Epidemiology & Population Hlth

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

Vascular risk factors, Framingham risk score, and

COVID-19: community-based cohort study

G. David Batty 1* and Mark Hamer 2

1Department of Epidemiology & Public Health, University College London, UK; and 2Division of Surgery & Interventional Science, University College London, UK

Received 1 June 2020; Revised 12 June 2020; editorial decision 17 June 2020; accepted 17 June 2020; online publish-ahead-of-print 16 July 2020

Introduction

Evidence from prognostic studies of groups of COVID-19 patients sug-
gest that those admitted to hospital with existing cardiovascular disease
(CVD) experience worse outcomes relative to those who are CVD-free,
as evidenced by their markedly increased risk of progression to intensive
care and death.1 While obesity also appears to be related to unfavourable
outcomes in COVID-19 patients,2 it is unknown whether this and other
established CVD risk factors, as distinct from CVD itself, are associated
with the occurrence of the infection, particularly in community samples.
More generally, there is evidence from cardiovascular research to suggest
that prognostic characteristics may reveal opposing relationships to those
apparent for aetiological factors. For instance, British people of South
Asian descent have a higher incidence of coronary disease but lower mor-
tality once diagnosed with the condition.3 For the first time to our knowl-
edge, we therefore examined whether unfavourable levels of classic CVD
risk factors, both individually and collectively within the Framingham
model,4 were implicated in the primary prevention of COVID-19 in a
community-based prospective cohort study. Assessing the predictive
value of the Framingham index for COVID-19 has potential clinical utility
as this tool is widely used in general practice in many countries.

Methods

We used data from UK Biobank, a prospective cohort study, the sampling and
procedures of which have been well described. Baseline data collection took
place between 2006 and 2010 across centres in the UK, giving rise to a sample
of 502 655 people (448 919 from England) aged 40–69 years (response rate
5.5%).5 Ethical approval was received from the North-West Multi-centre
Research Ethics Committee.

Cigarette smoking, physician-diagnosed diabetes, highest educational at-
tainment, ethnicity, and physical activity in the previous month were self-
reported using standard enquiries. Body mass index was based on direct
measurements of height and weight. Blood pressure was measured in the
seated position with the average of two readings used, and total cholesterol
and HDL cholesterol were assayed from a non-fasting blood sample. The
Framingham risk score was computed using sex-specific multivariable func-
tions comprising age, total and HDL cholesterol, systolic blood pressure,
smoking, and diabetes status.4 Provided by the Public Health England agency,
data on COVID-19 covered the period 16 March until 26 April 2020. Nose

and/or throat swabs were taken from hospitalized patients, and detection of
SARS-CoV-2 can be regarded as an indication of a severe manifestation of the
disease.

Results

In an analytical sample of 356 914 study members (194 167 women),
there were 700 hospitalizations (300 in women) for COVID-19.
Unfavourable levels of all individual CVD risk factors were related to an
elevated occurrence of COVID-19 in age- and sex-adjusted analyses
(Table 1), the only exception being blood pressure. The magnitude of
these effects was diminished somewhat after adjusting for a range of cova-
riates, but they remained statistically significant at conventional levels.
When we entered selected CVD risk factors into the Framingham model,
an increased likelihood of COVID-19 was apparent in the higher CVD
risk categories, whereas intermediate scores were not associated with
the infection. Again, adjustment for covariates attenuated this relation-
ship. With the Framingham index predicated upon predicting incident
(new cases) vascular events in a population initially free of CVD, we
recomputed effect estimates based on this subgroup (338 553 people,
634 COVID-19 events) and our results were essentially unchanged.

Discussion

Evidence from prognostic studies of hospitalized COVID-19 patients sug-
gests that a series of physical characteristics are linked to progression to
intensive care and death,1,2 and these relationships were also apparent in
the present analyses for age, being male, existing diabetes, and over-
weight/obesity in relation to hospitalization for the infection (Table 1).
Further, a history of CVD was related to an almost doubling of risk of sub-
sequent COVID-19 (1.82; 1.60–2.08). The replication of these relation-
ships in the present data set gives us confidence in our novel results for
CVD risk indices. People in the intermediate risk groups of the
Framingham algorithm experienced a lower risk of COVID-19. A similar
pattern of association was apparent for COVID-19 and age which itself is
the strongest predictor of CVD of those that comprise this index. It
maybe therefore that this ’J’-shaped relation between the Framingham
score and COVID-19 is largely generated by the impact of age.

*Corresponding author: David Batty, Department of Epidemiology & Public Health, University College London, 1–19 Torrington Place, London, WC1E 6BT, UK. Email: [email protected]
Published on behalf of the European Society of Cardiology. All rights reserved. VC The Author(s) 2020. For permissions, please email: [email protected].

Cardiovascular Research (2020) 116, 1664–1665 RESEARCH LETTER
doi:10.1093/cvr/cvaa178

D
ow

nloaded from

ic.oup.com
/cardiovascres/article/116/10/1664/5872319 by C

ollege of the N
orth Atlantic Libraries user on 12 M

arch 2024

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

..

.Follow-up for COVID-19 events was up to 14 years after baseline ex-
amination, and this can raise concerns about the stability of baseline data.
After a median of 4.4 years, a representative subgroup of study participants
were reassessed (n = 19 772). For those risk factors featured in the
Framingham algorithm that were captured at retesting, test–re-test corre-
lation coefficients were high for body mass index (0.93), systolic blood
pressure (0.65), and cigarette smoking (0.84). This suggests that risk factors
gathered at baseline have a high degree of stability. It is also the case that
the UK Biobank study sample is based on the recruitment of 5.5% of the
target population.5 As has been demonstrated,6 the data are therefore in-
appropriate for estimation of risk factor prevalance or disease occurrence.
These observations do not, however, seem to influence reproducibility of
the association of established risk factors for non-communicable disease
such as vascular disease,6 and we think that the same reasoning can be ap-
plied to relationships with communicable diseases.

In conclusion, in the present study, established CVD risk factors revealed
associations with hospitalization for COVID-19 at a magnitude similar to
those apparent for vascular outcomes. For people in the highest risk groups,
the Framingham Risk Score also offered some predictive utility and, if repli-
cated, this finding may have implications for clinical practice and the identifica-
tion of at-risk groups to be targeted when an effective vaccine is developed.

Data availability

Data from the UK Biobank ( are available
to bona fide researchers on application.

Authors’ contributions

G.D.B. and M.H. both generated the idea for the present analyses; G.D.B.
prepared a draft of the manuscript; M.H. analysed the data and edited the

draft manuscript. Part of this research has been conducted using the UK
Biobank Resource under Application 10279.

Conflict of interest: none declared.

Funding
G.D.B. is supported by the Medical Research Council (MR/P023444/1) and
the US National Institute on Aging (1R56AG052519-01; 1R01AG052519-
01A1); M.H. is supported through a joint award from the Economic Social
Research Council and Medical Research Council (RES-579-47-0001).

References
1. Wang B, Li R, Lu Z, Huang Y. Does comorbidity increase the risk of patients with

COVID-19: evidence from meta-analysis. Aging 2020;12:6049–6057.
2. Simonnet A, Chetboun M, Poissy J, Raverdy V, Noulette J, Duhamel A, Labreuche J,

Mathieu D, Pattou F, Jourdain M; LICORN and the Lille COVID-19 and Obesity study
group. High prevalence of obesity in severe acute respiratory syndrome coronavirus-2
(SARS-CoV-2) requiring invasive mechanical ventilation. Obesity 2020;doi: 10.1002/
oby.22831

3. Zaman MJ, Philipson P, Chen R, Farag A, Shipley M, Marmot MG, Timmis AD,
Hemingway H. South Asians and coronary disease: is there discordance between
effects on incidence and prognosis? Heart 2013;99:729–736.

4. D’Agostino RB SrVasan RS, Pencina MJ, Wolf PA, Cobain M, Massaro JM, Kannel WB.
General cardiovascular risk profile for use in primary care: the Framingham Heart
Study. Circulation 2008;117:743–753.

5. Sudlow C, Gallacher J, Allen N, Beral V, Burton P, Danesh J, Downey P, Elliott P,
Green J, Landray M, Liu B, Matthews P, Ong G, Pell J, Silman A, Young A, Sprosen T,
Peakman T, Collins R. UK Biobank: an open access resource for identifying the causes
of a wide range of complex diseases of middle and old age. PLoS Med 2015;12:
e1001779.

6. Batty GD, Gale CR, Kivimäki M, Deary IJ, Bell S. Comparison of risk factor associations
in UK Biobank against representative, general population based studies with conven-
tional response rates: prospective cohort study and individual participant meta-analysis.
BMJ 2020;368;m131.

…………………………………………………………………………………………………………………………………………………………………………………………….

Table 1 Individual CVD risk factors and Framingham risk score in relation to COVID-19 hospitalizations (n ¼ 356 914)

COVID-19 cases/number at riska Age- and sex-adjusted RR (95% CI) Adjustedb RR (95% CI)

Individual risk factors

Age (per 5 year increase) 700/356 914 1.07 (1.02–1.12) 1.05 (1.00–1.11)

Male 400/162 747 1.59 (1.37–1.84) 1.69 (1.45–1.98)

Current smoking 91/35 252 1.57 (1.24–1.99) 1.46 (1.15–1.86)

Physical inactivity 86/21 332 2.31 (1.84–2.84) 1.58 (1.25–2.00)

Obesity/overweight 544/237 440 1.63 (1.36–1.95) 1.45 (1.21–1.74)

Diabetes 63/17 266 1.77 (1.36–2.29) 1.31 (1.01–1.74)

Systolic blood pressure (per SD increase) 700/356 914 1.03 (0.96–1.12) 0.99 (0.92–1.08)

Total cholesterol (per SD increase) 700/356 914 0.82 (0.76–0.89) 0.86 (0.80–0.93)

HDL (per SD increase) 700/356 914 0.70 (0.64–0.77) 0.77 (0.70–0.85)

Framingham risk score (quintiles) Unadjusted RR (95% CI) Adjustedc RR (95% CI)

<_9 162/91 892 1.0 (ref) 1.0 (ref)

10–12 116/84 208 0.78 (0.62–1.0) 0.71 (0.56–0.90)

13–14 125/68 984 1.03 (0.82–1.30) 0.91 (0.72–1.15)

15–16 147/62 248 1.34 (1.07–1.68) 1.15 (0.91–1.45)

>_17 (highest risk) 150/49 582 1.72 (1.38–2.15) 1.35 (1.05–1.70)

aSample sizes correspond to full analytical sample for analyses of continuous risk factors, and the category of interest in analyses of categorical risk factors.
bAdjusted for age, sex, body mass index, physical activity, alcohol, education, and ethnicity.
cAdjusted for body mass index, physical activity, alcohol, education, ethnicity.
CI, confidence interval; RR, relative risk. SD, standard deviation.

Vascular risk factors, Framingham risk score, and COVID-19: community-based cohort study 1665
D

ow
nloaded from

ic.oup.com

/cardiovascres/article/116/10/1664/5872319 by C
ollege of the N

orth Atlantic Libraries user on 12 M
arch 2024