jails

There’s just a real lack of testing being done.

Six times in three weeks, Marci Rosenberg and her ailing husband and teenage children tried to get tested for the new coronavirus — only to be turned away each time, either for not meeting narrow testing criteria or because there simply were not enough tests available.

All the while, the Bellaire family of four grew sicker as their fevers spiked and their coughs worsened. They said they fell one by one into an exhaustion unlike any they had felt before.

By March 18, Rosenberg was desperate and pleaded with her doctor for a test. Dr. Lisa Ehrlich, an internal medicine physician, told Rosenberg to pull into her office driveway. But Ehrlich warned Rosenberg, “I can only test one of you.” She swabbed her throat through an open car window. The result came back the next day: positive.

The rest of her family was presumed to be positive but untested – and thus excluded from any official tally of the disease.

As the number of confirmed cases of the potentially deadly virus continues to explode across the Houston region – tripling from 1,000 to more than 3,000 in just the past week – there is mounting evidence that the true scope of the disease here could be far worse than the numbers indicate.

A Houston Chronicle analysis of testing data collected through Wednesday shows that Texas has the second-worst rate of testing per capita in the nation, with only 332 tests conducted for every 100,000 people. Only Kansas ranks lower, at 327 per 100,000 people.

In cities across Texas — from Houston to Dallas, San Antonio to Nacogdoches — testing continues to be fraught with missteps, delays and shortages, resulting in what many predict will ultimately be a significant undercount. Not fully knowing who has or had the disease both skews public health data and also hampers treatment and prevention strategies, potentially leading to a higher death count, health care experts say.

[…]

As the pandemic’s march quickened, Texas was slow to ramp up testing.

The first confirmed case in Texas, outside those under federal quarantine from a cruise ship, was March 4, striking a Houston area man in his 70s who lived in Fort Bend county and had recently traveled abroad. By month’s end, the Houston area had more than 1,000 confirmed cases. A week later, the number had pushed past 3,000.

Yet it was not until March 30 that the rate of testing per 100,000 people in Texas topped 100. As of Wednesday, the state was testing 327 per 100,000, according to a Chronicle analysis of data from The COVID Tracking Project, which collects information nationwide on testing primarily from state health departments, and supplements with reliable news reports and live press conferences.

Twenty-six states in the U.S. are testing at least double the number of patients per capita as Texas, in some cases six times more. New York, for instance, is testing 1,877 per 100,000 people while neighboring Louisiana is testing 1,622 per 100,000. Even smaller states, such as New Mexico, are testing triple the rate of Texas.

Texas officials defended the state’s response.

“We’ve consistently seen about 10 percent of tests coming back positive, which indicates there is enough testing for public health surveillance,” said Chris Van Deusen, a spokesman for the Department of State Health Services, in an email, “If we saw 40 or 50 percent or more of test coming back positive, we’d be concerned that there could be a large number of cases out there going unreported, but that has not been the case.”

It is unclear if that is a reliable measure. Nearly 41 percent of New York tests were positive, the second-highest rate in the country. In Texas, about 9.4 percent of tests were positive — roughly the same as Washington state, where one of the largest outbreaks of coronavirus has occurred.

Not the first time we’ve talked about this, and it won’t be the last. This also means that the official number of deaths attributed to coronavirus is likely too low. This has been the case globally, especially in the hardest-hit places, where the difference between the normal daily mortality rate and the observed mortality rate during the crisis is a lot bigger than the official count of COVID-19 deaths. The good news is that as yet our hospitals have not been overwhelmed, but we can’t say with confidence that that will continue to be the case.

The number of people hospitalized with COVID-19 in the Houston area is continuing a steady climb, not close to crisis levels but unnerving enough that experts still aren’t sure when the area’s grand experiment in social distancing will start showing up in daily counts.

After a week in which COVID-19 hospitalization numbers more than doubled in Harris County, epidemiologists and infectious disease specialists said it likely will be another week to 10 days before they know if the stay-at-home orders and closures are reducing the rate at which the coronavirus is spreading and keeping health care facilities from being overwhelmed.

“Even though we’ve been social distancing for three weeks, it’s too early to know when we’ll be on the downward slope,” said Catherine Troisi, a professor of epidemiology at UTHealth School of Public Health. “The numbers we’re seeing now reflect people who were exposed to the virus up to four weeks ago.”

Peter Hotez, dean of the National School of Tropical Medicine at Baylor College of Medicine and Texas Children’s Hospital, said the social distancing has paid off in terms of keeping hospital volumes under control so far but added that the pay-off in terms of ending the pandemic is unclear. He said that “we need to continue stay-at-home orders until the end of the month, then reassess whether to extend them longer.”

Hotez and others said that aggressive social distancing is more important now than ever, given modelers are projecting that the number of COVID-19 cases in the Houston area should peak in the next few weeks. They said people venturing out during the peak period will put themselves at high risk of contracting the virus.

[…]

The study, released on March 24, originally said the virus’ spread in the Houston area would peak April 7 and burn out by mid-May if stay-at-home orders are continued until May 12. It was not clear Tuesday when the study projects the virus will burn out now.

Eric Boerwinkle, the lead researcher, could not be reached for comment Tuesday and UTHealth officials had no update on the study. Boerwinkle, who did not make the original modeling publicly available, has briefed top local government officials on the work.

Another modeling study, conducted by the University of Washington’s Institute for Health Metrics and Evaluation, now projects that the Texas peak use of hospital resources for COVID-19 will be April 19, some two weeks earlier than it previously projected. The study, reportedly relied on by the Trump administration, foresees no bed shortage in the state, including in intensive care.

“That’s why you shouldn’t place too much weight on any one model,” said Dr. James McDeavitt, Baylor’s dean of clinical affairs. “They depend on assumptions plugged in and can show everything from Houston being able to handle the surge to a New York City-like situation.”

McDeavitt noted the wild cards that go into modeling — the number of people admitted to a hospital, the percentage that need intensive care, how long it takes to get patients off ventilators, how long they need to recover in a regular bed once they move out of intensive care. Those are the assumptions that drive models, he noted.

McDeavitt said he doesn’t think the number of cases will come down in the Houston area until the end of the month.

That story was from earlier in the week, so all of the numbers are a bit out of date by now. But the bottom line remains that we don’t know where we are on the curve because we don’t really know how many people are or have been sick. Models all rely on data, and we’re also not good with the data.

The information Texans are working with is too damn thin.

Where to start? Not enough tests have been completed, or taken, to really know who has or doesn’t have the disease, where the Texas hotspots are, or whether people who have died of respiratory problems had COVID-19. The relatively small number of test results also means we don’t know which people had the disease and recovered (and how many people have recovered) and whether the projections being made with that skimpy data are accurate enough to guide our public health decisions.

It’s not enough to say that the testing is getting better, that we know more than we knew just a few days ago. What we still don’t know overshadows what we do know.

We’re like pilots flying in clouds without instruments. We know a little bit, but not enough to make really solid decisions or to figure out what’s next. We’re learning as we go. As of Thursday, Texas was reporting 10,230 cases and 199 deaths, 1,439 hospitalized COVID-19 patients and 106,134 tests conducted.

Given the level of testing right now, it’s hard to know how many cases Texas really has. Because the best way to get tested for the new coronavirus is to show symptoms that a medical professional finds troublesome, it’s probably safe to say we’re not testing many people who are carrying the virus but don’t have symptoms.

It’s easier — because it’s more obvious — to map the institutional cases. When someone in a nursing home or a state supported living center or a prison tests positive, testing everyone in that location is simple and smart. It’s simple to figure out that everyone in a given building or campus might have been exposed.

Even that data isn’t always available. The state of Texas initially wasn’t sharing details about the data it has collected from nursing homes where COVID-19 cases have been found. But a few days after The Texas Tribune’s Edgar Walters and Carla Astudillo wrote about it, the state revealed 13% of nursing homes have at least one confirmed case.

We’re doing a lot of flying blind. If we want to make good decisions about things like when and how to restart the economy, we need a much better understanding of where we are, and where that means we’re likely to be going.