The sample size 1300 cyclists (and 69,000 non cyclists) is not that small. If you look at all-cause mortality, you can see that entire 95% confidence interval lies well to the left of the control group median. There is a ~5% chance that the protective effect is anything less than 25%.
With a larger sample of cyclists, the exact amount of protection may vary by a few percentage point and the confidence interval would be narrower, but the overall protective effect of cycling, is clearly demonstrated here.