Which COVID-19 positivity rate should you believe for Idaho? Why so many probable cases?
Among the overwhelming amount of numbers that have come from tracking the COVID-19 pandemic, positivity rate might be the most misunderstood.
The math is simple.
But identifying the right raw numbers to put into the formula has proved elusive for some popular national tracking websites — creating a “what source do you believe?” situation.
And that’s just one of many frequent questions the Idaho Statesman receives about Idaho’s COVID-19 data. Here, we try to answer those questions:
Is Idaho’s COVID-19 positivity rate 9% — or 21%?
If you want to know Idaho’s true positivity rate, you have to check the Idaho Department of Health and Welfare’s weekly update on Thursdays. Access the “COVID-19 Data Dashboard” at coronavirus.idaho.gov and click on “Laboratory Testing.”
The positivity rate statewide was 9.4% for Sept. 20-26 and 10.5% for Sept. 27-Oct. 3, although those numbers can change as more results are received for tests conducted that week. The rate also is available by health district and county.
Since the start of the pandemic, the weekly positivity rate never has been higher than 14.8% (July 5-11). The overall positivity rate from March 1 through Oct. 3 was 9.5% — that’s the number of positive results divided by the number of unique samples tested.
So why are different numbers published elsewhere?
One big reason is a misunderstanding of Idaho’s data.
At the top of the testing page, Idaho lists a total number updated daily for “laboratory tests completed.” However, that’s actually the number of Idahoans who have been tested one time — not the total number of tests. People who have been tested multiple times only get counted once. So if you do the positivity math based on that number, the percentage will be too high. Idaho’s number of tests is at least 90,000 higher than its number of people tested.
On the statewide cases page, Idaho lists a combined total for confirmed and probable cases. But for positivity, you need the number of lab-confirmed cases — probables don’t involve a positive PCR test. So if you use the total case count, you’re including thousands of cases (4,686 through Thursday) that should be excluded. If you hover over the case total, you get the confirmed/probable breakdown.
“We need to be a little bit more precise in our language,” said Kathryn Turner, Idaho’s deputy state epidemiologist, “to make sure we’re communicating what we intend to communicate.”
Another factor is this: IDHW’s data is de-duplicated. The state might get a positive report from the lab that conducted the test and the health care provider who ordered the test. IDHW compares records to ensure that case is only counted once.
However, the Centers for Disease Control and Prevention uses de-identified records. The CDC might get the same result reported by the state and the lab, Turner said, with no way to know which records are duplicates.
With all that in mind, here are three conflicting positivity numbers you might see online:
▪ Johns Hopkins, which is popular for nationwide COVID-19 data, showed Idaho with a 21.2% positivity rate for Sept. 20-26. The problem? Johns Hopkins is using the number of people tested instead of the number of total tests, and it’s using the total cases instead of just the confirmed cases. That creates a positivity rate that is far higher than reality.
The Washington Post also is using the wrong number for total tests.
However, since Idaho doesn’t publish a daily number of total tests conducted, there’s no way to produce an updated rate on a daily basis without using the people-tested number.
▪ The CDC data is less specific but lists Idaho as having a positivity rate of 11-20% during the pandemic. It says Idaho has conducted more than 490,000 tests — which is tens of thousands more than IDHW says Idaho has conducted, likely because of duplicated records. Idaho would need more than 54,000 positive test results for the CDC math to work — more than 12,000 more than the state has reported. The Idaho Statesman contacted the CDC but did not receive an explanation for its numbers.
This testing discrepancy varies by state: The CDC reports roughly the same number of tests as Nevada and California, fewer than Washington and far more than Utah.
“I think the main reason, frankly, is they don’t have access to the raw data,” Turner said of the national sites. “So if they’re using some aggregated numbers and doing some math, which is totally fine, as long as they’re doing it the same way all the time … they can see the trend just the same as we are. It’s just that their precision is off because they don’t have access to the raw data.”
▪ The Idaho Statesman reports a positivity rate every day that is the percentage of Idahoans who have tested positive among those who have been tested. That rate stood at 12.8% through Wednesday. It’s the most accurate rate that can be produced from the daily data, but is not the true test positivity.
One other wrinkle with positivity data: Idaho’s chart doesn’t include all positive cases because not all labs report negative results to the state. That situation has improved, Turner said, and only about 5% of positive cases are missing from the calculation. Only positives from labs that also report negative results are included.
Through Oct. 3, the state was including 39,447 positives and 417,024 tests in the positivity chart.
“If you want precision, you have to go with the state’s (positivity rate),” Turner said.
The COVID Tracking Project at The Atlantic published a blog post Thursday explaining why it doesn’t deal with positivity rates — and it’s because of the inconsistency in how states produce their data. The post’s title was “Test positivity in the U.S. is a mess.”
Why are we seeing so many probable cases now?
Idaho posted a record 1,521 probable cases in September, even though confirmed cases were down from the previous two months. We’re on pace to shatter that mark in October.
Why?
The nationwide definition of probable cases was updated Aug. 5 to include all positive antigen tests, Turner said. Antigen tests are rapid-response, point-of-care tests that aren’t as accurate as PCR tests. The previous definition included antigen tests only if the person had contact with a known case and compatible symptoms.
“We found that we could cast a wider net if we took off those additional criteria,” Turner said of the decision from the Council of State and Territorial Epidemiologists.
Antigen tests have been used in places like long-term-care facilities and corrections facilities that need to test large groups of people, Turner said. It’s unclear when they’ll be widely available to the public, but the Mountain West Conference made a deal with Quest Diagnostics to provide three antigen tests per week to members of the Boise State University football program and the conference’s other 11 teams so the 2020 season can begin later this month. Idaho also is providing tests to school districts.
People who test positive with antigen tests generally don’t also get a PCR test, Turner said. They are treated the same as a confirmed case by public health with contact tracing and information about how to isolate.
“We’re pretty sure a positive is a positive,” Turner said of antigen tests. “We’re not sure a negative is always a negative.”
Idaho has plans to eventually add antigen and antibody testing data to its dashboard.
How are positive results reported to the state? Why do state and health district numbers conflict?
With nearly all PCR tests, results are automatically reported to Health and Welfare and the health care provider, Turner said. IDHW directs the records to the appropriate health district in their shared, web-based system.
With antigen tests, the result might be provided by fax. Paper records are sent to the districts to input into the record-keeping system.
IDHW counts the number of records in the system for each district at about 2 p.m. each day. Those are the cases included in the daily report. A file that still is being worked on at that time but hasn’t been saved won’t be included.
The health districts report their numbers on their own websites at various times, though mostly between 4 and 6 p.m. So often the districts will report larger numbers than the state.
There also can be times that a record is in the system but the district is still investigating and might not be ready to count it as a case. If it’s been entered into the system, the state will count it. That sometimes leads to the state reporting higher numbers for a county than the district.
“Trying to do data in real time … those sorts of glitches are going to happen,” Turner said. “They have more information than we do.”
The Idaho Statesman reports its daily totals based on analysis of the data provided by the seven health districts and IDHW. In most cases, the Statesman uses the health district numbers when available. Through Wednesday, the Statesman’s analysis had reported 217 more confirmed cases and two more deaths than IDHW because of the lag in reporting.
What happens when a probable case gets tested?
A person counted as a probable case who later tests positive through a PCR test is reclassified as a confirmed case. If that person tests negative, the case is removed.
What happens when someone is tested multiple times?
Multiple positive tests for one person are considered one confirmed case, unless reinfection months later becomes an issue, according to IDHW. However, multiple positive and negative tests are counted toward the positivity percentage. Multiple positives are only counted if they are from different samples.
What is considered a COVID-19-related death in Idaho?
These are not just deaths of people who had COVID-19.
To count, COVID-19 must be listed as a cause of death on the death certificate. However, it’s possible for multiple causes to be listed. That’s why the state uses the “COVID-19 related” language.
“Somewhere along the line, COVID is related to the death,” Turner said. “Maybe they actually died of pneumonia as a result of their COVID infection.”
The death certificate is produced by the medical provider or coroner, according to an IDHW blog post. COVID-19 or the SARS-CoV-2 virus must be listed as either “an underlying cause of death or a significant condition contributing to death” for the death to be counted as COVID-19 related. Confirmed and probable COVID-19 cases are included.
Do we know what underlying conditions are contributing to COVID-19 deaths in Idaho?
IDHW doesn’t publish an accounting of underlying conditions. That data is collected by the Bureau of Vital Records and Health Statistics.
“Those data are available. We don’t publicize that,” Turner said. “… We don’t want to give the impression to anybody that just because a person has diabetes, or a person who perhaps was struggling with their weight and was classified as obese and died of COVID that COVID isn’t a big deal in that death. We certainly don’t want that to be the message.”
Some states do publish underlying conditions. Louisiana, for example, reports that hypertension was involved in 52.2% of its COVID-19 deaths, diabetes 31.1% and cardiac disease 20.4%.
Will college cases appear in Ada, Latah and other county counts?
Yes.
Students at Boise State University (Ada County) or the University of Idaho (Latah County) — and other schools — who test positive will be counted in their “usual place of residence,” Turner said. For students, that means they will be counted where they are living while attending school.
Case counts have been significantly impacted in Ada, Latah and Madison (BYU-Idaho) counties.
Does Idaho have a statewide tracker for K-12 school cases?
Yes.
IDHW launched the new report at coronavirus.idaho.gov last week. It will be updated Fridays with case counts by school building. It won’t be comprehensive — the flow of information is imperfect — but it paid immediate dividends with information on cases in the West Ada School District. Idaho’s largest school district previously declined to provide its numbers to the Idaho Statesman.
