Understanding the GA DPH Data

The Georgia Department of Public Health is providing a lot of statistics about the spread of coronavirus in the state, but not all of it is presented in a way that is easy to understand, so I am breaking down the Daily Status Report to help people understand how to read these numbers and graphs.

I don’t cover all the sections on this page, just the key sections that seem to be causing a lot of confusion.

Overall COVID-19 Status

This is the overview of four key numbers that Georgia is tracking:

Confirmed COVID-19 Cases

This is the total number of positive COVID-19 cases confirmed by PCR (viral) tests. It is a cumulative number, meaning it will only ever go up. It does not go down as people recover or die. Georgia, like many states, does not have a system for tracking recovered patients. See my estimates of recovered cases.

This number of cases is higher than the number of positive PCR tests shown on the site, because not all labs submit their data electronically, and those tests aren’t counted on the web site yet.

This number is also not very helpful on its own. The more important metric is how many new cases are reported each day, along with how many tests were performed. In addition, it’s important to note that reporting is often slower on weekends, so new cases appear to dip on weekends with unusually high numbers on Monday and Tuesday, as the weekend backlog of tests are processed and entered. For this reason, tracking newly reported cases as a 7-day average is most useful, as it smooths out the ups and downs caused by the weekends.

Deaths

This is a cumulative number of deaths from COVID-19 in the state – it will also only ever go up. Like the new case reports, reports are delayed on the weekend, so looking at the 7-day moving average is a better gauge than the increase on any given day. At it’s peak, our 7-day moving average of deaths was 42 deaths on April 21st and April 22nd. As of June 18, it’s down to 16 deaths per day on average.

Keep in mind, this number is the day the death is reported to DPH, not the day the person actually died, so they are always a bit delayed, as it takes time after a person dies for the coroner to issue a death certificate and report cause of death to the state, so reported deaths on any given day can include deaths that occurred even a month or more earlier. Learn more about how deaths are reported.

Hospitalizations & ICU Admissions

These numbers are less useful in my opinion, because they only indicate the patient status at the time of reporting their case to DPH. That means if a person tests positive for COVID-19, and then is later hospitalized or admitted to the ICU, they are not included in these numbers. It also includes patients hospitalized for other reasons but who were tested for COVID (patients are commonly tested before surgeries and other medical procedures).

But the other big problem with these numbers is that they are not *current* numbers in the hospital or ICU with COVID-19. They are cumulative numbers that do not ever go down. After someone is released from the ICU or the hospital, they are still included in these counts.

It’s not on the DPH web site, but GEMA (Georgia Emergency Management) reports current numbers of hospitalized patients, hospital bed capacity, and number of ventilators in use. See the current hospitalization numbers in Georgia.

COVID-19 Testing

This area has been expanded greatly. It now shows the number of viral (PCR) tests, antibody tests, and total tests that are reported via the electronic reporting system, as well as the number and percent positive for each category. Ignore the last column for total tests and total positives. There’s no point in mixing viral and antibody tests. My focus is on the viral tests, since that’s what tells us who is currently sick.

These numbers don’t tell you very much on their own – what matters most is how much they change many new tests are being performed each day/week. Many people (myself included) have been tracking testing numbers since they first started providing them in later March. In that time, testing has gone up substantially. In early April, we were only doing about 2000 tests per day. By late June, we’re up to about 15,000 tests per day.

The number of confirmed cases is higher than the number of positive viral (PCR) tests shown, because not all labs submit their data electronically, and those tests aren’t counted on the web site yet.

The number of positive tests divided by the number of tests gives the overall % positive. For viral (PCR) tests, this number should stay under 10%, and ideally will go below 5% if we are doing enough testing. On my Today in Georgia and Weekly Overview pages, I provide the percent positive overall as well as a 7-day % positive.

The percent positive of serology tests shows us what percentage of Georgians who got tested may be carrying COVID antibodies. These are considered possible cases, so positive antibody tests are not included in the confirmed cases total.

Georgia Cases by County

Screenshot from June 2

Maps of Cases & Deaths

These maps provide a visual of what counties have the most total confirmed cases (not active cases), and the most total confirmed cases per 100,000 people (adjusting for population), as well as the most total deaths and deaths per 100,000 people. Unfortunately, Georgia adjusts the color ranges inexplicably, so it’s hard to compare these graphs from day to day. However, they are helpful at spotting which counties are hot spots by total numbers or per capita.

The cases by county show the metro counties as hot spots since they have the bulk of Georgia’s population. When adjusted per 100K, southwest Georgia continues shows as a hot spot because these rural counties were hit hard early on, and they have such low populations compared to the metro counties. Keep in mind that the case maps are limited in their usefulness to determine current risk because it doesn’t show active cases, only total cumulative cases. Learn how to estimate active/recovered cases for your county.

Using the maps, you can view the following data for your county:

  • Confirmed Cases
  • Total Population
  • Case Rate (cumulative confirmed cases per 100K)
  • Total Deaths
  • Deaths per 100K of population
  • Hospitalizations

Again, keep in mind these numbers are based on total confirmed cases (not active cases), and total number of people who tested positive for COVID while hospitalized (not current hospitalizations).

You can also see some of this information on the Counties tab (below the maps), where you can sort all counties by confirmed cases, case rate, total deaths, hospitalizations.

COVID-19 “Over Time” Graphs

There are three graphs in this section: Cases, Cumulative Cases, Deaths. I generally focus on the Cases and Deaths graphs, as the cumulative cases and deaths don’t show the trends very well.

You can view these graphs for the statewide data, or click on your county on the map to see the curve for your individual county.

These graphs have caused a lot of confusion and frustration, as the 14-day window at the end always shows declining cases and deaths. These numbers are filled in as confirmed cases and deaths are reported, so the past 2 weeks are often very low and must be regarded as incomplete. What’s the problem with the DPH graphs?

My recommendation is to completely ignore everything in the 14-day window, as it’s often overly optimistic.

Cases Over Time

For the new cases graph, cases are added to the graph on the day symptoms appeared, or when the test was performed if the symptom onset is unavailable. (If both are unavailable, it is reported on the day the result was reported to DPH.) If you developed a cough on April 27th, you might have waited a few days before getting tested. Let’s say you got a test on April 30th. Then the test can take a few days to process at the lab, and another day or two to get reported to DPH, so your case may get reported and added to the total case count on May 5th (8 days later). On May 5th, your new case will be added to the total numbers, and placed on this graph for April 27th. It’s very unlikely you’d get tested, get results, and your results get reported to DPH on the same day. That’s why confirmed cases for the current day are always extremely low.

The orange line (showing the 7-day moving average) is the important part. The exact number of new cases per day is partly affected by test date (when symptom onset is not available), so weekend numbers are artificially low and Monday/Tuesday numbers are artificially high). The 7-day average smooths out those lows and highs.

Deaths Over Time

This graph shows the number of deaths on the day the person actually died. Again, ignore the entire shaded 14-day window, as it sometimes takes a week or more for deaths to be reported by DPH. As of now, the absolute peak day was April 16th with 58 deaths that day. We haven’t had a day higher than 45 deaths since then.

The orange line (showing the 7-day moving average) is what I focus on, because it helps smooth out day-to-day variations. Based on the 7-day moving average, it looks like deaths peaked on April 20th so far, and have been dropping since. Learn more about how deaths are reported.

COVID-19 Case Demographics

The updates to the site on June 2 added a nice new graph of Confirmed COVID-19 Cases, Hospitalizations, and Deaths by Age Group.

Screenshot from June 3

From this graph, you can see how many people of each age group make up the cases and the deaths. You can see from this that even though plenty of young and middle-aged adults may get COVID-19, it is rarely fatal to this age group. However, the fatality rate is much higher starting around age 60 and increasing with age. Learn more about deaths in Georgia by age group.