As states have reopened, the media has watched carefully for evidence of a "surge". Some naive analyses look only at cases/day, which count positive viral test results, that do not account for testing rates. If an area increases testing two-fold, even if the daily cases are stable, one might expect a doubling in daily case counts. This is the argument used by the president and governors in Texas, Florida and other places in trying to downplay upticks in cases. They are correct that we must look not only at cases but also the testing rate to accurately assess whether a case increase is actually evidence of a viral surge, but are extremely misguided in downplaying the current (undeniable) surge happening in Arizona, Texas, Florida, and South Carolina. The growth in cases is far far beyond the increase in testing, and this can be seen by the testing positivity rate.
The testing positivity rate is simply the proportion of SARS-CoV-2 viral tests that come up positive in a given place for a given time period. covidtracking.com has the most rigorously aggregated data on testing of states in the USA, and gives positive and negative test results daily for each that allow estimation and tracking of the testing positivity rate. We include this in our web app developed by my PhD student Emma Zohner that allows the user to simultaneously look at testing, cases, deaths for different states and counties, but when it comes to following the test positivity rate plotted along with cases, I like the interface on the Johns Hopkins site even better.
So why is this value important, and what does it tell us? There are two key valuable insights we can gain from looking at the testing positivity rate for a country, state, or county:
Assess testing adequacy: When tests are scarce they are typically reserved for those with more severe symptoms, and leading to high testing positivity rates. Conversely, in regions where community testing is broadly practiced or ample supply is available for all who want to be tested, the testing positivity rate is lower. Consistently high testing positivity over time potentially indicates that locale is under testing. Note that it is likely that actual new daily infections are about 10x the official case counts, so cases are just the tip of the infection iceberg, and undertesting suggests an even smaller part of the iceberg is showing.
Assessing potential surges: If test positivity increases over time, this potentially indicates an uptick of infections in that location, and if it sharply increases, it may indicate a major surge. If an increase of cases is simply due to increased testing, we would expect the testing positivity rate to remain stable or even decrease -- a sharp increase in testing positivity is one of the best indicators of an emerging surge.
Let's look at the 4 states that are undeniably surging right now:
We can see that the testing positivity rate has increased from below 10% to over 20% in the past 3 weeks, coinciding with the current surge. These are staggeringly high positivity rates. Also, the fact that Arizona's 7 day moving average testing positivity rate was never much less than 10%, even when case counts were low, indicating evidence that the state was undertesting throughout the pandemic and we may be seeing even less than 10% of infections in the counted cases.
With Texas, we also see a clear increase from test positivity from 4.7% at the beginning of June to 13.7% now, with Houston experiencing an even more alarming increase from 4% to weeks ago to 20% now.
In Florida, the test positivity rate in the first four weeks after reopening had actually decreased from above 5% down below 2.5%, suggesting the increase in cases during that time period could be explained by increased testing. If the governor made that point at the end of May, I would have agreed with him. However, the surge in testing positivity rate from 2.5% all they way near 15% across the state has coincided with the alarming accelerating increase in cases that is undeniably evidence of a surge. And these rates may be even higher in some of the most affected cities.
Similarly, South Carolina had very low testing positivity rate of about 3% at the end of May, showing no upticks since reopening, and then in early June this measure started creeping up and has increased over 15%, coinciding with an undeniable surge of cases. This is the quietest surging state because it has much smaller populations, but its per capita growth is on par or higher than the other three.
Looking carefully at these plots, you can see that the increase in testing positivity is a potentially useful harbinger of an uptick or surge of cases. Indeed, it may be one of the key influential measures in the PolicyLab projection model that successfully predicted the Texas and Florida surges. So are there other states primed to surge like these 4? Let's look at some of the states that show evidence of increasing test positivity rate:
The two states that jump out when you look at test positivity rate are Nevada and Utah:
Nevada, Utah (and Idaho and Oregon):
Here we see a sharp uptick from 3% three weeks ago to 14.2% most recently. Nevada seems prone to be the next surging location.
Since the beginning of June, the testing positivity rate has increased from 4-5% up to 12%, indicating a potential surge. Here I show neighboring Idaho, with low counts but test positivity rate increasing from 2.6% to 10% in the past 2 weeks.
Another western state Oregon had very low test positivity rates of <2% in late May, and has been trending upwards in June as cases have increased. The levels are still low, around 5%, but bears watching.
We see some increases in test positivity in Oklahoma, Missouri, Arkansas, Kansas and Tennessee.
Oklahoma is also at risk for a surge -- their testing positivity was very low, <2%, at the beginning of June but has increased to 6-7% in the past couple of weeks. Bears close watching. Missouri also showed a bit of an uptick from 3% to 6% in test positivity rate in the month of June.
Arkansas has showed a steady increase from 3-4% in early May up near 9% now, and Kansas has doubled from 3.5% up to 7% since the beginning of June.
Tennessee is also showing a potential increasing trend, recently up to 8%
Other Southern States.
Georgia, Alabama, and North Carolina are some other southern states worth watching
Georgia has increased from 6% to 12% since late May, North Carolina has flirted with upticks but seems relatively stable, and their testing positivity rate has also heard steady between 5% and 7%. Alabama has tended to be high, with a steady state of 6-7% indicating potential undertesting relative to other states, and recently increasing up to 12-14%.
Northeast and Midwest:
The northeast and midwest seem largely under control, and the stable and/or decreasing test positivity rate.
What about California?
I'll finish by saying some words about California. I've seen some commentators lump California in with Texas, Arizona and Florida as states with increasing covid-19 numbers. I've considered it more in the "stable" category, as their testing rates have increased proportionally to their new daily case counts. This is evident in the stable test positivity rate since April. There may be evidence of a slight uptick in the past couple weeks from 4.5% to 6% and bears watching, but as of now I consider it to have a stable infection rate.