Clinician by patient by psychiatric symptom data

Downloads: fifteen_clinicians.arff, fifteen_clinicians.csv, fifteen_clinicians.RData

If you use this data set in publications please cite
Van Mechelen, I., & De Boeck, P. (1990). Projection of a binary criterion into a model of hierarchical classes. Psychometrika, 55, 677-694.

Van Mechelen, I. (1991). Symptom and diagnosis inference from implicit theories of psychopathology: A review. European Bulletin of Cognitive Psychology, 11, 155-171.

Summary of the data

Contributor:
Iven Van Mechelen
License:
CC0 Creative Commons "No Rights Reserved"

General information about the data

Abstract:
Presence/absence ratings of 23 psychiatric symptoms in vignettes of 30 psychiatric inpatients made by 15 clinicians
Subject matter background:
The data have been collected in a study of 15 clinicians to identify the implicit taxonomies underlying their clinical judgements.
Data structure:
source x object x variable data array
Data objects and variables:
1. sources (15): clinicians (psychiatrists and clinical psychologists)

2. objects (30): vignettes of 30 psychiatric inpatients each of which consisted of the patient’s sex and age, a short description by the patient of his/her problem, and observational data from a nurse of the patient’s ward

3. variables (36):

3.1 technical variables (2): 1. number of clinician 2. number of vignette

3.2 internal variables (23): supposed presence of psychiatric symptoms

3.2.1 The first 19 internal variables were taken from the Psychiatric Evaluation Form (Endicott & Spitzer, 1972): 3. speech disorganization 4. agitation - excitement 5. hallucinations 6. inappropriate 7. disorientation - memory 8. depression 9. anxiety 10. suicide - self-mutilation 11. somatic concerns 12. narcotics - drugs 13. antisocial 14. retardation - lack of emotion 15. social isolation 16. daily routine - leisure time 17. alcohol abuse 18. belligerence - negativism 19. denial of illness 20. grandiosity 21. suspicion - persecution

3.2.2 The final 4 internal variables pertained to the supposed presence of several forms of impairment: 22. intellectual impairment 23. impulse control 24. social blunting 25. role impairment

3.3 external variables (11): All of these pertained to diagnostic judgements with regard to four overarching DSM-III Axis 1 categories.

3.3.1 overall measures (3): 26. certainty about diagnosis (diagnoses) 27. extent to which your set of DSM-III diagnoses correctly reflect the supposed problem of the patient 28. extent to which vignette includes sufficient information to make a diagnosis

3.3.2 specific measures (8): 29. supposed applicability of substance use disorder 30. certainty about preceding applicability judgement 31. supposed applicability of schizophrenic disorder 32. certainty about preceding applicability judgement 33. supposed applicability of depressive affective disorder 34. certainty about preceding applicability judgement 35. supposed applicability of anxiety disorder 36. certainty about preceding applicability judgement

Data values:
1. technical variables: 1: 1-15 (whole number: rank number of clinician); 2: 1-30 (whole number: rank number of vignette)

2. internal variables: symptoms 3-25: 0 (absent) - 1 (present, in part or in whole, to a small or a to large extent)

3. external variables: 26: whole number between 0 (very uncertain) and 9 (very certain); 27: whole number between 0 (very bad reflection) and 9 (very good reflection); 28: whole number between 0 (big lack of information) and 9 (fully sufficient information); 29,31,33,35: 0 (not applicable) - 1 (applicable); 30,32,34,36: whole number between 0 (very uncertain) and 9 (very certain)

no missing values

Preprocessing:
No
Other relevant papers:
result of two-mode overlapping cluster analysis (HICLAS) in relation with diagnostic judgements: Van Mechelen, I., & De Boeck, P. (1990). Projection of a binary criterion into a model of hierarchical classes. Psychometrika, 55, 677-694.

result of three-mode overlapping cluster analysis: Leenen, I., Van Mechelen, I., De Boeck, P., & Rosenberg, S. (1999). INDCLAS: A three-way hierarchical classes model. Psychometrika, 64, 9-24.

results of probabilistic two-mode overlapping cluster analyses and associated model checks: Maris, E., De Boeck, P., & Van Mechelen, I. (1996). Probability matrix decomposition models. Psychometrika, 61, 7-29. Meulders, M., De Boeck, P., & Van Mechelen, I. (2001). Probability matrix decomposition models and main-effects generalized linear models for the analysis of replicated binary associations. Computational Statistics & Data Analysis, 38, 217-233. Gelman, A., Van Mechelen, I., Verbeke, G., Heitjan, D. F., & Meulders, M. (2005). Multiple imputation for model checking: Completed-data plots with missing and latent data. Biometrics, 61, 74-85.

Justification for clustering:
A clustering of the patients may reveal the implicit taxonomic systems of the clinicians under study. A clustering of the symptoms may reveal the symptom patterns that constitute the basis of the implicit taxonomic systems of the clinicians under study. Optionally, a clustering of the clinicians may capture differences between the clinicians with regard to the implicit taxonomic principles they use.

External criteria for clustering quality

External variable that represents the underlying true clustering
One may expect the implicit taxonomies of the clinicians (and the location of the patients in these taxonomies) to be related to the explicit DSM-III diagnoses (29,31,33,35).
Substantive justification
The clinicians can be expected to respect more or less the "offical" nosological DSM-III rules; yet, significant idiosyncratic discrepancies between the implicit and explicit clusterings (reflecting, e.g., treatment-related concerns of the clinicians under study) should come at no surprise either.

Internal criteria for clustering quality: cluster membership

Number of clusters
At least two, reflecting the basic split between schizophrenia-like and affective disorders.
All objects clustered or not
Optionally, more healthy patients could be left out of the clustering.
Cluster overlap
Overlap between patient clusters may reflect implicit syndrome co-morbidity. Overlap between symptom clusters may reflect symptom overlap between implicit syndromes. Overlap between clinician clusters may reflect overlap in the taxonomic principles the clinicians use. In the three cases overlap is optional, and neither nestedness nor a full hierarchy is required.

Internal criteria for clustering quality: within and between cluster features

Ground for objects to belong to the same cluster:
Common pattern of symptoms or at least symptoms that meaningfully go together (e.g., related to a dysfunctioning in the same system such as the affective one)
Weight: 4, moderately important

Further aspects

-