The authors use a competing risk Cox proportional hazard model, which is a well-established method for this kind of time-to-event data. Simply looking at raw rates can be misleading because the background characteristics of the two groups are not the same, and one must adjust for confounding factors in order to make causal claims. The ca…
The authors use a competing risk Cox proportional hazard model, which is a well-established method for this kind of time-to-event data. Simply looking at raw rates can be misleading because the background characteristics of the two groups are not the same, and one must adjust for confounding factors in order to make causal claims. The case for bias or fraud is much more subtle, e.g. is the model well-formed + this data will likely never be made public by the Canadian government, so we have no direct way to reproduce their results and analyze the data ourselves.
The authors use a competing risk Cox proportional hazard model, which is a well-established method for this kind of time-to-event data. Simply looking at raw rates can be misleading because the background characteristics of the two groups are not the same, and one must adjust for confounding factors in order to make causal claims. The case for bias or fraud is much more subtle, e.g. is the model well-formed + this data will likely never be made public by the Canadian government, so we have no direct way to reproduce their results and analyze the data ourselves.