Elsevier

Journal of Psychiatric Research

Volume 32, Issue 6, 1 September 1998, Pages 353-360
Journal of Psychiatric Research

Diagnosing depression in the medically ill: validity of a lay-administered structured diagnostic interview

https://doi.org/10.1016/S0022-3956(98)00031-4Get rights and content

Abstract

Understanding the validity of structured psychiatric diagnostic interviews in medically ill patients will advance the ability to conduct research into the treatment and management of these disorders in general medical settings. We compared the University of Michigan version of the CIDI Fulop et al., 1987 (Composite International Diagnostic Interview) for major depression to a clinical gold standard, derived through Spitzers Longitudinal, Expert, All Data Fulop et al., 1987 (LEAD) criteria based on the SCID-III-R. A convenience sample of medical inpatients was administered the SCID-III-R and the CIDI for major depression in random order. A physician panel reviewed the SCID interview and other pertinent data and determined whether patients had a lifetime or current Fulop et al., 1987 (past month) diagnosis of major depression. The CIDI was scored with and without hierarchical exclusions for mania, hypomania, substance use, or medical illness. When the UM-CIDI was scored for a lifetime diagnosis of major depression without hierarchical exclusions, agreement above chance Fulop et al., 1987 (κ) was very good Fulop et al., 1987 (κ=0.67) between the CIDI and the physician panel and good Fulop et al., 1987 (κ=0.46) when the UM-CIDI was scored with exclusions. Agreement above chance for diagnosis of a recent disorder was better for UM-CIDI scoring with exclusions Fulop et al., 1987 (κ=0.51) compared to scoring without exclusions Fulop et al., 1987 (κ=0.43). Predictive value-positive was excellent in both scoring versions for a lifetime diagnosis Fulop et al., 1987 (82%) and good to very good for current depression Fulop et al., 1987 (46% and 62%). In all cases predictive value-negative was very good to excellent (77–93%). Discordant cases were almost uniformly due to difficulties in attribution of symptoms to medical illnesses. We conclude that the CIDI can perform acceptably as a research instrument to diagnose major depression in medically ill patients, potentially supplemented by clinician review of cases identified by the CIDI with current disorder.

Introduction

The past three decades have brought substantial advances in establishing the validity of diagnoses for psychiatric illness. Gelder et al., 1996 note two main roles for appropriate classification and diagnosis in psychiatry. The first is to serve as a means of communication among clinicians. The second is to provide a reliable and valid system to facilitate research. In the United States, classification of mental disorders was initially driven by the need to gather statistical data on the prevalence of different forms of mental illness Fulop et al., 1987 (American Psychiatric Association, 1994). Furthermore, the discovery of effective therapies for specific disorders necessitated the need for reliable and valid diagnostic tools.

While there has been great progress in the development of research methodology in general, there remain populations of patients for whom screening and diagnostic methods have not been well standardized. The depressed medically ill is one such population Fulop et al., 1987 (Kathol and Petty, 1981; Petty, 1987; Kathol et al., 1990a). Symptoms associated with a major depressive episode Fulop et al., 1987 (e.g. weight loss, fatigue, sleep difficulties, etc.) are frequently also present in medically ill patients Fulop et al., 1987 (Cameron, 1990; Pomerantz et al., 1992). Thus, diagnosing major depression is a difficult task even for skilled diagnosticians. In research efforts, this difficulty is compounded by the need to establish a diagnosis in medically ill patients with a minimum of resources. Kathol and coworkers recognized these difficulties and noted the need for the validation of diagnostic tools for the depressed medically ill Fulop et al., 1987 (Kathol et al., 1990b). Psychiatric disorders, particularly depression, are common in medically ill patients Fulop et al., 1987 (Fulop et al., 1987; Levenson et al., 1990; Katon and Schulberg, 1992; Coyne et al., 1994) and are associated with increased costs and negative outcomes Fulop et al., 1987 (Fulop et al., 1987; Levenson et al., 1990; Simon et al., 1995; Henk et al., 1996; Unutzer et al., 1997). Therefore, diagnostic tools are needed in order to conduct valid research on outcomes and interventions in this high-risk and clinically challenging population.

The assignment of a psychiatric diagnosis can be achieved through several different methods, including the traditional psychiatric interview, the lay-administered structured clinical interview and the semi-structured clinical interview. The traditional psychiatric interview is almost universally used to arrive at a clinical diagnosis in routine practice. In addition to the information obtained from the psychiatric interview, the psychiatrist may also use data obtained from a medical record, laboratory values and other secondary sources in the diagnostic process. As would be expected, the resulting diagnosis can be affected by many factors including the clinicians interviewing skills, diligence and experience Fulop et al., 1987 (Spitzer and Fleiss, 1974; Spitzer et al., 1978). As a result, studies have documented that clinical interviews have low reliability and low validity.

Lay-administered structured diagnostic interviews were developed, following a demand for more standardized methods of diagnostic assessment, especially for research purposes and the need for diagnostic data in large community samples. Such studies required the availability of brief and structured interviewing tools, capable of being administered by non-clinicians. Lay interviewers administer these instruments word-for-word and the diagnostic decision is made on the basis of the respondents answers alone, not on interviewer judgment. Frequently the diagnostic decision is made by computer scoring. These interviews generally emphasize reliability, or the likelihood of two interviewers obtaining the same diagnosis, rather than validity. Among these are the National Institute of Mental Health Fulop et al., 1987 (NIMH) Diagnostic Interview Schedule Fulop et al., 1987 (DIS) Fulop et al., 1987 (Robins et al., 1981) and the Composite International Diagnostic Interview Fulop et al., 1987 (CIDI)Fulop et al., 1987 (Robins et al., 1988). Developers of the original DIS Fulop et al., 1987 (DSM-III version for depression) reported a κ of 0.69, sensitivity of 80% and specificity of 84% between lay interviewers and psychiatrist judgment Fulop et al., 1987 (Robins et al., 1981).

Developed from the DIS, the CIDI incorporates criteria from the International Classification of Diseases Fulop et al., 1987 (ICD) Fulop et al., 1987 (Robins et al., 1988) diagnosis as well as DSM criteria. Version 1.2 of the CIDI, utilizing DSM-III-R and ICD-10 diagnoses, has demonstrated good overall agreement Fulop et al., 1987 (κ=0.78) with a DSM-III-R checklist, with highest agreement for the diagnosis of depressive disorders Fulop et al., 1987 (κ=0.84) Fulop et al., 1987 (Janca et al., 1992). Interrater reliability for the diagnosis of major depression is high Fulop et al., 1987 (κ=0.97) Fulop et al., 1987 (Wittchen, 1994). In addition, the CIDI provides the ability to make a standard psychiatric diagnosis across different cultures and nationalities. Psychometric properties of the CIDI in various settings and studies are described in a comprehensive review article by Wittchen, 1994. The CIDI was further modified for the National Comorbidity Survey Fulop et al., 1987 (NCS) Fulop et al., 1987 (Kessler et al., 1994), a recent national prevalence study, to improve lifetime recall as well as to provide a more detailed assessment of the clinical significance for affective episodes. This modified interview is subsequently referred to as the UM-CIDI.

The advantages of utilizing a structured interview for psychiatric diagnosis in research are many and well-known. Interviews such as the DIS and the CIDI have been utilized in large national epidemiological studies and can therefore allow for national comparisons. In addition, because they are administered by nonclinicians they are less expensive to administer than interviews by clinicians. Limitations to this type of diagnostic interview include reliance on the subject as the only source of information and the inability of the non-clinician lay interviewer to evaluate other sources of data including the medical record and family informants.

The semi-structured clinical interview combines the systematic criteria coverage of the structured interview with the clinicians interviewing skills in an effort to increase the validity of the diagnosis while maintaining high reliability. The Structured Clinical Interview for DSM-III-R Fulop et al., 1987 (SCID) is a frequently used semi-structured clinical interview developed by Spitzer and coworkersFulop et al., 1987 (Spitzer et al., 1992; Williams et al., 1992) in an attempt to integrate the skills of an experienced clinician into a standard diagnostic assessment procedure. The SCID allows the interviewer to phrase questions in a manner appropriate to the patients cognitive abilities and to call upon other sources of information Fulop et al., 1987 (e.g. medical history). The SCID has been implemented extensively in both research and clinical settings and has established reliability and validity through multiple trials Fulop et al., 1987 (Skre et al., 1991; Williams et al., 1992). Though widely accepted as a diagnostic tool, the SCID has a source of potential error, in that it is based on the assessment of one clinician and therefore influenced by individual interviewing skills and style Fulop et al., 1987 (Spitzer et al., 1992). In addition, it is costly to administer on a routine basis, especially in research, because it requires an interviewer with clinical experience and substantial knowledge of psychopathology.

Attempts to develop a definitive diagnostic criterion to act as a gold standard have continued. In his presidential address at the 73rd Annual Meeting of the American Psychopathological Association, Spitzer proposed his LEAD standard Fulop et al., 1987 (Spitzer, 1983). This method incorporates three critical components Fulop et al., 1987 (1) Longitudinal evaluation of symptomatology,Fulop et al., 1987 (2) Expert clinicians making a consensus diagnostic decision and Fulop et al., 1987 (3) All Data from multiple sources utilized in the diagnostic process. Inclusion of these important components can provide a standard against which less elaborate and less expensive diagnostic tools can be compared for diagnostic validity. Use of the LEAD criteria can be particularly helpful as a standard for evaluating difficult diagnostic issues, such as depression in the medically ill.

We undertook this study to understand the validity of a lay-administered structured diagnostic interview for major depression in medically hospitalized inpatients. We compared the UM-CIDI to a clinical gold standard, derived through the LEAD method.

Section snippets

Study sample

A non-random sample of 54 patients was enrolled in the study from medical and surgical inpatient units at a university-affiliated midwestern Department of Veterans Affairs Medical Center between February 1993 and January 1994. All subjects were required to have been hospitalized for a medical diagnosis, rather than for psychiatric or neurological reasons. Potential subjects were referred to the study research assistants by their medical and surgical inpatient physician providers on designated

Results

The SCID-informed physician panel, or LEAD diagnosis, determined that 35% had a lifetime diagnosis of major depression and 17% of the sample were currently depressed. Using the most general UM-CIDI diagnosis without hierarchy Fulop et al., 1987 (DEP1), 31% were determined to have lifetime depression and 24% were determined to have current depression. Twenty percent of the sample had a lifetime diagnosis and 15% had a current disorder by the UM-CIDI scoring with exclusions for medical illness

Discussion

Agreement between the UM-CIDI and a LEAD diagnosis was very good Fulop et al., 1987 (κ=0.67) for the diagnosis of lifetime major depression in medically ill inpatients when omitting the scoring hierarchy for medical illness and other comorbid conditions Fulop et al., 1987 (DEP1), but less strong Fulop et al., 1987 (κ=0.46) once the hierarchy was imposed Fulop et al., 1987 (DEP7). This finding is consistent with a previous study by Knauper, who found that the elderly more often attribute

Acknowledgements

Funding for this project was provided by the Department of Veterans Affairs Health Service Research Development Service Fulop et al., 1987 (IIR-91-077) to BMB and Career Development Award to JEK Fulop et al., 1987 (CD97-308.A) and by the National Institute of Mental Health to the Center for Rural Mental Health Services Research Fulop et al., 1987 (MH48197) to BMB, GRS, JEK and Research Scientist Award to GRS Fulop et al., 1987 (K02 MH00843).

References (33)

  • JC Coyne et al.

    General Hospital Psychiatry

    (1994)
  • RG Kathol et al.

    Diagnosing depression in patients with medical illness. Psychosomatics

    (1990)
  • W Katon

    General Hospital Psychiatry

    (1992)
  • American Psychiatric Association. Diagnostic and statistical manual of mental disorders: IV (DSM-IV). 4th ed....
  • B.M. Booth et al.

    Functional impairment and co-occurring psychiatric disorders in medically hospitalized men. Archives of Internal Medicine

    (1998)
  • Cameron OG. Guidelines for diagnosis and treatment of depression in patients with medical illness. Journal of Clinical...
  • Fleiss J.L. Statistical methods for rates and proportions. 2nd ed. New York: John Wiley,...
  • Fulop G, Strain JJ, Vita J, Lyons JS, Hammer JS. Impact of psychiatric comorbidity on length of hospital stay for...
  • Gelder M., Gath D., Mayou R., Cowen P. Classification in psychiatry. In: Gelder M., Gath D., Mayou R., Cowen P.,...
  • HJ Henk et al.

    Medical costs attributed to depression among patients with a history of high medical expenses in a health maintenance organization. Archives of General Psychiatry

    (1996)
  • Hohmann A.A. Measurement sensitivity in clinical mental health services research: recommendation for the future. In:...
  • A Janca et al.

    Comparison of Composite International Diagnostic Interview and clinical DSM-III-R criteria checklist diagnoses. Acta Psychiatrica Scandinavica

    (1992)
  • RG Kathol et al.

    R. Diagnosis of major depression in cancer patients according to four sets of criteria. American Journal of Psychiatry

    (1990)
  • Kathol RG, Petty F. Relationship of depression to medical illness: a critical review. Journal of Affective Disorders...
  • RC Kessler et al.

    Lifetime and

    (1994)
  • Knauper B, Wittchen HU. Diagnosing major depression in the elderly: evidence for response bias in standardized...
  • Cited by (51)

    • A Randomized Controlled Trial of Sertraline for the Treatment of Depression in Persons With Traumatic Brain Injury

      2009, Archives of Physical Medicine and Rehabilitation
      Citation Excerpt :

      This interview resulted in a dichotomous score indicating the presence or absence of current and lifetime depressive disorders, lifetime psychotic disorders, and current anxiety disorders and substance abuse disorders in accordance with DSM-IV criteria. Reliability scores for the SCID are high, with the kappa value between .70 and 1.00,59 and the interview is considered a criterion standard in research studies evaluating most clinical populations.61-65 The 17-item HAM-D54 is a widely used clinician-rated measurement of depression severity.66,67

    • Death or hospitalization of patients on chronic hemodialysis is associated with a physician-based diagnosis of depression

      2008, Kidney International
      Citation Excerpt :

      The SCID was used as the gold standard for the diagnosis of depressive disorder and was administered to all patients at study entry by the same nephrologist, who was blinded to patient medical history and scores on self-report measures. The SCID is a DSM IV-based structured clinical interview for the diagnosis of depression and has been tested for reliability and validity in previous studies.75,76,77 Depression was defined as a diagnosis of major depressive disorder, dysthymia, or minor depression based on SCID.

    • Gender differences in depression and chronic pain conditions in a national epidemiologic survey

      2007, Psychosomatics
      Citation Excerpt :

      As with all research that studies the association between chronic pain conditions and depression, the issue of “criterion contamination” exists.34 Although the prevalence of depression in this study may be inflated because of the somatic symptoms, it should also be underscored that the CIDI has performed well in detecting depression in medically-ill populations.22 Despite this fact, the CCHS–1.1 used the abbreviated version of the full diagnostic interview, which contained no probe questions to determine the source of the symptoms.

    View all citing articles on Scopus
    View full text