Saturday, May 2, 2020

What's the Goal?

As we try to figure out what to do next on COVID, we need a better public discussion of the goal we are trying to achieve. Is the goal to:
(a) flatten the curve to prevent overwhelming the healthcare system
(b) minimize deaths from COVID (before a vaccine or silver bullet treatment?)
(c) minimize total excess mortality, including non-COVID causes... recognizing that massive amounts of healthcare are being deferred, with implications of mortality
(d) balance excess death with benefits of opening. Not just economic benefits, but personal human flourishing that comes from interacting with others: graduation ceremonies, youth athletics, small dinner parties, limited travel, and the million other benefits that "social animals" get from interacting with others.
(e) something else? 

The original intent of lockdowns was (a). If (a) is the goal, Sweden looks good so far, and it seems there has been enough time to judge. Taiwan and South Korea look even better. Opening beaches seems obvious. Radically different policies by geography make sense - rural vs. urban, different baseline rates, sub-geographic actions by age group. "One size fits all" shutdowns don't.

In both online and personal conversations, I hear people slip between these goals. You can't say (a) is the goal and then react negatively when cases rise but stay below hospital capacity. You can't say (b) is the goal without recognizing the cost is increased poverty and negative non-COVID healthcare impacts. Goal (c) seems strictly better than (b), except that it is harder to measure. Goal (d) is not well-enough specified... what is that balance?

Personally, I think (a) is the right goal. This disease is going to take its course. We can focus on vulnerable populations, using testing capacity wisely with that group in mind. Perhaps (d) is also arguable.

But my point here is different: we need to push people - politicians, epidemiologists, commentators, media, and each other - into clarity on the goal.

Tuesday, April 21, 2020

Three High Leverage Coronavirus Actions We Should Take Now

Many people are working on the hardest problems of COVID-19, from therapeutics and vaccines to economic stimulus packages. But there are activities that would require a lot less time and effort and could drive huge benefits. I’d like to explore these high leverage actions with the hope that someone will pick up the ball and run. Here are three; I’d like to add more over time. I’d like to push the envelope, meaning some might be bad ideas.

1) Test Different Recovery Levers (Rather Than Fully Baked Plans)

Competing plans are emerging (Opening Up America Again, state efforts, AEI, Harvard Center for Ethics). While valuable frameworks to consider, they all rely on beliefs about the intertwined health, human wellbeing, and economic impact of different combinations of actions. But this is a new world. Epidemiological models might provide guidance but the parameters these models rely on depend on the actions we take, and we have zero historical data on that relationship. We haven’t ever done a shutdown like this before.

We need to know which actions drive benefit and at what cost. Rather than gates and phases where we do multiple things at once, can we test and measure the actual impact of the most potentially impactful levers? By “test” here, I don’t mean test people for COVID. I mean try types of loosening in some places and not in other places, while mitigating risk.

My guess for the best levers to test includes: (a) reopening schools, (b) reopening all retail, (c) reopening dine-in restaurants, (d) allowing small gatherings (e.g., less than 25 people: kid’s birthday parties, dinner parties, kid athletics without spectators, etc.), (e) allowing physically-semi-distant, non-elderly services (construction, house cleaning, childcare, etc.).

You could test these by county. States should be enabling and supporting these tests; smart, data-driven governors should lead the way. Higher potential risk ideas can be tried in more smaller or lower risk geographies first.

Perhaps the best test right now is opening schools. With summer break coming, there’s a time window with a natural risk-limiting device at the end. You could test in geographies where fewer people live with grandparents. Open only elementary schools. Of course, allow students who live with high risk individuals to stay home (pay teachers extra, if allowed, to deal with this complexity). Ask residents with children not to leave the county until 2 weeks into summer break. If it works, we would feel better about summer camps, kids sports, and going back to school in August. If it doesn’t, the risk is limited, as COVID-19 doesn’t hurt kids and hurts very few parents (who are generally non-elderly). Our kids’ lives are far from normal right now. Getting closer to normal would be a massive benefit.

We can do this soon. It has been important to realize at every stage of this particular disease that the next 2 weeks are already mostly determined by past actions. It’s the time period after that which you are impacting. A month or two ago, that fact combined with exponential growth meant that acting “too early” had disproportionate impact (see Northern California). Now it means we need to build that delay into the reopening plans, given the massive human wellbeing and financial costs of an extra week of shutdowns. We should start testing these levers at a time that feels “too early”.

2) Serosurvey of New York City Healthcare Workers

Several serosurveys have been released in recent days, and all have been aggressively challenged (correctly). We can do better if we focus on a high exposure population, which will get around many of the statistical issues from a less-than-perfect specificity test of a rare condition.

Conduct a serosurvey of New York City healthcare workers now. Most have likely been exposed and they encompass a wide range of ages and health conditions, at least among working age people. They’d be willing to do it, easily reachable, and better at recalling and reporting symptoms. You could get a good measure of attack rate in the population, especially if you could assess likelihood of exposure from when/where they worked. Outcome data could include IFR, but also hospitalization rates, asymptomatic infection rates, etc.

It is possible this may overestimate IFR given higher initial doses of virus from workers exposed during intubation or other high risk activities. However, one could segment the analysis on this factor, and even without this segmentation the study would provide a bookend versus other (flawed) serosurveys to date.

Maybe this is survey is underway already… does anyone know?

3) Suspend Regulations to Promote Flow Between Commercial and Consumer Supply Chains

By now, many have read Will Oremus‘s excellent Medium piece on toilet paper. This same effect happens all over the economy: supply chains tuned for one channel face radical shifts in demand by channel. That’s why dairy farmers are pouring out milk while I’m limited to 2 milk products at my local Whole Foods.

Private companies will sort this out, but we are making it hard for them. Now is not the time for labeling laws and the like to interfere with flow of goods. Let any retailer sell commercially packaged and labeled goods. Create simple standardized consumer waivers of liability, if needed. Gavin Newsom took some excellent steps on analogous issues in healthcare. The same should be done in consumer goods and retail.