Elsevier

Biological Psychiatry

Volume 57, Issue 1, 1 January 2005, Pages 96-101
Biological Psychiatry

Original articles
Reliability of psychiatric diagnosis in postmortem research

https://doi.org/10.1016/j.biopsych.2004.10.016Get rights and content

Background

Postmortem human brain research is an important approach for identifying the cellular, molecular, and genetic pathways involved in the pathophysiology of psychiatric disorders. One critical component in postmortem research is the reliability of psychiatric diagnoses used to define study cohorts. Finding reliable methods for assessing lifetime psychiatric diagnoses in subjects after death is extremely challenging.

Methods

Two commonly used approaches were compared: psychiatric record reviews and postmortem family interviews. We hypothesized that these two methods would lead to more diagnostic agreement for subjects with schizophrenia than those with mood disorders. For 37 cases, psychiatric records were reviewed retrospectively using the Diagnostic Evaluation After Death, and family members were interviewed using the Structured Clinical Interview for DSM-IV.

Results

Comparison of diagnoses derived from these two approaches generated an overall kappa coefficient of .67. Kappa coefficients for the schizophrenia cohort were .94, .68 for the major depressive disorder cohort, and .58 for the bipolar disorder cohort.

Conclusions

Thus, although it may be sufficient to establish the postmortem diagnosis of schizophrenia using one of the two methods, the best method for reaching an accurate postmortem diagnosis for mood disorders is more difficult to determine and requires further study.

Section snippets

Subjects

The majority of human brain specimens were obtained following autopsy through the Offices of the Chief Medical Examiner of the District of Columbia or Northern Virginia. Four cases were obtained from hospitals or funeral homes through direct donations by the next of kin. Cases were referred directly from the medical examiner by telephone communication, and initial telephone contact with families was made within 48 hours of death. Informed consent was obtained for all subjects by either a

Results

The kappa coefficient of diagnostic reliability for the overall group (N = 37) was .67 (p < .001; see Table 2). Agreement between the two diagnostic methods generated a kappa coefficient of .94 (p < .001) for the schizophrenia cohort, which was a collapsed group including all subtypes of schizophrenia and schizoaffective disorder. Disagreement for schizophrenia occurred in only one subject, who had a psychiatric record diagnosis of psychosis not otherwise specified (which for the purposes of

Discussion

While the rate of agreement between psychiatric diagnoses derived from record reviews and diagnoses derived from postmortem family interviews was relatively high for subjects with a diagnosis of schizophrenia, the rate of agreement for subjects with mood disorders was only moderate. This replicates the finding in a previous study by Kelly and Mann (1996) in which 100% agreement was achieved for nine subjects with schizophrenia. Because individuals with schizophrenia generally have more frequent

References (19)

There are more references available in the full text version of this article.

Cited by (0)

View full text