Impact of personality disorders in a sample of 212 homeless drug users

The impact of the mental disorders (axis I and II, according to DSM IV) on psychosocial problems (axis IV) is now a well-known fact, notably when substance abuse disorders are encountered on axis I. This leads to the conclusion that personality disorders increase the risk of substance abuse, that substance abuse increases the risk of homelessness, that dual diagnosis has a high impact on homelessness and underlines interactions between personality disorders (PD), drug abuse (DA) and homelessness. The aim of this paper is also to study these interactions. We will process the classical epidemiological measures, which have already produced interesting findings on other substance-linked disorders. We will study the multiplicative interaction (I(AB)) and the etiological fraction (EFi) linked to interaction, which evaluate the effects of two factors on another. According to the authors, the I(AB) determines whether the co-occurrence of two risk factors in a group induces more cases than each factor acting together; also if the I(AB) is greater than 1 it is possible to estimate the EFi, that proportionally measures the number of cases of the third factor that can be attributed to the co-occurrence. We will also study the interactions of homelessness and PD on DA, of the PD and DA on homelessness, and of this association DA and homelessness on PD. The data we will use in the paper deal with the prevalence of PD in general, drug users (n=226), homeless (n=999) and homeless drug abusers (n=212). The two last data are extracted from the same population and have been collected through clinical interviews, and the diagnosis follows the DSM criteria. They are comparable to Kokkevi et al.'s sample regarding the drug (heroin), the mean age (28 years for Kokkevi et al., 27 years in our sample), and the geographic origin of the populations (Mediterranean basin). The repartition of PD differs significantly (0.001) in the homeless population and the homeless drug abusers (chi(2)=70.5, df=20). Therefore, the intensity of this link is low (rphi=0.30), which is a consequence of the high prevalence of PD in the homeless population (80% versus 10% in general population). On the other hand, the reparation of PD in the homeless drug abusers sample and Kokkevi et al.'s drug abusers is different at 0.001 (chi(2)=92.64, df=20). The link is high (rphi=0.45) and could be interpreted as a supplementary effect of PD's linked to homelessness and in the PD linked to DA, thus motivating furth