Title: Formal Diagnostic Assessment
1Formal Diagnostic Assessment
- Based on
- life history facts
- signs (objectively observable abnormalities)
symptoms (subjective abnormalites), present
past - abnormal personality traits
- 3 ways to get diagnosis
- Inattention to criteria unstructured interview
- Criteria unstructured interview
- Criteria structured interview
2Example Diagnosis (No Criteria, DSM-II)
3Example Diagnosis (DSM-IV Criteria)
4DSM-IV Criteria, contd
5DSM-IV Criteria, contd
6DSM-IV Criteria, contd
7Traditional Diagnostic Interview
- Patient identification
- Chief complaint
- History of present illness
- Past illnesses
- Social history
- Medical history
- Family history
8Diagnostic Interview, contd
- Mental Status Exam (see next slide)
- Impression
- definite, or rule outs
- Plan (treatment, referral, etc.)
- The unstructured interview takes 4590 min
- Only pursuing 1 or 2 problem areas
- Only asking enough questions to establish likely
diagnosis
9Mental Status Exam
- Appearance
- Consciousness level
- Cooperativeness
- Speech abnormalities
- Cognitive abnormalities
- Judgment abnormalities
- Content of thought
- Mood affect
- Psychotic features
10Structured Interviews
- Tied to particular sets of criteria (e.g.,
Research Diagnostic Criteria, DSM-IV) - Such interviews spell out
- What questions to ask
- What order to ask them in
- (Often) how to ask them, how to follow them up
- (Often) how to interpret (code) the answers
11Example Structured Clinical Interview for DSM-IV
- About 100 pages
- Covers affective, psychotic, psychoactive
substance use, anxiety, somatoform, eating,
adjustment disorders - Nominally takes 1 hr, usually 24 hr
- Used in much research, some clinical work
12Disadvantages of Structured Interviews
- Long
- Boring for clinician
- Some allege they interfere with client rapport
(balk data refute this) - May not cover all disorders appropriate for
particular clients
13Advantages (?) of Structured Interviews
- May be more reliable than unstructured (scant
evidence) - Much better ability to follow criteria
- More symptoms, diagnoses found
- Higher clinician confidence in assessment
- No evidence that it leads to better client
outcomes/better prediction of future behavior
14Reliability of Diagnoses
- Vary over wide range, from zero to .90, even
within studies - Somewhat better with criteria than without, on
average - For diagnoses made without criteria, reliability
not as bad as often claimed - Reliability at best mediocre for Axis II
diagnoses (.6 down to 0), and for some Axis I
diagnoses as well
15Validity of Diagnoses
- Difficult to assess, absent gold standard
- Need well-replicated data to assess
- Clinical studies of discreteness of syndromes,
overlap with other syndromes - Genetic/Family studies
- Etiological/Laboratory studies
- Course outcome studies
- Treatment response studies? (hard to interpret)
16Spitzer on Establishing Validity
- To establish a yardstick in diagnostic validation
studies - Cant use RDC or DSM as criterion, since its
these we wish to test - LEAD rather than gold standard
- Longitudinal (course of illness over time)
- Expert (clinicians)
- All Data (multiple informants, family history,
etc.)
17Problems With Nomenclatures
- Exuberant growth of diagnostic lists
- Feighner (St. Louis group) criteria 14
diagnoses, importantly including Undiagnosed - Research Diagnostic Criteria 26 diagnoses,
including two other diagnoses - DSM-III 207 (!) diagnoses, including NOS
diagnoses - DSM-IV 400 diagnoses
18Nomenclature Problems, contd
- Inclusion of many totally unvalidated disorder
categories (e.g., dependent personality disorder) - Existence of many, only weakly validated,
categories (e.g., multiple personality disorder) - Existence of many wastebasket categories
other, atypical, especially not otherwise
specified
19Problems with Classifications
- Question of true structure of phenomena of mental
disorders - Categorical?
- Dimensional?
- Some of both?
- Which disorders are of which kind?
20Problems with Diagnostic Process
- Errors/fallacies
- Ignoring base rates, failing to properly adjust
for them Barnum statements (Meehl) - Biases e.g., confirmatory overweighting (,)
cell in mis-estimating correlations
overpathologizing acceding to authority - Heuristics e.g., representativeness
availability anchoring adjustment - Illusory correlation illusion of validity in
the absence of meaningful feedback
21Problems with Diagnostic Process, contd
- Lack of demonstrable relationship of
- Experience to judgment accuracy
- (Almost) all training to judgment accuracy
- Impact of non-rational factors on diagnosis
- Stereotyping
- Under/overpathologizing
- Overstatement of diagnosis in third-party
reimbursement diagnostic creep
22Problems with Diagnostic Process, contd
- Reliance on client report for too many crucial
data - Near impossibility of substantiating many
reported data (e.g., PTSD trauma) - Consequent vulnerability to faking bad
(exaggeration, malingering) faking good
(defensiveness, denial)
23In Defense of Diagnosis
- The better is the enemy of the good
- Critics offer no replacement for nomenclature
that - Will likely have notably better validity, etc.
- Will cure diagnosiss drawbacks, without causing
equally vexing problems of some other kind - Will be practical
- Ready to go (not pie-in-sky plan for potential
replacement) - User-friendly for clinical practice
24In Defense of DSM
- Compare DSM not to pre-drawing-board
alternatives compare to predecessors - Advantages of DSM-III et seq. over DSM-II
- Clearer, more communicative
- More informative (manual diagnoses)
- Somewhat more reliable, for many diagnoses
- If followed strictly, eliminates various biases
(e.g., diagnostician race bias in diagnosing
schizophrenia)