On March 23rd we pointed out positive signs of progress in Italy’s tragic war against COVID-19: (1) a bend in the confirmed case curve, and (2) the ability of small towns in Italy to control their outbreak with widespread testing and targeted quarantines.  We predicted that clear signs of a turn would occur in early April due to the imposition of stronger national travel restrictions on March 8th. More than a few people were skeptical about this “peaking” or “bending” of the curve. 

Moreover, there is a lot of confusion about the terminology bandied about in the press and by the President. Peaking is often not the sharp rise and fall connoted by the colloquial use of the word, but more like a broad topping of the case count or death count lasting weeks rather days.  In fact, a broad topping or a flattening of the curve is what is preferred to ease the load on medical facilities.  In the figure above the plot of daily confirmed case count (brown squares) seemed to spike to 6,557 on March 21st but really the peaking occurred on Mar 24th and it was a broad peak lasting more than a week (best fit brown curve) with several subpeaks as the infection peaked in one area (Lombardy), then another and then a third and so on.  We expect the overall case count to continue to fall, but with additional smaller peaks and dips as cities get infected and gain control over their infections.

Secondly, there is a marked difference between peaking in terms of confirmed case count and death count.  Deaths usually occur 3–7 days after a case of infection is confirmed and can range from -1 to 21 days before resolution – the short end is due to cases confirmed after death and the long end is due to the time that it takes for many cases to resolve themselves into deaths and cures.  Thus our preference to use a lagged5 mortality rate calculated from the current death count divided by the number of cases averaged 3–7 days prior.  In the figure above, the death count (blue diamonds) spiked to 919 on March 27th but exhibit a broad peak (best fit blue curve) centered around March 29th — 5 days after the broad peak in confirmed case count.  Again this broad peak shows waves of subpeaks and dips as individual towns experience and manages their outbreak.  One other thing to note is that the death count plotted against the right axis shows that about 15% of all confirmed cases result in death so the mortality rate is close to 14.8% rather than the 12.6% calculated from the current death count divided by current case count. The next important milestone to look for is when the net number of daily cases falls below zero, ie there are fewer new cases than resolved cases (deaths plus cures) so that the hospital facilities will have passed peak utilization.  We forecast this should happen in the next 7–10 days.  By the time Italy can be considered to have gained control of this outbreak in early May (daily count below 1% of the peak ~65) we expect 160,000 cumulative cases and 24,000 deaths.  When and if travel restrictions are lifted infection might reoccur so diligent testing, tracking, and treating will still be required until a vaccine is made available next year.

This pattern of broad peaking in the confirmed daily case count and the subsequent daily death count is what we expect to see in the US later this month.