Correction to response on starting point for productive NHS Race and Health Observatory
In an earlier rapid response to this article that explained that in the US having high income reduced self-rated health-less-than-good (HLG) proportionately more for whites than blacks but increased self-rated health-good-or-better (HGB) proportionately more for blacks than whites, I stated. “In the US during 2001-05, black and white rates of self-rated health-less-than-good (HLG) were 36.1% and 30.8% among the highest income group and 9.8% and 6.2% among the lowest income group.”
The response should have stated that the former (higher) HLG rates were for the lowest income group and the latter (lower) HLG rates were for the highest income group. I apologize for the error, especially since the contrasting pattern as to the comparative size of proportional effects of a factor on an adverse outcome and the corresponding favorable outcome is often quite difficult to understand even when the data are correctly presented.
The pattern whereby a factor that affects an outcome rate for two groups tends to cause a larger proportionate effect on the rate for the group with the lower baseline rate for the outcome while causing a larger proportionate effect on the opposite outcome rate for the other group (which is the group with the lower baseline rate for that outcome) is also well illustrated in survival analysis data in the recent report by Public Health England titled “Disparities in the risk and outcomes of COVID-19”  that has been discussed several times in the BMJ. For example, data in the report's Tables A.2 and A.3 (at 85-88) show that mortality rates for men and women seriously ill from COVID-19 were 8.60% and 3.19% among persons age 20 to 64 and 44.67% and 33.30 among persons over 65. Such data thus show that, while the relative gender difference in mortality was greater in the younger group than the older group, the relative gender difference in survival was greater in the older group than the younger group.
Such pattern means that being older increased mortality proportionately more for women than men, but decreased survival proportionately more for men than women. It correspondingly means that being male increased mortality proportionately more in the younger group than the older group but reduced survival proportionately more in the older group than the younger group.
Such patterns may not always be observed. But it is difficult to usefully interpret data on demographic differences or the effects of factors on outcomes rates of different groups without understanding them. As I explained here almost a decade ago, understanding the patterns is also crucial for making sound clinical decisions.
1. Scanlan JP. Essential starting point for a productive NHS Race and Health Observatory
BMJ (6 June 2020) (responding to Kmietowicz K. NHS Launches Race and Health Observatory after BMJ’s call to end equalities. BMJ 2020;369:m2191) https://www.bmj.com/content/369/bmj.m2191/rr-0
3. Scanlan JP. Re: The number needed to treat: a clinically useful nomogram in its proper context. BMJ (8 Dec 2011). https://www.bmj.com/content/312/7028/426/rapid-responses
Competing interests: No competing interests