stuwrighthealthecon
commented
2 years ago We believe, having discussed this, that women with higher 10 year risk do actually have a higher risk of getting cancer in the next 10 years in the model. Women with higher 10 yr risk have higher lifetime risk. This means that in any time period a woman with higher 10yr risk has a probability of getting cancer. The probabilities noted in the comment for assigning a year in which the cancer occurs are conditional probabilities. While it is assumed in the model that the probability of getting cancer in any year (conditional on getting cancer) is the same regardless of risk, a higher risk woman's higher lifetime risk will essentially mean that she will have a higher risk in the next 10 years (and across all years in the model)
When the cancer incidence time is calculated the TYR value is ignored, and the time is sampled from the included ONS dataset. The TYR calculated by the model is not included in the calculation and the TYR value does not correspond to the ten-year risk for any simulated patient (with the model start_age set to 38, given that cancer is present there is around an 11% chance of the cancer occurring in the first ten years, whereas the risk data suggests that 24% of all cancers occur in the first 10 years {TYR/Lifetime Risk}). Furthermore, the risk matrix suggests that the likelihood of a cancer occurring within the first ten years given that cancer occurs varies between women, whereas the current model structure does not have the ability to vary this conditional probability between patients. In summary, the model uses TYR to stratify screening strategies, but does not then generate TYR values that match.