Cancer and mental illness

The relationship between mental illness and neoplastic disease has been the subject of controversy for decades. 1–3 Early studies using proportionate mortality rates (i.e., the proportion of cancer deaths to total deaths in a given population) were interpreted as evidence that psychotic patients have a significantly lower rate of cancer than the general population. 3–5 These studies have been extensively criticized on the grounds that since psychiatric patients tend to have increased mortality from a variety of causes (e.g., pneumonia) relative to the general population, if cancer mortality were more equal in incidence between patient and general population groups, the proportionate mortality data would spuriously suggest reduced cancer risk among patients. 4–6 Other reports using age specific or age adjusted mortality data, with three exceptions, 7–9 have indicated an equal or slightly increased incidence of cancer mortality among psychiatric patients as compared to the general population. 4–6,10–15 However, several of these studies have suggested that the relative risk of cancer is greater among female patients than among male patients, with male patients having relative risk equal to or less than the male general population. 11–13,15 In addition, a number of reports have commented that paranoid schizophrenics have an increased risk for cancer while other categories of schizophrenia are associated with reduced risk. 1,4,5 To our knowledge, this hypothesis has not been further replicated. To complicate the picture, recent reports have raised the question of whether neuroleptic treatment might predispose to breast cancer through its dopamine blocking effect of raising serum prolactin. 16,17 Virtually none of the studies examines cancer rates by age, sex, and diagnosis, however; thus there is little opportunity to identify more subtle trends. The present study, part of a larger study of mortality among psychiatric inpatients, provided the opportunity to investigate these issues in a relatively large sample.