Dear Editor,

We recently published a study on mobility and mortality around the COVID-19 pandemic (J Gen Intern Med 2020 35, 3100–3101). In the analysis, we estimated correlation coefficients and p-values comparing changes in population mobility and COVID-19 mortality across multiple countries over time. Upon reviewing our computer code, we have discovered an error that leads to changes in our estimated p-values.

The primary analysis of driving and COVID-19 mortality (r = 0.41, p < 0.001) changes now to (r = 0.41, p = 0.014). The sensitivity analysis of specific countries (r = 0.42, p = 0.01) changes now to (r = 0.45, p = 0.052). The sensitivity analysis of countries with the top public health systems (r = 0.38, p = 0.02) changes now to (r = 0.38, p = 0.11). The walking and COVID-19 mortality analysis (r = 0.43, p < 0.001) changes now to (r = 0.43, p = 0.009). No other changes are necessary.

In summary, the corrections do not change the direction of our findings or the qualitative interpretation of results. The primary analysis is essentially unchanged. We draw attention to these corrections for investigators elsewhere interested in pursuing more studies of mobility and mortality.

Sincerely,

Jonathan S. Zipursky, MD FRCPC

Clinical Associate, Division of General Internal Medicine & Clinical Pharmacology

Eliot Phillipson Clinician-Scientist Training Program, University of Toronto

Donald A. Redelmeier, MD, FRCPC, MS(HSR), FACP

Canada Research Chair in Medical Decision Sciences

Division of General Internal Medicine, Sunnybrook Health Sciences Centre