Active cases in isolated places
When comparing the Covid-19 experiences in different countries, there’s always the “apples and oranges” problem. Here I’d like to compare the number of active Covid-19 cases in 4 relatively isolated places:
New Zealand: 37 active cases
Canadian Maritimes: 15 active cases
Alaska: 3239 active cases
Hawaii: 2307 active cases
I can see three possible reasons for why isolated parts of the US have more cases:
1. Perhaps it’s easier to travel to these areas from the outside.
2. Perhaps local governments in the US have less effective policies for containing the virus (which doesn’t necessarily mean less optimal policies.)
3. Perhaps there are cultural differences.
The differences are so large that they are probably not reflective of genetics or population density or degree of geographical isolation. It’s still quite difficult to travel to Hawaii, although that may soon change.
It’s also noteworthy that this huge divergence did not happen right away. As late as mid-May, New Zealand had more active cases than Alaska or Hawaii. So it wasn’t a question of some regions being hit hard at first, before policymakers had time to react.
There’s a bit of a puzzle here, and I’d be interested in theories that might explain this pattern.
Tags:
8. October 2020 at 01:41
Don’t know about the other locations, but here in Hawaii there was essentially no preparation completed during a first stay-at-home order in March. I suppose that’s a state capacity problem. Hospitals didn’t increase staffing or beds, data wasn’t gathered and shared, and the Dept of Health hid tracing capability:
“[Former Health Director] Anderson had been blasted for grossly exaggerating the state’s capacity to conduct contact tracing while refusing to accept help in the form of additional manpower. He has also been rebuked for his failure to roll out widespread coronavirus testing or release comprehensive data about COVID-19 hotspots.”
https://www.civilbeat.org/2020/10/were-in-a-really-tough-place-right-now-can-hawaiis-new-health-director-make-a-difference/
It’s a real tragic comedy in that the state and county governments are performing so poorly. Even though my priors are classical liberal, I really don’t think it’s my biases that are seeing real discontent and frustration towards state/local leaders for their constant mistakes here.
8. October 2020 at 05:12
Isn’t Alaska a popular vacation spot in the summer for people who like the outdoors? Could be that nobody wants to go to Alaska in the spring but they do in the summer, bringing COVID with them.
8. October 2020 at 05:37
in no particular order:
1. bad data
2. different testing regimes
3. different strains – causing milder/no symptoms
4. different strains – causing change in level of contagion
5. different levels of social cohesion enabling restrictions to be effective for longer
6. timing of initial introduction (later introduction allowed restrictions to have greater overall effect at minimizing foothold)
8. October 2020 at 08:34
Laron, Thanks for the info.
Garrett, Don’t know about Alaska, but I know that Hawaii shut off tourism.
Chris, No, it’s obviously not those factors. Not a 100-1 difference. Except perhaps #5.
8. October 2020 at 09:30
Obviously the correct thing to do would be to take all of the places on Earth with similar characteristics to Alaska and Hawaii, in terms of “relative isolation,” and see how Alaska and Hawaii are doing.
This exercise might not be very valuable – “relative isolation” is just one characteristic, so maybe at least also maybe also places of similar population? – but at least it would be something.
Anyway, The TMI approach – cherry-picking two places which may or not have similar “relation isolation” scores to Hawaii and then claiming “aha!” – is obviously of zero value. It shouldn’t make anyone (except an idiot) revise their view on anything. We all know the US isn’t doing very well, the interesting question(s) is just how not well and why.
8. October 2020 at 09:31
A recent Atlantic article[1] argues that COVID-19 is characterized by a very high overdispersion[2], i.e. a small number of infected accounts for the majority of the spread. Apparently this combined with exponential spread can lead to very heterogeneous outcomes that we simply cannot attribute to concrete measures or policies.
Maybe we are in the unlucky situation that very different outcomes are really mostly down to bad luck in the initial conditions.
8. October 2020 at 09:31
This perhaps similar attempt from the (I think) last TMI Coronavirus Stats post was better, but still:
“The best America state (Alaska) has 62 death per million, while in Canada there are provinces with no deaths at all. Indeed they have 8 provinces with a combined death toll of 45, the same as Alaska, but with many times more people than Alaska.”
If you had asked me for my advice on how to make this point (not that in general I’m claiming this would improve the blog), I would have suggested this:
“If Alaska was a Canadian province, it would be one of 6 with a population between .5 and 1.4 million people. It would be one of two with a death rate of about 65 per million, whereas the other four are all between 3 and 20. In other words, a high-performing US state looks a lot like a low-performing Canadian province.”
See? Instead of making the point with the “larding” inherent in the omission of Nova Scotia, you could have included Nova Scotia and still made the point. In fact now (in my view) it’s an interesting point! Alaska happens to be roughly the same size as several Canadian provinces, of which one has done about as well as Alaska, but the others have done much better.
Comparing Alaska to the Canadian provinces made sense to me. But things like specifically disincluding Nova Scotia in the previous post (becuase they had a big outbreak earlier on) and then including Nova Scotia in this post (because cases there are low now) makes no sense to me. It almost seems like an effort to make the point as poorly as possible.
8. October 2020 at 09:32
Forgot the references:
[1] https://www.theatlantic.com/health/archive/2020/09/k-overlooked-variable-driving-pandemic/616548/
[2] https://pubmed.ncbi.nlm.nih.gov/32685698/
8. October 2020 at 11:09
anon/portly. If I compare America to New Zealand, people say “but New Zealand is small and isolated, so that doesn’t count”. If I compare NZ to parts of America that are small and isolated, I’m accused of cherry picking. I can’t win.
My point is that we certainly could have controlled Covid in places like Alaska and Hawaii, but did not. Which is fine if that’s our choice. But you have lots of people ridiculing the notion that Covid was controllable. It was–we simply chose not to.
8. October 2020 at 13:19
My understaning is that August saw a big jump in cases in Hawaii following an outbreak on Oahu in a prison and a homeless shelter. The homeless might have been superspreaders event.
June 16-19 they allowed interisland travelers without a quarantine period. They reopened gyms, bars, and recreation areas at this time and saw a small spike in cases.
July 23 gyms are tied to a spike in new cases.
July 31 bars closed for three weeks. Restaurants must stop drinks at 10AM. Cases mounting plus children
Sept 15 cases have continued to climb, 11 deaths at nursing home.
Even under quarentine hundreds, sometimes thousands of people arrived daily
Outbreak in Alaska in July might be tied to person working as a ticket agent for Alaska Airlines
More recent outbreaks are amongst the homeless and a nursing home
https://ktla.com/news/nationworld/despite-coronavirus-concerns-hawaii-moves-forward-with-oct-15-date-to-reboot-tourism/
http://www.honolulumagazine.com/Honolulu-Magazine/August-2020/A-COVID-19-Timeline-How-Honolulu-Got-To-This-Point/
https://www.adn.com/alaska-news/anchorage/2020/10/02/anchorage-reports-
large-cluster-in-assisted-living-facility-and-shrunken-icu-capacity-as-virus-cases-rise/
8. October 2020 at 14:25
Scott, if you take a look at the IHME site with the charts of daily infections and mobility data for New Zealand, the daily infections there skyrocketed at the end of March and fell just as sharply at the beginning of April. New Zealanders seem to have taken the shelter in place order seriously with mobility dropping by 80% at the beginning of April and staying below 70% through most of the month.
8. October 2020 at 16:04
I live in the USA but have close family in New Zealand. It really is not comparable. Every single person entering New Zealand must spend at least 14 days in a government managed quarantine facility. This include citizens, though in general they are exempt from the NZ$3100 quarantine fee. Work visa holders (H1B equivalents) travelling outside of New Zealand have been locked out of the country, recently some who are normally resident in New Zealand are now being allowed to return. Quite a number of people who had no other residence outside of New Zealand have been stranded outside of the country for an extended period of time. New immigration and work visas are essentially suspended. There is no non-essential travel allowed into New Zealand.
Basically the New Zealand deal is this, everyone does everything they can to eliminate the virus and the government ensures that not a single case enters the country, this is why the recent outbreak was a fairly serious scandal. I have seen a lot of praise for New Zealand’s results without any considerations of the tradeoffs.
9. October 2020 at 09:11
Gordon, I agree.
Luke, You said:
“I live in the USA but have close family in New Zealand. It really is not comparable. Every single person entering New Zealand must spend at least 14 days in a government managed quarantine facility.”
The whole point of this post is that the US and NZ had very different outcomes, so I’m not sure what point you are making.
You say they are not comparable, and in the very next sentence you start comparing them! I’m not sure you actually disagree with me.
9. October 2020 at 11:53
“anon/portly. If I compare America to New Zealand, people say “but New Zealand is small and isolated, so that doesn’t count”. If I compare NZ to parts of America that are small and isolated, I’m accused of cherry picking. I can’t win.”
You can’t win because you *shouldn’t* win!
It’s true that comparing USA to NZ is not necessarily very meaningful, given they have different characteristics, two of which are “size” and “degree of isolation.” This means (I would say) that an “equally good response” will not necessarily yield an equal number in terms of some Covid statistic.
Then if one is going focus on “small and isolated” parts of America, the obvious comparison is to other places that are similarly small and isolated. (From there we may want to incorporate other characteristics into our analysis, of course). Just comparing the US to NZ is, yes, “cherry-picking.”
To me the Alaska vs. Canadian Province formulation, especially in my suggested version, avoids these problems. There are natural and obvious reasons to make this particular comparison. Now you can win!
9. October 2020 at 12:20
“My point is that we certainly could have controlled Covid in places like Alaska and Hawaii, but did not. Which is fine if that’s our choice. But you have lots of people ridiculing the notion that Covid was controllable. It was–we simply chose not to.”
And the level to which NZ was able to control Covid makes this point …. how?
It doesn’t make that point! I think it’s a good point, but the NZ vs. AK/HI Covid Stats comparison doesn’t actually advance this point *one tiny inch*.
Now maybe a more detailed analysis of what NZ did and when, with an eye to whether AK and HI could have pursued a similar strategy, could help advance the argument.
But notice you’re going to end up with an argument like “if Alaska had just down all travel into and out of the state on March 1st, I estimate that its current [Covid Stat} would be x% lower.” Maybe the NZ example is helpful to jog our thinking in a particular direction, but ultimately it’s irrelevant to the AK/HI situation.
To me making simple Covid Stat comparisons, as if it by itself is really meaningful (or at least without carefully teasing out what the meaning actually is), is ultimately unhelpful.
9. October 2020 at 12:22
I liked this Alan Cole twitter thread:
https://twitter.com/AlanMCole/status/1314313981008412679
9. October 2020 at 13:39
One possible factor wrt AK and HI that I’m surprised no one has mentioned is the military. Are service men and women moving in and out and through much like normal, or much less?
9. October 2020 at 16:32
I’m in the canadian maritimes and for sure the fact that it came later here and that we have little public transit (even though I’m in one of the larger cities) helps.
Rules have also been pretty aggressive. Masks are mandatory in all public spaces here (though this rule is new and put in place after a small outbreak in a care home a few days ago but a lot of places like grocery stores had mandatory masks before it was mandated at the government level). Of course, I can’t compare experiences across regions since we’re not allowed to go anywhere.
10. October 2020 at 05:47
Canada vs US. If you remove the first wave that hit Italy, Spain, and the New York area, the Canada and US numbers look closer. The initial response in the New York area led to horrific death totals. Canada shutdown the border and was able to delay cases in Canada. Thay delay allowed time to find some things that worked better than others and gave Canada some time to find some protection measure for health care workers. They avoided some of the nursing home nightmare. Rural communities in Canada and the States avoided much of the spread until recently. Now cases in Canda and rural states are climbing. Luck, timing, geographic concentration, all played a role. Perhaps Canadaians with long isolating winters just adjusted to shutdowns better than the US. A poor initial response in the New York region (as in Italy and Spain) seems to be the biggest difference in numbers. In any case Canada seems to be see a spike current;y with new high totals for them
10. October 2020 at 12:36
Anon/portly. Don’t agree on NZ. I’ve used the Alan Cole graph in many posts, and made some similar points.
Benoit, It also came late to Alaska and Hawaii.
LB. I strongly disagree. There’s a clear “break” at the border, and the gap is most noticeable in recent months. So it’s not caused by the initial effects being different. Look at the Alan Cole graph that Anon/Portly links to.
And there’s no reason to remove NY unless you also plan to remove Quebec. Both are big outliers. The US is doing far worse than Canada, it’s not even close.
10. October 2020 at 15:22
” I’ve used the Alan Cole graph in many posts, and made some similar points.”
Huh? Sure, you’ve used the same graph. But did you read the tweets? “Cole points” and “Sumner points” seem like chalk and cheese to me.
E.g. this one:
“Similar to the “authoritarian” thesis but phrased more nicely is “state capacity,” but I find this is being applied entirely ex-post in unconvincing ways.
We’re talking about a spectrum where Belgium is the worst and Thailand and Vietnam are the best.”
Or this one:
“Here, you take that same group of eight countries, and a geographic sort looks much better than an income sort.
I have no idea why though.”
11. October 2020 at 09:06
anon/portly. Recall my post very early in the epidemic, claiming this was a white country problem. Even today, death rates in Asian and black countries remain far lower. (Oddly, black death rates in the US are quite high.)
11. October 2020 at 16:08
The New York response was the worst in the country. Slow to respond, late and o close schools, nursing homes, etc. The response inside of hospitals spread the disease. Just as the initial response in Italy and Spain say awful out comes. Canada shut their borders, stopped international flights, closed schools, rather quickly. That extra time let them learn from the mistakes of others. Canada has a population that quickly goes into versions of shutdowns seasonally. They adjusted quickly. The United States did bend the curve. Look at The NY Times cases in United States. We had bent the curve until the Floyd demonstrations. You see a rapid spike following the demonstrations. One study said it didn’t have an impact. That is highly questionable. Most academics have avoided look at the issue. If you want to avoid political attacks avoid the topic. Canada had less politics involved in their responses. Police in Canada enforced shutdowns with fines. The hat wasn’t an option in the United States. Aside from the police issue don’t forget we had a polarizing impeachment going on at the start. Lastly look at numbers for Canada the last two weeks. They are seeing new highs.
12. October 2020 at 08:50
ssumner
“You say they are not comparable, and in the very next sentence you start comparing them! I’m not sure you actually disagree with me.”
I did not state that well, I was not not so much arguing with your question as responding to people who wish to isolate one or two aspects of NZ’s response without considering the broader implications.
Point is, NZ took a very high cost elimination approach to the virus with the understanding that there will be no new cases introduced into the country, even if it meant an very high degree of international isolation, as I described earlier. If we do not want that degree of international isolation then I believe there is little point in using them as an example to follow.
14. October 2020 at 09:28
Many factors could and do affect the spread of Covid-19, but in the end none of them really matter. Ultimately the society decides what level of disease it’s comfortable with, and behaviors adjust to hit that level, bounded by a few parameters.
The dynamics of closed-loop systems explains everything; or, more accurately, the math and language can be used to describe everything.