Introduction: Reverse causation is a major bias potentially affecting the association of CT scan radiation with brain cancer. To minimize the effect of reverse causation bias, the pre-diagnostic symptomatic interval (PSI) can be used to inform the choice of exclusion periods.
Methods: We describe a method to calculate the percentiles of the PSI using the cumulative distribution function of the exponential distribution. We then present illustrative examples of the method.
Results: In twenty studies reporting PSIs for children diagnosed with brain cancers (total of 3,223 children), the sample-size weighted mean of the 95th percentile is 10.1 months. Benign or low grade astrocytomas had a of 21 months.
Conclusion: A one year exclusion period for brain cancer should correctly separate the majority of those patients who are causally exposed from those who appear to be exposed because of reverse causation. For slow-growing tumors such as low-grade gliomas, an exclusion period of at least two years is warranted.