Hospital comorbidity bias and the concept of schizophrenia

Abstract

The comorbidity bias predicts that if disease definition is based on observations of patients in the hospital, spurious comorbidity of psychopathological dimensions that increase the probability of hospital admission will be included in the disease concept, whereas comorbid dimensions that are not associated with admission will be excluded. The direction of any dimensional comorbidity bias in psychotic illness was assessed in a longitudinal analysis of the psychopathology of patients assessed both inside and outside the hospital. Four hundred and eighty patients with broadly defined psychotic disorders were assessed between one and nine times (median two times) over a 5-year period with, amongst others, the Brief Psychiatric Rating Scale. Dimensional comorbidities between positive symptoms, negative symptoms, depression/anxiety, and manic excitement were compared, in addition to their associations with current and future admission status. Higher levels of psychopathology in all symptom domains were associated with both current and future hospital admissions. Associations between the positive, negative, and manic symptom domains were higher for patients in the hospital than for patients outside the hospital, in particular, between positive symptoms and manic excitement (beta=0.28, p<0.001). However, associations between depression and other symptom domains were higher in out-patients as compared to in-patients (positive symptoms and depression, beta=-0.26; p<0.002). The current analyses suggest that, to the extent that disease concepts of psychosis do not take into account effects of dimensional comorbidity biases occasioned by differential psychopathology according to treatment setting, "florid" psychotic psychopathology may be overrepresented, whereas depressive symptoms may be spuriously excluded.