Covid and Immunity
In the study of Covid-19 we are learning as we go. Since it has only been a year (is that possible?) that this plague has been upon us, one thing we cannot yet know is the duration of immunity to re-infection with the SARS-CoV-2 virus. Other coronaviruses (several of which cause versions of the common cold) stimulate an immune responses in their hosts that last only a matter of a few months, so it is logical to worry that our immune response to SARS-CoV-2 (or even our immune response to a fancy vaccine) might also wane over a few months. That's worrisome to consider.
In the December 23, 2020, issue of the New England Journal of Medicine (NEJM) a large and (to me) fascinating study was published entitled "Antibody Status and Incidence of SARS-CoV-2 Infection in Health Care Workers." This work provides an interim answer to the question of durability of the immune response to SARS-CoV-2. It is a great example of how modern science works. I love this stuff and I have read such articles for years, but with all the jargon I still find it challenging to digest the details.
Here is the study conclusion: "The presence of anti-spike or anti-nucleocapsid IgG antibodies was associated with a substantially reduced risk of SARS-CoV-2 reinfection in the ensuing 6 months." Dang. That's really dry and unexciting. That conclusion barely made the news, even though this is a peer-reviewed study involving 32 authors, 12,541 participants, a huge effort and cost (funded by the government of the United Kingdom) that was published in one of the most prestigious medical journals in the world. Dry and unexciting it might be, but this is the stuff of which scientific truth is made.
Let me re-phrase and expand on the study's conclusions and implications. If your blood shows anti-spike antibodies to SARS-CoV-2, whether as a result of asymptomatic or symptomatic Covid-19 disease, then, at least for six months, it appears that your protection against re-infection is really, really good. We didn't know that. Instead, what we've seen in the news and on social media is the occasional worrisome anecdote, "Joe Blow was sick with Covid-19 in April and, OMG, he's suffering with it again!" Basic details in these anecdotes are usually lacking. Was Mr. Blow PCR test positive in April? Was his first illness really Covid or something else? How often does this happen?
The results of this study make the Joe Blow anecdote doubtful--or at least suggest Mr. Blow's experience is very rare. The study enrolled a special group of participants, 12,541 health care workers at Oxford University Hospitals in the United Kingdom, a relatively captive audience with more exposure to Covid-19 than most. All participants were tested for the presence of two different and very specific antibodies in their blood: anti-spike and anti-nucleocapsid IgG (immunoglobulin of "class" G).  This special group of participants (called a cohort) was assembled back in April, 2020. (How's that for forethought?) Of the 12,541 participants, 11,364 had no detected anti-spike antibodies when first tested on enrollment in the study, whereas 1177 did have anti-spike antibodies.
During the study 88 of those who did not have antibodies at baseline "seroconverted," that is, they developed anti-spike antibodies during the followup period and were then counted in the anti-spike antibody positive group, bringing the number of seropositive participants up to 1265.
Of those 1265 people anti-spike antibody seropositive participants not a single one developed symptomatic Covid-19 in the following six months. Even more encouraging (and striking), of those 1265 participants only 2 turned up with a positive PCR test for the virus in routine followup testing. Since anti-spike antibodies are the same type of antibody produced in response to the Pfizer and Moderna vaccines it seems likely that immunity induced by these vaccines will be similarly durable.
Of course, the lack of symptomatic Covid-19 in the antibody positive participants would be meaningless if those folks had no exposure to the disease during the followup period, so reporting the results among the antibody negative participants (the "control group") is critically important. Among the 11,364 originally antibody negative participants, 223 had one or more positive PCR tests for Covid-19 during followup. Of those 223 with a positive PCR test 106 experienced symptomatic disease. (PCR="polymerase chain reaction" test, the standard, highly sensitive, test for presence of a virus particle or part of a virus particle in the nose.) 
After adjustments for total days followed in the study the ratio between the risk of having a positive PCR test for a person with anti-spike antibodies over six months is 0.11, i.e. 1 in 10, of the risk of having a positive PCR test for a person without baseline antibodies. Moreover, in this study, speaking only of symptomatic Covid-19, the risk ratio is zero. The takeaway is that immunity is durable at least up to six months once one demonstrates anti-spike antibodies in one's blood.
Keep to the high ground,
 The anti-spike IgG antibody is produced in response to Covid-19 infection but it is also the antibody type our immune systems are asked to produce in response to the Pfizer and Moderna vaccines. In contrast, the anti-nucleocapsid IgG antibody (the other antibody tested in this Oxford study) is not produced in response to the two currently approved vaccines. That suggests that anti-nucleocapsid antibodies can be thought of as a marker for actual infection with Covid-19. That offers a possible means of determining whether a individual's immunity is based on a response Covid-19 disease or a response to the vaccine.(The "spike protein" is the corona of the coronavirus. It is essential to the binding of the virus to the surface of our cells. The "nucleocapsid" is the genetic material of the virus particle contained in the particle's interior. It consists of the virus' nucleic acids (in this case it's RNA) and associated enveloping and associated proteins, its "capsid")
 Notice that only 26 of the 88 participants who seroconverted during the study (developed anti-spike antibodies) ever demonstrated a positive PCR test for the presence of virus. Message: unless one is tested very frequently (more than once every two weeks) the testing may miss a brief period of viral shedding--or maybe some folks never shed virus in their noses). Also notice that only 26 of the 233 originally seronegative folks who then had a positive PCR test went on to seroconvert (and be counted among the 88). Tantalizing (but not nearly statistically significant) is the fact that among 24 originally seronegative participants who had had a positive PCR test for the virus (and symptoms) before enrollment, none of them had another positive PCR test during the study (and 5 of those 24 seroconverted during the study). That raises some small hope that some folks with symptoms and a positive PCR even without seroconversion (at least to the threshold titer used in this study) may be more immune than this study otherwise suggests.
P.S. Asymptomatic infection: Of the 1177 participants with anti-spike antibodies at baseline, 864 (68%) recalled having had symptoms consistent with Covid-19, that is, 32% recalled no symptoms. (This is consistent with current estimates that about 40% of those infected never have symptoms.)
P.P.S. One final note on the scientific endeavor. Immunity is complex. Other studies have been a bit worrisome because antibody titers (concentration of antibodies in the blood) naturally wane with time, but the immune system tends to keep a record of the threat and the antibody response to that threat in the form of "memory cells" of various kinds (T and B). This Oxford Study is, at this time, unique in providing us with actual data of viral shedding and re-infection--not just measurement of waning titers of antibodies. (It turns out that the assays for immune memory are expensive, challenging, and not always predictive. That's why most of what gets quoted in the media are studies that worry us by reporting antibody titers alone as a proxy for immunity--and that can be misleading.)