician chooses to do. Needless to say, I would think that decisions like this drive the DSM categorization of mental illness further from carving nature at its joints for the benefit of the patients in these categories, toward a rather philosophically inconsistent manual streamlined for use by clinicians, insurers, and organizations.

There are essentially three ways to respond to philosophically traumatic DSM-5 changes: detach from them, forget them, or separate the memory of them from one’s present [professional] identity.



Worry as an Affective Insurance Policy

Newman, M. G., Llera, S. J., Erickson, T. M., Przeworski, A., & Castonguay, L. G. (2013). Worry and generalized anxiety disorder: A review and theoretical synthesis of evidence on nature, etiology, mechanisms, and treatment. Annual Review of Clinical Psychology, 9, 275-297. doi: 10.1146/annurev-clinpsy-050212-185544

All graduate students are familiar with generalized anxiety, and most would be lying if they told you they hadn’t furtively tried on the formal criteria for generalized anxiety disorder like a shoe you hope won’t fit. Differential diagnoses of GAD in the average graduate student include (but certainly are not limited to) obsessive compulsive disorder (OCD), and obsessive compulsive personality disorder (OCPD). I actually remain unconvinced that OCPD is not in fact a requirement for success in graduate school. For the sake of completeness, here are the DSM-IV criteria.

A pervasive pattern of preoccupation with orderliness, perfectionism, and mental and interpersonal control, at the expense of flexibility, openness, and efficiency, beginning by early adulthood and present in a variety of contexts, as indicated by four (or more) of the following:

(1) is preoccupied with details, rules, lists, order, organization, or schedules to the extent that the major point of the activity is lost

(2) shows perfectionism that interferes with task completion (e.g., is unable to complete a project because his or her own overly strict standards are not met)

(3) is excessively devoted to work and productivity to the exclusion of leisure activities and friendships (not accounted for by obvious economic necessity)

(4) is overconscientious, scrupulous, and inflexible about matters of morality, ethics, or values (not accounted for by cultural or religious identification)

(5) is unable to discard worn-out or worthless objects even when they have no sentimental value

(6) is reluctant to delegate tasks or to work with others unless they submit to exactly his or her way of doing things

(7) adopts a miserly spending style toward both self and others; money is viewed as something to be hoarded for future catastrophes

(8) shows rigidity and stubbornness

Hmm. I’d like to say that the shoe doesn’t fit, but I’ve just re-typed a detailed and orderly list of rigidly defined criteria. Ok, but back on track in the name of perfectionism and scrupulous overconscientiousness: GAD and worry.

Newman et al. begin by laying out some of the primary characteristics of GAD and worry (features, comorbidity, prevalence, course etc.) before moving on to an explanation of the Contrast Avoidance Model. According to this model, individuals with GAD are more reactive to emotional experiences, and use worry as a way of shifting their interpersonal experiences toward negativity in order to avoid large swings or contrasts in their emotional experiences. Or stated more succinctly: “…individuals with GAD prefer to feel chronically distressed in order to prepare for the worst” (p. 286)

In effect, this model posits that worry is used by those with GAD as a kind of emotional insurance policy – exacting relatively small costs in positive affect in order to stave off or mitigate the effect of unexpected large decrements in mood or emotionality. Interestingly, the authors discuss evidence that GAD is experienced primarily verbally/linguistically, which is an interesting contrast to the apparently more visually oriented experiences of those suffering from social anxiety (as discussed by Morrison & Heimberg).

In terms of risk factors for the development of GAD, Newman et al. discuss the environment, attachment/parenting style, and temperament. While the negative impacts of environment and temperament seem to be fairly straightforward (e.g., maltreatment, loss etc.), the implication of parenting and attachment styles seems somewhat more problematic to me (perhaps not surprisingly). I have elsewhere (see p.20) discussed my views on the purported impact of parenting on child outcome, and will limit my thoughts here to just the central concern – the oft neglected impact of genetically transmitted tendencies and traits from parent to child. To say that it is the attachment style of the child to the parent (i.e., just the behaviours) misses the whole possibility of the symptoms being a product of the parent’s genetic contribution to the child. This will strike some as hugely controversial, and so it should as its strongest case is essentially a repudiation of decades of psychological inquiry into attachment styles, their meaning, and their utility in research and clinical practice. I’ll stop short of developing that strongest case here, but I’m certain that I’ll have more to say about it in future posts. For now, let’s just say that I tend to think of attachment styles within the realm of normal parenting (i.e., barring abuse and neglect) as the anti-endophenotype. The most distal expression of the root cause of behaviour X or Y. This is particularly interesting given that Newman et al. begin their review by suggesting that worry may actually constitute a transdiagnostic process, based on its extremely high comorbidity.

Moving to treatment, Newman et al. discuss the implications of the Contrast Avoidance Model and recommend more research be conducted to investigate treatment based on this etiological view. Essentially, if individuals are maladaptively using worry as a mechanism for softening the blow of sharp affective shifts, perhaps they could benefit from increased tolerance to these shifts, making worry redundant.

Of what’s been investigated thus far for treatments, it would seem that CBT is superior to non-specific therapies or a waitlist control, as are SSRI’s. However, there does not appear to be an added effect for the combination of these treatments, however, motivational interviewing in conjunction with CBT appears to hold some promise. Interestingly, duration of disorder was found to moderate treatment effect of combined versus individual components of CBT. Longer duration GAD was better treated by purely cognitive or purely behavioural therapy, whereas shorter duration GAD was better treated with these elements combined. Additionally, change in client resistance/expectancy/worry/rigidity of symptoms were found to mediate treatment outcome.

Overall, I think Newman and colleagues produced a good review of GAD, the Contrast Avoidance Model, and future directions. My one concern regarding the Contrast Avoidance Model relates to the notion of teleology – or the idea of explaining phenomena via their purpose(s). Here, they suggest that the purpose of worry is to avoid sharp affective contrasts, and proceed to construct their etiological model of GAD on this foundation. Important questions moving forward should concern other purposes of worry – perhaps for example as a motivational factor in prompting future adaptive behaviour. Simply stated, to say that the purpose of something is X, often serves to mask an entire alphabet of other purposes, some of which deserve a place in our etiological theories.

A related issue that I don’t believe is addressed in the article is the extent to which the use of worry as a kind of affective insurance policy is conscious or unconscious. Do individuals with GAD realize that this is the ultimate purpose of their worry? Or would they give it a different interpretation? Similarly, a hugely important question is, does adopting the perspective of worry as an affective shift avoidance mechanism make recovery easier, regardless of its truth or falsity? All questions for future graduate students to worry about.



Social Anxiety – A Decidedly Non-Anxious Review

Morrison, A. S., & Heimberg, R. G. (2013). Social anxiety and social anxiety disorder. Annual Review of Clinical Psychology, 9, 249-274. doi: 10.1146/annurev-clinpsy-050212-185631

“The process begins with the perception of an audience…” (p. 251). From here Morrison & Heimberg explicate their theory of how this perception precipitates social anxiety and social anxiety disorder (SA/SAD). The article is very well organized and guides the reader through the relevant aspects of this condition. First, they outline information processing biases including, attention bias, interpretation bias, implicit associations, imagery and visual memories. Next they discuss the concepts of self-focused attention, emotion regulation, safety behaviours (i.e., avoidance and impression management), and finally post-event processing.

As much as I found the article to be a great overview of the topic, I found very little to sink my teeth into in terms of philosophical questions or methodological confusions. Essentially, people who suffer from SA/SAD tend to perceive the world in biased ways, and then go on to interpret those perceptions in biased ways leading to the feeling of intense scrutiny and ultimately inadequacy.

The most surprising aspect of the article for me was that the visual modality was the most salient modality for self-focus, worry etc. For example, many of the research studies used a form of imagery as a manipulation for an analogue SA condition. I would have thought that the self-doubts of many people would be more verbally oriented. It seems like many of the downward comparisons that people might make of themselves in relation to other people would even require some kind of verbal or linguistic component in order to convey the complexity of the self-effacing thought. I would be interested to see how a purely visual manipulation or intervention would compare with a purely verbal one. As identified by the authors, this will be an area of future research.



Fear Return, Renewal, Recovery, Reinstatement, Retrieval, Reacquisition, Reconsolidation, and Relapse…Really?

Vervleit, B., Craske, M. G., & Hermans, D. (2013). Fear extinction and relapse: State of the art. Annual Review of Clinical Psychology, 9, 215-248. doi: 10.1146/annurev-clinpsy-050212-185542

Setting aside the fact that there are enough r-words to describe the return of fear to substantiate an entire season of Monty Python skits, Vervliet et al. provide a nice summary of the landscape of fear extinction and recovery. I want to first outline a number of these r-words in order to help keep the concepts straight for myself (something this paper doesn’t readily do for the reader), before turning to a few issues and questions I had about some of the points made in the paper.

From what I can gather, ‘return of fear’ in an umbrella term used to describe the following three concepts: renewal of fear, recovery of fear, and reinstatement of fear. Renewal of fear appears to be describing the case wherein the return of fear is augmented by the removal of the context in which the fear was originally extinguished. Examples of this would include ABA renewal, in which the fear is instantiated in context A, extinguished in context B, and tested again in context A. ABC, and AAB renewal being additional examples.

In contrast, spontaneous recovery of fear is the phenomenon in which fear returns over time without further artificial intervention – as if the animal (or participant) simply forgets about the extinction of the fear.

Finally, reinstatement of fear is similar to spontaneous recovery, except that further aversive stimuli are experienced between extinction and testing. Another term for it might be non-spontaneous recovery of fear.

Ok, now that we’re straight on these definitions, what can we say about the state of the art re: research on the topic? The authors discuss fear extinction and (insert r-word of choice here), in relation to Learning Theory. Quite simply (maybe too simply), the application of an aversive stimuli creates three components in the mind of the shocked: a mental representation of the conditional stimulus (CS), a mental representation of the uncondtitional stimulus (US), and an association between them. Importantly, in this conceptualization fear is not a reaction to the aversive stimulus, but rather a negatively valenced emotional state related to the anticipation of some future aversive stimuli. The strength of the fear is determined by 1) the strength of the CS-US association, and the intensity of the memory of the US. As a result of this simple model, extinction of fear can be conceptualized as 1) the weakening of the CS-US association; 2) the weakening of the memory of the US; or 3) a deactivation of the US memory.

Following this introduction to simple concepts of learning theory relating to fear extinction, the authors turn to an explication of the ‘retrieval model of extinction’ (via Bouton and colleagues) in which fear extinction is completely moderated by the development of an inhibitory CS-US association (depicted in figure 3).

At this point I wanted to interject with a few thoughts and questions about the concepts raised thus far. First, regarding ‘context’ as discussed in renewal studies, is it not the case that context and stimulus are inextricably linked? Or, stated differently, are context and stimuli not synonymous in these situations? For example, if a rat were presented with some CS, perhaps a bright light, which acted as a CS altering it to a forthcoming US, lets say a shock, in one context shouldn’t we expect there to be some difference in response if that light is presented in another context? In other words, shouldn’t we consider the stimuli to be the sum of the CS and the context? I’m not sure that I’m explaining myself particularly well here, but it would seem to work to simplify the relevant models of fear recovery if the context were not treated as a separate variable from the stimulus which is presented in it. This insight, in conjunction with an evolutionary perspective on the ultimate purpose of fear in all creatures seems to make the empirical results that follow simply commonsense: 1) reinstatement of fear in humans is context dependent; 2) reinstatement of fear in humans can occur with various timings; 3) reinstatement of fear in humans causes the return of attentional biases; 4) reinstatement of fear in humans is larger to negative stimuli etc. etc. None of this seems particularly surprising. Particularly unsurprising is that people with clinical anxiety disorders display some odd patterns of fear, avoidance, and attentional bias. Indeed, in some sense, this is exactly what these disorders are.

Another question I have about the literature presented in this paper is that it appears to offer up the learning theory and retrieval models of extinction with qualifying words like ‘may’ but then move on to apply these models with much more certainty than that with which they were introduced.

Finally, it seemed that despite the research amassed in rats, and the handful of translational studies that have been conducted in human populations, the final section of this paper on “Future Research” essentially serves to remove much of the validity of the conclusions drawn thus far. What constitutes a context? What role(s) are played by individual differences? How does the return of fear relate to the construct of, or predict, relapse? Are the processes of renewal/reinstatement actually at work, or is it actually new learning of fear that is taking place? This, of course is not to say that the research is not meaningful or important, but rather just makes me question that given this literature, do we have any more insight into how we ought to treat Bob for his fear of elevators?




Miller, G. A., & Rockstroh, B. (2013). Endophenotypes in psychopathology research: Where do we stand? Annual Review of Clinical Psychology, 9, 177-213. doi: 10.1146/annurev-clinpsy-050212-185540

Let’s start with some definitions. Actually, let’s start with my perhaps naïve definition of ‘endophenotype’, before refining that concept with snippets of Miller et al.’s excellent summary of the concept.

My understanding has been that endophenotype is essentially synonymous with the term ‘biomarker’ – some element of one’s biology that impacts behaviour relevant to psychopathology to a measurable or at least observable degree. In this sense, an endophenotype would seem to be the biological phenomenon, or cluster of phenomena, closest to the disorder itself in the grand causal chain from the disorder itself to eventual symptom expression. Thus, the concept of an endophenotype is inherently reductionistic, an attempt to capture the ‘rootest’ causes of expression of psychopathology, or I suppose any other expression of the phenotype.

Ok, so how does my naïve conceptualization compare to that in the established literature? The original Gottesman & Gould (2003) article, defines endophenotypes as “measurable components unseen by the unaided eye along the pathway between disease and distal genotype” (p. 636). Several years later, these authors confirmed this conceptualization: “Endophenotypes represent more defined and quantifiable  measures that are envisioned to involve fewer genes, fewer interacting levels and ultimately activation of a single set of neuronal circuits…” (Gould & Gottesman, 2006, p. 115). Formally, an endophenotype must meet the following criteria:

  1. 1. An endophenotype must be associated with illness in the population
  2. 2. An endophenotype is heritable
  3. 3. An endophenotype is state independent but age normed and might need to be elicited by a challenge
  4. 4. Within families, endophenotypes cosegregate
  5. 5. An endophenotype identified in probands is found in their unaffected relatives at a higher rate than in the general population
  6. 6. The endophenotype should be a trait that can be measured reliably, and ideally is more strongly associated with the disease of interest than with other psychiatric conditions


Setting aside the small irony that the concept of an endophenotype has formal criteria, just like the diagnostic categories which it aims to reconceptualize as dimensional constructs, I think these definitions match up fairly nicely with my previous thinking on the topic. However, one crucial difference between my naïve defining of the concept and those offered above, is the reliance on its rooting in biology. Miller notes that enodophenotypes need not be biologically based, but could include observable behaviours such as speed or accuracy on cognitive tests. One final definition of the concept is offered in the NIMH RDoc policy: “Endophenotypes are relatively well specified psychiological or behavioural measures that are considered to occupy the terrain between disease symptoms and risk genotypes” (Insel & Cuthbert, 2009, p.988).

Miller and Rockstroh go on to make some criticisms of the rather loose utilization of the endophenotype concept in the present literature. In my view, this seems warranted given the haphazard nature of its usage (e.g., X could be a biomarker for Y etc.), however, it also seems that such statements are to some degree precipitated by the lack of conceptual clarity about the construct itself. While the authors greatly clarify the construct, there still remains much wiggle-room in the construct’s definition. For example, Miller & Rockstroh mention that future research may need to focus on ‘endophenotypes of endophenotypes’. In such a case is the secondary endophenotype still considered an endophenotype? Or, does the definition of an endophenotype hinge on some concept being the ‘rootest’ cause of some condition or symptom, such that finding a prime endophenotype could thereby downgrade the definition of the contingent process? Essentially, my concern here is that the term endophenotype could in this sense be continually applied to whatever that current most reductionistic root cause is at the time, completely contingent upon current technologies and techniques. In this sense, the term endophenotype would be more a useful label for our nomenclature of causal chain (or network) involved in the disease process than a reliable label for some phenomenon in nature. A question I had about the formal definition of the concept is regarding Gottesman & Gould’s (2003) first stated criteria (…must be associated with illness). Why must it be associated with illness? One could easily imagine there being conceptually similar endophenotypes for all phenotypic expressions including personality traits, and even superior intelligence. Finally, it seems to me that the second criteria (…an endophenotype is heritable) adds little to the construct as everything is essentially heritable. In other words, because all psychological phenomena are heritable to some degree or other, adding this as a criteria seems to add little in the way of specificity while on the hunt for additional determinants of disease. As an exercise, try to think of something that might be considered a good candidate for endophenotype status, for which a heritability coefficient could not be calculated… not an easy task!

After defining the somewhat nebulous concept of endophenotype, Miller and Rockstroh go on to evaluate the truth of three propositions: 1) that endophenotypes are useful in understanding the origins of disorder; 2) that the endophenotype construct challenges the categorical view of psychiatric illness proffered by the DSM; and 3) that a network approach to genes, endophenotypes and symptom expression is preferable to a single causal chain. They note that ‘the decade of the endophenotype’ has accomplished much, but far from cemented the case for the truth of these propositions.

What is clear to me now is that I have a lot more thinking to do about endophenotypes, and especially their role in disrupting the neat and tidy order of psychiatric categorization of mental illnesses. I think I’ll let those thoughts simmer for a while, and perhaps update this post later on after these definitions sink in a little bit more.




Psychopathology? Yup, There’s an App for That

Trull, T. J., & Ebner-Priemer, U. (2013). Ambulatory assessment. Annual Review of Clinical Psychology, 9, 151-176. doi: 10.1146/annurev-clinpsy-050212-185510

Ambulatory Assessment (AA) – no, not completing standardized psychological assessments while walking, but rather prompting people about their psychological experiences while immersed in their real-world environs. Why? To minimize retrospective biases in assessment, and to gather the most ecologically valid data possible. Specifically, the authors discuss how data gathered via AA procedures might be used to augment the clinically relevant processes of investigating mechanisms and symptom dynamics, predicting symptoms, monitoring treatment effects, predicting treatment success, and/or preventing relapse. It is interesting to note however, that traditional assessment methods can already speak to the above areas, perhaps just in a more blunt, or delayed fashion. So it might be a bit of stretch to say that it’s the ambulatory component of AA which allows it to aid in these processes, rather than the assessment component.

That being said, I do find the concept of AA quite impressive. The fundamental idea is quite straightforward: collect data about people’s psychological states before they have time to reinterpret or otherwise bias these raw experiences (as we all do) before traditional assessment would have otherwise taken place. Great idea. What really impresses me about AA however, is its innovation in implementation. Whereas some areas of psychological assessment have remained quite stagnant over the last several decades, the area of AA appears to be capitalizing on the latest technologies to capture people’s mental experiences accurately. I believe that if psychology is to stay relevant in the quickly changing technological and social landscape, we need to keep up with current technologies, and meet patients and research participants on their terms. For example, monitoring symptom levels via text message and delivering prompt treatments based on critical symptom level cut-offs strikes me as a particularly effective strategy in combating needless suffering in mental illness. As with many medical disorders, proactive prevention trumps reactive care – and early detection through effective assessment is the key.

However, at this point the AA literature seems to have largely produced proof-of-concept type investigations rather than stringently controlled trials. Many of the investigations reviewed by Trull and Ebner-Priemer lacked control groups thus leaving the beneficial effects of the AA trials completely open to alternate explanations, including placebo effects and the Hawthorne effect. On a similar note, I was interested to see the idea that people might change their behaviours based on the expectation of being watched ‘reactivity’ when referred to in the research context (p. 155), but ‘self-monitoring’ and ‘self-management’ (p.162) in the clinical context.

A final issue I wanted to briefly touch on regarding AA is the validity of the assessment results it might produce. Traditional diagnostic criteria for psychiatric disorders were created without the use of AA methodology. For example, ‘loss of interest or pleasure in things one usually enjoys’ is one of the core criteria for major depressive disorder. Presumably, this criterion was validated for clinical relevance via the very types of retrospective reports that AA aims to supplant in psychological assessment. Is loss of interest or pleasure the same psychological phenomenon when considered in-the-moment versus after-the-fact? Thus, an important consideration for the AA literature would appear to be the empirical convergence or divergence of retrospective versus ambulatory assessments of symptoms. Similarly, following the logic of De Los Reyes et al., perhaps there is some kind of ‘meaningful divergence’ in retrospective versus ambulatory assessment results. This is not to say that traditional diagnostic criteria and categories themselves are completely valid (more on this later, I’m sure), but rather more just a consideration such that diagnostic apples may still be compared to apples rather than apps.




Divergence in Informants’ Reports is All Measurement Error, Right?

De Los Reyes, A., Thomas, S. A., Goodman, K. L., & Kundey, S. M. (2013). Principles underlying the use of multiple informants’ reports. Annual Review of Clinical Psychology, 9, 123-149. doi: 10.1146/annurev-clinpsy-050212-185617

The current paradigm of psychological assessment research, according to De Los Reyes and colleagues is to methodologically and theoretically account for divergence in informant reports as measurement error. Needless to say, sometimes that paradigm fails to capture or otherwise account for meaningful divergence between informant reports. I very much agree.

Once this point is established, the remainder of the article lays out a nomenclature for the taxonomy of all possible options for convergence or divergence of informant reports, and in the case of divergence, whether or not the divergence is statistically significant or represents something theoretically meaningful.




On Saddling Unicorns

Borsboom, D., & Cramer, A. O. (2013). Network analysis: An integrative approach to the structure of psychopathology. Annual Review of Clinical Psychology, 9, 91-121. doi: 10.1146/annurev-clinpsy-050212-185608

At issue in this article are fundamental questions about the nature of mental disorders. In some ways, mental disorders are comparable to medical maladies. For example, both represent some pathological process within the body that is (hopefully) amenable to change via the application of some therapy or another. However, as Borsboom and colleagues point out, the similarities may not run much deeper than that. Importantly, they note, that medical sicknesses are often discernible from their symptoms – wherein it would be theoretically possible to have the root medical problem, without overtly expressing the associated symptoms. In contrast, they highlight the fact that psychiatric diagnoses are not readily discernible from their symptoms. For example, what is it to have major depressive disorder, or generalized anxiety disorder if one is not expressing the symptoms such as problems with sleep, concentration, or fatigue? Is it even theoretically possible to have a psychiatric disorder without expressing the DSM symptoms? I’ll return to these questions later.

The authors suggest that the problem here is in mistakenly pinpointing the cause of the symptoms of a psychiatric disorder to be ‘the disorder itself’. Because of the failure of psychology to substantiate this view, Borsboom and colleagues propose the self-described ‘potentially radical’ framework of Network Analysis to better account for the inherently symptom-based nature of psychiatric illnesses: “…there is now a new game in town, and it is called network analysis… In such approaches, disorders are conceptualized as systems of causally connected symptoms rather than as effects of a latent disorder.” (p. 93). Following this logic, trying to ‘cut out’ the core of a psychiatric illness like major depression as one would cut out a tumor is rather like trying to “saddle a unicorn” (p. 98).

From here, the authors lay out what I would consider a good case for thinking of disorders more in terms of dynamic and interacting symptoms than stable entities. For example, it makes sense to think that in major depression, sleeping problems might cause fatigue, rather than considering both of these symptoms independently caused by some (unsaddled?) unicorn named major depression. I also thought that using approaches like the experience sampling method to get a better idea of how symptoms interact over time was a great idea, regardless of the philosophical or etiological nature of mental illness. I also liked the idea of risky network structures and disproportional symptoms influence on those structures at an individual level. However, these ideas are simply recapitulations of fairly basic clinical knowledge, recast in network analysis terms. It should come as no surprise that certain people might be more predisposed to certain symptoms, or that certain symptoms might be more salient for certain people.

Despite these good ideas, as I read through the article I started thinking that the two systems (network analysis versus the traditional approach) need not be wholly incompatible with each other. While the network analysis approach lays out the logic for focusing on symptoms alone quite explicitly, I felt that the authors may have been guilty of erecting somewhat of a straw man (unicorn?) when it comes to the traditional view of mental disorders. That is, in my experience researchers and clinicians largely understand that disorders are merely labels or titles for a cluster of symptoms and would not advocate for a literal interpretation of the medical model wherein the corpus of the disorder needs to be excised from the body or mind. To this end, even when discussing these clusters of symptoms Borsboom and colleagues refer to them by their DSM shorthands – MD, GAD etc. highlighting the usefulness of the labels. Additionally, it struck me that the DSM symptom space displayed in Figure 3 seemed to align nicely with where decades of research carved nature at its fuzzy joints. Symptoms cluster together – and we call those clusters by name. In other words, unicorns do not exist in the physical sense, but we can still use the concept to make useful points about the nature of phenomena that have real import to research and clinical practice.

Finally, I wanted to comment on a potential inconsistency I found in the authors’ treatment of the implications of network analysis. On page 94-95 they go as far as to say that the root causes of psychiatric disorders will likely never be found. They note that attempts to localize these root causes in several prominent theories throughout the history of psychology have failed, including repressed desires, learned helplessness, hormonal imbalances, neural abnormalities, and genetic defects. However, in concluding they bring up the subject of endophenotypes (“heritable phenotypes that are internal to the organism and that promote development of psychiatric syndromes”). My concern here is that the concept of an endophenotype is simply the latest advancement in the localization of the root cause of an illness – a theoretical manifestation of abnormality, independent of symptoms. Certainly, in resolving the problem of the unsaddlability of a unicorn, one should not attempt to swap the unsaddleable unicorn with a promising Pegasus.




Integrate the Data, and Let the Core Principles of Science Fall Where they May

Hussong, A. M., Curran, P. J., & Bauer, D. J. (2013). Integrative data analysis in clinical psychology research. Annual Review of Clinical Psychology, 9, 61-89. doi: 10.1146/annurev-clinpsy-050212-185522

I’ve come to see the process of science as a very black and white process. Well, the process of formal hypothesis testing anyway. Granted there is plenty of room for creativity and theoretical grey areas in literature review and hypothesis generation, but, the process using science to decide whether or not some proposition is true is quite starkly the opposite. The proposition is either true or its not, and there are both good and poor methods of evaluating the truth of each proposition. The blurring of this basic logic leads to things like claiming a dataset has a “trend toward significance” etc. etc. See here for a rather entertaining collection of ways to blur this line when reporting on the ‘significance’ of a finding.

Indeed, the hallmark of science is essentially experimentally manipulating only one variable (let’s say X) to see its effects on some other variable (for creativity’s sake, let’s say Y) such that a causal inference may be drawn about X affecting Y to some degree and in some fashion. This idealized version of science is extensively complicated within the social sciences (and especially in clinical psychology) as each individual person brings with them a plethora of individual differences across thousands of variables. Thankfully however, enterprising psychologists have been engineering solutions to this problem for over a century, finely tuning the implementation of standardized psychometric measures, consistent study methodology, and erudite interpretation of results. All this in an effort to gain a fleeting empirical glimpse of truth, or at least provisional truth that stands until proven wrong by subsequent study or scientific paradigm.

I think it must be for this reason that I find myself rather strongly (if yet naively) opposed to the idea of Integrative Data Analysis (IDA) as described by Hussong et al. The most succinct version of my argument against such data integration is simply that on some philosophical level, condoning the combination of divergent measures and samples between studies, must be the same as condoning haphazard sampling, inconsistent measurement, and ad hoc hypothesis testing within the same study. For example, if within a single study different measures of some psychological construct were given to the control and experimental groups, reviewers would consider the differences in measurement to be a potential confounding variable – one that could seriously undermine the validity of causal implications for the effect of X on Y. However, in the name of economic efficiency, and scientific collectivism Hussong and colleagues seem to skim over these issues and claim that assessment measures (for example) can be modeled in various ways to test for equivalency such that this difference in measurement would not affect any causal inferences that might be made from combining the data. This may in fact be the case, I just tend to think that this methodology would essentially open the flood-gates to another several hundred potential confounding variables that the investigators of each of the individual studies ostensibly worked so hard to excise from their final estimation of effect size. In this respect, the process of meta-analysis retains this hard work by taking a weighted average of the final effect sizes of many studies, whereas IDA returns to the raw data and combines them in a less controlled fashion.

Aside from the statistical benefits of increasing power, and the ability to address novel research questions from extant data (a.k.a. data mining?), Hussong and colleagues list economy (i.e., efficiency with research dollars) as one of the main benefits of the IDA process. However, I would argue that IDA is less a tool of expunging raw inefficiency from the politico-economic landscape of modern clinical psychology, and more a tradeoff between efficiency and scientific certainty. Or stated differently, if something seems too good to be true, it probably is.

However, on a more optimistic note, I should add that what excites me about this method is its creativity in answering questions about the human experience in a novel way. I’ve long thought that clinical psychology should move toward analysis of ‘big-data’ as a way of understanding the human condition. For example, what might we be able to learn about human cognition from Lumosity’s database? To that end, the most basic tenets of the IDA framework are certainly a step in the right direction.




Eschewing Obfuscation in Drug Addiction Remission Statistics

Heyman, G. M. (2013). Quitting drugs: Quantitative and qualitative features. Annual Review of Clinical Psychology, 9, 29-59. doi: 10.1146/annurev-clinpsy-032511-143041

As my ever-so-slightly sarcastic title might suggest, Heyman’s review of drug/substance remission left me with more questions than answers. The grist for the mill here is four National-level surveys of psychiatric disorder prevalence in the United States:

  1. 1. The Epidemiological Catchment Area (ECA; 1980-1984)
  2. 2. The National Comorbidity Study (NCS; 1990-1992)
  3. 3. The National Comorbidity Study – Revised (NCS-R; 2001-2003)
  4. 4. The National Epidemiological Study of Alcohol and Related Conditions (NESARC; 2001-2002)


Each of the above surveys offer the requisite data to calculate the remission rate for each of several types of drugs, over a given time period via the following formula %Remitted = (Lifetime% – 12-Month%) / Lifetime%. However, the surveys varied greatly in terms of diagnostic criteria used to define remission, as well as the type of sampling that was conducted. For instance, where the ECA lumped substance abuse and dependence together, the others did not, and where the ECA required a complete absence of DSM symptoms to qualify for remission, the others did not. In terms of sampling, where the ECA collected data from prison populations, the others did not, and were subsequently charged with underestimating drug dependence prevalence by 25%!

On top of these methodological inconsistencies, Hayman lists several other factors that should give the concerned reader pause in making hasty conclusions from the data. First, people that are calculated to have remitted from one drug, may have moved on to another drug – a possibility that the above surveys cannot track or quantify. Second, the actual reporting of drug usage statistics may be skewed to begin with starting at the level of basic data collection. Interviewees may have exaggerated or more likely downplayed their use of drugs. Third, those drug users which are harder for researchers to contact may be those that are in fact the heaviest users, or at least those less likely to stop using drugs. Finally, Heyman states that remission may be temporary.

All these methodological problems aside, I wanted to focus here on just this final point – that ‘remission may be temporary’. Hayman sets up his review by suggesting that present data is sufficient to resolve the controversy surrounding Winick’s claim that people largely “mature out” of addiction. Early on in the review he spoils his conclusion by mentioning in a section heading that “Maturing Out and Relapse Are Not Incompatible Results”. However, through reading through the arguments, I found just became more confused about what conclusions could solidly be drawn from the available survey data.

The foremost problem in my way of thinking about this topic is the nature in which the remission data is collected and calculated at a population level, and then used to describe probably life trajectories on an individual level. That is, if remission statistics are calculated on a population level, does this not completely smooth over the periods of usage and remission for each individual? These patterns of usage and remission, by the way, I should think hold key insights into treatment or at least ‘nudging’ people into more sustained remission. Stated differently, so long as statistics are being collected about the population, the results will appear “orderly” (as Heyman states) though likely represent the aggregation of thousands of completely disorderly cycles of usage and remission.

An apt analogy here might be population levels of binging on junk food versus periods of relative remission (defined by some DSM-esque set of criteria, of course). Of what value would it be to say that a certain subset of the population is currently in remission at some point, relative to some other point at which a completely different set of people may uphold that statistic? That is, the ‘rolling continuity’ or ‘orderliness’ of the prevalence rates of remission seem to completely obscure the perhaps more important individual trajectories of individual people through the peaks and valleys of drug use and addiction.

Having written this, I can tell that my thinking on this article is not well formed yet. I’d like to re-read this article, or get a second opinion on the value of these statistics in treating substance dependence.




Eysnck Repudiated with Evidence of Clinical Progress, but Validated by Future Challenges

Barlow, D. H., Bullis, J. R., Comer, J. S., & Ametaj, A. A. (2013). Evidence-based psychological treatments: An update and a way forward. Annual Review of Clinical Psychology, 9, 1-29. doi: 10.1146/annurev-clinpsy-050212-185629

Some sixty years on, it would seem that the discipline of clinical psychology is still dealing with the repercussions of Eysnck’s 1952 charge that psychotherapies are ineffective, and (if I remember correctly from my psyc. 101 days) offer similar rates of recovery as those observed for spontaneous recovery. Clearly unsatisfied with Eysnck’s conclusion, Barlow et al. set out to show just how far we have come in the intervening years. In doing so, the authors discuss diverse treatment literatures and gently compare the blunt or unspecific nature of Eysnck’s original question (Does psychotherapy work?) to similar unspecific questions the field is facing at the present.

After reviewing the progression of psychological therapy across a wide variety of specific psychopathologies Barlow and colleagues provide five reasons for the broad success of psychotherapy. They are:

  1. 1. Greater understanding of the psychopathologies themselves leading to more precisely targeted psychological treatments (e.g., interoceptive triggers for panic disorder)
  2. 2. Improved research methodology (e.g., improved design methodology and data management)
  3. 3. Support and emphasis of evidence-based therapy by governments
  4. 4. Precedence of evidence base of a treatment regardless of its theoretical origins (e.g., use of motivation interviewing for anxiety and mood disorders)
  5. 5. Preference for psychological treatments over drug treatments by consumers


While I grant the first three points, I have a degree of discomfort in accepting the final two. First, it seems that co-opting exiting treatments for use in other domains (point #2 above) might well be a consequence or a side-effect of the success of psychotherapies as much as it is a driver of that success. It might simply be the case that more therapies exist now than several decades ago, such that the amount of ‘cross-pollination’ in psychotherapy application grows in proportion to the number available. Regarding the fifth point from above, it seems unlikely to me that consumers would prefer psychological treatments to drug treatments across all contexts. In my anecdotal observations, people have a tendency to look for the path of least resistance. Psychotherapy takes a lot of hard work and dedication. If the same benefits could be garnered by taking a pill, I would think that most people would opt for the pill. Perhaps people simply like the idea of psychological therapies over drug therapies – which is wholly different from a non-context-specific preference for psychological therapies. Indeed, in the very next section, Barlow et al. discuss the relative decline in prevalence of psychotherapies, and the concomitant relative increase in prescription medication treatment of mental health problems. How can we square the idea of people generally preferring psychological therapies with empirical data suggesting their relative decline?

Barlow and colleagues balance their discussion of the reasons for the success of psychotherapy with a discussion of the barriers to evidence-based practice moving forward. They list the following factors as potential barriers:

  1. 1. ‘Less than desirable treatment effects’
  2. 2. Ubiquitous comorbidity in the real world vs. single diagnoses in clinical research
  3. 3. Poorly understood ‘mechanisms of action’ for treatments, and the nomothetic approach to research
  4. 4. Lack of marketing efforts re: dissemination and implementation of psychotherapeutic approaches


While I agree that all of these factors are reasonable candidates for barriers to the future use of psychological therapies, I can’t help but think that the first point is really the only one of any importance. Regardless of comorbidity, regardless of mechanisms of action, psychological therapies would have more impetus for dissemination and implementation if they clearly had more desirable effects. Lack of understanding regarding comorbidity and mechanisms of action seem more appropriately considered ‘barriers to desirable treatment effects’. In a world where these factors were understood, and treatment effects increased accordingly, I would think that issues presently surrounding dissemination and implementation would largely take care of themselves.