To the Editor: Nelson's response (1) to the article in the May issue by Draine and his colleagues (2) illustrates the stubborn persistence of clinical illusions about the relationship between mental illness and crime, unemployment, and poverty. Nelson states that the article "does not succeed in showing that untreated mental illness is not substantially associated with these conditions." Draine and colleagues' extensive documentation that the effects of mental illness are considerably smaller than have been implied in the psychiatric services literature would cause most readers to turn Nelson's response on its head: "Psychiatrists have not succeeded in showing that untreated mental illness per se is substantially associated with these conditions."
Perhaps because of their clinical roles, psychiatrists more easily identify the proximal causes of these conditions. Thus Nelson points to "the paranoia often associated with homelessness." Indeed, mental illness may appear to be the dominant factor at the point of entry into homelessness, unemployment, or prison. However, a scientific approach necessitates that we ask questions about distal causes (3). For example, what is it about persons with mental illness that make them more likely to become homeless? The article by Draine and colleagues provides evidence that factors such as poverty, unavailability of low-cost housing, and the inaccessibility of services more strongly explain why a mentally ill person shows up at the shelter door than does his or her psychosis.
Despite the excellence of the article by Draine and his colleagues, they have relied primarily on one conceptual model, whereas a multitiered approach must be considered. The effects of the relationship between poverty (P) and mental illness (M) on outcomes (O) such as homelessness, unemployment, or criminality can be understood by at least three different models, which are not mutually exclusive (4). The first model postulates that the relationship is additive: P + M=O. Thus M may have an independent effect on O, and then P further increases the likelihood of O. However, this model also allows for P to modify the effects of M. In other words, if there is a correlation between the two, adding P to the analysis would diminish the original effect of M on O.
The Draine article focused primarily on the implications related to this model. However, a second model proposes that the relationship may also be interactive: P × M=O. Hence, the likelihood of O increases appreciably as the level of P or M increases, and conversely, if the level of either P or M is low, the risk of O is much less.
The third model hypothesizes that the relationship between P and M are dialectical so that they are mutually transforming. Thus P alters M so that the mentally ill person who becomes poor is more vulnerable to O, or conversely, a poor person who becomes mentally ill is more vulnerable to O.
Although I agree with Severson and Lieberman (5) that it is time to put money into solving these problems, they minimize the necessity of combining theoretical research with practice as well as the role research plays in refuting those who neglect the social concomitants of mental illness in favor of biomedical solutions. The wheel keeps being reinvented because scientific research is not a dispassionate enterprise. Too often, the questions posed and whether results are acted upon depend on sociopolitical forces.
Dr. Cohen is professor in the department of psychiatry and director of the division of geriatric psychiatry at SUNY Downstate Medical Center in Brooklyn, New York.