November 14, 2011

oral contraceptives and prostate cancer

This analysis accompanied a roundup which can be viewed here.

Title, Date of Publication & Journal

Oral contraceptive use is associated with prostate cancer: an ecological study, BMJ Open, 14th November 2011

Summary 

• The study design and methods used are appropriate to answer the question being asked.
• The paper is well written and is very clear on the first page that this study design has significant limitations with respect to showing a causal relationship between oral contraceptive use (OC) use and prostate cancer (PCa) and as such must be considered hypothesis generating. The press release also states this but it is less obvious than the full paper.
• Inclusion of other methods of contraceptive was important to see if OC use itself was different to other methods and that it wasn’t just a marker for sexual activity which may be associated with prostate cancer.
• Inclusion of nations GDP in the analysis was important due to its likely confounding effect.
• Discussion is well written. Plausible explanations are discussed with reference to the results from this paper and from others, including research that doesn’t support the hypothesis that PCa is related to endogenous androgens.
• Authors state that ‘additional confounding does exist and should be explored in future studies’.

Objective

To examine association between prostate cancer (PCa) incidence and mortality and oral contraceptive use.

Study design

An ecological study comparing population levels of oral contraceptive use and PCa incidence and mortality. Data from 87 nations were included with an aim to include equal representation from developed and under-developed countries.

Methods

Authors used age-standardised rates of prostate cancer from the International Agency for Research on Cancer and contraceptive use data from the United Nations World Contraceptive Use report. A Pearson correlation and multivariate linear regression were used to associate the use of difference methods of contraception to prostate cancer incidence and mortality. Analyses performed by nation and worldwide. The multivariate analysis adjusted for the nations GDP due to its likely confounding effect.

Results

OC use was significantly associated with PCa incidence and mortality in individual nations worldwide. PCa incidence was also associated with OC use in Europe and by continent but PCa mortality was not associated with OC use in Europe or by continent. No significant associations between PCa and use of condoms, IUD or vaginal barrier methods of contraception were found. The results for condom or IUD use were not as consistent across the different analyses as those for OC use and vaginal barrier method.

Strengths

This was a large study using appropriate methods and analysis to look at this association at a population level.
The authors aimed to include an equal representation of developed and under-developed countries and also controlled for the nation’s GDP which was very important as it is already known that countries with higher standards of living have higher rates of PCa and also have more PCa screening.
The discussion of the results is well written and it is made clear that no causal relationship can be inferred from these results and that they should be considered hypothesis generating.

Limitations

This type of study design cannot show a causal link between OC use and incidence or mortality of PCa.
Data are from reports which are based on survey and registry data which is unlikely to have been reported consistently accurately across nations.
Other methods of contraception and nations GDP were the only two confounders included in the analyses although the authors are quite clear in the discussion that ‘additional confounding does exist’.

Conclusions of the Paper

The discussion highlights a few plausible explanations for this which should be investigated further. These include the fact that PCa has been associated with sexual transmission but that the results for the different methods of contraception in this study do not really support that conclusion. The authors discuss other recent research that both support and contradict the results of this study. They also point out that the true levels of Endocrine Disrupting Compounds (EDC) in the water supply and food is not available, but hope that it will be in the future – this will be the key data to help support or refute the results from this study. The fact that the association is seen in the incidence of PCa but not consistently in the mortality from PCa is not discussed.

Before The Headlines is a service provided to the SMC by volunteer statisticians: members of the Royal Statistical Society (RSS), Statisticians in the Pharmaceutical Industry (PSI) and experienced statisticians in academia and research. A list of contributors, including affiliations, is available here.