Notes for week 4 – 4.3C SMI
WESTEN – THE EMPIRICAL STATUS OF EMPIRICALLY SUPPORTED PSYCHOTHERAPIES
Intro
- empirically supported therapies (ESTs)
- this paper argues that: unqualified statement and dichotomous judgments about
validity or invalidity in complex areas are unlikely to be scientifically or clinically
useful, and we should attend more closely to the conditions under which certain
empirical methods are useful in testing certain interventions for certain disorders
- examining the empirical basis of the assumptions that underlie the methods used to
establish empirical support for psychotherapies
retelling the story: the assumptions underlying ESTs
- ESTs gained momentum in 1995 with the publication of a task force by APA saying
that clinicians should only be trained in ESTs and other methods are outdated
- ESTs and research methods to validate them include:
o treatments are designed for Axis 1 disorders
o patients are screened to maximize homogeneity of diagnosis and minimize
comorbidity
o treatments are manualized
o treatments are for brief and fixed durations to minimize within-group
variability
o outcome assessment focuses on the symptoms of focus
- these characteristics make sense to maximize internal validity
- goal is to draw unambiguous conclusions about cause and effect by randomly
assigning participants, manipulating a small set of variables, controlling confounding
variables, and standardizing procedures
- this method assumes that:
o psychopathology is highly malleable
o most patients can be treated for a single problem
o psychiatric disorders can be treated independently of personality factors
unlikely to change in brief treatments
o experimental methods provide a gold standard for psychotherapy
- author argues that these assumptions of EST methodology are generally not valid
1. PSYCHOLOGICAL PROCESSES ARE HIGHLY MALLEABLE
- assumption of malleability is obvious in the treatment lengths, typically 6-16 sessions
- when researchers wanted to compare psychotherapies to meds, to avoid
confounding they had to make it same length as medication studies
- the longer the therapy, the more variability within experimental conditions, and the
less one can draw causal conclusions
- brief treatments are a consequence of wanting to standardize treatments and have
them under experimental control
- brief treatments work for some disorders but for many relapse rates are high
- malleability assumption is inconsistent with the dose-response relationships (longer
treatment means more effect)
,- evidence shows enduring rehabilitation requires substantially longer treatment
depending on the patient’s degree and type of impairment
- most psychopathological vulnerabilities are highly resistant to change and may be
rooted in personality and temperament
- evidence shows that changes in state may not be accompanied by changes in
diathesis for those states encoded in implicit networks
2. MOST PATIENTS HAVE ONE PRIMARY PROBLEM OR CAN BE TREATED AS IF THEY DO
- reflects on admixture of methodological constraints
- best to start with pure cases to avoid confounds presented by comorbid patients
- requirement for funding: research proposals are tied to the categories of DSM
the pragmatics of DSM5 diagnosis
- linking treatment research to DSM has advantages: generalizing across diff. settings
and identifying process that might alter psychopathology BUT
- 1. DSM categories are created by committee consensus based on available evidence
rather than by strictly empirical method (e.g. factor analysis)
- 2. the assumption that patients typically present a specific axis I diagnosis that can be
identified at the start of treatment is not valid
- treatment research has gone independently of systematic needs assessment of the
reasons the average patient presents for psychopathology
- data from naturalistic and community studies suggest that almost half of help-
seeking patients present a diffuse picture that can’t be diagnosed using the DSM bc it
doesn’t fit or cross threshold
- 3. it’s unrealistic for clinicians to learn disorder-specific manuals for more than a
handful of disorders
- given that 40-60% of patients don’t respond to first line ESTs, clinicians would have to
learn 2-3 manuals per disorders
- this is problem for shifting manuals to tools for clinical practice
the problem of comorbidity
- evidence suggests that single-disorder presentations are the exception rather than
the rule
- most Axis I disorders are comorbid with Axis I or Axis II disorders
- many advocates for ESTs argue that the best way to approach polysymptomatic
pictures is to use sequential manuals, but this is not optimal bc:
o Axis I disorders reflect common underlying causes
o Axis I disorders arise in the context of enduring personality patterns that crate
vulnerability to future episodes
o presence of multiple symptoms can’t be reduced independently
- one can’t assume that psychopathology is additive and can be treated like that
the function of comorbidity assessment and generalizability to everyday clinical practice
- function of assessing for co-occurring disorders differs in research and practice in a
way that affects the generalizability of ESTs
- researchers screen patients first, and clinicians assessing the EST don’t conduct their
own evaluation and assume that the diagnosis by researchers is accurate and primary
, - function of assessment in the lab: eliminate participants who don’t meet criteria
- assessment in practice/reality: clinicians don’t assume a symptom to be primary and
inquire broadly about the patient’s symptoms, history etc.
- the requisites of experimental control in ETS methodology limit the variation
permitted in case formulation
- in clinical practice, symptoms initially identified as primary don’t often remain the
focus of treatment over time, weeks/months can pass before the patient recognizes
the true source of distress
- ESTs make this^ irrelevant bc:
o testing treatments need to be brief enough, for experimental control and
comparability
o manualization assumes that same techniques should work for the same Axis I
symptom regardless of etiology, the circumstances that caused the
symptoms, personality etc.
3. PSYCHOLOGICAL SYMPTOMS CAN BE UNDERSTOOD AND TREATED IN ISOLATION
FROM PERSONALITY DISPOSITIONS
- ^bc of brief, focal nature of treatment required to max. control, and bc of focus on
syndromes rather than processes/diatheses
- the only treatment considered as EST for PDs is Linehan’s DBT, which takes a year to
complete just for the first stage
- complications of this assumption:
independence of symptoms and personality processes
- most Axis I disorders are not independent of personality and personality often
moderates treatment response
- Axis I mood disorders are related personality variables (high neg, and low pos. affect)
- the prevalence of Axis I disorders is even a proxy for presence of Axis II disorders
(PDs) (more Axis I, greater the likelihood of Axis II)
- the same Axis I symptoms may have different implications in the presence of diff.
personality structures e.g. borderline and non-borderline depression
- researchers might want to develop treatment for neg. affectivity and emotional
dysregulation rather than focusing on DSM-defined syndromes
the paradox of pure samples
- if researchers include patients with personality psychopathology, they have the risk
of ambiguous conclusions, unless the sample is big enough to permit co-variation or
moderator analyses
- but if they exclude these patients, they can’t assume generalizability to a target
population bc pure symptomology is rare
- going from pure to impure samples, from lab to community, might seem like an
appropriate strategy BUT
o this hold the assumption that the polysymptomatic conditions seen in the
community have no emergent properties that might need diff. interventions
o these emergent properties can never be identified if we start with less
complex cases
WESTEN – THE EMPIRICAL STATUS OF EMPIRICALLY SUPPORTED PSYCHOTHERAPIES
Intro
- empirically supported therapies (ESTs)
- this paper argues that: unqualified statement and dichotomous judgments about
validity or invalidity in complex areas are unlikely to be scientifically or clinically
useful, and we should attend more closely to the conditions under which certain
empirical methods are useful in testing certain interventions for certain disorders
- examining the empirical basis of the assumptions that underlie the methods used to
establish empirical support for psychotherapies
retelling the story: the assumptions underlying ESTs
- ESTs gained momentum in 1995 with the publication of a task force by APA saying
that clinicians should only be trained in ESTs and other methods are outdated
- ESTs and research methods to validate them include:
o treatments are designed for Axis 1 disorders
o patients are screened to maximize homogeneity of diagnosis and minimize
comorbidity
o treatments are manualized
o treatments are for brief and fixed durations to minimize within-group
variability
o outcome assessment focuses on the symptoms of focus
- these characteristics make sense to maximize internal validity
- goal is to draw unambiguous conclusions about cause and effect by randomly
assigning participants, manipulating a small set of variables, controlling confounding
variables, and standardizing procedures
- this method assumes that:
o psychopathology is highly malleable
o most patients can be treated for a single problem
o psychiatric disorders can be treated independently of personality factors
unlikely to change in brief treatments
o experimental methods provide a gold standard for psychotherapy
- author argues that these assumptions of EST methodology are generally not valid
1. PSYCHOLOGICAL PROCESSES ARE HIGHLY MALLEABLE
- assumption of malleability is obvious in the treatment lengths, typically 6-16 sessions
- when researchers wanted to compare psychotherapies to meds, to avoid
confounding they had to make it same length as medication studies
- the longer the therapy, the more variability within experimental conditions, and the
less one can draw causal conclusions
- brief treatments are a consequence of wanting to standardize treatments and have
them under experimental control
- brief treatments work for some disorders but for many relapse rates are high
- malleability assumption is inconsistent with the dose-response relationships (longer
treatment means more effect)
,- evidence shows enduring rehabilitation requires substantially longer treatment
depending on the patient’s degree and type of impairment
- most psychopathological vulnerabilities are highly resistant to change and may be
rooted in personality and temperament
- evidence shows that changes in state may not be accompanied by changes in
diathesis for those states encoded in implicit networks
2. MOST PATIENTS HAVE ONE PRIMARY PROBLEM OR CAN BE TREATED AS IF THEY DO
- reflects on admixture of methodological constraints
- best to start with pure cases to avoid confounds presented by comorbid patients
- requirement for funding: research proposals are tied to the categories of DSM
the pragmatics of DSM5 diagnosis
- linking treatment research to DSM has advantages: generalizing across diff. settings
and identifying process that might alter psychopathology BUT
- 1. DSM categories are created by committee consensus based on available evidence
rather than by strictly empirical method (e.g. factor analysis)
- 2. the assumption that patients typically present a specific axis I diagnosis that can be
identified at the start of treatment is not valid
- treatment research has gone independently of systematic needs assessment of the
reasons the average patient presents for psychopathology
- data from naturalistic and community studies suggest that almost half of help-
seeking patients present a diffuse picture that can’t be diagnosed using the DSM bc it
doesn’t fit or cross threshold
- 3. it’s unrealistic for clinicians to learn disorder-specific manuals for more than a
handful of disorders
- given that 40-60% of patients don’t respond to first line ESTs, clinicians would have to
learn 2-3 manuals per disorders
- this is problem for shifting manuals to tools for clinical practice
the problem of comorbidity
- evidence suggests that single-disorder presentations are the exception rather than
the rule
- most Axis I disorders are comorbid with Axis I or Axis II disorders
- many advocates for ESTs argue that the best way to approach polysymptomatic
pictures is to use sequential manuals, but this is not optimal bc:
o Axis I disorders reflect common underlying causes
o Axis I disorders arise in the context of enduring personality patterns that crate
vulnerability to future episodes
o presence of multiple symptoms can’t be reduced independently
- one can’t assume that psychopathology is additive and can be treated like that
the function of comorbidity assessment and generalizability to everyday clinical practice
- function of assessing for co-occurring disorders differs in research and practice in a
way that affects the generalizability of ESTs
- researchers screen patients first, and clinicians assessing the EST don’t conduct their
own evaluation and assume that the diagnosis by researchers is accurate and primary
, - function of assessment in the lab: eliminate participants who don’t meet criteria
- assessment in practice/reality: clinicians don’t assume a symptom to be primary and
inquire broadly about the patient’s symptoms, history etc.
- the requisites of experimental control in ETS methodology limit the variation
permitted in case formulation
- in clinical practice, symptoms initially identified as primary don’t often remain the
focus of treatment over time, weeks/months can pass before the patient recognizes
the true source of distress
- ESTs make this^ irrelevant bc:
o testing treatments need to be brief enough, for experimental control and
comparability
o manualization assumes that same techniques should work for the same Axis I
symptom regardless of etiology, the circumstances that caused the
symptoms, personality etc.
3. PSYCHOLOGICAL SYMPTOMS CAN BE UNDERSTOOD AND TREATED IN ISOLATION
FROM PERSONALITY DISPOSITIONS
- ^bc of brief, focal nature of treatment required to max. control, and bc of focus on
syndromes rather than processes/diatheses
- the only treatment considered as EST for PDs is Linehan’s DBT, which takes a year to
complete just for the first stage
- complications of this assumption:
independence of symptoms and personality processes
- most Axis I disorders are not independent of personality and personality often
moderates treatment response
- Axis I mood disorders are related personality variables (high neg, and low pos. affect)
- the prevalence of Axis I disorders is even a proxy for presence of Axis II disorders
(PDs) (more Axis I, greater the likelihood of Axis II)
- the same Axis I symptoms may have different implications in the presence of diff.
personality structures e.g. borderline and non-borderline depression
- researchers might want to develop treatment for neg. affectivity and emotional
dysregulation rather than focusing on DSM-defined syndromes
the paradox of pure samples
- if researchers include patients with personality psychopathology, they have the risk
of ambiguous conclusions, unless the sample is big enough to permit co-variation or
moderator analyses
- but if they exclude these patients, they can’t assume generalizability to a target
population bc pure symptomology is rare
- going from pure to impure samples, from lab to community, might seem like an
appropriate strategy BUT
o this hold the assumption that the polysymptomatic conditions seen in the
community have no emergent properties that might need diff. interventions
o these emergent properties can never be identified if we start with less
complex cases
