Living Dangerously

May 3, 2020

Previous | Next

On a week-to-week basis, the United States as a whole continues to see some decline in new deaths from the coronavirus. The worst increase was during the week beginning April 19, with 18,482 new deaths reported, increasing total mortality by 90%. The following week’s increase was 14,224, increasing total mortality 36%. This past week, we saw 12,527 new deaths, increasing total mortality 23% (from 53,314 to 65,841).

Because mortality is not dropping as fast as expected, the Institute for Health Metrics and Evaluation has increased its total projected mortality from this wave of the virus from 67,641 to 72,433. That means that only 6,592 more deaths would occur, a very low number considering that 12,527 occurred just this week. The institute acknowledges, however, an important limitation of its projection:

Our present modeling framework does not yet capture how the risk for more COVID-19 cases – and potentially deaths – could increase due to increased interaction among individuals. This is particularly true if locations have not fully instituted strong containment strategies like widely available testing and contact tracing.

In other words, the modeling does not yet take into account the rush to re-open the economy. Although the country as a whole has experienced slowing mortality, the trend has been very uneven from state to state. Most of the states that are moving to lift restrictions are doing so sooner than recommended, according to both IHME and CDC guidelines. They are not waiting until new cases in their state have declined for two weeks, or until they have “strong containment strategies like widely available testing and contact tracing.” That suggests three scenarios:

  1. The warnings from epidemiologists are exaggerated, and the lifting of restrictions will have little effect on the mortality trend
  2. The lifting of restrictions will raise the infection rate somewhat, but not enough to overwhelm medical systems or increase mortality dramatically
  3. The lifting of restrictions will touch off serious outbreaks with exponential increases in mortality in at least some states

I don’t pretend to know which of these will occur. We do know that just a few weeks of careful preparation and prudent restraint can make a big difference when dealing with something that moves as quickly as an infectious disease. We should find out very soon which states are doing the right thing.


Ready or Not, Here We Come

April 26, 2020

Previous | Next

As of this morning, the official death toll from covid-19 in the United States stood at 53,314, an increase of 14,224 in the past week. However, the rate of increase slowed to 36%, compared to 90% the previous week and 142% the week before that.

Passing the peak

This week the Institute for Health Metrics and Evaluation reported its latest estimates of peak mortality for countries and states. For the US as a whole, deaths are thought to have peaked on April 15 at over 2,600 a day. That should now fall pretty quickly, but additional deaths are projected to bring the total deaths to 67,641 by the time this wave of the virus peters out in August. Such projections have a large margin of error, and so total deaths could reach as high as 123,157.

A number of individual states have not yet reached their peak daily mortality, but are expected to by the end of May. These include Arkansas, Georgia, Iowa, Kansas, Kentucky, Montana, Nebraska, North Dakota, Rhode Island, South Dakota, Vermont and Wyoming. Most of these states are farther inland than the coastal cities which experienced the epidemic first. All of them except Rhode Island are red states, which may also have been slower to issue stay-at-home orders. Where the mortality has been greatest, the restrictions on public activity are credited with bending the exponential growth curve downward.

When to reopen?

The IHME also suggested dates on which the states could safely lift their restrictions and let people go back to work. The assumption is that if states would keep their restrictions in place until these dates, the rate of infection would then be very low. States could theoretically manage the remaining infections with testing and quarantines, avoiding a serious resurgence of cases and deaths.

Here are the suggested dates, by week:

May 4-10: Montana, West Virginia, Arkansas, Hawaii, Vermont

May 11-17: Idaho, North Carolina, Ohio, New Hampshire, Maine

May 18-24: California, Nevada, New Mexico, Louisiana, Alabama, Tennessee, Illinois, Indiana, Michigan, Wisconsin, Delaware

May 25-31: Washington, Oregon, Wyoming, Colorado, Minnesota, Mississippi, Pennsylvania, New Jersey, New York

June 1-7: Maryland, Virginia

June 8 or later: Arizona, Utah, Texas, Oklahoma, Nebraska, Kansas, North Dakota, South Dakota, Arkansas, Missouri, Iowa, Kentucky, Florida, Georgia, South Carolina, Massachusetts, Rhode Island, Connecticut

The dates reflect both how many cases states have had, but also when they had them. Although the states with the earliest reopening dates have had low mortality, the states with the latest reopening dates are not, for the most part, the states with the highest mortality. Of the ten states with the highest covid-19 death rates per 100,000 population, only three—Massachusetts, Rhode Island and Connecticut—have reopening dates this late. But of the states whose death rates have not yet peaked, all except two—Montana and Wyoming—have the latest reopening dates. That means that some of the states with the highest mortality, especially New York and New Jersey, can reopen before states with relatively low mortality, such as Utah and South Dakota. That will make the later reopening dates a tough sell in many states.

The suggested reopening dates also come with important conditions. They work only if restrictions remain in place up to the reopening date, and they assume that the state is then ready to implement containment strategies including “widely available testing, contact tracing and case-based isolation, restrictions on mass gatherings.” Otherwise surges in cases can occur, even in places that have not experienced them already.

States that are not prepared to do these things will be strongly tempted to try and reopen anyway. Why accept the real hardships of economic recession, with the national unemployment rate already around 16% and climbing, in order to avoid a spike in mortality that is only a mathematical projection? A case in point is Georgia, which has seen only 9 deaths per 100,000 population (a little less than one death per 10,000). And yet epidemiologists do not see it as a good candidate for reopening, since its death rate has not yet peaked and its testing rate is among the lowest in the nation.

The fact that most of the late-peaking states recommended for late openings are red states adds a political component to the mix. If the climate-change debate is any indication, most Republicans would rather risk future catastrophic change than disrupt the existing economy. It should be no surprise that “the easing will not be universal and is cleaved largely along party lines, with some Republican governors moving to reopen key sectors and Democrats moving more slowly” (The Washington Post).

So here we come, ready or not. Some states will throw caution to the winds and forge ahead recklessly. Others may be too cautious and hurt more people than they help. That is where the national mismanagement of the crisis from the beginning has left us. We can only hope that we manage the recovery from this wave of the virus better than we managed the initial spread. And commit to being better prepared for the next one.

Virus Poses Dilemma for States

April 19, 2020

Previous | Next

In the past week, total reported Covid-19 deaths in the United States increased from 20,608 to 39,090, an increase of 89%. However, the rate of increase was once again lower than the previous week, when it was 142%. The rate of increase slowed in every state except three, Montana, New Mexico and Vermont, all states with relatively few cases to begin with.

Also encouraging is the fact that death rates per 100,000 population remained low in most states. Death rates of 10 per 100,000 or more have only been seen in nine states—New York, New Jersey, Connecticut, Louisiana, Michigan, Massachusetts, District of Columbia, Rhode Island and Illinois. Their rates range from 90 per 100,000 in New York to 10 per 100,000 in Illinois. In the other states fewer than 10 per 100,000—or fewer than one person in 10,000—have died from this disease.

“Opening Up America Again”

Now that the mortality rate is slowing overall, as well as remaining low in most states, is it time to roll back stay-at-home orders and put the country back to work? This week the Trump administration released a set of guidelines for the states to follow in making that decision. It included some “gating criteria” for states to “satisfy before proceeding to phased opening.” One criterion was a “downward trajectory of documented cases within a 14-day period.” Another was that hospitals have a “robust testing program in place for at-risk healthcare workers, including emerging antibody testing.” The guidelines also listed “core state preparedness responsibilities,” one of which was the “ability to quickly set up safe and efficient screening and testing sites for symptomatic individuals and trace contacts of COVID+ results.”

Where do we stand on new case trajectories and testing capabilities? For the country as a whole, new cases per day peaked on April 10 and fell for the next four days. Then on April 15 they jumped back almost to the April 10 peak and leveled off. We have not yet seen a downward trajectory for 14 days. With regard to testing, the country is falling hundreds of thousands short of the 500,000-750,000 tests per day that epidemiologists recommend. States that are most anxious to ramp up testing, such as New York, maintain that they need far more money and materials from the federal government to be able to do so.

Looking state by state, I was unable to find any state with a sustained decline in new cases over a two-week period. Washington state, whose largest number of new cases came on April 6, comes closest. Some of the higher-mortality states—New York, New Jersey, Connecticut, Louisiana, and Michigan—are starting to decline, but without a very consistent trend. In Massachusetts daily cases are fairly level; in Rhode Island and the District of Columbia they are still rising; and in Illinois they fell for a few days and then rose again. And to my knowledge, no U.S. state has succeeded in implementing a comprehensive testing program yet, although some foreign countries have done so.

Although the situation is very fluid, states are well advised not to rush into lifting restrictions before the virus is under better control. Even states that have not yet had a serious outbreak must be cautious, since they may owe their success largely to the number of people who have been staying home. We have just seen in South Dakota—a state with few cases before last week—how quickly the disease can spread in a workplace, if we are unable to test workers, quarantine the sick, and track their contacts.

The politics of “liberation”

This was also the week that the virus issue became most politicized. Many states saw the beginnings of a protest movement led by Fox News commentators, other conservative media outlets, Trump supporters, and now the President himself. The ink had hardly dried on the “Opening Up America Again” guidelines when Trump expressed his support for the protesters clamoring for an end to the stay-at-home orders, tweeting out a call to “liberate” Michigan, Minnesota and Virginia. None of those states have met the administration’s own stated criteria for reopening. New cases per day in Michigan are only starting to subside, while they are level in Minnesota and rising in Virginia. Then why single out those states? Because they have large populations of Trump supporters, but Democratic governors. The main objective seems to be to portray them as the enemies of their own people, when they are doing basically the same things that other governors are doing.

President Trump’s role in this matter is deeply frustrating to the state and local politicians and hospital workers who are on the front lines of this battle. He wants to get credit for liberating the economy, but he refuses to take responsibility for the federal initiatives needed to halt the epidemic, especially in the areas of testing and hospital supplies. He wants to be the star of the show, scoring political points while someone else does the heavy lifting. This is, after all, the president whose promise of a cheaper but better replacement for Obamacare ended in his plaintive whine, “Who knew that healthcare could be so complicated?”

A president fit to lead in a national emergency would be great. But even more is at stake here. We need a national culture that respects science and expertise, balances individual freedom and the public interest, defines a strong role for the federal government, and stops blaming our problems on foreigners. The epidemic started in China, but our lack of preparedness despite repeated warnings is something we did to ourselves.

Inevitably, states will differ in their approach to lifting restrictions, and we will have a natural experiment in economic and health policy. I only hope that the experiment is not lethal in many parts of the country.

Flatter but Wiser?

April 12, 2020

Previous | Next

Last week I said that the United States was poised to become the world leader in coronavirus cases very soon. That milestone has now been reached, as our 20,608 deaths have surpassed every other country. That includes China, despite the fact that it is less economically developed, has four times our population, and was first to be hit by what became a pandemic.

However, this week we also had evidence that our steeply rising mortality curve is starting to flatten. Total deaths did increase 142% during the week—that is, they more than doubled—but they had nearly quadrupled the previous week. Deaths are still rising, but the rate of increase declined in almost every state. (The exceptions were Idaho, Missouri and Oregon.)

Nevertheless, the spread of the disease remains alarming. With over 20,000 deaths already, any more weekly doublings would result in astronomical death tolls. Epidemiologists have developed much more sophisticated models for the spread than I can describe here. One of them that has gotten a lot of attention projects that we can hold this year’s ultimate death toll to 60,000. Whether we can do that while quickly putting the country back to work is not as clear.

What have we learned?

What have we learned from experiencing this pandemic so far? Are we drawing the right lessons? In particular, what has President Trump learned? We know that he initially underestimated the problem, and that he continued to treat it dismissively for weeks after being briefed by both intelligence officials and medical experts. We also know that previous administrations—both Republican and Democrat—had developed plans and proposals for dealing with a pandemic, and that this administration chose to ignore them. How much has the President wised up since then?

The first lesson most of us have learned is that a pandemic requires a quick and decisive response, since every week counts. Locating cases and quarantining the infected are crucial.

The second lesson is that a country can slow the transmission of a contagious disease by telling most people to stay home, although that is a crude way of doing it. At the very least that strategy can spread the caseload over a longer time, easing the burden on medical facilities. Hopefully, few states will now experience what New York has just been through.

But then what? How does a country permanently limit the number of people who contract the disease and the number who die from it? The ideal solution is general vaccination, but that appears to be at least a year away. Another possibility is a breakthrough in treatment, to make the disease less life-threatening, but that also appears a number of months off.

Advocates of a quick return to business as usual seem to be relying heavily on “herd immunity,” the idea that once a lot of people have survived the disease and developed immunity, new cases will peter out. We can also use the antibodies in the blood of survivors to treat those who do get sick. But how large would such a “herd” of survivors be? Right now, we have about one death for every 25 confirmed cases. That means that we could get to 60,000 cases with only 1.5 million Americans having contracted the disease, leaving the vast majority of our 330-million population still at risk with no immunity. Of course, the reported numbers of deaths and cases could be wrong. Maybe a lot of deaths have yet to show up, since it is a “lagging indicator,” and some deaths occur at home without being correctly classified. The number of cases could be even more seriously underestimated, since people can have mild symptoms without reporting it as covid-19 at all. But even if the real ratio of deaths to cases is only one in 100, we could have 60,000 deaths from only 6 million cases, still leaving most of the population unprotected.

Most medical experts are dubious about the President’s eagerness to “reopen” the economy. We would be ending the strategy we’ve been relying on to stop the dying, and sending people back into society with no more protection than homemade face masks.

Test, test, test

What the experts do recommend is what should have been done earlier on. Scale up testing to the point where new cases can be quickly identified and selectively quarantined, while other people start to feel safer going about their business. That’s what countries with the most success in halting the epidemic have been doing. In contrast, President Trump’s consistent pattern of over-promising and under-delivering testing is a national embarrassment.

I doubt that the economy can restart with one big rush of people back to work. It’s going to be a while before people are comfortable in crowded workplaces, shopping malls, stadiums, or on buses, subways or airplanes. Given the great variety of workplaces, shopping areas, entertainments and methods of transportation, I suspect we will have a patchwork economy for a time, with some places a lot safer than others. That should make the overall recovery a bit sluggish, especially since the various parts of the economy are interdependent. One business cannot thrive if some other business it depends on cannot operate safely.

Another widespread prediction is that technological means of interacting and producing without face-to-face interaction—automated production, teleconferencing, online shopping, etc.—will get a permanent boost from this experience. But reorganizing along those lines will take time, and it will require some upgrading of the skills of many workers if they are to remain employed.

Instead of a quick return to “normal”, we should expect a painstaking transition to a new normal.

Beyond the Sheer Numbers

April 5, 2020

Previous | Next

This has been a week in which hospitals in New York and many other states began to be overwhelmed by the exponential growth of coronavirus cases. When every patient counts and every death is a tragedy, the sheer number of cases is daunting.

Having said that, epidemiologists have to look at rates and growth trends as well as sheer numbers to understand an epidemic, and so do the rest of us. In last week’s post, I used the concept of doubling time to compare mortality trends in different countries, based on data from Our World in Data. The good news this week is that the doubling times in days are increasing a little in most countries, meaning that the rate at which deaths are multiplying is slowing. The bad news is that exponential growth curves are still very steep in many places. Deaths are doubling every four days in the United States and United Kingdom, apparently the fastest growth in the world. The doubling times are five days for Germany, six days for France and Netherlands, seven days for Spain, and ten days for Italy.

To appreciate the implications, if Italy’s deaths would keep doubling every 10 days for the next 30 days, its 15,362 deaths would double three times, increasing by a factor of 8 to 122,896. But if US mortality would keep doubling every 4 days in the same period, its 8,501 deaths would double at least 7 times, increasing by a factor of 128 to 1,088,128. To put that in perspective, 405,000 Americans died fighting World War II. No one knows what the real numbers will be, since no one knows how much and how fast we can bend the growth curve. What is clear is that the United States is poised to become the world leader in coronavirus deaths very soon. How the country that prides itself on the world’s most advanced health care system could accomplish that feat is a topic for another time.

State mortality rates

Within the United States, death rates also provide additional perspective to raw numbers. As of this morning, the ten states with the highest number of deaths are New York, New Jersey, Michigan, Louisiana, Washington, California, Illinois, Massachusetts, Georgia, and Florida, based on data from the Washington Post. Taking into account state size by using deaths per 100,000 population changes the picture somewhat. California’s 289 deaths no longer look so large, and Vermont’s 20 deaths become more significant. California’s rate of less than 1 death per 100,000 drops it down to 30th in death rate, while Vermont’s 3 per 100,000 brings it up to 7th. Illinois, Georgia and Florida also drop out of the top ten, to be replaced by Connecticut, Colorado and the District of Columbia. The number of deaths in a small state may get less attention, but it can have a large proportional impact on the smaller number of medical personnel and hospital beds.

Not only do states differ greatly in total deaths and death rates per 100,000 to date, they are also adding deaths at very different rates. Most of the states with the most deaths—either raw numbers or deaths per 100,000—have also had relatively large percentage increases over the past week. Increases of 300% or more are common—New York’s is 331%—but Michigan and New Jersey have seen increases over 500%. One notable exception is Washington, which has the sixth highest death rate so far, but one of the slower rates of weekly growth, 67%. The virus hit Washington first, but stay-at-home measures seem to be working. California has both a low rate of death and a below-average rate of weekly increase, having been the first state to issue a stay-at-home order.

Meanwhile, other states have had relatively low numbers and rates of death so far, but now have above-average rates of growth. Tennessee’s mortality rate is less than 1 per 100,000, but its deaths increased from 7 to 50 in a week, an increase of 614%. Other states that experienced significant jumps from low beginnings were Alabama, Kentucky and Maryland.

Given the potential for exponential growth to change the situation with dizzying speed, current low numbers are no excuse for complacency, anywhere in the country.