OToPS/ABACAB (NIH R01 MH066647)/LEAD Diagnoses

Overview of LEAD

edit

(Lead section)

Data sources

edit

(stuff)

Source Documents

edit

The KSADS version used in ABACAB was an amalgam of two other versions. The source documents ("binders") were a photocopy of the master document. Each rater received their personal copy as a 3 ring binder, which they used for each interview.

 
Screenshot of diagnosis codes sorted in ascending numerical order (more helpful for people using the dataset).
 
This is a screenshot of a portion of the record sheet for consensus diagnosis conferences that compared findings from a semi-structured diagnostic interview (KSADS) with the diagnoses made by the intake clinician using regular unstructured interview, and a LEAD process that added clinical observations, family psychiatric history, and prior treatment history.

This file contains the record sheet for the LEAD consensus conferences (p. 1), along with the values and labels for the axis I and axis II diagnoses (sorted in alphabetical order by diagnosis name, to simplify the search for a human rater), followed by consensus meeting guidelines, and then the diagnosis values and labels sorted numerically (which is more useful for data cleaning, scoring, and analysis).

 
Screenshot of the start of an alphabetical list of DSM-IV diagnoses and their corresponding codes for variables used in the dataset.

This spreadsheet was used to record scores for each item on the KSADS interview. If the interview was for rater training or inter-rater reliability, then a second (and sometimes multiple) raters would be entered. There is SPSS code to compare the rater responses and calculate a form of weighted kappa, using methods described in Castellan & Siegel.

Scoring instructions to raters in NIH R01MH066647 (ABACAB)

edit

LEAD Scoring Instructions:

  1. KSADS diagnosis is default.
  2. Additional evidence can trump the KSADS, but it should follow the spirit of Meehl’s “broken leg” in clinical decision making. Examples:
    • Rater 1 having a different clinical impression than the KSADS, and corroborating evidence found in chart and/or treatment history.
    • Treatment history may resolve diagnostic ambiguity (e.g., adverse response to stimulant and positive response to mood stabilizer may suggest a bipolar diagnosis instead of ADHD).
    • Family history information not considered in the KSADS may also provide important information potentially changing the diagnostic formulation.
  3. LEAD can add diagnoses not assessed on the KSADS (this will be especially common with V-codes, pervasive developmental disorders, and possibly adjustment disorders). For these diagnoses, LEAD would add the diagnosis based on chart information unless the symptoms are better accounted for by a diagnosis based on the KSADS. Examples:
    • A KSADS diagnosis of major depression might eliminate a chart diagnosis of an adjustment disorder.
    • A KSADS diagnosis of ODD might bump off parent-child conflict V-code.
  4. Applewood sometimes labels Axis I diagnoses as “provisional.” Note that this is a “provisional diagnosis” in the ACI text box, but go ahead and treat this as a formal diagnosis (i.e., assign it the appropriate code number, and treat it as an “over the line” diagnosis and not a “rule out”).
  5. When combining GAFS, go ahead and average them and round towards the KSADS GAF. For example, KSADS GAF = 40 and intake GAF = 55 yields a LEAD GAF of 47 (rounding the average of 47.5 down, towards the KSADS score). The LEAD team can further adjust the GAF score, but a note documenting that there was a variance and the rationale for it should be added. If multiple raters complete KSADS (either from training or for reliability), then average their GAF ratings to come up with a KSADS GAF estimate, and then follow the procedure above.
  6. SUBTRACTING DIAGNOSES: If the rater feels that a diagnosis met at threshold on the KSADS actually is inappropriate or the symptoms are subsumed in another diagnosis, then this should be documented in the “KSADS Dx” column under the double line “Rater 1 Clinical Dx (if Different)” as “subtract ADHD.” Then add a notes in the “Workspace and Notes” explaining why the diagnosis was subtracted. The subtracted diagnosis should NOT be reported in the LEAD column (unless countervailing data suggests adding it back).

Data and analysis

edit
 
(See previous)

This is a screenshot of the codebook that superimposes the variable names and codes on the source document.