Reply to the John Speakman critique of “Impact of obesity on COVID‐19 related mortality: A comment on estimates in Popkin et al (2020)” published in Obesity Reviews

The author falls into a common problem for individuals not well schooled in epidemiological methods. We used for this study: odds ratios. These are different from relative risks. Moreover, we combined all obesity categories (often individuals with obesity and individuals with morbid obesity were presented separately). We combined them. In addition, for articles where the authors did not calculate RR's or ORs, we calculated them. Odds Ratios and Relative Risks are often confused despite being unique concepts. ... The basic difference is that the odds ratio is a ratio of two odds, whereas the relative risk is a ratio of two probabilities. (The relative risk is also called the risk ratio.) “Risk” refers to the probability of occurrence of an event or outcome. Statistically, risk = chance of the outcome of interest/all possible outcomes. The term “odds” is often used instead of risk. “Odds” refers to the probability of occurrence of an event/probability of the event not occurring. At first glance, though these two concepts seem similar and interchangeable, there are important differences that dictate where the use of either of these is appropriate. The reasons that Dr. John Speakman did not find some results or found different results in the citations are as follows: (1) Some studies reported relative risks (RR), while we presented odds ratios (OR). (2) Some studies did not report OR or RR. (3) Some studies reported OR for individuals with obesity (BMI ≥ 30) and OR for individuals with morbid obesity (BMI ≥ 35 or BMI ≥ 40) separately, while we combined those individuals with obesity and morbid obesity in our analyses. (4) Some studies reported OR without original data. We had to recalculate standard errors. Due to rounding numbers, our results may be different from those in the original citations. (5) And A few studies reported adjusted OR and unadjusted OR without original data for calculation. We chose unadjusted OR for consistency. Results from the four papers questioned by Speakman surprised us as well, because our hypothesis was that being an individual with obesity would increase the risk of adverse outcomes. We recalculated all ORs and found identical results as reported in our metanalysis. Specifically, Goyal et al. reported hazard ratios (HR) of 0.75 (0.56–1.00) for individuals with overweight (BMI, 25.0–29.9 kg/m), 0.98 (0.70–1.36) for individuals with mild to moderate obesity (BMI, 30.0–39.9 kg/m), and 1.41 (0.74–2.70) for individuals with morbid obesity (BMI ≥ 40). The authors concluded that “the association between individuals with obesity and adverse outcomes could differ in other settings and thus merits additional investigation.” Based on the data provided in the paper, we recalculated OR and its 95% confidence internal, which was 0.59 (0.43–0.81) (Table 1). For ICNARC report (we apologize for the typo in the online supplemental tables), our results were different from those presented in Figure 20, because HRs were reported relative to the median age of 60 years. However, our results were consistent with those presented in Table 19 in the report, which included all patients aged 16 years and older. When we combined individuals with obesity and morbid obesity, the results showed that obesity did not increased the risk of mortality [OR = 0.82 (0.74–0.92)] (Table 1). Kim et al (2020) published their study on the Clinical infectious diseases after we published our metanalysis on the Obesity Reviews. Their newly published paper presented an adjusted OR of 1.33 (1.15–1.50, visually guess) but did not present data for