Twin studies of psychopathology: why do the concordance rates vary?

Many current discussions of hereditary factors in psychopathology focus on twin studies as the primary source of evidence supporting the importance of genetic determinants. In summarizing the results of these studies, authors often derive estimates of concordance rates by collapsing across studies and presenting mean or median rates. This practice implicitly assumes that the concordance rates yielded by different studies represent equally reliable estimates of the population mean. The present study evaluates the validity of this assumption. Reports of twin studies of schizophrenia and affective disorder were reviewed. A meta-analysis was conducted to examine the influence of methodological factors on concordance rates. Analyses indicated that both sample-selection and zygosity-determination procedure are systematically associated with concordance rates. For schizophrenia, MZ concordance rates are significantly lower when samples are selected from a twin register as opposed to a psychiatric facility. Lower MZ concordance rates are also yielded by studies that employ laboratory procedures to determine zygosity. Implications of the findings for future research are discussed.