Opens in a new tab

SCID-5 Clinical Interview:
Research Use and Analysis

The Structured Clinical Interview for DSM-5 (SCID-5) is a semi-structured, clinician-administered interview that records whether DSM-5 diagnostic criteria are met. The SCID-5 clinical interview produces categorical diagnoses and specifiers and no numerical total, so this guide explains the versions, how diagnostic outcomes are analyzed, rater agreement, and reporting wording for researchers.

ITEMSVersion and module dependent
SUBSCALESDiagnostic modules
RESPONSE SCALEClinician-coded diagnostic criteria
SCORE RANGEDiagnoses and specifiers, no total
DEVELOPEDFirst, Williams, Karg and Spitzer, 2016 (SCID-5-CV)

What the SCID-5 Records

First, Williams, Karg and Spitzer (2016) authored the Clinician Version (SCID-5-CV), one of several SCID-5 products; the others serve research, clinical trial and personality disorder assessment and differ in coverage and time frame. In every version, a trained interviewer judges each criterion as the interview proceeds, and the record of those judgments is what later reaches the dataset.

This post is a methods note for researchers who analyze data that come from the interview. It is not a guide to making diagnoses, which is the work of trained clinicians applying the criteria and the version’s own procedures. For the research analyst, the practical issues are which version was used, how outcomes were coded and how agreement between interviewers was assessed.

Structure of the SCID-5

  • Versions: the Clinician Version (SCID-5-CV), the Research Version (SCID-5-RV), the Clinical Trials Version (SCID-5-CT) and the Personality Disorders version (SCID-5-PD) differ in coverage, detail and intended use. Author lists and publication years differ between products, so cite the exact product administered.
  • Modules: diagnostic modules and their branching depend on the version. There is no fixed item count.
  • Output: diagnostic status, often current and lifetime, and specifiers as defined for the module. There is no universal numerical total and no cutoff.
  • Administration: carried out by a trained interviewer, and the interviewer’s background and training belong in the methods.
  • Translations: name the language of the interview and the validation evidence for that translation. If none exists, say so and describe how the version was adapted.

The Clinician Version is described in First et al. (2016), and each other version has its own manual.

How SCID-5 Outcomes Are Recorded and Analyzed

  1. Record the product, modules administered and reference period (current, lifetime or both) for each participant.
  2. Code each diagnostic outcome as a categorical variable, with an explicit category for not assessed, indeterminate or missing, and do not recode these as negative.
  3. Report prevalence or group comparisons for the categories, with confidence intervals, and do not construct a total from the number of criteria met unless the study protocol defines and justifies one.
  4. For agreement between interviewers, use a statistic suited to categorical ratings, such as Cohen’s kappa for two raters, on cases rated independently.
SPSS SYNTAX

* Illustrative syntax: dx_rater1 and dx_rater2 hold one diagnostic status coded 0 = not met, 1 = met by two raters.
CROSSTABS
  /TABLES=dx_rater1 BY dx_rater2
  /STATISTICS=KAPPA
  /CELLS=COUNT.
R

# data: one row per participant; dx_rater1 and dx_rater2 coded 0 = not met, 1 = met.
library(irr)
kappa2(data[, c("dx_rater1", "dx_rater2")])

Interpreting SCID-5 Results

Result Interpretation
Criteria met The assessed disorder’s criteria were met under the stated version, module and reference period
Criteria not met The criteria were not met under the same conditions
Indeterminate or not assessed Report separately; do not recode as not met

Source: First et al. (2016); the interview gives categorical outcomes and no numerical total.

A diagnostic status comes from the interview and the applicable criteria, not from a numerical sum. For analysis, treat each status as a categorical variable defined for a specific version, module and time frame, and avoid comparing outcomes across versions without checking that their coverage matches.

This post does not provide diagnostic guidance. Diagnosis requires a trained clinician, and research classifications should not be used to inform decisions about an individual.

Psychometric Properties

Property Published finding Source
Clinical validity and reliability Clinical validity against reference diagnoses was reported for the SCID-5-CV in a study of 180 patients, together with intrarater and test-retest reliability estimates Osório et al. (2019)
Interviewer agreement Compute and report agreement on cases rated independently in your own study Study-specific
Coverage Depends on the version and module First et al. (2016)
Training Interviewer training and experience affect outcomes and should be reported Study-specific

These are source-specific findings and cannot be assumed to hold as reliability or accuracy estimates in a new sample.

Using the SCID-5 in a Study

Report the edition and modules, reference periods, interviewer credentials and training, and whether interviewers were blind to group or hypotheses. Analyze diagnostic outcomes as categories, using logistic or log-binomial models for binary outcomes, and report how indeterminate or missing assessments were handled. Where the interview defines the outcome, a diagnostic-accuracy question about another instrument is analyzed against it as a reference standard.

Agreement should be estimated on independently rated cases and reported with a confidence interval, using Cohen’s kappa for two raters or an intraclass or multi-rater statistic when the design has more. Kappa depends on prevalence, so report the observed agreement and the category prevalence alongside it. Plan sample size from the primary comparison, bearing in mind that rare disorders may need larger samples to estimate agreement.

Reporting the SCID-5

Diagnostic status was assessed with the Structured Clinical Interview for DSM-5 [version, e.g., Clinician Version (SCID-5-CV); First et al., 2016], covering [modules] for [current, lifetime or both] diagnoses. Interviews were conducted by [interviewer role and training] who were [blind or not blind to group]. Interrater agreement was [kappa = XX, 95% CI XX to XX] on [number] independently rated cases. Indeterminate or missing assessments were [handled as stated]. The statistical model was [model].

References

First, M. B., Williams, J. B. W., Karg, R. S., & Spitzer, R. L. (2016). Structured Clinical Interview for DSM-5 Disorders, Clinician Version (SCID-5-CV). American Psychiatric Association Publishing.

Osório, F. L., Loureiro, S. R., Hallak, J. E. C., Machado-de-Sousa, J. P., Ushirohira, J. M., Baes, C. V. W., Apolinario, T. D., Donadon, M. F., Bolsoni, L. M., Guimarães, T., Fracon, V. S., Silva-Rodrigues, A. P. C., Pizeta, F. A., Souza, R. M., Sanches, R. F., dos Santos, R. G., Martin-Santos, R., & Crippa, J. A. S. (2019). Clinical validity and intrarater and test-retest reliability of the Structured Clinical Interview for DSM-5, Clinician Version (SCID-5-CV). Psychiatry and Clinical Neurosciences, 73(12), 754-760. https://doi.org/10.1111/pcn.12931

Frequently Asked Questions

No. It records categorical diagnostic outcomes and specifiers for the modules administered, so there is no universal total or cutoff.

They are different products with different coverage and detail. The Clinician Version is shorter, while the Research Version is more detailed. Cite the exact version you administered.

A trained interviewer familiar with DSM-5 criteria. Report the interviewers’ background and training in the methods.

Use Cohen’s kappa for two raters on independently rated cases, with its confidence interval, and give the observed agreement and prevalence alongside it.

NEXT STEP

Need SCID-5 Outcome Data Analyzed
or Checked?

OnlineSPSS is a statistical consultancy for researchers. Share your design and dataset, and a statistically trained Research Coordinator reviews it. The scope and one fixed project fee are confirmed before work begins.

No Commitment Required

Strictly Confidential

Fixed Project Fee

In-House PhD-Level Statisticians

Same-Day Response

Serving Researchers Since 2014