Friday, June 19, 2020

A Problem With COVID-19 Virus Antibody Testing

While going through each state's COVID-19 virus reports, I have noted that, among those states that report antibody testing, the antibody positive rates are generally substantially lower than the confirmatory test rates. 

Why is this significant? Three possibilities come to mind. 1) Antibody testing is of a substantially different population than is the test for the presence of virus. 2) There is little community immunity or little community exposure. 3) The antibody tests are flawed. I will discuss these possibilities further below. 


There are two general ways to test for COVID-19 virus/exposure. The PCR test (sometimes called the molecular test) is the gold standard and says that someone is currently infected. It should only show a positive result if a person is secreting the virus. Typically this is performed from an oral or nasal swab. 

The antibody test (sometimes called the serology test) says whether the person has been exposed to COVID-19 to the extent that he or she produces antibodies. This can be due to a full-fledged infection or due to a silent, asymptomatic infection. The person may have recovered and may no longer be infectious--or else not. This person may be immune to future infection--or else not. Antibodies take a period of time to generate in a body and will not appear immediately after infection, so this test does not account for the very recently infected.

Now to the Data. 

Although several additional states report the total number of antibody tests, seven states report both test numbers and how many are positive. This latter group is, in alphabetical order, Arizona, Georgia, Iowa, Maine, Missouri, North Dakota, and South Carolina.

Arizona, which is undergoing a massive increase in the number of confirmed (PCR) infections, has been providing antibody positivity data since May 7. Unfortunately, they provide only a percent positive figure, not the actual number. Here is a brief table with some of their antibody results.

Arizona   Ab tests   positive   PCR tests   positive
May 7 14652           3.7%       96434   10.31%      
June 1 93586           3.0%       228070   8.82%      
June 17 136117           3.0%       365846   11.19%      
The above numbers are cumulative. For example, a total of 14652 antibody tests were performed up to May 7, the day they were first reported, and 3.7% (Arizona calculations, meaning 3.65 to 3.749%, they do not provide an exact count) of those were positive at that time. 

Six weeks is a long time in the history of this pandemic. Rather than cumulative, let's break these down in two time periods. 

Arizona Ab tests Ab pos min Ab pos max Ab % PCR tests pos.%
May 7 to June 1 78934           2215      2319      2.81 - 2.94 131636       7.73            
June 1 to June 17 42531           1162      1390      2.73 - 3.27 137776      15.1            
May 7 to June 17 121465           3470      3616      2.86 - 2.98  269412      11.5            

Even during the month of June, when the PCR tests passed 15% positive, the antibody tests remained around 3%.

The other states provide numbers rather than vague percentages, so their data are more straightforward without the need for ranges. They have, however, been reporting their antibody positivity rates for a briefer period.

Georgia Ab total Ab pos Ab % PCR tests PCR pos    PCR %
June 2 91325           5395             5.91           474287           48207        10.16         
June 17 118426           6912             5.84           644723           60030        9.31         
June 2 to 17 27101           1517             5.60           170436           11823        6.94         

Since Georgia began reporting recently, I included cumulative and time period data in one table. Again, the PCR rate is higher than the antibody rate.

Iowa places their antibody testing data on an inner page of their report. There is not an archival record of these data, so to access the data you have to visit on a particular day. They began reporting antibody tests results on June 9.

Iowa     Ab total      Ab pos      Ab %     PCR tests     PCR pos     PCR %
June 9 24290         2046         8.42        198123       22238    11.22    
June 18 27575         2157         7.82        240657       24805    10.31    
June 9 to 18 3285         111         3.38        42534       2567    6.04    

Again, the first two rows are cumulative data to that date, the last row is over the past nine days since Iowa began testing.

Maine began reporting antibody results on May 27. It has a low new infection rate, 45th out of 51 in the nation (including DC) in new weekly cases per million population. When the PCR rates are very low, it seems that the antibody rates can be slightly higher.

Maine     Ab total      Ab pos      Ab %     PCR tests     PCR pos     PCR %
May 27 3661     198     5.41      46043     2137 4.64     
June 17 6137     294     4.79      74648     2836 3.80     
May 27 to June 17 2476     96     3.88      28605     699 2.44     

Missouri also has a report on antibody testing on an internal page. Along with daily data, they provide a chart of daily tests with numbers positive and negative. When the negatives are more than a thousand in a day, they present the number as "k," for example, 1705 becomes 1.7 k. In spite of this inexactness, the daily information is useful.

Missouri     Ab total      Ab pos      Ab %     PCR tests     PCR pos     PCR %
May 24 18868       692       3.67   151619     11988    7.91 
June 20 37210       1387       3.73   311115     17590   5.65 
May 24 to June 20 18342       695          3.79   159496     5602    3.51 

Like with Maine, low PCR numbers coincide with low antibody positivity rate.

North Dakota began reporting antibody testing on May 29. The PCR positivity rate is low, the antibody testing rate, lower.

North Dakota     Ab total      Ab pos        Ab %       PCR tests      PCR pos     PCR %
May 29 2546     86     3.38   69453       2520   3.63  
June 17 4125     128     3.10   89674       3166   3.53  
May 29 to June 17 1579     42     2.66   20221       646   3.19  

Like Arizona, South Carolina is undergoing a surge in confirmed COVID-19 cases. Like Arizona, the antibodies have barely register an uptick.

South Carolina     Ab total      Ab pos      Ab %     PCR tests    PCR pos     PCR %
June 11 27609 1337 4.84 236456 16441 6.95
June 17 31376 1538 4.90 272749 20556 7.54
June 11 - June 17 3767 201 5.34 36293 4115 11.34

These seven states, since the beginning of the pandemic, have performed a total of 360,966 antibody tests with a positivity rate of 4.40 to 4.44%. These states have performed a total of 1,979412 PCR tests with a positivity rate of 7.53%. They have a combined population of 34.4 million.

Over the course of the time in which they reported their data, the states performed 192,667 antibody tests with a positivity rate of 3.47 to 3.54%. Over that same period, the states performed 726,997 PCR tests with a positivity rate of 7.76%. These are the better figures for comparison: the cumulative data includes a higher positivity rate period near the beginning to the pandemic, when antibody tests were not in common use.

So, we have roughly 7.8% by PCR and 3.5% by antibody. In theory, the antibody tests should find a larger group and include those who had mild disease. Something is off here.

High and Low Antibody Testing Rates.

In a highly publicized study conducted by Stanford, approximately 50 times the number of people in the community of Santa Clara, California tested positive for COVID-19 virus antibodies than did for PCR testing. If accurate, this study could have far-reaching implications. It suggests widespread immunity, many people having minor or no symptoms, and a far lower death rate for those who do have infections.

That study, which did have glaring faults, went on to be criticized, perhaps even demonized. In a more recent study, Stanford found an antibody positivity rate of 0.3% among 11,000 asymptomatic  university employees.

Spain was an epicenter of the infection from March to May, for a time leading Europe in cases. A study of 70,000 individuals in mid-May found that only 5 percent of Spaniards possessed antibodies to the virus with one region as high as 14.2%.

Spain: On May 15, had a cumulative 274,367 cases (population 46.94 million) for a nationwide positivity rate of 0.58%. In this same time period, the antibody positivity rate was 5.0%. This suggests the total case number is underestimated by about 8.6 fold.

What's Going On?

One piece of the puzzle is that different groups are being tested. The PCR, confirmatory, test for infection is more likely to be used for those who show symptoms and those who are directly exposed to the infected individuals. The antibody studies are more likely to survey others, for example, the entire workforce of health care facilities.

Still, I find this explanation unsatisfactory, or, at least incomplete. In Spain's broad survey, the positivity rate for antibodies was several-fold higher when trying to select a representation of the total population. Why would the American antibody surveys choose a population with such a lower fraction of those positive?

Another possibility is that there is simply very little community exposure. This explains the low numbers in the antibody test, but not why it is even lower than confirmatory tests or why it is not in line with other studies.

A final possibility is that the antibody tests are flawed. The positivity rate per test continues to linger around 3% in Arizona even as the PCR rate doubled.

Of course, the answer may be due in part to combinations of the three possibilities.

Martin Hill Ortiz is a Professor of Pharmacology at Ponce Health Sciences University and has researched HIV for over thirty years.


Post a Comment