Respondents’ report of a clinician-diagnosed depression in health surveys: comparison with DSM-IV mental disorders in the general adult population in Germany

In many large-scale health surveys, the prevalence of depression is assessed by asking participants whether they have been diagnosed with depression by a health professional in the past 12 months [1‐4]. This simple measure has various preconditions such as previous health care utilization of the respondent, reporting depressive symptoms to a health professional, understanding the question asked in the survey, and admitting the diagnosis in the survey interview. In addition, the accuracy of this measure in determining the existence of depression depends on whether health professionals correctly diagnose their patient with depression. The latter can be at least partly questioned considering the notable extent of over- and under-diagnosis of depression in primary care found in clinical and community studies [5‐8].

Despite these preconditions, a recent general population survey found that 12-month prevalence estimates based on the respondent’s report of a clinician-diagnosis of depression and an interviewer assessment of major depression with a comprehensive standardized diagnostic interview based on the Diagnostic and Statistical Manual of Mental Disorders-IV (DSM-IV) were on a similar level [1]. Yet, the overlap of the two measures was only moderate, and, in agreement with previous research [2, 4, 9, 10], prevalence of major depression was highest in younger age adults and of clinician-diagnosed depression in middle-aged and older adults [1]. These findings suggest that there are substantial differences between DSM-IV-based major depression and a clinician-diagnosed depression and, as a consequence, that the presence of major depression may not be inferred with accuracy from the respondent’s report of a previous clinician diagnosis. However, there is a lack of specific knowledge about differences between both measures at an individual level.

Considering the fact that depression is still quite often assessed with a single question about clinician diagnoses in health surveys, this study aims to contribute to a better understanding of survey results based on the respondent’s report of a clinician-diagnosed depression. Therefore, we investigate the association between such reports and independent diagnoses of depression in a large general population survey. Specifically, we examine the proportion of mental disorders based on a comprehensive standardized diagnostic research interview among survey participants who report a clinician-diagnosed depression as a function of sex, age, and depression severity. Moreover, in order to examine the false negative rate based on the question about a previous clinician diagnosis, we examine the proportion of survey participants with major depression determined by the diagnostic interview who do versus do not report that a clinician diagnosed them with depression. Finally, socio-demographic, health-related and mental health characteristics are reported separately for survey participants with different combinations of depression diagnoses based on the cross-classification of diagnoses according to a diagnostic interview administered in the survey and the respondents’ reports of being diagnosed by a clinician.

Of n = 4483 participants of DEGS1-MH, n = 101 (2.3%) had missing values in at least one of both measures, resulting in a study sample of n = 4382. In the sample, 50.9% (95%CI 48.9–52.9) were female, 19.2% (95%CI 17.9–20.5) were 18–29 years old, 24.9% (95%CI 23.3–26.6) were 30–44 years old, 36.0% (95%CI 34.2–37.8) were 45–64 years old and 20.0% (95%CI 18.7–21.3) were 65–79 years old. Overall, there were 78.8% (95%CI 76.8–80.7) married and living with the partner or in a steady relationship. The distribution of CASMIN educational level was 35.0% (95%CI 32.5–37.5) for low, 50.8% (95%CI 48.7–52.9) for middle and 14.2% (95%CI 12.6–16.0) for high.

Based on a nationally representative sample of the general adult population, this study shows that a large proportion of survey participants who report a previously clinician-diagnosed depression meets criteria of at least one DSM-IV mental disorder. At the same time, there were major differences between participants reporting a clinician-diagnosed depression and participants who meet criteria for major depression in a comprehensive standardized diagnostic interview.

The findings of this study should be considered in the context of several limitations. First, the time lag between DEGS1 and DEGS1-MH might have led to an underestimation of the overlap of both measures examined and of the prevalence of mental disorders among participants with a clinician-diagnosed depression due to remission or to the time difference between both assessments. Analyses limiting the time lag to ≤25 weeks led to a slight increase of overlap, but the results did not substantially change. The time-lag stratified analyses of the overlap of both measures indicates that the time lag might have led to an age- or sex-specific under- or over-estimation of the overlap. However, the confidence intervals are large due to the small numbers of cases in the upper quartile. Second, the measure of a clinician-diagnosed depression is diffuse in two ways. The wording of the question leaves it unclear whether the depression diagnosis was made in the past 12 months or whether the respondent had received a diagnosis previously and interprets depressive symptoms in the past 12 months as a continuation. Further, the measure depends on the respondent’s ability to recall a depression diagnosis, which however also applies for symptom reports in the CIDI. Third, the mental disorders assessed in DEGS1-MH does not include all possible disorders, hence the prevalence of any other mental disorder might be underestimated. Fourth, the small number of cases results in large confidence intervals and restricts analyses with respect to the number of strata.

Despite these limitations, this study contributes to a better understanding of survey results based on the measure of a clinician-diagnosed depression. There are four noteworthy findings. First, there was a remarkable age pattern: Among participants with a clinician-diagnosed depression, the proportion of those meeting the criteria of MDD was 62.6% in those aged 18–29 years and decreased per age group to 29.8% in those aged 65–79 years. A similar yet less pronounced age pattern was found in a large community-based study in the US [5]. Interestingly, our results show that 96.1% of young adults who report a clinician-diagnosed depression had any mental disorder, while 39,8% of adults aged 65–79 years did not have any disorder even though reporting a health professional-diagnosed depression. The relatively high proportion of 62.6% MDD among those aged 18–29 years is higher than the 42.0% that was reported in a meta-analysis on the accuracy of unassisted depression diagnosis in primary care [30] and equal to the proportion of 66.5% which was reported for the optimal threshold for the CIDI screening scale [31]. Our study indicates that a clinician-diagnosed depression reported by this age group in a health survey might include to a large part participants who actually fulfilled the MDD criteria. A possible explanation for this finding is the higher prevalence of severe depression among depressed young adults [10] which might lead to more accurate diagnoses by health professionals. Also, health professionals might be more rigorous in diagnosing depression when it comes to younger patients because of its possible psychological impact on the life young people. Further, the high prevalence of depression in young adults [9, 10, 14, 32] might result in a higher number of “true positives” than in populations with lower depression prevalence. The high proportion of middle-aged and older participants with a clinician-diagnosed depression who did not meet the criteria of MDD implies a substantial amount of over-diagnosis of depression for these groups, replicating results from primary care in Italy [6] and the US [5]. Increased physical complaints in older age could be misinterpreted as symptoms of depression by health professionals [33, 34]. Additionally, underreporting of depressive symptoms by older participants in the CIDI due to its complex questioning and its multiple time frames [35] and due to problems recalling the symptoms are possible methodological explanations. Research has suggested that older adults more frequently show clinically significant depressive symptoms without fulfilling all diagnostic criteria [36]. In this context, our finding could at least partly be explained by the fact that the depression diagnosis based on the CIDI might exclude older people due to the diagnostic algorithm who might have reported significant symptoms to the health professional, resulting in a depression diagnosis.

The second notable result is the striking age pattern of the proportion of clinician-diagnosed depression among participants with MDD (low in young age groups and highest in those aged 45–64 years). Obviously, these findings cannot be interpreted as the percentage of actual unrecognition of depression by health professionals. Instead, it reflects the specific preconditions of the measure of a clinician-diagnosed depression reported by the respondent. Thus, our findings supposedly mirror to a large part the lack of any health professional contact in people with mood disorders [37] and the fact that young adults are less likely to use professional health services due to mental health problems [38‐40] than older adults, but prefer seeking help rather from family or friends [41]. Additionally, low numbers of physician contacts are reported for younger adults [42], reducing the odds for a physician to detect depression [8, 43]. This is supported by the fact that the age pattern disappeared after controlling for the number of outpatient physician visits. In contrast to the US, where financial barriers have been reported the most common reasons for not seeking professional help for mental health problems [44], personal help-seeking barriers such as self- and perceived stigmatization [45‐47] are likely more relevant in Germany, as diagnosis and treatment of mental disorders are covered by mandatory health insurance. As to depression recognition itself, our results reflect the considerable amount of unrecognized depression in primary care in Germany and internationally [8, 30, 48], which was suggested of being the most pronounced in younger people [7]. In addition, health professionals might not always fix and communicate a depression diagnosis to the patient [49]. Not least, self-stigmatization might lead to a reporting bias of a depression diagnosis in the survey interview.

Third, it is remarkable that no differences were found between men and women. On one hand, concerning the similar level of the proportion of MDD among men and women with a clinician-diagnosed depression, our results agree with findings from the US [5]. On the other hand, the fact that we did not find any sex difference regarding the proportion of clinician-diagnosed depression among those with MDD is contra-intuitive as research has suggested sex-specific barriers to help seeking [50], utilization of somatic and mental health services [42, 51], symptom reporting [52] and depression recognition rates [8].

Fourth, we found that participants with severe depression according to different severity indicators were more likely to be classified as cases by both measures. This is plausible considering that research found an association of seriousness of mental disorder with service use [38] and higher proportions of depression recognition in primary care when depression was severe and when suicidal ideation or tendency was reported by the patient [8]. This finding indicates that a considerable part of participants with severe depression is included in the measure of a clinician-diagnosed depression. In addition to the explanations given above, this finding can also be interpreted in the context of discussions about the DSM-IV potentially assigning a depression diagnosis partly to people without relevant disability [53]. Following this discourse, participants with depression without relevant disabilities might have qualified for a depression diagnosis in the CIDI, but were unlikely to seek for professional help due to their lack of disability.

This study contributes to a better understanding of survey results based on the respondent’s report of a clinician-diagnosed depression. It shows that a large proportion of survey participants who report a clinician-diagnosed depression meets criteria of at least one DSM-IV mental disorder (most frequently any affective and any anxiety disorder) according to the CIDI. However, substantial differences exist between participants who report a clinician-diagnosed depression and participants who meet criteria for depression in a comprehensive standardized diagnostic interview. Furthermore, concordance of both measures varies with age and severity of depressive symptoms. Thus, respondent reports of clinician-diagnosed depression cannot be considered a short alternative for CIDI-based major depression in studies of the general population. This needs to be considered when interpreting survey results based on reports about clinician-diagnosed depression.

Instead, this study underlines the importance to assess a range of depression indicators in health surveys in order to cover a wide spectrum. For many research questions, information about depression diagnosis in the health care system is important. But if information on depressive symptoms or on the presence of diagnostic criteria is needed, further indicators such as standardized diagnostic interviews or one of the established short screening scales, which have been shown to have good concordance with diagnoses based on independent research diagnostic interviews [21, 31], should be additionally assessed.

Regarding clinical practice, our findings first imply that if patients report that they have not been diagnosed with depression, they most likely do neither have a history of major depression. Second, relating to younger patients, health professionals should be especially attentive and keep a possible major depression in mind when depressive symptoms are reported in order to provide adequate treatment. Third, relating to older patients, our results imply that health professionals should on one hand be more rigorous when diagnosing depression in order to reduce over-diagnosis and over-treatment as a consequence. On the other hand, however, health professionals should consider determining a differential depression diagnosis when depressive symptoms are reported by elderly in order to increase recognition and adequate depression treatment in this age group.