Year: 2024

Commentary, COVID-19

When to go into Outbreak mode

Recently I described how I monitor the disease indicators published by government agencies, and how I’ve set a threshold of 6 people hospitalized per hundred thousand residents for deciding when to relax out of “outbreak mode” and eat and sing indoors.  So then, when to go back into outbreak mode and start taking precautions again?

I need to stress again here that I am not a doctor or an epidemiologist.  I’m not qualified to make these decisions, and I wish we had guidance from experts.  Unfortunately, the experts are assuming that everyone cares only about themselves and provide no guidance for people who care whether they pass these diseases on or develop long term conditions.  I do care, so I’m figuring this out as best I can.

Hospitalizations and deaths are trailing indicators: after infection it can take weeks for people to become sick enough to go to the hospital, and longer to die.  That means we need to look at leading indicators like case counts and wastewater concentration.  I’ll talk about wastewater in a future post.

Case counts are less reliable than hospitalizations and deaths, because they depend on the number of tests administered.  As we know, the number of tests dropped precipitously when President Biden declared an end to the State of Emergency and stopped reimbursing for tests. Providers tend to administer tests when they think people might be sick, so if they don’t anticipate infections they may not find them.

Case counts could be inaccurately low, so we should always keep an eye on hospitalizations.  There’s a chance that case counts could be inaccurately high, but if that happens, the worst that could result is that we think an outbreak is more severe and take too many precautions.  Of course, it’s better to err on the side of caution.

So if we’re looking at case counts, what is a good threshold for going into outbreak mode?  This turned out to be a lot more complicated than I thought, but let’s start with the basics and assume that all the data is three weeks old.

I looked at the seven outbreaks we’ve had since the the first wave (when we didn’t have good tests) and found the point when the 7-day average of COVID hospitalizations in New York City went above 6 per lakh per day.  Then I looked at the case rates for the day three weeks earlier:

Outbreak First day with hospitalizations > 6.2 Cases 3 weeks earlier
Fall 2020 2020-11-10 55.4
Fall 2021 2021-08-03 30.8
Winter 2021 2021-12-01 70.8
Spring 2022 2022-05-01 172.8
Fall 2023 2023-09-05 85.5
Winter 2023 2023-12-14 55.4
Summer 2024 2024-07-12 57.5

As you can see, there’s a fair range of variation.  But let’s pick a rate that’s in the more common range, say 60 cases per lakh population per day.  How much warning would that give us?

Outbreak First day with hospitalizations > 6.2 First day with cases > 60 Days notice
Fall 2021 2021-08-03 2021-07-18 16
Winter 2021 2021-12-01 2021-11-03 28
Spring 2022 2022-05-01 2022-03-13 49
Fall 2023 2023-09-05 2023-08-02 34
Winter 2023 2023-12-14 2023-11-20 24
Summer 2024 2024-07-12 2024-06-20 22

It looks like anywhere from 2-7 weeks, usually about three weeks, which is basically what we want.  If we see a case rate above 60, that means that the hospitalization rate is likely to go above 6, and may already be above 6.  That’s a sign that it’s time to go into Outbreak Mode.  For me, that means moving karaoke online, avoiding indoor dining and social events, and wearing a mask in all indoor public spaces.

Keep in mind that COVID is only one of many infectious respiratory diseases that can kill or disable.  I chose the rate of 6 hospitalizations per lakh per day because that was what we were prepared to tolerate during the 2018-2019 influenza season.  The hospitalization and death rates we care about are for all these diseases, but flu and RSV are not as well documented as COVID.  Hospitalization rates for flu and RSV are reported nationwide, but the local reports for New York only give absolute numbers for cases and hospitalizations for those diseases.

What I tend do do is to estimate a ratio based on the total case counts for all three diseases, or on the nationwide hospitalization rates.  If RESP-NET is reporting that there are as many flu and RSV hospitalizations as COVID hospitalizations nationwide, I’ll assume that that applies to New York.  If New York’s flu and RSV report shows that there are half as many positive tests for flu and RSV as for COVID, I’ll go into outbreak mode at 40 COVID cases per lakh instead of 60.

Stay tuned for that post about wastewater concentrations!

Commentary, COVID-19, Queens

How do we know it’s safe?

Hospitalizations per 100,000 people (for week ending on listed date) All ages 03/09 2.06 To prevent unreliable rates, rates are suppressed if the underlying count is between 1 and 4.

I shouldn’t be writing this.  I have no training in medicine or epidemiology.  I’m just some random person.  And if you have something from a better trained source that tells you how to manage your exposure to airborne infectious diseases like COVID, the flu or RSV in order to avoid passing it on to others and perpetuating the pandemic, you should probably go with that.

Unfortunately, our expert doctors and epidemiologists at organizations like the United States Centers for Disease Control and Prevention, and the World Health Organization, haven’t provided any guide for people who want to avoid passing COVID or other airborne diseases on to others.  Their guides focus on telling people how to minimize the risks to themselves.  They assume that everyone is a selfish asshole.

I’ll talk about what I’ve tried, but again, I’m just one person, with three close people in my family.  I have no way of doing an exhaustive study of the transmission of COVID or the flu.  My priorities are different from those of many other people.  So are the strengths and weaknesses of my body, my family’s bodies, and our risk tolerance.  So what works for me – or doesn’t – may well work differently for you.

We’ve also only had four years of COVID.  Our understanding of it is constantly evolving, and the disease itself is constantly evolving, so that what works one year may not work in the next.

With that in mind: a year ago I articulated a provisional strategy for balancing my wants and needs, and those of my family, with our desire to avoid catching COVID (and other infectious diseases), spreading it to others, and perpetuating the disease.

I plan on doing the following for the rest of my life:

  • Wearing an N95-type mask in medical settings, including pharmacies
  • Monitoring outbreak warnings
  • Monitoring hospitalization rates for COVID, the flu and RSV
  • Getting tested regularly during outbreaks

During an outbreak, I plan on:

  • Wearing an N95-type mask in indoor public spaces
  • Eating outdoors
  • Organizing events online/outdoors
  • Avoiding risky activities like singing

When I’m sick, I plan on:

  • Staying home as much as possible
  • Isolating from my family

This much is fairly straightforward, but the key questions are when to switch between regular mode, outbreak mode and sick mode, and back.  First of all: which indicators should we watch, and what will tell us that it’s a good time to change our behavior?

Citywide: deaths Data from the most recent days are incomplete 7-day average 03/05 3

Hospitalizations and deaths per population are the indicators that seem to fit most closely with what we care about with COVID.  The mild cases I’ve experienced are no fun, but they’re not much worse than what I’ve had for colds, flu, strep or other respiratory infections.  Death is the worst outcome, but we want to prevent people from getting infections that are so bad they are admitted to the hospital.  We really need a good measure of Long COVID, but as of writing we don’t have one.

Hospitalizations and deaths are trailing indicators – they tell us what happens after infections – so they are good for conservative estimates about when to relax our precautions.  They can be compared across time and geographic area by counting the number of deaths or hospitalizations for a fixed number of residents of that geographic area.  Dividing by 100,000 gives us nice easy numbers that are usually between 1 and 100.  100,000 is a standard quantity in Indian arithmetic: one lakh.

Last year I found out from the Centers for Disease Control and Prevention that the nationwide peak of hospitalizations during flu seasons before COVID was around 6 per lakh, flu and RSV combined.  That means that before COVID we were tolerating six people in the hospital with flu and RSV every day, and not taking extraordinary measures like lockdowns or working from home.  I decided that that was a good benchmark: when the combined hospitalizations for flu, RSV and COVID are below 6 per lakh, it’s no worse than what we used to tolerate in 2019.

We can tell when we’re below 6 hospitalizations per lakh nationwide, but how easy is it to check that locally?  I’m fortunate that the New York City Department of Health and Mental Hygiene publishes regular reports of citywide COVID cases, hospitalizations and deaths.  The data is compiled by day; in 2021 and 2022 these reports were updated daily, but as of writing in May 2024 they are updated weekly.

Influenza positive laboratory test results reported to NYC DOHMH by season
In the 2019-2020 season, positive test results start to rise in late november, peak around 8500 at the end of January, and fall to zero by the end of March.
In 2020-2021, they remain consistently low.
In 2021-2022, they start to rise in December, peak around 3,000 in late December and fall back to negligible amounts by the end of January. Another wave starts in March and remains around 3,000 through the end of the chart in mid-May.
In 2022-2023, positive results start in November, peak around 18,000 in mid-December, drop to around 1,000 by the end of January and remain around that number through May.
In 2023-2024, positive results start to rise in late November, peak around 14,000 at the New Year, and slope gradually down to their most recent count around 1,000 on May 11.

Data for influenza and respiratory syncytial virus is much less comprehensive.  Both the CDC and New York City release weekly reports during the infectious seasons for these diseases, but while the CDC measures hospitalizations, New York City does not. That means there is a nationwide way of measuring the cumulative risk of catching or spreading any of the three diseases, but locally we can only measure COVID and guess at flu and RSV.

RSV positive laboratory test results reported to NYC DOHMH by season
In the 2019-2020 season, positive test results start to rise in early november, peak around 1500 at the end of January, and fall to zero by the end of March.
In 2020-2021, they start to rise in early March, peak around 1,000 in late April, and decline through May.
In 2021-2022, they are already around 1,000 when the chart begins in October.  They peak around 1,300 in late December and fall to around 200 by the end of January, where they remain through the end of the chart in mid-May.
In 2022-2023, positive results start at about 950 at the beginning of the chart in October, peak around 5,000 in mid-November, drop to around 1,300 by the end of December and slowly decline to the teens through May.
In 2023-2024, positive results begin the chart around 800, peak around 4,900 in early December, drop to around 1,300 by mid-January and slope gradually down to their most recent count around 200 on May 11.

Eyeballing the data from the most recent winter wave, it looks like COVID cases constituted roughly half of hospitalizations nationwide.  It’s quite possible for flu and RSV hospitalizations to outpace COVID or vice versa, but as a first approximation we can say that if COVID cases drop below three hospitalizations per lakh residents per day, we are no longer in an outbreak and can relax some precautions, like eating and singing indoors.

That is the principle I used to determine when to start organizing in-person meetups for the New York Tech Karaoke Meetup, where I am an organizer, and in general to switch from outbreak mode to normal mode. Switching from normal mode to outbreak mode is a different challenge that deserves a separate blog post. Spoiler alert: I failed to take adequate precautions in December 2023 and was sick with COVID, so I’ll talk about some lessons learned from that experience.