Probability of major depression diagnostic classification based on the SCID, CIDI and MINI diagnostic interviews controlling for Hospital Anxiety and Depression Scale – Depression subscale scores: An individual participant data meta-analysis of 73 primary studies

https://doi.org/10.1016/j.jpsychores.2019.109892Get rights and content

Highlights

  • Fully structured diagnostic interviews may misclassify major depression.

  • Compared to the SCID, the MINI may diagnose depression too often.

  • Compared to the SCID, the CIDI may not be adequately responsive to symptom severity.

Abstract

Objective

Two previous individual participant data meta-analyses (IPDMAs) found that different diagnostic interviews classify different proportions of people as having major depression overall or by symptom levels. We compared the odds of major depression classification across diagnostic interviews among studies that administered the Depression subscale of the Hospital Anxiety and Depression Scale (HADS-D).

Methods

Data accrued for an IPDMA on HADS-D diagnostic accuracy were analysed. We fit binomial generalized linear mixed models to compare odds of major depression classification for the Structured Clinical Interview for DSM (SCID), Composite International Diagnostic Interview (CIDI), and Mini International Neuropsychiatric Interview (MINI), controlling for HADS-D scores and participant characteristics with and without an interaction term between interview and HADS-D scores.

Results

There were 15,856 participants (1942 [12%] with major depression) from 73 studies, including 15,335 (97%) non-psychiatric medical patients, 164 (1%) partners of medical patients, and 357 (2%) healthy adults. The MINI (27 studies, 7345 participants, 1066 major depression cases) classified participants as having major depression more often than the CIDI (10 studies, 3023 participants, 269 cases) (adjusted odds ratio [aOR] = 1.70 (0.84, 3.43)) and the semi-structured SCID (36 studies, 5488 participants, 607 cases) (aOR = 1.52 (1.01, 2.30)). The odds ratio for major depression classification with the CIDI was less likely to increase as HADS-D scores increased than for the SCID (interaction aOR = 0.92 (0.88, 0.96)).

Conclusion

Compared to the SCID, the MINI may diagnose more participants as having major depression, and the CIDI may be less responsive to symptom severity.

Introduction

Different types of standardized diagnostic interviews are commonly used to classify major depression in research. Semi-structured interviews, for example, the Structured Clinical Interview for DSM (SCID) [1], are designed to be administered by clinically trained professionals with experience in diagnosis; they allow evaluators to ask additional questions and to use their judgement to determine whether or not symptoms are present [[2], [3], [4]]. Fully structured interviews, on the other hand, such as the Composite International Diagnostic Interview (CIDI) [5], were designed specifically to address the costliness of using clinician-administered interviews in epidemiological surveys and can be administered by trained lay interviewers. The CIDI is fully scripted, and thus interviewers are instructed not to explain or rephrase symptoms; its developers emphasized that they were hoping to achieve a high level of reliability for large-scale survey work with the possible loss of validity of diagnoses [5]. The Mini International Neuropsychiatric Interview (MINI) [6,7] is a very brief fully structured interview that was originally designed for potential use as a screening instrument [7]. As described by its developers, it is intended to be over-inclusive in classifying disorders [7].

Despite the different designs and intended uses of semi-structured interviews, fully structured interviews (MINI excluded), and the MINI, these instruments are typically treated as equivalent reference standards for major depression classification in research, including in evidence syntheses [8]. Only five small studies, which each included only 6–22 cases of major depression based on semi-structured interviews and 8–61 cases based on fully structured interviews, have directly compared different types of diagnostic interviews for major depression [3,[9], [10], [11], [12]]. In the three studies that included >100 participants, prevalence of major depression was substantially higher based on fully structured interviews compared to semi-structured interviews [3,9,12]. Only in a study of patients from an alcoholic treatment unit, where depressive symptoms would be expected to be much more severe, major depression prevalence was similar when assessed with semi-structured and fully structured interviews [11].

Recently, we used an individual participant data meta-analysis (IPDMA) approach in two studies to compare the probability of major depression classification across diagnostic interviews [13,14]. In the first, which included 17,158 participants from 57 primary studies, participant characteristics and depressive symptom severity were controlled using Patient Health Questionnaire-9 (PHQ-9) scores. Among fully structured interviews, the MINI classified depression approximately twice as often as the CIDI. Compared to semi-structured interviews, fully structured interviews (MINI excluded) classified more patients with low-level depressive symptoms but fewer participants with high-level symptoms as depressed [13]. Similar findings were observed in a second IPDMA of 46 studies that included 12,759 women who were pregnant or had recently given birth [14]. Controlling for Edinburgh Postnatal Depression Scale (EPDS) scores, the MINI classified more participants as having major depression than the CIDI, while as EPDS scores increased, both the CIDI and MINI classified fewer participants as having depression than the SCID [14]. These findings highlight that different diagnostic interviews may classify different proportions of patients with major depression or be more or less responsive to symptom levels in samples comprised of a range of participants, including women in pregnancy and postpartum.

Neither of the two previous IPDMAs focused on diagnosis primarily in people with medical conditions. Because only two large studies have been conducted to date it is important to test the generalizability of findings in different populations, including people with medical conditions. The Depression subscale of the Hospital Anxiety and Depression Scale (HADS-D) [15] is commonly used to assess depressive symptom severity in medically ill patients. The HADS was designed specifically for use in people with physical health problems and to avoid somatic items that are common in both depression and many other medical conditions [15]. The objective of the present study was to use an IPDMA approach to examine patterns between diagnostic interviews and the proportion of participants classified as having major depression among studies that administered the HADS-D. As in previous studies [13,14], first we compared major depression classification odds within fully structured interviews (MINI vs. CIDI), and then between fully structured and semi-structured interviews (CIDI vs. SCID and MINI vs. SCID), to determine if different interviews influenced the odds of being classified as having major depression. In each case, we controlled for participant characteristics and depressive symptom severity based on HADS-D scores. Second, we tested whether differences in the probability of classification across the three types of interviews were associated with depressive symptom severity by including an interaction term.

Section snippets

Methods

We registered the main analyses of the HADS-D IPDMA in PROSPERO (CRD42015016761) and published a protocol [16]. We reported the results of the present study following PRISMA-DTA [17] and PRISMA-IPD [18] reporting guidelines. We did not plan at the time of registration and publication of our protocol to conduct analyses that compared diagnostic interviews, but results from previous studies [13,14] indicated that there may be important differences between interviews and that this should be tested

Results

Of 10,015 unique titles and abstracts identified from the database search, 9584 were excluded after title and abstract review, and 264 were excluded after full text review, leaving 167 eligible articles with data from 116 unique samples, of which 69 (59% of datasets; 71% of participants) contributed data (Fig. 1). Reasons why articles were excluded at the full-text level are provided in Appendix B. Authors of included studies contributed data from an additional five unpublished studies and

Discussion

We compared the odds of being classified as having major depression according to three diagnostic interviews, controlling for participant characteristics and depressive symptom severity using IPDMA. Although different types of diagnostic interviews are used in research, semi-structured interviews, which allow queries with clinical judgement, such as the SCID, most closely replicate standard diagnostic criteria administered by a trained evaluator [[2], [3], [4]]. Our study found that, first,

Conclusion

Among primary studies that administered the HADS-D, we found that compared with the SCID, the MINI and CIDI may misclassify major depression, which is generally consistent with findings from previous studies that were conducted with similar methods in other populations [14, 15]. The MINI and CIDI are the most commonly used fully structured interviews for major depression. They are fully scripted and can be administered by lay research staff, but they may not perform equivalently to SCID, which

Contributors

YW, BLevis, JTB, PC, SG, DM, JPAI, LAK, SBP, IS, RCZ, MHenry, ZI, CGL, NDM, MT, ABenedetti and BDT were responsible for the study conception and design. JTB and LAK designed and conducted database searches to identify eligible studies. SA, ABeraldi, APBMB, NBD, ABunevicius, GCarter, CKC, GCheung, KC, RMC, DC, CED, ED, FMD, ED, MGD, AF, PPF, FHF, AJF, MF, PG, MG, SG, LG, MHärter, JJ, NJ, MJ, MKeller, SK, JMK, SWK, MKjærgaard, BLöwe, WLL, RMS, LMassardo, YM, AM, IM, LMisery, RN, MLO, MO, JP, LP,

Roles of the funding source

This study was funded by the Canadian Institutes of Health Research (CIHR, KRS-144045 & PCG 155468). Dr. Wu was supported by an Utting Postdoctoral Fellowship from the Jewish General Hospital, Montreal, Quebec, Canada. Ms. Levis was supported by a CIHR Frederick Banting and Charles Best Canada Graduate Scholarship doctoral award. Ms. Rice was supported by a Vanier Canada Graduate Scholarship. Mr. Bhandari was supported by a studentship from the Research Institute of the McGill University Health

Declaration of competing interest

All authors have completed the Unified Competing Interest form at http://www.icmje.org/coi_disclosure.pdf and declare that: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous 3 years with the following exceptions: (1) Dr. Patten declares that he has received a grant, from the University of Calgary Hotchkiss Brain Institute, which was jointly funded by the Institute and

Acknowledgement

We thank Dr. Linda Kwakkenbos for providing assistance with translation.

References (28)

  • J.C. Anthony et al.

    Comparison of the lay Diagnostic Interview Schedule and a standardized psychiatric diagnosis: experience in eastern Baltimore

    Arch. Gen. Psychiatry

    (1985)
  • V. Hesselbrock et al.

    A comparison of two interview schedules: the Schedule for Affective Disorders and Schizophrenia-Lifetime and the National Institute for Mental Health Diagnostic Interview Schedule

    Arch. Gen. Psychiatry

    (1982)
  • V. Jordanova et al.

    Validation of two survey diagnostic interviews among primary care attendees: a comparison of CIS-R and CIDI with SCAN ICD-10 diagnostic categories

    Psychol. Med.

    (2004)
  • B. Levis et al.

    Probability of major depression diagnostic classification using semi-structured versus fully structured diagnostic interviews

    Br. J. Psychiatry

    (2018)
  • Cited by (33)

    • External validation of a shortened screening tool using individual participant data meta-analysis: A case study of the Patient Health Questionnaire-Dep-4

      2022, Methods
      Citation Excerpt :

      For these analyses, we modeled sensitivity and specificity separately among studies that used each reference standard category (semi-structured, fully structured, or MINI) as well as pooled together. We present accuracy results for the PHQ-Dep-4 separately by reference standard type because previous studies have found that there are important differences in the design and performance of different types of diagnostic interviews used as reference standards [27–30], and that PHQ-9 sensitivity and specificity vary across different reference standards [8,18]. For each reference standard category, we constructed an empirical receiver operating characteristic (ROC) plot for the PHQ-Dep-4 based on pooled sensitivity and specificity estimates from each cutoff.

    • Cognitive impairment in patients with rheumatoid arthritis: A systematic review and meta-analysis

      2022, Joint Bone Spine
      Citation Excerpt :

      This could have contributed to at least two errors: a) the tools used were self-report, so they do not necessarily indicate that the respondent actually suffered from depression; rather we should talk about the self-reported severity of depressive symptoms; and b) in some cases, the tools used examined somatic symptoms of depression (e.g., the BDI) that overlap with the health difficulties due to RA symptoms and treatment (e.g., difficulty sleeping due to pain, weight changes). These limitations suggest that, in the case of RA, one short screening tool that does not take somatic symptoms into account should be used to assess the severity of depressive symptoms (e.g. PHQ-9, HADS) or, if possible, the presence of depression should be assessed with a structured interview (e.g., the SCID [34]; see also [40]). Many studies also did not control for other factors that may be related to CI or report them in a way that did not allow them to be included in the analyses.

    • Journal of psychosomatic research 2021 year in review

      2022, Journal of Psychosomatic Research
    View all citing articles on Scopus
    1

    Co-senior authors.

    View full text