In March 2006, the Archives of General Psychiatry published an article by Bulik et al1 on the heritability of anorexia nervosa based on the Swedish Twin Registry. Table 4 of the article (p 310) provided descriptive data for univariate analysis of anorexia nervosa. In this table, column headings were divided into “T1-No/T2-No”; “T1-Yes/T2-No”; and “T1-Yes/T2-Yes.” The table note stated that T1 equaled twin 1, and T2 equaled twin 2. For the row designated “DZ-MF [dizygotic–male/female]” under the “Narrow DSM-IV AN [anorexia nervosa]” heading, 2212 twin pairs were “T1-No/T2-No,” meaning that neither twin was affected; 0 were “T1-Yes/T2-Yes,” meaning no dizygotic twinship discordant for sex had both affected; and 16 were “T1-Yes/T2-No.” As there is a DZ-FM row above this one, the most logical interpretation is that T1 refers to the first twin by sex in a dizygotic twinship. This interpretation of Table 4 was used by Procopio and Marriott2 in their reanalysis of the data. Their reanalysis indicated that for a male with a female co-twin, the risk for anorexia nervosa was nearly 10-fold the risk in twinships consisting of males alone and approached the risk for females.
Erratum: “Prevalence, Heritability, and Prospective Risk Factors for Anorexia Nervosa” (2006;63:305-312). Arch Gen Psychiatry. 2008;65(9):994. doi:10.1001/archpsyc.65.9.994