Post #550: Looking one week ahead.

Posted on March 15, 2020

This began as a compendium of a few things I found most informative about the current coronivirus pandemic.  But in the end, it’s a look at the week ahead.  How will this probably look one week from now?  Short answer:  Cumulative US cases should rise to about 10,000 (from about 3000 currently), but the US hospital system should have no major difficulties through the end of the week.

Treatment protocols for hospitalized patients

There is no medical treatment, other than providing support.  If you want to know all there is to know about the epidemic in China, start with this World Health Organization report (.pdf). For example, that report says this (page 32):

That said, this did not stop Chinese physicians from administering Tamiflu (Oseltamvir) to about one-third of the most severely ill cases, and giving inflammation-reducing steroids to about 20% of cases, per this New England Journal of Medicine article.  Nor could one reasonably blame them, as those might help, and this has a high mortality rate and has not been studied previously.

But the upshot is that the only tools effective against this disease are the ones that keep it from spreading.  So it’s back to the pre-WWII-era of control of infectious disease (Post #549).  It’s back to public health as it was practiced in the US a century ago.  Wash your hands, cover your cough, quarantine the sick, and avoid large gatherings of people.  The entire course of this epidemic, in this country, will depend on how well those things work here.

Disease spread is following near-identical curves in Western countries.

Predictions are hard, particularly about the future, but at least you can say that the early spread of the infection seems to follow more-or-less the same growth curve across most countries.  These are cumulative case counts, starting from the time each county had 100 reported cases.  To read case counts, multiply by 100 (so 30x = 3000).  These are all county-specific counts (dots) projected against Italy’s experience (gray line).

Source:  “The United States Is Not a Coronavirus Outlier”, by Kevin Drum, in Mother Jones.

I thought it was well worth looking at these handful of graphs in this Mother Jones article because they tell such a clear and simple story.  The graphs are dated 3/13/2020, and so are now (3/15/2020) two days old.

(And, sure enough, if you extend the US figures by two days, you get just about 3000 cases, which is just about where things stand this morning in the US.  And the current count in Italy is about 21,000, again, right on the predicted curve.)

Whether or not various actions will pull us below that growth curve remains to be seen.  Presumably, those would only have an effect to the extent that we impose restrictions faster, or more broadly, than Italy/France/Germany did.

So, in terms of case counts, we’re maybe two days behind Germany, and four days behind France.  And (as noted in Post #549) maybe a week and a half behind Italy.  And, again, while the future is hard to predict, if you had to take a guess — well, just have a look at those graphs.  As of right now, that looks like a pretty good guess.  At least for the near term.

The point being that as the US numbers rise over the next couple of weeks, that’s only to be expected.  That’s not unusual.  That’s just average.  If we have 3000 today, reading the graph, we should expect a case count of about 10,000 one week from today.  And, to repeat, that’s just normal.  That’s if we’re average or typical.  And we test people so that cases are identified.  And our avoidance strategies are no more effective or stringent or timely than was the case in those European nations.

On the positive side, it only took about a month for the Chinese to hit the “inflection point” on the curve, from the time they passed a few hundred total cases.  That is, it only took a month before the number of new cases began to decline.

The graph below is from a World Health Organization report on China’s experience with coronavirus (.pdf).   This one is new cases reported each period, not cumulative (see cumulative next).  So the cumulative count continues to rise throughout this period, but China broke the back of the epidemic right around the end of January.  That’s when the number of new cases stopped growing.


Source:  a World Health Organization report on China’s experience with coronavirus (.pdf).

In other words, by their extreme measures of isolating the affected population, the Chinese appear to have more-or-less brought the epidemic under control in a month.  The question for us is, will we do better or worse than the Chinese?

Going back to cumulative cases,here’s a different take on comparing Italy to China.  (These curves start a few days later than the ones above, when the cumulative case count hit 400 cases, not 100 as above.)  Again, if you add in a few more days of data, the Italian growth curve in cumulative cases remains quite close to the Chinese curve.

Source:  The Guardian.

The upshot is that at this point, it’s all a waiting game.  We know cumulative case counts will continue to climb.  It’s just a question of if and when our “social distancing” measures have an effect.  Will ours be as effective as those used by China?

We’ve closed the schools, but we still have half-a-million daily ridership on the Metrorail system.  We have not, as other countries have done, closed businesses.

That said, other countries contained this without such drastic measures.  But if you read up on the success stories such as Singapore, it’s like a laundry list of what the US did not do.  Everything they did made perfect sense, and none of it was done here in the US.

So if you had to guess, you might guess that our measures may not be as effective as those undertaken in China and elsewhere.  There are ways to shut this down without draconian measures, but that train left the station for the US about a month ago.

This next week will not strain average US hospital capacity.

The big system-wide worry in this is exceeding the capacity of the hospital system.  (Because, again, all they can do for these patients is provide support, which may mean putting them on a ventilator if the case is bad enough.)

Here, I’m just going to compare to flu.  How burdened with coronavirus will US hospitals be, next Sunday, if the US reaches that projected 10,000 cases by next Sunday.

To cut to the chase, by the end of next week, in terms of bed-days associated with coronavirus admissions, this won’t yet have as large an impact as a typical flu season.  By the end of next week, hospitals still will not be swamped.  But it will be a whole lot more dangerous for hospital staff and medical personnel in general, compared to our more normal flu season.

Details follow.

In a typical flu season, the US gets an additional half-million-ish hospitalizations due to consequences of flu.  (That’s the CDC’s estimate.)  It’s enough that it adds a distinct and clearly visible seasonality to Medicare hospital discharge data, when plotted by month, as many of those hospitalizations are for the elderly.

The exact number of flu hospitalizations is a surprisingly squishy statistic.  Not just for potential imprecision on the diagnosis, but to the very large difference between those hospitalized for flu, and those hospitalized with flu.  So, e.g., if an elderly person with congestive heart failure (CHF) is hospitalized for an exacerbation of CHF brought on by flu, the reason for the hospitalization (principal diagnosis) is CHF.  According to the 2016 AHRQ HCUP (US hospital discharge) database, in 2016, the number hospitalized with flu (any diagnosis of) was about 2.5 times those hospitalized for flu (principal diagnosis of).

The same uncertainty will occur with coronavirus hospitalization statistics.  Hospitalizations and deaths in other countries have been concentrated among the elderly with significant comorbidities.  Those who want to downplay the impact will count admissions for (with principal diagnosis of) coronavirus, not admissions with coronavirus.

So, if you do the math, that typical annual flu case load is an additional 500,000 discharges spread over the flu season (typically, Dec-Feb), or an average of about 40,000 extra hospital discharges per week.  But that’s in the context of a US health care system that has about 36 million hospital discharges annually, for all payers.  Or roughly 700,000 discharges per week.  A bad flu season can result in some hospital crowding, certainly at peak periods, but what I’m trying to get across is that the US health care system is set up to handle an extra 40,000 hospital admissions per week, nationwide, on a routine basis, for flu.

Based on limited information, both the hospitalization rate and Intensive Care Unit (ICU) use rate differ substantially between the Chinese and Italian outbreaks.  China reported that about 5% of diagnosed cases required an ICU level of care, while in Italy, preliminary data suggest that 13% were given that level of care (Reference).  From that same reference, you get the sense that it was ICU beds, not hospital beds in general, that are in short supply in parts of Italy.

That said, in the US, of individuals who are sick enough to be admitted to the hospital for flu, an estimated 10% spend at least some time in the ICU (reference).  Unsurprisingly, it was individuals with severe comorbidities (e.g., cardiovascular disease) who were most likely to require ICU use when admitted for flu.

In terms of the overall hospitalization rate, I could not find a firm figure, but the Chinese definition of “serious” illness includes conditions that, I think, would normally warrant hospitalization in the US.  In China, about 20% of cases fell into the severe or critical categories (reference).  Separately, a theoretic estimate from Johns Hopkins suggested that potentially 25% of diagnosed cases would require hospitalization (reference).  Between the Italian experience (13% required an ICU bed) and these other estimates, it would not be implausible that, of those diagnosed (i.e., severely ill enough to seek a diagnosis), 20% might require hospitalization.

At that level, this coronavirus pandemic would only add 2000 cases to US discharges, by the end of this coming week.


Bed days per case.  First, these cases have much longer lengths-of-stay than typical US flu cases.  A typical US flu discharge is a four-day stay (calculated directly from the AHRQ HCUPnet database, at this link).  In China, average length of stay was 12 days.  Thus, each coronavirus case effectively takes up as many hospital bed-days as three standard flu cases.

Adjusting for longer average stays, the expected hospital case load from coronavirus, by the end of next week, will be about 6000 flu-discharge equivalents.

Impact on hospital staff and required staffing levels.  The outstanding difference between typical flu and coronavirus is that nobody has immunity yet.  The Chinese experience saw thousands of hospital workers and medical staff inflected with coronavirus.  The need for isolation makes these cases more difficult and resource-intensive than standard flu admissions.

The upshot of all of that is that a) in all likelihood, case counts are going to climb in the US this week, and b) the US health care system will be able to handle this weeks cases. 

Beyond that, it’s hard to say.  It all depends on how effective our efforts are at preventing the spread of disease.