Antibody serosurveillance is an essential tool for monitoring the COVID-19 pandemic,
offering a more comprehensive picture of who has been infected than swab testing of
symptomatic individuals alone. In recent months, several countries have done large-scale
seroprevalence surveys, including the USA,1, 2 China,
3
Brazil,
4
England,
5
and Spain.
6
These studies have confirmed that the world is still in the early stages of the COVID-19
pandemic, with the majority of the populations surveyed testing negative for severe
acute respiratory syndrome coronavirus 2 (SARS-CoV-2) antibodies.
The surveys carried out so far have two major limitations. The first is the use of
sampling strategies prone to selection bias, including non-random, unrepresentative
sampling,3, 4 postal sampling with substantial dropout,2, 6 or convenience sampling.
1
This is problematic for an infection that disproportionately affects some ethnic groups
and deprived communities who are less likely to participate in research.5, 7 The second
is the use of antibody tests with inadequate performance characteristics. Most large
surveys have sought to avoid costly laboratory testing by using point-of-care lateral
flow assays.2, 4, 5, 6 These tests are often poorly validated on a handful of samples,
2
can be subject to inter-batch variation, and even when thoroughly assessed have inferior
sensitivity to laboratory assays (<90%).
8
This adds uncertainty and necessitates substantial adjustment of raw data to account
for false-negative results.
9
In The Lancet, Shuchi Anand and colleagues describe an inventive, practical, and scalable
strategy for conducting SARS-CoV-2 seroprevalence surveys, which overcomes these limitations.
10
By testing the remainder plasma of 28 503 randomly selected patients receiving dialysis
in the USA, they were able to test an unbiased sample of an important patient group
across the entire country.
Importantly, Anand and colleagues chose a good test for their survey. The Siemens
lab-based spike-protein-receptor-binding domain total antibody chemiluminescence assay
adopted by the authors was the best-performing platform in the largest external appraisal
of commercial assays to date, in terms of both sensitivity and specificity.
11
Their choice negates the need for major adjustment of the raw data to obtain reliable
prevalence estimates.
The authors standardised data by age, sex, region, and race and ethnicity to provide
the first nationally representative estimates of SARS-CoV-2 seroprevalence in the
US dialysis and US adult populations, with samples taken in July, 2020. Using anonymised
demographic data, residence, postal codes, census data, and publicly available COVID-19
burden and community mobility data, the authors provide estimates for differences
in seroprevalence by neighbourhood, race and ethnicity, poverty, population density,
and mobility restriction.
The findings are striking. 2292 dialysis participants had SARS-CoV-2 antibodies, comprising
970 (42·3%) women and 1322 (57·7%) men, the majority of whom (1765 [77·0%]) were aged
45–79 years. This translated to a seroprevalence of 8·0% (95% CI 7·7–8·4) in the sample,
rising to 9·3% (8·8–9·9) when standardised to the US adult population. There was a
remarkable variation in seroprevalence by state in the sampled participants, with
early pandemic hotspots such as New York (33·6%, 95% CI 31·7–35·6), Louisiana (17·6%,
10·8–28·7), and Illinois (17·5%, 15·2–20·2) recording substantially higher seroprevalence
than their respective neighbouring states of Pennsylvania (6·4%, 4·7–8·8), Arkansas
(1·9%, 1·0–3·5), and Missouri (1·9%, 0·9–3·8).
By comparing sample seroprevalence data from July, 2020, with Johns Hopkins University
estimates of cumulative PCR-diagnosed cases as of June 15, 2020, the authors estimate
just 9·2% (95% CI 8·7–9·8) of seropositive cases were diagnosed. Given antibodies
take days rather than weeks to appear, this might underestimate the true proportion
of patients diagnosed by swab testing. However, this finding still points to a high
number of people with the virus never being tested. In the absence of clinical data,
it is not clear whether this is because of asymptomatic infection or difficulty accessing
testing, or other reasons.
The study also estimated substantially higher seroprevalence in residents of predominantly
Hispanic (11·3%, 95% CI 9·8–12·9), non-Hispanic Black (13·9%, 12·1–16·0), and Hispanic
and Black (16·3%, 14·3–18·5) neighbourhoods compared with predominantly non-Hispanic
white neighbourhoods (4·8%, 4·1–5·5), when standardised to the US adult population.
This alarming discrepancy is in keeping with trends identified in the largest survey
from Europe
5
and demands urgent attention.
As the authors point out, patients receiving dialysis might be considered an ideal
sentinel population in which to study the evolution of the pandemic, given the guarantee
of regular blood tests, established vascular access, and a high proportion of patients
with multiple risk factors for SARS-CoV-2 infection and COVID-19, including older
age, non-white ethnicity, hypertension, diabetes, and poverty. Importantly, end-stage
kidney disease is considered a qualifying condition for Medicare in the USA, such
that patients effectively enter a universal health-care system.
Extrapolation of seroprevalence in the dialysis population to the general population
is inevitably problematic. Despite adjustments for age, sex, region, and race and
ethnicity, the dialysis population's risk of exposure to SARS-CoV-2 is unlikely to
be representative of the general population: attending a health-care facility three
times a week would seem like a good way to encounter SARS-CoV-2, as has been shown
elsewhere.
11
However, concerns over sample applicability are bidirectional: patients with end-stage
kidney disease and associated comorbidities might be less likely to mount a detectable
antibody response.
12
They are also more likely to die from COVID-19,
13
increasing the chance of unexposed, seronegative survivors being over-represented
in the sample. Although general population estimates from dialysis sampling are imperfect,
they at least remain consistent across the country and from one survey to the next,
permitting longitudinal surveillance.
Despite the massive burden of COVID-19 in the USA, Anand and colleagues show that
a small minority of the population has evidence of humoral immunity to SARS-CoV-2.
Questions remain around the longevity of the immune response and correlates of protection,
but high-quality longitudinal serosurveillance with accompanying clinical data can
help to provide the answers. Anand and colleagues deserve credit for pioneering a
scalable sampling strategy that offers a blueprint for standardised national serosurveillance
in the USA and other countries with a large haemodialysing population.
© 2020 Irfan Khan/Getty Images
2020
Since January 2020 Elsevier has created a COVID-19 resource centre with free information
in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre
is hosted on Elsevier Connect, the company's public news and information website.
Elsevier hereby grants permission to make all its COVID-19-related research that is
available on the COVID-19 resource centre - including this research content - immediately
available in PubMed Central and other publicly funded repositories, such as the WHO
COVID database with rights for unrestricted research re-use and analyses in any form
or by any means with acknowledgement of the original source. These permissions are
granted for free by Elsevier for as long as the COVID-19 resource centre remains active.