Pages

Saturday, March 21, 2020

COVID-19: data, information and opinion

I've been following reporting on the coronavirus COVID-19 both for specific personal reasons and out of the general interest in data-driven communication I've tried to practice on this site. Realizing the shelf life of this post is likely to be incredibly short, I thought I'd take some time to write down which media is proving useful to me in explaining the situation, some sources of data on the spread of the virus, and what actions I'd hope to see taken.

I am the spouse of a paramedic - which is among the professions most likely to contact the coronavirus. We are both past our half-century mark, and we're fortunate that all of our parents are alive. We can't visit them without being confident we don't have the virus. I've felt unwell, with mild systems that could fit this virus, or a number of other things: my wife's symtoms were bad enough that we were directed to drive over an hour (each way) for testing this weekend, not so bad they performed the testing, but bad enough she was to quarantine for 2 weeks.

I won't attempt to communicate how the virus operates (there is a video for that) - but the specifics aren't necessarily what is driving the actions of, in Ontario anyway, the past 8 days. One of the most-viewed pages ever on the Washington Post site explains exponential growth with a model of a fictitious virus spreading. The paper, using four scenarios to address the speed at which the virus spreads, is credited with making "'social distancing' easy to understand." From that post: "If the number of cases were to continue to double every three days, there would be about a hundred million cases in the United States by May." It doesn't mention that the number would grow to include the entire US population within 5 more days, but ...

Maybe the most impactful graphic of the growth in the impacts to Canadians is the graphic on number of confirmed cases from Our World in Data reporting, filtered to show only Canada:





I don't like these graphs given without context, and the missing context is that the impacts of the actions taken one day don't show in the statistics for a week or more. This will be explained in the remainder of the post.

I've noticed a number of graphics, accompanying advice to isolate, a.k.a. socially distance, that show the growth in cases without measures to reduce the spread, and the time bought with introducing measures - with a line added to the graphic, usually a straight horizontal one,  demonstrating there are limits to the healthcare system capacity. The graphics specific to jurisdictions are more alarming as that capacity is usually much lower than the demands that will be made due to the spread of the virus even with protective measures.

Graphic from Samantha Lee/ Business Insider Australia

Tomas Pueyo's posts at Medium have been remarkable. His March 10th post,
Coronavirus: Why You Must Act Now, having received 40 million page views (and being translated into 30 languages), and his recent Coronavirus: The Hammer and the Dance is the best summary I've read. Both of those works contain a graphic that differentiates the number of cases in Hubei (where the coronavirus allegedly originated), by "date of onset" and by "date of diagnosis." It's a very important distinction which provides the context for the current communication of Canadian case data - which is only by the date of diagnosis.



A drop in new cases may be close in Ontario, and elsewhere in Canada. Puevo communicates the true challenge is to reduce the transmission rate (R) to below 1. I think the exponential growth idea overwhelms most, and concentrating on R is a much better method to achieve improvement. As long as one carrier passes the virus on to more than one other person, total cases will grow (and swamp the health care system) - but once R is below 1 active cases will start to shrink.

The immediate challenge is to drop R below 1. Hopefully aggressive actions preventing socialization in-real-life will soon get R below 1, but it would be nice to do that without doing too much damage to all the people currently relying on the economy for money, goods, and services.

I notedpaper from the Imperial College COVID-19 Response Team on social media 3 days ago - and subsequently found that team's work cited as impacting policy actions taken in North America and the U.K. I didn't read the full paper but was struck by the table demonstrating the variance in health impacts by age:


There are issues with data quality, but the general trend is pretty obvious: the COVID-19 is not very dangerous for children, few in their twenties will be hit hard, and then things worsen to where people my age group are expected to to have 6 deaths per thousand infections, increasing to over 90 per 1000 infections for those over 80.

The uneven impacts of infection may very well impact adherence to regulations - such as maintaining 2 meters separation the rare time one leaves their shelter. The impact of the economic slowdown may very well be disproportionately borne by those least likely to be significantly harmed by the virus.

I have not yet seen a data set with the statistics I was tempted to dive into despite all the very fine data reporting being done on COVID-19. Most significantly, in my opinion, are the number of tests and the transmission rate. Until the transmission rate is reduced below 1 the number of beds, ventilators and personal protection gear for health care workers are also important measures. I noted above I was uncomfortable with the straight line often being presented to indicate that "healthcare system capacity" is flat. I find that unlikely, living with a quarantined emergency response worker and knowing of others I imagine the line could move lower.

More tests would help in a number of ways: people without the virus would not be quarantined for weeks accidentally, and return dates would be based on something more concrete than the calendar. Also - some of us might find when we could visit our elderly relatives in real life.  The test I hope to see developed is a serological test that would determine if a person has ever had the virus - which wouldn't guarantee immunity but provide some confidence to those who'd already carried whatever strains of COVID-19 are in their communities.

It may also be necessary to expand the healthcare's system capacity if the transmission rate is not dropped below 1 quickly enough. I believe we should be identifying spaces, and equipment, should secondary sites need to be set up to care specifically to those carriers of the virus requiring hospitalization. I suspect these are not hard to find at this time of year as arenas and curling facilities are mostly vacant - not that anybody would prefer these solutions outside of a crisis.

I'm hopeful there are people with better ideas than I, but not unrealistically hopeful. I find most mainstream reporting relatively dreadful. Yes, people die who are under 60, but reports I find on those deaths de-emphasize the deceased's pre-existing conditions. The emphasis seems to be on fear - the old "if it bleeds it leads" impulse persists.

Mostly, I hope restrictions on retailers, and restaurants, are lifted shortly after R drops below 1. I do not think we should ask young people to sacrifice socializing for the security of the rest of us, nor do I think we should assume the economy can recover from whatever scars we put on it.

The discovery of a treatment would be nice, but we might have to take steps to restore commerce before COVID-19 is eliminated.


Addendum/s



March 30th: found the Ontario government data feed. Hopefully this quick Power BI report I've done will update around noon each day.


No comments:

Post a Comment