The article claimed that the (original) article's conclusions about strenuous running were suspect because they were based on 2 deaths out of a sample of 40. B

The article claimed that the (original) article's conclusions about strenuous running were suspect because they were based on 2 deaths out of a sample of 40. But is that really the case? I get sick of hearing that "the sample size was too small" when no analysis is done of the statistical significance of the results. In this case, the difference between death rates for strenuous vs moderate runners was significant (by my calculation based on their 95% intervals). Given the small sample size, there are still questions to be asked about the methods they used to correct for age, sex, etc. But still, outright dismissal based on small sample size is not reasonable. If the sample is big enough to give statistically significant results, that should be an a priori reason to accept the results.