In four U.S. states (Colorado, Minnesota, Ohio, and Virginia), cumulative COVID-19 hospitalizations from March 2020 to April 2020 fit an exponential growth curve leading up to the effective stay-at-home date better than a linear fit: exponential $R^2$ = 0.973 (Colorado), 0.965 (Minnesota), 0.98 (Ohio), and 0.994 (Virginia), versus linear $R^2$ = 0.695, 0.865, 0.803, and 0.775 respectively. The exponential model was used to estimate 95% prediction bands for post-order hospitalizations, per Table 1 of Sen et al. (2020, JAMA).