Avoidant PD

[mistafab]

Advertisement - Members don't see this ad

Have had a handful of cases in my time so far.

Fundamentally, I am having difficulty helping them.

I am viewing my role primarily as assessment/diagnosis, psychoeducating - providing prognostication. If they accept the diagnosis, connect them with further steps (intensive personality focused psychotherapy, preferably psychoanalysis).

I am wondering if there is anything else I can do as the psychiatrist to assist in these cases? Of my 4 cases thus far, 2 continued to do very, very poorly - one did poorly, and one did well. The person that did well engaged in psychoanalysis and fundamentally wanted to change her life systems to improve her QOL.

What can I do different?

 

[Sushirolls]

I'm not seeing anything different you can do.

I've seen similar number, too. Most just stop showing up to appointments.

 

[comp1]

I know that people describe narcissistic and OCPD as more common, but I have a suspicion that avoidant is much more common than all of them, just not seen clinically or reported in surveys. The issue is that other personality disorders create external pressure on someone to address their maladaptive coping. In fact, that external pressure is the only way you can treat narcissistic PD. Avoidant PD creates no such pressure and in fact, actively avoids its creation. I mean I can conceptualize something where a young adult is too anxious to start a job and parents create external pressure, but the external pressure is usually nothing like other PDs. Ultimately even skilled psychotherapists have many, if not most, patients leave during treatment for avoidant PD. The typical advice is to turn on your own hypervigilance regarding shaming. This is hard because the patient's concept of what you're saying and what your own or a typical understanding would be are likely world's apart. Every word is going to be under a microscope during and between appointments. I do think psychopharmacology has a huge role here given the challenges with psychotherapy. Pills are relatively incapable of rejecting you, although there are psychodynamics with medications of course too. In addition to vulnerable narcissist, the more practical conceptualization of avoidant PD is an extremely pervasive, disabling and long term social anxiety. We have meds for that and the patients should generally be on a SSRI with possible other adjunctive medications.

 

[Stagg737]

I’ve had a few Avoidant PD patients who have done fairly well. The common string was they all really wanted greater socialization and had good insight that their avoidance was actively harmful. I did a little MI with one or two to help them consciously connect those dots, but the underlying motivation was there. I think if you’re good at MI you may be able to make solid headway for some of these patients, especially if they’re distressed.

That said, if their comfort in avoidance is great enough or they don’t care that much, I agree the prognosis isn’t going to be great and I’ve also had a few who just didn’t want to continue care. I also think technology and the ability to easily have pseudo-socialization likely impedes this significantly more today, especially after COVID when “social distancing“ was not only acceptable, but encouraged.

 

[smalltownpsych]

Interesting that the base rate in outpatient is about 15%. I typically have not diagnosed it and tend to diagnose GAD. Thinking about it now, part of the differential is probably the patients who come back and engage in therapy have GAD and the ones who ghost me have Avoidant PD. Leads to a bit of a skewed experience. It does make me think that I should develop some more skill or awareness around this disorder. In other words, if I was to identify and diagnose it more clearly, then perhaps that would improve the outcome.
A quick glance at the literature, it seems I’m already doing what needs to be done to maximize engagement, but I would think that more intentionality would help. Also, often we feel the pressure to “help” or alleviate the presenting distress or fix the problem, and I’m pretty good at slowing that roll and getting to the dialectic of acceptance vs change, but in this population I need to recognize that is even more important and is in essence the treatment itself.

 

[clozareal]

I avoid them. They avoid me.

Unless they are coming to me for GAD, social anxiety, depression, substance use, etc. I see a lot of these patients have dependent personality disorder too. I wonder how comorbid they are.

These patients do poorly in therapy because of the high level of perceived criticism. They do even worse in group therapy. Both are highly transformative for them if they are willing to engage and persist although I imagine some of that persistence is temperamental rather than learned.

 

[calvnandhobbs68]

I know that people describe narcissistic and OCPD as more common, but I have a suspicion that avoidant is much more common than all of them, just not seen clinically or reported in surveys. The issue is that other personality disorders create external pressure on someone to address their maladaptive coping. In fact, that external pressure is the only way you can treat narcissistic PD. Avoidant PD creates no such pressure and in fact, actively avoids its creation. I mean I can conceptualize something where a young adult is too anxious to start a job and parents create external pressure, but the external pressure is usually nothing like other PDs. Ultimately even skilled psychotherapists have many, if not most, patients leave during treatment for avoidant PD. The typical advice is to turn on your own hypervigilance regarding shaming. This is hard because the patient's concept of what you're saying and what your own or a typical understanding would be are likely world's apart. Every word is going to be under a microscope during and between appointments. I do think psychopharmacology has a huge role here given the challenges with psychotherapy. Pills are relatively incapable of rejecting you, although there are psychodynamics with medications of course too. In addition to vulnerable narcissist, the more practical conceptualization of avoidant PD is an extremely pervasive, disabling and long term social anxiety. We have meds for that and the patients should generally be on a SSRI with possible other adjunctive medications.

Yeah I mean there's such a high overlap between symptom criteria for avoidant PD and social anxiety disorder that I'd be hard pressed to say any of these patients don't meet criteria for an anxiety disorder of some type. I'm actually surprised the reported overlap is variable considering like all the criteria for avoidant PD have to do anxiety around social situations....

https://onlinelibrary.wiley.com/doi/10.1002/da.20475

 

[mistafab]

Yeah I mean there's such a high overlap between symptom criteria for avoidant PD and social anxiety disorder that I'd be hard pressed to say any of these patients don't meet criteria for an anxiety disorder of some type. I'm actually surprised the reported overlap is variable considering like all the criteria for avoidant PD have to do anxiety around social situations....

https://onlinelibrary.wiley.com/doi/10.1002/da.20475

When I worked at a personality disorder focused practice setting - the trend we had with diagnosis was that if we had AvPD, we would not diagnose comorbid SAD given we believed the SAD was better explained by another condition (AvPD). So I think just practice patterns related to that are likely the driving factor for the variance in the comorbidity.

@clozareal We also saw that in almost all cases of AvPD on a SCID, the patient had dependent traits or also met full criteria for Dependent PD. The thinking generally was that in order to sustain an AvPD lifestyle, there invairably had to be a dependent component to survive. An enabler somewhere had been performing critical functions for the avoider in order for the personality disorder to sustain itself. I bet the comorbidity on a formal study would be north of 50%.

 

[clausewitz2]

When I worked at a personality disorder focused practice setting - the trend we had with diagnosis was that if we had AvPD, we would not diagnose comorbid SAD given we believed the SAD was better explained by another condition (AvPD). So I think just practice patterns related to that are likely the driving factor for the variance in the comorbidity.

@clozareal We also saw that in almost all cases of AvPD on a SCID, the patient had dependent traits or also met full criteria for Dependent PD. The thinking generally was that in order to sustain an AvPD lifestyle, there invairably had to be a dependent component to survive. An enabler somewhere had been performing critical functions for the avoider in order for the personality disorder to sustain itself. I bet the comorbidity on a formal study would be north of 50%.

It's almost like most personality disorders have terrible psychometric properties as diagnostic categories.

 

[R. Matey]

It's almost like most personality disorders have terrible psychometric properties as diagnostic categories.

 

[comp1]

In my experience they are just under diagnosed because there isn't a great PD to pill pipeline. Actually avoidant PD might be somewhat unique there, but good luck with the study recruitment.

 

[SeniorWrangler]

A lot of these patient do just fine until they get sick or otherwise need help from other humans, and then they’re in a pickle.

 

[FatherPsychiatry]

It's almost like most personality disorders have terrible psychometric properties as diagnostic categories.

Do you think the "living document" DSM-6 will finally change this?

 

[comp1]

The DSM is primarily a billing guide. So to answer any questions about what will be included you only have to ask if helps billing. Billing killed the personality profile thing they tried in DSM 5 and it will continue to do so in 6, IMHO. I also do not believe it will be "living" in any meaningful way. Constantly changing things does not assist billing. The DSM trails even the ICD and it will continue to do so.

 

[Stagg737]

Do you think the "living document" DSM-6 will finally change this?

I know you asked Clause, but I think it's unlikely.

The AMPD is already in the DSM-5, has been for the past several revisions. Experts in PDs have been campaigning to do away with the cluster system and switch to the AMPD which is more aligned with the HiTOP model for probably 20 years. The ICD-11 actually doesn't have unique personality disorders, the diagnosis would just be "personality disorder, mild/moderate/severe" with specifiers based on domains of functioning like the AMPD uses. The only specific disorder related specifier is "with borderline pattern" which is based on the classic cluster B criteria for BPD. So really ICD 11 only directly acknowledges BPD as personality disorder, the rest of their criteria align with the AMPD.

Personally, I wouldn't count on it changing despite the dramatic change in ICD-11 classification. The cluster system is entrenched, and the powers that be seem to have little motivation to revamp this on a larger clinical scale. I say this as someone whose previous research area was PDs and specifically the AMPD and who has spoken with some of the international leaders in PDs at conferences. They seemed fairly resigned to the idea that including the AMPD in the DSM under the "emerging models" was as good as it would get, but were still strongly advocating to move to a dimensional model.

 

[clausewitz2]

Do you think the "living document" DSM-6 will finally change this?

I think it really depends on how far they plan to go with making diagnostic categories more dimensional. If they go hard with this idea of a "major category" being the more typical category (e.g., psychosis) while at the same time permitting more precise specification as warranted (e.g., schizophrenia), it is easy to see how this would bring them much more in line with the ICD-11's approach and indeed what the PD literature supports as more reflective of reality. If we can actually utilize a "personality disorder" major category in practice without it becoming devalued by insurance like "unspecified" and "NOS" categories were, this could be a major step forward. Especially given they also seem to be driving more towards making all diagnoses multiaxial across several different domains (although not the same as the AMPD's domains per se).

 

[R. Matey]

Maybe a bit of thread hijack, but I would imagine that physicians (i.e., not psychiatrists) would struggle the concept of a dimensional model rather than a categorical disease state. A categorical disease state seems actionable. There are clear signs of its presence, delineated as symptoms, that are indicative of a set of treatment protocols. Dimensional traits would require much more nuance in a treatment approach and some familiarity with statistics (particularly psychometrics) to understand what a trait expression means. Inherent in the shift to dimensional approaches also are the problems of measurement error and regression to the mean; something we as clinicians encounter all the time on severity measures or outcome rating scales (e.g., wild swings in PHQ-9 scores taken a week apart). There's also the theoretical issue that research supports trait expression of illness because we use psychometrics to study them, which philosophically assumes the presence of a latent trait. It's not quite a hall of mirrors because dimensional approaches do correlate strongly with categorical ratings for many forms of psychopathology (in ADHD for instance, I think it's like 0.97).

At the aggregate, you can use advanced statistical modeling procedures to minimize measurement error to make predictions so researchers tend to handwave these problems away. But making cut scores that warrant specific protocols seems more fraught. For one, the psychometric properties of the instruments used to determine trait expression would need to really good (i.e., high internal and external validity/reliability), accessible to clinicians working with vulnerable populations (i.e., not behind a paywall), and scores would likely need to operate within a range. Afterall, patients may not understand questions or have secondary gains for responding one way or another and clinicians may not still understand what these scores really mean. So while the categorical system is very flawed, I am wondering if we're running quickly towards something that seems to solve one set of problems only to take on another.

 

[Stagg737]

Maybe a bit of thread hijack, but I would imagine that physicians (i.e., not psychiatrists) would struggle the concept of a dimensional model rather than a categorical disease state. A categorical disease state seems actionable. There are clear signs of its presence, delineated as symptoms, that are indicative of a set of treatment protocols. Dimensional traits would require much more nuance in a treatment approach and some familiarity with statistics (particularly psychometrics) to understand what a trait expression means.

This x1000. While the categorical model is significantly flawed, having a checklist of criteria is both far easier to understand as well as evaluate. It's a lot easier to screen for a checklist of symptoms vs interpret domains with specific traits which require larger batteries. Our research project that was presented at a major conference was creating an revised/abridged screen for a specific trait (the AMPD has 25 traits/facets) and it was a still a 26 item assessment. PCPs don't have the time or energy to be assessing for that and most don't have the desire to learn about it. Not really helpful when PCPs are the ones seeing the highest volume of psych patients.

Just another reason why I don't think the PD criteria is going to move away from the Cluster system any time soon.

 

[clausewitz2]

This x1000. While the categorical model is significantly flawed, having a checklist of criteria is both far easier to understand as well as evaluate. It's a lot easier to screen for a checklist of symptoms vs interpret domains with specific traits which require larger batteries. Our research project that was presented at a major conference was creating an revised/abridged screen for a specific trait (the AMPD has 25 traits/facets) and it was a still a 26 item assessment. PCPs don't have the time or energy to be assessing for that and most don't have the desire to learn about it. Not really helpful when PCPs are the ones seeing the highest volume of psych patients.

Just another reason why I don't think the PD criteria is going to move away from the Cluster system any time soon.

Timely and relevant:

https://www.awaisaftab.com/uploads/9/8/4/3/9843443/aftab_et_al_hitop_foundational_assumptions_ppp_2026.pdf

 

[R. Matey]

Timely and relevant:

https://www.awaisaftab.com/uploads/9/8/4/3/9843443/aftab_et_al_hitop_foundational_assumptions_ppp_2026.pdf

Thanks for sharing the paper. Idk if it read this way to you, but it felt really like the authors were trying to have it both ways. Many of the theoretical assumptions of both CTT and IRT assume an unobserved score and/or latent trait. They're technically right that the math doesn't require it to identify the model, but this is a bit misleading. It's hard to determine what scores mean without a concept of score reliability, for instance.

The analogy of hypertension or diabetes management is interesting, but these are also not behavioral observations dependent on subjective experience. I don't think we need to look further than the perils of pain management based on subjective rating scales for issues with this treatment strategy.

The fact that the HiTOP model seems far away from a definition of psychopathology or cut points to determine psychopathology makes this model seem less ready for prime time. I'm all for change and progress, but this seems a long way off.

 

[clausewitz2]

Thanks for sharing the paper. Idk if it read this way to you, but it felt really like the authors were trying to have it both ways. Many of the theoretical assumptions of both CTT and IRT assume an unobserved score and/or latent trait. They're technically right that the math doesn't require it to identify the model, but this is a bit misleading. It's hard to determine what scores mean without a concept of score reliability, for instance.

The analogy of hypertension or diabetes management is interesting, but these are also not behavioral observations dependent on subjective experience. I don't think we need to look further than the perils of pain management based on subjective rating scales for issues with this treatment strategy.

The fact that the HiTOP model seems far away from a definition of psychopathology or cut points to determine psychopathology makes this model seem less ready for prime time. I'm all for change and progress, but this seems a long way off.

I think a degree of waffling and prevarication is an inevitable consequence of a paper that is written as a broad consensus by a large group of people who were originally selected to represent a spectrum of viewpoints rather than all being members of the same research group.

It's important to bear in mind that this paper is more interested in the philosophical and conceptual underpinnings of this and does not really have as its object the production of clinically useful instruments derived from HiTOP. I think they do clearly say however that we should be explicit in saying that whatever cut points we decide to use are based on considerations external to the particulars of the symptoms/behaviors/experiences we are measuring. They are going to be inevitably and always somewhat arbitrary with respect to our behavioral observations alone. They say they are fine with the idea of setting score ranges for severity categories for helping to clarify descriptions, but any other cuts, in their view, are based on whatever the pragmatic and clinical needs of the setting the scores are being used in.

They would reject the idea that there is any difference in kind between just below threshold and just above threshold scores and that we should not reify this line we have drawn for our own utility. Our medical colleagues do not pretend that the boundaries they have for, say, an A1C of 6.4 v 6.5 is some step change or transformation in the pathophysiological processes involved in poor glucose regulation, but simply that it is felt based on the best available data that the risk v benefit ratio of certain interventions becomes favorable there. If some new, much cheaper and more benign medication that improved blood sugar regulation became available, you would actually expect this boundary to be revised downward. Similarly if new long-term data suggested a higher risk of bad outcomes from higher A1Cs per se than previously understood.

Here is a blog post from the corresponding author laying out more of his thoughts on this:

The Philosophical Foundations of HiTOP

A new paper examines the assumptions behind a new way of classifying mental illness

www.psychiatrymargins.com

He notes that HiTOP is not attempting to provide a definition of psychopathology, taking more the pragmatic approach of considering this to be isomorphic with the set of conditions, disorders, and problems that mental health clinicians treat. This fits with the basic inductivist approach it is trying to take and resonates with a respectable minority of opinion in philosophy of psychiatry as to how one might ultimately define mental illness.

 

[R. Matey]

It's important to bear in mind that this paper is more interested in the philosophical and conceptual underpinnings of this and does not really have as its object the production of clinically useful instruments derived from HiTOP. I think they do clearly say however that we should be explicit in saying that whatever cut points we decide to use are based on considerations external to the particulars of the symptoms/behaviors/experiences we are measuring. They are going to be inevitably and always somewhat arbitrary with respect to our behavioral observations alone. They say they are fine with the idea of setting score ranges for severity categories for helping to clarify descriptions, but any other cuts, in their view, are based on whatever the pragmatic and clinical needs of the setting the scores are being used in.

Fair enough though some of their other writings (Key Papers) where it does seem they have intent on integrating the HiTOP model into clinical care. There's even a manual for a digital tracker that relies on several assessments. They also mention several instruments in the original article that are behind paywalls (e.g., MMPI) and/or have very, very paltry psychometrics even if the original author has a lot of influence (i.e., the CBCL).

I wonder if that perspective invites a lot of subjectivity and puts clinicians in the position of relying on functional impairment. If so, how is that different than what we're doing now? FWIW, I'm highly supportive of a dimensional model of psychopathology and do agree that a symptom profile approach is likely the way to go (similar to some psychoanalytic conceptualizations of psychopathology). But it also needs to be carefully developed and researchers/theoreticians in my field can't do what they typically do with measurement issues, which is move the goal posts or wave away results they don't like. I see that happening in this article. I would be totally for a free MMPI-type instrument based on the IPP or some other free resource provided the psychometrics are good enough to warrant high-stakes decision making. Of course, nothing will be 100% specific and sensitive and clinician judgment wouldn't be sidelined in such a model.

They would reject the idea that there is any difference in kind between just below threshold and just above threshold scores and that we should not reify this line we have drawn for our own utility. Our medical colleagues do not pretend that the boundaries they have for, say, an A1C of 6.4 v 6.5 is some step change or transformation in the pathophysiological processes involved in poor glucose regulation, but simply that it is felt based on the best available data that the risk v benefit ratio of certain interventions becomes favorable there. If some new, much cheaper and more benign medication that improved blood sugar regulation became available, you would actually expect this boundary to be revised downward. Similarly if new long-term data suggested a higher risk of bad outcomes from lower A1Cs per se than previously understood.

That's a good point, but also score by definition is an observation of a positionality of a examinee's true score on a latent trait (or delta in IRT) after accounting for measurement error. The HiTOP folks are trying a avoid a lot of criticism on the p factor, which according to some is conceptually and statistically a weak argument. They avoid the critique by saying the HiTOP model is not a latent variable model nor does it rely on one specific form of latent variable modeling. But if that's the case, then how do we know that psychopathology is continuous? Indicators can reflective of categories or Possion distributions too. By contrast, there's less ambiguity about what an A1C means. So the argument by analogy doesn't feel strong to me.

He notes that HiTOP is not attempting to provide a definition of psychopathology, taking more the pragmatic approach of considering this to be isomorphic with the set of conditions, disorders, and problems that mental health clinicians treat. This fits with the basic inductivist approach it is trying to take and resonates with a respectable minority of opinion in philosophy of psychiatry as to how one might ultimately define mental illness.

Someone has to define psychopathology in order to treat it. If HiTOP wants to position itself as an alternative to the DSM, should they be pulling their punches?