On 2017 May 30, Thomas Heston commented:

The data presented in this study do not support ageism but rather illustrate a primary feature of statistics: as the sample size increases, you are more likely to find a statistically significant p-value. First of all, look at admission rates to the coronary care unit (CCU). If there was a gender bias, then a greater proportion of men (or women) without acute myocardial infarction would be admitted to the CCU compared to the other sex. This study shows clearly that sexism in CCU admission rates is not present. In 1990, 53% of men admitted to the CCU did not have an infarction compared to 52% of women, i.e. over half of the people admitted to the CCU did not have a myocardial infarction yet received CCU level care. This difference is not statistically significant. Then, in 1992 it was found that 48% of men admitted to the CCU did not have infarction compared to 54% of women. This difference was statistically significant, indicating that a greater proportion of women vs men admitted for suspected myocardial infarction received CCU level care. If anything, this represents sexism against men, not against women. But if we look at the data closer, we find that quite a few more people were admitted for suspected infarction in 1992 compared to 1990. This larger sample size means that we are more likely to find statistically significant differences. So while no statistically significant difference was found with the lower sample size (n=1473) in 1990, the higher sample size in 1992 (n=1967) resulted in a statistically significant difference. This is not surprising, this is just how statistical analysis works: large sample sizes are much more likely than small sample sizes to show statistically significant differences. To highlight this further, look at when the authors broke down their data by age (see Table 4). When broken down by age, no significant differences were found, but when all ages were combined, a statistically significant gender difference was found in 1990 and in 1992 in terms of thrombolytic therapy. This does not indicate that the differences are explained by age, but rather indicates that small sample sizes are less likely to show statistically significant differences compared to large sample sizes. The data does show a difference in rate of thrombolysis that is interesting, with an associated difference in mortality rates. My interpretation of this data is that a higher percentage of men vs women with confirmed infarction received thrombolytic therapy, and a lower percentage of men vs women with confirmed infarction died. Unlike the CCU admission rates which are not very illuminating and do clear show gender or age bias, the rate of thrombolysis suggests a bias towards more aggressive treatment of men with infarction (likely starting with differences in their emergency room care Heston TF, 1992). Furthermore, the data also suggest that thrombolysis works, as a lower in-hospital mortality rate was seen for men vs women. However, mortality was not broken down by whether or not the patient had received thrombolysis, so a meaningful analysis of mortality by gender and thrombolytic therapy is not possible.

On 2017 May 30, Thomas Heston commented:

The data presented in this study do not support ageism but rather illustrate a primary feature of statistics: as the sample size increases, you are more likely to find a statistically significant p-value. First of all, look at admission rates to the coronary care unit (CCU). If there was a gender bias, then a greater proportion of men (or women) without acute myocardial infarction would be admitted to the CCU compared to the other sex. This study shows clearly that sexism in CCU admission rates is not present. In 1990, 53% of men admitted to the CCU did not have an infarction compared to 52% of women, i.e. over half of the people admitted to the CCU did not have a myocardial infarction yet received CCU level care. This difference is not statistically significant. Then, in 1992 it was found that 48% of men admitted to the CCU did not have infarction compared to 54% of women. This difference was statistically significant, indicating that a greater proportion of women vs men admitted for suspected myocardial infarction received CCU level care. If anything, this represents sexism against men, not against women. But if we look at the data closer, we find that quite a few more people were admitted for suspected infarction in 1992 compared to 1990. This larger sample size means that we are more likely to find statistically significant differences. So while no statistically significant difference was found with the lower sample size (n=1473) in 1990, the higher sample size in 1992 (n=1967) resulted in a statistically significant difference. This is not surprising, this is just how statistical analysis works: large sample sizes are much more likely than small sample sizes to show statistically significant differences. To highlight this further, look at when the authors broke down their data by age (see Table 4). When broken down by age, no significant differences were found, but when all ages were combined, a statistically significant gender difference was found in 1990 and in 1992 in terms of thrombolytic therapy. This does not indicate that the differences are explained by age, but rather indicates that small sample sizes are less likely to show statistically significant differences compared to large sample sizes. The data does show a difference in rate of thrombolysis that is interesting, with an associated difference in mortality rates. My interpretation of this data is that a higher percentage of men vs women with confirmed infarction received thrombolytic therapy, and a lower percentage of men vs women with confirmed infarction died. Unlike the CCU admission rates which are not very illuminating and do clear show gender or age bias, the rate of thrombolysis suggests a bias towards more aggressive treatment of men with infarction (likely starting with differences in their emergency room care Heston TF, 1992). Furthermore, the data also suggest that thrombolysis works, as a lower in-hospital mortality rate was seen for men vs women. However, mortality was not broken down by whether or not the patient had received thrombolysis, so a meaningful analysis of mortality by gender and thrombolytic therapy is not possible.