Effective containment explains sub-exponential growth in confirmed cases of recent COVID-19 outbreak in Mainland China

Benjamin F Maier; Dirk Brockmann

In Hubei, we find that the initital increase in confirmed cases is followed by an algebraic scaling with time, i.e. C(t) ∝ t µ , with a scaling exponent µ ≈ 2.3 that persisted until Feb. 9th, see Fig. 1A . On Feb. 12th, the case definition was changed by Chinese authorities which labeled a large number of previously unconfirmed cases as confirmed, leading to a discontinuity in the curves. We will therefore only consider data prior to Feb. 12th, 6am UTC. Fig. 1B illustrates the cumulated case count in all affected provinces except Hubei province. In the period Jan. 21st until Feb. 2nd the curve follows an algebraic scaling law t µ with µ ≈ 1.9 lacking the initial exponential phase observed in Hubei province. Starting on Feb. 2nd, the observed case count starts deviating towards lower case counts. This poses the question whether the observed behavior is an averaging affect introduced by accumulating case counts across provinces. Interestingly, Fig. 1C provides evidence that this is not so. The confirmed case counts in the most affected provinces all exhibit a scaling law with exponents close to µ = 2. Among the most affected provinces only Chongqing Province exhibits a significantly lower exponent. Furthermore, all provincial case count curves start deviating from the algebraic curve around Feb. 2nd.