Credibility of the CDC on SARS-CoV-2


One of the main places Amer­i­cans look for in­for­ma­tion on coro­n­avirus is the Cen­ter for Disease Con­trol and Preven­tion (ab­bre­vi­ated CDC from the days be­fore “and Preven­tion” was in the ti­tle). That’s nat­u­ral; “han­dling con­ta­gious epi­demics” is not their only job, but it is one of their pri­mary ones, and they po­si­tion them­selves as the au­thor­ity. At a time when so many things are un­cer­tain, it saves a lot of anx­iety (and time, and money) to have an ex­pert source you can turn to and get solid ad­vice.

Un­for­tu­nately, the CDC has re­peat­edly given ad­vice with lots of ev­i­dence against it. Below is a list of ac­tions from the CDC that we be­lieve are mis­lead­ing or oth­er­wise in­dica­tive of an un­der­ly­ing prob­lem. If you know of more ex­am­ples or have in­for­ma­tion on any of these (for or against), please com­ment be­low and we will in­cor­po­rate into this post.


Dis­missed Risk of In­fec­tion Via Packages

On the CDC’s coro­n­avirus FAQs pages on 2020-03-04, they say, un­der “Am I at risk for COVID-19 from a pack­age or prod­ucts ship­ping from China?”:

“In gen­eral, be­cause of poor sur­viv­abil­ity of these coro­n­aviruses on sur­faces, there is likely very low risk of spread from prod­ucts or pack­ag­ing that are shipped over a pe­riod of days or weeks at am­bi­ent tem­per­a­tures.”

How­ever, this metare­view found that var­i­ous coro­n­aviruses re­mained in­fec­tious for days at room tem­per­a­ture on cer­tain sur­faces (card­board was not tested, alas) and po­ten­tially weeks at lower tem­per­a­tures. The CDC’s an­swer is prob­a­bly cor­rect for pack­ages from China, and it’s pos­si­ble it’s even right for do­mes­tic pack­ages with 2-day ship­ping, but it is in­cor­rect to say that coro­n­aviruses in gen­eral have low sur­viv­abil­ity, and to the best of my abil­ity to de­ter­mine, we don’t have the ex­per­i­ments that would prove de­liv­er­ies are safe.

Blinded It­self to Com­mu­nity Spread

As late as 2020-02-29, the CDC was re­port­ing that there had been no “com­mu­nity spread” of SARS-CoV-2. (Com­mu­nity spread means that the per­son hadn’t been trav­el­ing in an in­fected area or as­so­ci­at­ing with some­one who had). At this time, the CDC would only test a per­son for SARS-CoV-2 if they had been in China or in close con­tact with a con­firmed COVID-19 case.

Test­ing Cri­te­ria as of 2020-02-11

This not only left them in­ca­pable of de­tect­ing com­mu­nity spread, it ig­nored po­ten­tial cases who had trav­el­led to other coun­tries with known COVID-19 out­breaks.

By 2020-02-13, this had been amended to in­clude

The crite­ria are in­tended to serve as guidance for eval­u­a­tion. Pa­tients should be eval­u­ated and dis­cussed with pub­lic health de­part­ments on a case-by-case ba­sis. For severely ill in­di­vi­d­u­als, test­ing can be con­sid­ered when ex­po­sure his­tory is equiv­o­cal (e.g., un­cer­tain travel or ex­po­sure, or no known ex­po­sure) and an­other etiol­ogy has not been iden­ti­fied.

(The CDC de­scribes this change as hap­pen­ing on 2020-02-12, how­ever the Way­back Ma­chine did not cap­ture the page that day).

Based on this an­nounce­ment on 2020-02-14, when test­ing that could de­tect com­mu­nity ex­po­sure was hap­pen­ing it was in one of 5 ma­jor cities. How­ever as of 2020-03-01 only 472 tests had been done, so no test could have been hap­pen­ing very of­ten.

Between 2020-02-27 and 2020-02-28, the pri­mary guidelines on this page were amended to

How­ever guidance went out on the same day (the 28th) that only listed China as a risk (and even then, only medium risk un­less they had been ex­posed to a con­firmed case or trav­el­led to Hubei speci­fi­cally).

Test­ing Kits the CDC Sent to Lo­cal Labs were Unreliable

They gen­er­ated too many false pos­i­tives to be use­ful.

Ham­strung De­tec­tion by Ban­ning 3rd Party Test­ing (HHS/​FDA, not CDC)

One rea­son the CDC used such stringent crite­ria for de­ter­min­ing who to test was that they had a very limited abil­ity to test, ham­strung fur­ther by the faulty tests sent to lo­cal labs. Nor­mally pri­vate test­ing would fill the gap, but the de­part­ment of Health and Hu­man Ser­vices in­voked emer­gency mea­sures that cre­ated a re­quire­ment for spe­cial ap­proval of tests, and the FDA didn’t grant it to any­one (source).

There are mul­ti­ple har­row­ing sto­ries of peo­ple with ob­vi­ous symp­toms and ex­po­sure to the virus be­ing turned away from test­ing, of­ten against a doc­tor’s pleas:

There is also a ru­mor that the first case caught in Seat­tle, which has since turned into the US epi­cen­ter of the dis­ease, was caught by a re­search lab us­ing a loop­hole to perform unau­tho­rized test­ing (rais­ing the pos­si­bil­ity that it’s worse el­se­where and sim­ply hasn’t been caught).

Ceased to Re­port Num­ber of Tests Run

Un­til 2020-03-02, the CDC re­ported how many tests SARS-CoV-2 tests it had run. On March 2nd, it stopped (be­fore, af­ter). There are many po­ten­tial rea­sons for this, none of which in­spire con­fi­dence. The offi­cial rea­son for this as told to re­porter Kel­sey Piper is that the num­ber would no longer be rep­re­sen­ta­tive now that states are run­ning their own tests. So, best case sce­nario, the CDC can not co­or­di­nate enough to count tests performed by other labs.

Gave False Re­as­surances About Re­cov­ered Individuals

As of this writ­ing (2020-03-05), the CDC’s “Share Facts” page states that “Some­one who has com­pleted quaran­tine or has been re­leased from iso­la­tion does not pose a risk of in­fec­tion to other peo­ple.”

While it is cer­tainly true that be­ing re­leased from quaran­tine im­plies a sig­nifi­cantly re­duced risk, the quaran­tine that is typ­i­cally performed is not stringent enough to say that peo­ple re­leased pose no risk. The quaran­tine pro­ce­dure performed by the CDC lasts 14 days, af­ter which if symp­toms have not ap­peared, they can be re­leased.

There are case re­ports of in­di­vi­d­u­als with in­cu­ba­tion pe­ri­ods of 27 days and 19 days. There was a case in Texas where a per­son tested pos­i­tive af­ter be­ing re­leased from quaran­tine and vis­it­ing a mall.

While an epi­demic is still con­tained, safely quaran­tin­ing at-risk peo­ple means choos­ing a quaran­tine pe­riod long enough to be con­fi­dent that, if they haven’t shown symp­toms, they don’t have the dis­ease. When a dis­ease is still con­tained, this should be risk averse, since a sin­gle in­fected per­son could start an out­break. The CDC’s 14-day quaran­tine pe­riod was not long enough to catch the cases de­tailed above.

This was fore­see­able. This pa­per, pub­lished Feb 6, es­ti­mated the dis­tri­bu­tion of in­cu­ba­tion pe­ri­ods, in­clud­ing the in­cu­ba­tion pe­ri­ods of out­liers.

The rele­vant row is the 99th per­centile row, which es­ti­mates the longest in­cu­ba­tion pe­riod per 100 peo­ple. If you quaran­tined 100 peo­ple, one of them would have an in­cu­ba­tion pe­riod at least that long. The pa­per es­ti­mates this us­ing three differ­ent meth­ods; two of those es­ti­mates are greater than 14 days, and all three es­ti­mates put sig­nifi­cant prob­a­bil­ity on in­cu­ba­tion pe­ri­ods longer than 14 days.

There are also re­ports of the virus re-emerg­ing in pa­tients who were be­lieved to have re­cov­ered.

Con­flated Ge­net­ics and En­vi­ron­men­tal Exposure

This is a tough topic to write about.

Cru­elty to peo­ple be­cause they have or might have a dis­ease is never okay. And the vast ma­jor­ity of peo­ple who were cruel to Asian-ap­pear­ing peo­ple in the early days of an epi­demic were do­ing it to healthy peo­ple out of knee jerk fear and an­tag­o­nism, not a mea­sured, well-in­formed cost-benefit anal­y­sis. When the CDC claimed on 2020-02-29 that “Peo­ple of Asian de­scent, in­clud­ing Chi­nese Amer­i­cans, are not more likely to get COVID-19 than any other Amer­i­can.” they were surely try­ing to dampen at­tacks on peo­ple who had done noth­ing wrong and were hurt­ing no one.

But the state­ment is false. Chi­nese-Amer­i­cans are more likely to travel to China or as­so­ci­ate with peo­ple who have, and thus were more likely to catch SARS-CoV-2. This doesn’t mean they are more likely to catch it given ex­po­sure, but they were more likely to be ex­posed.

The CDC ad­mits this in the page speci­fi­cally on stigma (2020-02-24), say­ing “Peo­ple—in­clud­ing those of Asian de­scent—who have not re­cently trav­eled to China or been in con­tact with a per­son who is a con­firmed or sus­pected case of COVID-19 are not at greater risk of ac­quiring and spread­ing COVID-19 than other Amer­i­cans.”

How­ever that same anti-stigma page goes on to say “Viruses can­not tar­get peo­ple from spe­cific pop­u­la­tions, eth­nic­i­ties, or racial back­grounds.” This is also false. About 10% of Euro­peans are im­mune to HIV, an im­mu­nity not found peo­ple origi­nat­ing from other ar­eas. So we know it is tech­ni­cally pos­si­ble for a virus to have differ­en­tial effects based on race.

Does SARS-CoV-2 in par­tic­u­lar have race-re­lated effects? There are peo­ple claiming Asian men are more sus­cep­ti­ble to SARS-CoV-2 than oth­ers due to a higher ex­pres­sion of a cer­tain pro­tein (ex­am­ple). Other peo­ple dis­pute this (ex­am­ple). Right now it is very much an open ques­tion.

We can see why the CDC pri­ori­tized calming racially-mo­ti­vated vi­o­lence over fully ex­plain­ing their con­fu­sion over an unan­swered ques­tion. It might have been the high­est-util­ity thing to do. But it is im­por­tant to know that “mis­rep­re­sent­ing data in or­der to pro­duce bet­ter ac­tions from the pub­lic” is a thing the CDC does.

Dis­cour­aged Use of Masks

Which brings us to the CDC’s state­ment on masks:

CDC does not recom­mend that peo­ple who are well wear a face­mask to pro­tect them­selves from res­pi­ra­tory dis­eases, in­clud­ing COVID-19.

The Sur­geon Gen­eral (who is not di­rectly part of the CDC) takes a stronger tack:

While we can’t hold the CDC re­spon­si­ble for the Sur­geon Gen­eral, they are be­ing con­flated in a lot of news ar­ti­cles say­ing or im­ply­ing that masks are use­less for healthy peo­ple. They’re (prob­a­bly) not.

Our best guess is that the CDC is try­ing to con­serve masks for health care pro­fes­sion­als and oth­ers with the high­est need, in the face of a loom­ing mask short­age. That could eas­ily be the op­ti­mum mask al­lo­ca­tion. I can’t prove the lie wasn’t jus­tified for the greater good. But it is an­other ex­am­ple of the CDC plac­ing “get­ting the out­come it wants” over “tel­ling peo­ple the literal truth.”

What Does This Mean?

Th­ese er­rors we’ve high­lighted tend to­wards er­rors of omis­sion: say­ing some­thing is com­pletely safe when it’s not, say­ing some­thing is un­helpful when it is, say­ing the cur­rent state is less dan­ger­ous than it is. You should in­clude that bias when pro­cess­ing new in­for­ma­tion from the CDC. Notably we’re not say­ing any of the things they do recom­mend are bad: to the best of our knowl­edge, you should be wash­ing your hands and not touch­ing your face. Vac­cines are (mostly) great. But I would not take the CDC say­ing an ac­tivity is safe or un­nec­es­sary as the last word on the sub­ject.