Schizophrenia and cancer: an epidemiological study
Goldacre MJ, Kurina LM, Wotton CJ, Yeates D and Seagroat V;
Commented by , 23 Oct 2005
Aim of the study
Research in the association between schizophrenia and cancer has a very long history, and so has the hypothesis of decreased cancer risk in schizophrenia. An early review by Martin du Pan and Müller in 1977 (ref. 1) quoted numerous studies, the first one having been published already in 1909.
The results were conflicting with some studies showing a decreased risk of cancer in schizophrenia, others no difference in risk and again others even an increased risk. These early studies had, however, a number of methodological limitations, especially by far too low numbers given the rarity of both schizophrenia and cancer.
Therefore, a number of population based incidence studies have been carried out which showed again conflicting results. Goldacre and colleagues add the first English population based study to this body of literature.
Methods
Data from the Oxford Record Linkage Study which mainly includes records of all hospital admissions in National Health Service (NHS) hospitals and all deaths within the former Oxford NHS Region from 1963 to 1999 were analysed.
Until 1994 the data-set also contained all contacts with psychiatric services in Oxfordshire. A schizophrenia cohort and a reference cohort were constructed by identifying the first admission to an NHS hospital or other service.
Both cohorts were compared for any subsequent NHS hospital care for, or death from, cancer. It was assumed that the rates of cancer in the reference cohort would be similar to that in the general population.
Results
The schizophrenia cohort involved 9649 individuals and the comparison cohort nearly 600 000. The risk ratio (RR) for any cancer was virtually identical to that of the control cohort (RR 0.99).
The only kinds of cancer that showed statistically significant differences compared to the comparison group were cancer of the oesophagus (significantly raised, RR 1.61) and skin cancer (RR 0.56, significantly decreased).
In contrast to a number of previous studies the risk ratio (RR) for cancer of the lung was raised (RR 1.18), although not statistically significantly so.
Dr Leucht's comments
The scientific literature about the association between schizophrenia and cancer is confusing. Numerous articles had already been published until the late 1970s, but due to methodological limitations of these studies population based incidence studies have been demanded and published.
In the beginning a Danish population-based study showed a decreased risk for cancer in schizophrenia in general and maybe even more surprisingly also of lung cancer, although it is known that many people with schizophrenia smoke.
Some of the subsequent studies e.g. in Australia and Israel partly confirmed these findings, while a large Finnish population based study found increased rates of lung cancer.
A new finding of the current study is a lower risk for skin cancers that the authors explain by decreased exposure to sun by people with schizophrenia.
In summary, after 100 years of research the epidemiological puzzle of the association between schizophrenia and cancer remains unsolved. A joint discussion paper of the different groups of their results and methods might be useful.
References
1. Martin du Pan R and Müller C. La mortalité par cancer chez les patients psychiatriques hospitalisés. Schweizer Medizinische Wochenschrift 1977; 107; 597-604