Original Article
High agreement of self-report and physician-diagnosed somatic conditions yields limited bias in examining mental–physical comorbidity

https://doi.org/10.1016/j.jclinepi.2009.08.009Get rights and content

Abstract

Objective

To quantify the misclassification bias of self-reported somatic diseases and its impact on the estimation of comorbidity with mental disorders.

Study Design and Setting

Data were drawn from the German National Health Interview and Examination Survey (N = 7,124), which assessed both self-reported and physician-diagnosed somatic diseases. Eight chronic diseases were examined: coronary heart disease, heart failure, asthma, chronic bronchitis, diabetes, cancer, arthrosis, and arthritis. Mental disorders were assessed by means of the Munich-Composite International Interview.

Results

The agreement of case ascertainment by patient self-report and physician diagnosis was high (kappa: 0.74–0.92), except for arthritis (0.53). False-positive and false-negative disease statuses were partly associated with age, sex, socioeconomic status, somatic comorbidities, marital status, and mood and anxiety disorders. In most conditions, the odds ratios (ORs) of comorbid mental disorders based on self-reported diseases were slightly overestimated with regard to mood disorders (relative OR: 0.91–1.38), whereas there proved to be no such trend regarding anxiety disorders (0.82–1.05). Substance disorders were partly biased without showing an interpretable trend across diseases (0.49–2.58).

Conclusions

Evaluation of mental–physical comorbidity based on self-reported and physician-diagnosed physical conditions yielded similar results, with modestly inflated ORs for mood disorders for several self-reported physical conditions.

Introduction

Over the last few years, the association between somatic diseases and mental disorders has received increasing attention as a focus of research. Based on large-scale epidemiological surveys on mental health prevalence rates, comorbid mental disorders in patients with several somatic diseases have been examined [1], [2], [3], [4], [5], [6], [7], [8], [9]. Recent reviews have shown that comorbid mental disorders are associated with increased mortality and morbidity, intensified health care utilization and lowered quality of life in patients with chronic somatic diseases [10], [11], [12], [13], [14]. A precondition of these research activities is the use of a reliable and valid instrument for recording mental disorders, such as the Composite International Diagnostic Interview (CIDI) [15]. Such types of standardized interviews can be regarded as the gold standard for assessing mental disorders, and are thought to overcome the validity problems of mental disorder assessments used in the past [16].

Studies on the prevalence rates of mental disorders in patients with vs. without a somatic disease emphasize the importance of a valid assessment of mental disorders. The operationalization of the index disease, however, is often based solely on patients' self-reports [4], [5], [6], [7], [8], [9]. Although there are studies which reported at least moderate to high agreements, for example, diabetes [17], [18], [19], [20], [21], [22], [23], coronary heart diseases [18], [19], [20], [21], [22], [23], [24], or cancer [18], [19], [22], the reliability and validity of self-reported somatic disease status proved to be limited for diseases, such as arthritis [19], [20], other cardiovascular diseases [17], [18], [19], [20], [21], [22], or lung diseases [18], [20], [22]. As a result, the measurement errors in index diseases may lead to biased estimates of prevalence rates of comorbid mental disorders [25].

The present study aims to quantify the misclassification bias of self-reported somatic diseases and to predict the impact of this bias on the estimation of comorbidity with mental disorders based on eight frequent, chronic somatic diseases: coronary heart disease (CHD), heart failure, asthma, chronic bronchitis, non–insulin-dependent diabetes, cancer, arthrosis, and arthritis. For this purpose, data from the German National Health Interview and Examination Survey (GHS) were used [26], which assessed both self-reported and physician-diagnosed somatic diseases. The following questions will be answered.

  • 1.

    What are the misclassification rates of self-reported somatic diseases?

  • 2.

    What are the sociodemographic and medical correlates of the misclassification rates of self-reported somatic diseases?

  • 3.

    To what extent are mental comorbidity estimations in patients with vs. without a somatic disease biased when they are based on self-reports of somatic diseases rather than physicians' diagnoses?

Section snippets

Study design and samples

Data were drawn from the GHS, a stratified, multistage, cross-sectional, nationally representative sample of subjects aged 18–79 years from the noninstitutionalized population of Germany. Aims, design, and methods have been described in greater detail in a separate publication [26]. Therefore, design and sample characteristics are discussed only briefly here.

The GHS consisted of a stratified random sample from 113 communities throughout Germany with 130 sampling units. A representative gross

Misclassification rates of self-reported somatic diseases

The agreement rates between self-reported and formally diagnosed disease cases are reported in Table 2. The specificity (92.76–99.70%) and the NPV (94.52–99.73%) proved to be very high for all somatic diseases. The sensitivity was lowest for arthritis (84.45%) and highest for asthma (94.99%). A misclassification of self-reported and formally diagnosed cases was most frequent for the condition of a false-positive disease status report. The PPV was lowest for arthritis, at 42.26%, and highest for

Discussion

The present study examined the impact of a misclassification of patients with somatic diseases based on self-report data on risk calculations of comorbid mental disorders. Most large-scale epidemiological studies that have analyzed the association of somatic and mental comorbidities were based on self-reported somatic disease status [4], [5], [6], [7], [8], [9]. Those studies have often become landmark studies determining future research efforts and public health topics, making it all the more

Acknowledgments

We would like to thank the Max Planck Institute for Psychiatry and the Department of Psychology and Psychotherapy of the TU Dresden for providing the GHS-MHS survey data [34] and the Robert Koch Institute for providing the GHS-CS survey data [35].

References (35)

  • S.S. Merkin et al.

    Agreement of self-reported comorbid conditions with medical and physician reports varied by disease among end-stage renal disease patients

    J Clin Epidemiol

    (2007)
  • K. Wada et al.

    Self-reported medical history was generally accurate among Japanese workplace population

    J Clin Epidemiol

    (2009)
  • K. Yamagishi et al.

    Self-reported stroke and myocardial infarction had adequate sensitivity in a population-based prospective study

    J Clin Epidemiol

    (2009)
  • H. Baumeister et al.

    Somatoform disorders in patients with musculoskeletal and cardiovascular diseases in comparison to the general population

    Acta Neuropsychiatr

    (2003)
  • H. Baumeister et al.

    Mental disorders in patients with obesity in comparison with healthy probands

    Int J Obes

    (2007)
  • M. Härter et al.

    Increased 12-month prevalence rates of mental disorders in patients with chronic somatic diseases

    Psychother Psychosom

    (2007)
  • Y. He et al.

    Mental disorders among persons with arthritis: results from the World Mental Health Surveys

    Psychol Med

    (2008)
  • Cited by (154)

    • Disability and life satisfaction in neurological disorders: The role of depression and perceived cognitive difficulties

      2021, General Hospital Psychiatry
      Citation Excerpt :

      Secondly, the study utilised an online convenience sample and participants self-reported their confirmed diagnoses of a neurological disorder, which were not cross-checked with medical records. However, this study was interested in assessing community dwelling adults, where self-report is common [9] and has been found to be accurate and comparable with physician report [62]. Lastly, there were several potentially important factors that were not assessed or controlled for including anxiety, fatigue, pain, and disease severity [10], which may also contribute to poorer outcomes in this cohort.

    View all citing articles on Scopus
    View full text