Development of Diagnostic Variables

1
Development of Diagnostic Variables

• Diagnoses have been created for both the DSM-IV
  and ICD-10 systems
• WMH CIDI Advisory Committee instrumental in this
  process
• Instrument Development Phase
• During development phase of survey, CIDI
  questions were designed to assess each criterion
  necessary for a diagnosis.
• Experts in each field were consulted for best way
  to assess each aspect of the diagnosis
• Studied existing CIDI 2.1 as well as all standard
  research instruments for assessing diagnoses

2
Diagnostic Algorithms

• Algorithm Development Phase
• Once CIDI 3.0 was finalized, a team of
  researchers and programmers developed SAS code to
  operationalize each diagnostic criterion from
  questions in instrument
• Clinical Calibration
• Validity studies have been done in
• US, Italy, France, Spain, China, Nigeria, India
• Iterative process continues to date
• Updates/Improvements to the recent versions of
  the CIDI 3.0
• Revisions to the diagnostic algorithms are made
  based on this analysis.
• Minor revisions when a particular item does not
  work or a threshold should be modified to improve
  concordance (SO)
• Major revisions when analysis proves that the
  cidi was grossly overestimating a particular
  disorder (bipolar I and bipolar II)
• Algorithms released are the most recent as of
  Feb, 2006. We will not be updating the
  diagnostic data file available for public
  release, however, changes will be posted in the
  diagnostic algorithm section of the ncs website.

3
Diagnostic Variables available through the Public
Release dataset

• ATTENTION DEFICIT DISORDER
• AGORAPHOBIA
• ALCOHOL ABUSE with or without dependence
• ALCOHOL DEPENDENCE with Abuse
• ADULT SEPARATION ANXIETY DISORDER
• BIPOLAR I
• BIPOLAR II
• BIPOLAR SUBTHRESHOLD
• CONDUCT DISORDER
• DRUG ABUSE with or without dependence
• DRUG DEPENDENCE with abuse
• DYSTHYMIA
• GENERALIZED ANXIETY DISORDER
• HYPOMANIA
• INTERMITTENT EXPLOSIVE DISORDER
• MANIA
• MAJOR DEPRESSIVE DISORDER
• MAJOR DEPRESSIVE EPISODE

4
Diagnostic Hierarchy Rules

• Some DSM-IV diagnoses contain a criterion called
  a hierarchy rule.
• E.g. If meet GAD but only during a mood disorder
  do not receive gad diagnosis. (SEE DSM-IV GAD
  criterion F)
• The disturbance does not occur exclusively
  during a Mood Disorder, a Psychotic Disorder, or
  a Pervasive Developmental Disorder
• In these cases, we create two diagnostic
  variables one with hierarchy (narrow
  definition) and one without hierarchy (broad
  definition that does not operationalize the
  hierarchy criterion).
• Important for studies of comorbidity.
  Researchers discretion which version to use but
  version must be clearly stated in all reports.

5
Hierarchy Example

• Criterion F.
• Part 2. The disturbance does not occur
  exclusively during a Mood Disorder, a Psychotic
  Disorder, or a Pervasive Developmental Disorder.
• Note Psychotic Disorder and Pervasive
  Developmental Disorder hierarchies are not
  operationalized.
• (Major Depression No(5) AND Minor
  Depression No(5) AND Dysthymia No(5) AND
  Mania No(5)) OR
• ((Major Depression Yes(1) OR Minor
  Depression Yes(1) OR Dysthymia Yes(1) OR
  Mania Yes(1)) AND ( (GAD onset lt Mood onset) OR
  (GAD recency gt Mood recency) OR (GAD persistence
  gt Mood persistence))) OR
• G10e No(5)

6
Diagnostic Variables w/Hierarchy
DSM-IV Disorder Hierarchical Disorder Alcohol
Abuse with hierarchy Alcohol dependence Drug
Abuse with hierarchy Drug dependence Dysthymia
with hierarchy MDE, Mania, Hypomania GAD with
hierarchy MDE, MND, DYS, Mania IED with
hierarchy Mania, ALA, ALD, DRA, DRD, MDE,
hypomania MDD with hierarchy Mania,
hypomania ODD with hierarchy MDE, MND, Mania,
CD
7
Organic Exclusion

• Many DSM-IV diagnoses contain a criterion called
  organic exclusion.
• This criterion has been operationalized using a
  standard format across CIDI sections.
• DXA. Episodes of this sort sometimes occur as the
  results of physical causes such as physical
  illness or injury or the use of medication,
  drugs, or alcohol. Do you think your episodes
  ever occurred as the result of such physical
  causes?
• DXB. Do you think your episodes were always the
  result of physical causes?
• In any interview schedule where this question
  (DXB) is yes we ask the follow-up question
• DXC. Briefly, what were the physical causes?
• All open ended text from the organic exclusion
  item DXC have been reviewed by a psychiatrist for
  the NCS-R and a determination has been made as to
  whether it is a qualifying organic exclusion.
• If, it was determined that there is no qualifying
  organic exclusion, and the respondent meets all
  other criteria, then the respondent has be
  hard-coded as meeting the diagnostic criteria for
  the disorder in the SAS code .

8
Programming Conventions (1)

• Diagnostic assignment accomplished by a series of
  SAS macros
• We provide word documents that give detailed
  descriptions of the sas code but we do not
  release the code.
• Onset and Recency are determined by looking at
  the Minimum of any onset item and the Maximum of
  any recency item

9
Programming Conventions (2)

• Standard notation
• Lifetime Diagnosis dsm_dx icd_dx
• 12 M Diagnosis d_dx12 i_dx12
• 30 day Diagnosis d_dx30 i_dx30
• Onset Age dx_ond dx_oni
• Recency Age dx_recd dx_reci
• Hierarchy example
• Dx w/out hierarchy dsm_dx
• Dx w/ hierarchy dsm_dxh d_dxh12 d_dxh30
  icd_dxh i_dxh12 i_dxh30
• Presence/absence of each criterion established
  first
• Allowable values yes/no/dont know/refuse
• Hierarchy of yes/no/dk/ref for Criteria
  variables if not yes
• if any no ? no, if any dk ? dk, if any ref ?
  ref
• Standard rules to compile criteria into final
  diagnosis (yes/no)

10
Imputations of missing data

• No imputations of Diagnostic Disorders
• Imputation of onset and recency
• hot deck imputation
• Rational imputation
• Imputation of demographic/constructed variables.
• Regression based, hot deck and rational

11
(No Transcript)