COVID-19 and Diabetes: A Collision and Collusion ofTwo DiseasesEva L. Feldman,1,2 Masha G. Savelieff,1,2 Salim S. Hayek,3 Subramaniam Pennathur,4 Matthias Kretzler,4
and Rodica Pop-Busui5
Diabetes 2020;69:2549–2565 | https://doi.org/10.2337/dbi20-0032
The coronavirus disease 2019 (COVID-19) pandemic hasinfected >22.7 million and led to the deaths of 795,000people worldwide. Patients with diabetes are highly sus-ceptible to COVID-19–induced adverse outcomes andcomplications. The COVID-19 pandemic is superimpos-ing on the preexisting diabetes pandemic to create largeand significantly vulnerable populations of patients withCOVID-19 and diabetes. This article provides an over-view of the clinical evidence on the poorer clinical out-comes of COVID-19 infection in patients with diabetesversus patients without diabetes, including in specific pa-tient populations, such as children, pregnant women, andracial and ethnicminorities. It also draws parallels betweenCOVID-19 and diabetes pathology and suggests that pre-existing complications or pathologies in patients with di-abetesmight aggravate infection course. Finally, this articleoutlines the prospects for long-term sequelae after COVID-19 for vulnerable populations of patients with diabetes.
The coronavirus disease 2019 (COVID-19) pandemic hasinfected.22.7 million and killed.795,000 people world-wide, as of 21 August 2020 (1). COVID-19 infection iscaused by severe acute respiratory syndrome coronavirus2 (SARS-CoV-2), a single-stranded RNA b-coronavirus (2).Patients with diabetes are highly susceptible to adverseoutcomes and complications of COVID-19 infection (3).The COVID-19 pandemic is superimposing on the preex-isting diabetes pandemic to create large and significantlyvulnerable populations of patients with COVID-19 anddiabetes. Other comorbid conditions frequent in patientswith type 2 diabetes, e.g., cardiovascular disease (CVD) and
obesity, also predispose COVID-19 patients to adverseclinical outcomes (4,5).
SARS-CoV-2 pathophysiology remains incompletely un-derstood, but evidence suggests it triggers hyperinflam-mation in certain patients (6) and that tissue tropism isexhibited (7), pathologies shared with chronic inflamma-tion and multitissue damage in diabetes (8). COVID-19infection disrupts glucose regulation, rendering glycemiccontrol difficult and necessitating particularly careful man-agement in patients with diabetes (9). Moreover, earlyindicators and comparison with the previous severe acuterespiratory syndrome coronavirus (SARS-CoV) outbreak(10) suggest that survivors may face sequelae, which willrequire long-term care. Currently, the U.S. and some othercountries are experiencing surges in COVID-19 cases (1).This article will review the current state of knowledge ofCOVID-19 and diabetes to address nine critical questions,some of which remain unanswered (Fig. 1).
Review MethodologyWe initially performed our literature search on PubMedwithout any filters on publication date and completed itby 10 July 2020. The search keywords varied by section.For the diabetes and comorbidities section, we searched“COVID-19” or “SARS-CoV-2” with “clinical characteris-tics,” “clinical cohort,” “clinical,” or “cohort,” and priori-tized clinical, high-quality medical studies. We did notgenerally include meta-analyses and excluded preprints,since we had sufficient peer-reviewed material. To the bestof our ability, we selected studies that appeared to reportdifferent patient cohorts, considering some cohorts may
1Department of Neurology, University of Michigan, Ann Arbor, MI2NeuroNetwork for Emerging Therapies, University of Michigan, Ann Arbor, MI3Division of Cardiology, Department of Internal Medicine, University of Michigan,Ann Arbor, MI4Division of Nephrology, Department of Internal Medicine, University of Michigan,Ann Arbor, MI5Division of Metabolism, Endocrinology and Diabetes, Department of InternalMedicine, University of Michigan, Ann Arbor, MI
Corresponding author: Eva L. Feldman, [email protected]
Received 15 July 2020 and accepted 10 September 2020
This article is part of a special article collection available at https://diabetes.diabetesjournals.org/collection/diabetes-and-COVID19-articles.
E.L.F. and M.G.S. were equal contributing authors.
© 2020 by the American Diabetes Association. Readers may use this article aslong as the work is properly cited, the use is educational and not for profit, and thework is not altered. More information is available at https://www.diabetesjournals.org/content/license.
Diabetes Volume 69, December 2020 2549
PERSPECTIVES
INDIA
BETES
https://doi.org/10.2337/dbi20-0032http://crossmark.crossref.org/dialog/?doi=10.2337/dbi20-0032&domain=pdf&date_stamp=2020-11-07mailto:[email protected]://diabetes.diabetesjournals.org/collection/diabetes-and-COVID19-articleshttps://diabetes.diabetesjournals.org/collection/diabetes-and-COVID19-articleshttps://www.diabetesjournals.org/content/licensehttps://www.diabetesjournals.org/content/license
have been duplicated without reporting it (11). However,we may have included studies from the same cohort if thestudy focus was different. We focused on China, U.S., andEurope as the early epicenters. We also repeated the searchwith the keyword “diabetes,” “acute kidney injury,” or “acutecardiac injury.”We read all abstracts to select relevant manu-scripts, which we searched for the term “diabetes” and allrelevant information. During the revision process, weupdated the review with relevant literature (same criteria)published up until 18 August. For the pediatric section, wesearched “COVID-19” or “SARS-CoV-2” and “diabetes” with“pediatric,” “childhood,” “children,” “youth,” or “adolescent.”For the pregnancy section, we searched “COVID-19” or“SARS-CoV-2” and “diabetes” with “pregnant,” “pregnancy,”or “gestational.” For the race section, we searched “COVID-19” or “SARS-CoV-2” and “race,” “black,” “African American,”“Hispanic,” or “Asian” and prioritized high-quality clinicalstudies. We also performed a subsearch using “diabetes.”
Diabetes and COVID-19
General COVID-19 Patient CohortsAlthough the COVID-19 pandemic evolved quickly, therewere clear early warning signs that comorbidities, includingdiabetes, predisposed patients to adverse outcomes (Table1). The first reports that emerged from Wuhan, China,documented that diabetes raised the risk of dangerousinfection-induced adverse outcomes and complications,leading to acute respiratory distress syndrome (ARDS),intensive care unit (ICU) admission, mechanical ventilationuse, and greater risk of death (12,13). In univariate logisticregression analysis, diabetes had an odds ratio (OR) of 2.85for in-hospital death (13). At the national level, severalChina studies found association of diabetes with severedisease (ICU, mechanical ventilation) (14) and death (14,15).
These findings are replicated in the U.S., where diabetesis one of the three most common comorbid conditionsnationwide, with total comorbidity prevalence as high as78% among ICU COVID-19 admissions (n 5 457 total)(16). In New York City (NYC), patients with diabetes weremore likely to need mechanical ventilation or ICU admis-sion (17,18). In a different NYC cohort, the diabetesunivariate hazard ratio (HR) for in-hospital mortalitywas 1.65, which did not persist in multivariate analysisafter adjustment for age, sex, and seven additional param-eters (5). In Detroit (n5 463), diabetes was more frequentin hospitalized versus discharged and ICU versus non-ICUpatients but was not a risk in multivariate analysis (19).Diabetes was an independent risk for hospital admission(OR 2.24, with full adjustment for patient characteristicsand comorbidities) but not for critical disease or death ina large NYC cohort (n 5 5,279) (20).
In other countries, a German study (n 5 50) found nodifferences in diabetes frequency in ARDS versus non-ARDS patients (21), though these outcomes contrast withthose of another study in China (22). An observationalU.K. study (n5 1,157) found that diabetes had an age- andsex-adjusted HR of 1.42 for critical care and could beintegrated into a 12-point prognostic risk score (criticalcare admission, death) (23), similar to another 10-variablerisk score (24). Collectively, these general cohort studiessuggest that patients with diabetes have a higher likeli-hood of adverse outcomes, although other mitigating riskfactors likely exist, contributing to the varying conclusions.
Cohorts of Patients With COVID-19 and DiabetesSeveral reports have focused specifically on cohorts ofpatients with diabetes. The multicenter French Coronavi-rus SARS-CoV-2 and Diabetes Outcomes (CORONADO)
Figure 1—Outstanding questions on diabetes in the context of COVID-19.
2550 COVID-19 and Diabetes: Collision and Collusion Diabetes Volume 69, December 2020
Tab
le1—
Ove
rview
ofad
ultCOVID
-19clinical
coho
rts
Study
Loca
tion
Partic
ipan
ts(n)
Diabetes
find
ings
Com
orbidities
find
ings
*Selec
tlaboratoryfind
ings
**
Wan
get
al.
(12)
Wuh
an,China
138
Patientswith
diabetes
cons
tituted
22.2%
ofICU
patientsvs.5.9%
ofno
n-ICU
patients,
P5
0.00
9
CVD,hy
pertens
ion,
cerebrova
scular
disea
sepredispos
edto
ICU
Eleva
tedWBC,n
eutrop
hils,A
LT,A
ST,
CK-M
B,
Cr,
D-dim
er,hs
-TnI,LD
H,PCT,
and
lympho
pen
iain
ICU
vs.no
n-ICU
patients
Zho
uet
al.(13
)Wuh
an,China
191
Patientswith
diabetes
cons
tituted
31%
ofno
nsurvivo
rsvs.14
%of
survivors(P
50.00
51);OR
2.85
(95%
CI1.35
–6.05
;P5
0.00
62)forin-hos
pita
ldea
thin
aun
ivariate
mod
el
CVD,24
%no
nsurvivo
rsvs.1%
survivors(P
,0.00
01),OR
21.40(95%
CI4.64
–98
.76;
P,
0.00
01)in
univariate
mod
el;hy
pertens
ion,
48%
nons
urvivo
rsvs.23
%su
rvivors(P
50.00
08),OR
3.05
(95%
CI1.57
–5.92
;P5
0.00
10)in
univariate
mod
el
Eleva
tedWBC,ALT
,CK,Cr,
D-dim
er,ferritin,
hs-TnI,IL-6,LD
H,PT,
andPCTha
dsign
ifica
ntHR
.1fordea
thin
univariate
mod
el;D-dim
erha
dsign
ifica
ntHR
.1for
dea
thin
multiv
ariate
mod
el
Gua
net
al.
(14)
China
1,09
916
.2%
ofpatientswith
seve
revs.5.7%
with
nons
evereCOVID-19infections
had
diabetes
,an
d26
.9%
that
met
vs.6.1%
that
did
notm
eetthe
prim
aryco
mpos
iteen
dpoint
(ICU,mec
hanica
lven
tilationus
e,dea
th)ha
ddiabetes
;no
Pva
lues
5.8%
ofse
vere
vs.1.8%
ofno
nsev
ereCOVID-
19patientsha
dCHD,a
nd9.0%
that
met
vs.
2.0%
that
did
notmee
ttheprim
ary
compos
iteen
dpoint
hadCHD;23
.7%
ofse
vere
vs.13
.4%
ofno
nsev
ereCOVID-19
patientsha
dhy
pertens
ion,
and35
.8%
that
met
vs.13
.7%
that
did
notmee
ttheprim
ary
compos
iteen
dpoint
hadhy
pertens
ion;
noPva
lues
Eleva
tedWBC,ALT
,AST,
CRP,D-dim
er,LD
H,
PCT,
andlympho
pen
iain
seve
revs.
nons
evereinfectionan
din
patientsthat
met
vs.did
notmee
ttheprim
aryco
mpos
iteen
dpoint,no
Pva
lues
Wuan
dMcG
ooga
n(15)
China
72,314
total,
44,672
confi
rmed
(factored
into
CFR
)
CFR
7.3%
inpatientswith
diabetes
vs.2.3%
fortheen
tireco
hort
CFR
10.5%
forCVD,6.0%
forhy
pertens
ion
Not
exam
ined
Richa
rdso
net
al.(17)
NYC
area
5,70
0Diabetes
oneof
threemos
tco
mmon
morbidities
.Patientswith
diabetes
more
likelyto
need
mec
hanica
lven
tilationor
ICU
Hyp
ertens
ionan
dob
esity
twoof
threemos
tco
mmon
morbidities
.Hyp
ertens
ivepatients
less
likelyto
need
mec
hanica
lven
tilationor
ICU;88
%of
COVID-19patientsha
dtw
oor
moreco
morbidities
compared
with
one
(6.3%
)or
none
(6.1%
).
Eleva
tedALT
,AST,
BNP,C
RP, D
-dim
er,ferritin,
LDH,PCT,
andlympho
pen
iain
hosp
italized
COVID-19patients
Goy
alet
al.
(18)
NYC
393
Diabetes
was
morefreq
uent
inpatients
requirin
gmec
hanica
lven
tilation(27.7%
)vs.no
t(24.0%
)(P
valueno
tstated
)
Hyp
ertens
ion,
CAD,an
dob
esity
weremore
freq
uent
inpatientsrequirin
gmec
hanica
lve
ntilatio
n(P
values
notstated
)
Majority
ofpatientsha
dlympho
pen
ia(90.0%
),thromboc
ytop
enia
(27%
);man
yha
delev
ated
liver
func
tionva
lues
andinflam
matory
marke
rs(CRP,D-dim
er,ferritin,
PCT),which
werefurthe
rincrea
sedin
patientsrequirin
gmec
hanica
lven
tilation
Cum
mings
etal.(5)
NYC
1,15
0Diabetes
oneof
threemos
tco
mmon
morbidities
.Univa
riate
HR
1.65
(95%
CI
1.11
–2.44
),no
tsign
ifica
ntin
multiv
ariate
HR
1.31
(95%
CI0.81
–2.10
)forin-hos
pita
lmortality
Hyp
ertens
ionan
dob
esity
twoof
threemos
tco
mmon
morbidities
.Hyp
ertens
ion
univariate
HR2.24
(95%
CI1
.40–
3.59
);CCD
univariate
HR
2.21
(95%
CI1.44
–3.39
),multiv
ariate
HR
1.76
(95%
CI1.08
–2.86
);BMI$
40kg
/m2no
tsign
ifica
ntun
ivariate
HR
0.76
(95%
CI0.40
–1.47
)forin-hos
pita
lmortality;
CKD
was
notaris
kforin-hos
pita
ldea
th
Asidefrom
othe
ralte
redmarke
rs,IL-6un
ivariate
HR1.12
(95%
CI1
.04–
1.21
)and
multiv
ariate
HR
1.11
(95%
CI1.02
–1.20
)an
dD-dim
erun
ivariate
HR
1.18
(95%
CI1.10
–1.27
)an
dmultiv
ariate
HR
1.10
(95%
CI1.01
–1.19
)for
in-hos
pita
lmortality
Con
tinue
don
p.25
52
diabetes.diabetesjournals.org Feldman and Associates 2551
Tab
le1—
Continue
dStudy
Loca
tion
Partic
ipan
ts(n)
Diabetes
find
ings
Com
orbidities
find
ings
*Selec
tlaboratoryfind
ings
**
Suley
man
etal.(19)
Detroit,
MI
463
Diabetes
was
morefreq
uent
inho
spita
lized
(43.4%
)vs.
disch
arge
d(20.4%
)patients(P
,0.00
1).Itwas
also
morefreq
uent
inICU
(51.8%
)vs.no
n-ICU
(38.8%
)patients(P
50.02
)but
was
notaris
kin
multiv
ariate
analysis
forICU
ormec
hanica
lven
tilation.
African
American
race
was
notm
orefreq
uent
inad
mitted
orICUvs.d
isch
arge
dpatientsor
aris
kformec
hanica
lven
tilationor
dea
th
Hyp
ertens
ion,
CVD,ob
esity
,an
dCKD
were
morefreq
uent
inho
spita
lized
vs.disch
arge
dpatients.
Hyp
ertens
ionan
dCKD
werealso
morefreq
uent
inICU
vs.no
n-ICU
patients.
CKD
andse
vere
obes
itywereris
ksin
multiv
ariate
analysis
forICU
ormec
hanica
lve
ntilatio
n
Eleva
tedAST,
Cr,an
dhs
-TnI;lower
WBC;an
dlympho
pen
iain
hosp
italized
vs.disch
arge
dpatientsbyun
ivariate
analysis.Eleva
ted
WBC,A
ST,
Cr,
D-dim
er,ferritin,h
s-Tn
I,LD
H,
PCT,
andlympho
pen
iain
ICU
vs.no
n-ICU
patientsbyun
ivariate
analysis
Petrilliet
al.
(20)
NYC
5,27
9Diabetes
hadmultiv
ariate
OR
2.24
(95%
CI
1.84
–2.73
;P,
0.00
1)forh
ospita
ladmission
,with
adjustmen
tforpatient
charac
teris
tics,
comorbidities
Allmultiv
ariate:he
artfailure
OR
4.43
(95%
CI
2.59
–8.04
;P,
0.00
1),hy
pertens
ionOR
1.78
(95%
CI1.49
–2.12
;P,
0.00
1),CKD
OR
2.6(95%
CI1.89
–3.61
;P,
0.00
1),
hyperlip
idem
iaOR
0.62
(95%
CI0.52
–0.74
;P,
0.00
1),B
MI25.0–
29.9
kg/m
2(ove
rweigh
t)OR
1.3(95%
CI1.07
–1.57
;P5
0.00
7),BMI
30–39
.9kg
/m2(obes
eclas
sIan
dII)
OR
1.8
(95%
CI1.47–
2.2;
P,
0.00
1),B
MI$
40kg
/m2
(obes
eclas
sIII)O
R2.45
(95%
CI1
.78–
3.36
;P,
0.00
1);allfor
hosp
itala
dmission
,ad
justed
forsa
meva
riablesas
diabetes
Eleva
tedCr,CRP,D-dim
er,PCT,
trop
onin,an
dlympho
pen
iain
critica
lCOVID-19
Drehe
ret
al.
(21)
Aac
hen,
German
y50
Diabetes
did
notraisetheris
kforARDS;no
Pva
lues
Obes
ity,but
nothy
pertens
ion,
raised
theris
kforARDS;no
Pva
lues
Eleva
tedWBC,CK,CRP,D-dim
er,IL-6,LD
H,
andPCT;
noPva
lues
Wuet
al.(22)
Wuh
an,China
201
Diabetes
was
morefreq
uent
inARDS(19.0%
)than
non-ARDS(5.1%)patients(P
50.00
2);
riskforARDS(HR
2.34
[95%
CI1.35
–4.05
];P5
0.00
2)but
notdea
th(HR
1.58
[95%
CI
0.80
–3.13
];P5
0.19
)
Hyp
ertens
ionwas
morefreq
uent
inARDS
(27.4%
)than
non-ARDS(13.7%
)patients
(P5
0.02
),ris
kforARDS(HR
1.82
[95%
CI
1.13
–2.95
];P5
0.01
)but
notdea
th(HR1.70
[95%
CI0.92
–3.14
];P5
0.09
)
Eleva
tedne
utrophils,AST,
CRP,D-dim
er,
ferritin,
LDH,and
PTha
dsign
ifica
ntHR.1for
ARDS;ne
utrophils,D-dim
er,IL-6,an
dLD
Hha
dsign
ifica
ntHR
.1fordea
th
Galloway
etal.
(23)
U.K.
1,15
7Diabetes
hadHR
(adjustmen
tforse
x,ag
e)of
1.42
forcritica
lcare(95%
CI1.04
–1.95
;P5
0.02
9)
Hyp
ertens
ionha
dHR(adjusted
forse
x,ag
e)of
1.53
forcritica
lcareor
dea
th(95%
CI1.24
–
1.90
;P5
0.00
0)
Neu
trop
hils,C
r,an
dCRPha
dsign
ifica
ntHR.1
forcritica
lcareor
dea
th
Lian
get
al.
(24)
China
1,59
0,disco
very
coho
rt;71
0,va
lidation
coho
rt;ris
ksc
oreforcritica
lillne
ss
6.8%
nonc
ritical
vs.23
.7%
critica
ldisea
seam
ongpatientswith
diabetes
3.2%
nonc
ritical
vs.9.9%
critica
ldisea
seam
ongCVD
patients;
14.8%
nonc
ritical
vs.
40.5%
critica
ldisea
seam
onghy
pertens
ion
patients.
Num
ber
ofco
morbidities
hadOR
1.60
(95%
CI1.27
–2.00
;P,
0.00
1)in
multiv
ariate
analysis
Asidefrom
othe
ralte
redmarke
rs,n
eutrop
hil-to-
lympho
cyte
ratio
(OR
1.06
[95%
CI1.02
–
1.10
];P5
0.00
3)an
dLD
H(OR1.00
2[95%
CI
1.00
1–1.00
4];P,
0.00
1)integrated
into
a10
-point
risksc
oreforcritica
lillnes
s
Cariouet
al.
(25)
Fran
ce1,31
7,of
who
m1,16
6with
T2D
Diabetes
type,
HbA1c,gluc
ose-lowering
therap
yus
edid
notaffect
prim
aryou
tcom
e(m
echa
nica
lven
tilationan
d/ordea
thwith
in7day
sof
admission
)in
univariate
analysis
Micro-(OR
2.14
[95%
CI1.16
–3.94
];P5
0.01
53)a
ndmac
rova
scular
(OR2.54
[95%
CI
1.44
–4.50
];P5
0.00
13)co
mplications
indep
enden
tlyas
sociated
with
7-day
mortality;
BMImultiv
ariate
OR
1.28
(95%
CI
1.10
–1.47
),P,
0.00
10forco
mpos
ite
AST(OR2.23
[95%
CI1
.70–
2.93
];P,
0.00
01)
andCRP(O
R1.93
[95%
CI1.43
–2.59
];P,
0.00
01)indep
enden
tlyas
sociated
with
prim
aryou
tcom
e;high
erlympho
cyteswere
protective(O
R0.67
[95%
CI0.50
–0.88
];P5
0.00
50)
Con
tinue
don
p.25
53
2552 COVID-19 and Diabetes: Collision and Collusion Diabetes Volume 69, December 2020
Tab
le1—
Continue
dStudy
Loca
tion
Partic
ipan
ts(n)
Diabetes
find
ings
Com
orbidities
find
ings
*Selec
tlaboratoryfind
ings
**
Barronet
al.
(26)
U.K.
23,698
COVID-
19dea
ths,
364T1
DCOVID-19
dea
ths,
7,43
4T2
DCOVID-19
dea
ths
T1DOR2.86
(95%
CI2
.58–
3.18
;P,
0.00
1)an
dT2
DOR1.80
(95%
CI1
.75–
1.86
;P,
0.00
1)fordea
th,with
adjustmen
tforag
e,se
x,dep
rivation,
ethn
icity
,CVD,ce
rebrova
scular
disea
se
CVD
andce
rebrova
scular
disea
semore
freq
uent
inT1
Dan
dT2
Dvs.no
ndiabetes
inCOVID-19dea
ths
Not
exam
ined
Zha
nget
al.
(27)
Wuh
an,China
258,
ofwho
m63
with
diabetes
Diabetes
hadmultiv
ariate
HR
3.64
(95%
CI
1.09
–12
.21;
P5
0.03
6);elev
ated
FBG
(.7.54
mmol)h
admultiv
ariate
HR1.19
(95%
CI1.08
–1.31
;P,
0.00
1);bothfordea
thad
justed
forag
e,CVD,CKD,inflam
matory
marke
rs
CVD
morefreq
uent
inpatientswith
diabetes
(23.8%
)vs.patientswith
outdiabetes
(12.3%
),P5
0.02
7;CKD
morefreq
uent
inpatientswith
diabetes
(8.8%)vs
.patients
with
outpatients(2.1%),P5
0.02
7
Eleva
tedWBC,ne
utrophils,CK-M
B,D-dim
er,
TTin
patientswith
diabetes
vs.patients
with
outdiabetes
Guo
etal.(28)
Wuh
an,China
174,
overall
analysis;50
,su
bgrou
pan
alysis
Patientswith
diabetes
with
outan
yothe
rco
morbidities
(16.5%
)diedmoreoftenthan
patientswith
outdiabetes
with
out
comorbidities
(0%)(P
50.03
);ho
wev
er,the
latter
patientswereyo
unge
r
CVD
was
morepreva
lent
inpatientswith
diabetes
,P5
0.01
3Eleva
tedne
utrophils,D-dim
er,an
dESR,an
dlympho
pen
iain
patientswith
diabetes
vs.
patientswith
outdiabetes
;ne
utrophils,ALT
,CRP,D-dim
er,ESR,ferritin,
IL-6,LD
H,an
dlympho
pen
iain
patientswith
diabetes
vs.
patientswith
outdiabetes
with
out
comorbidities
;ho
wev
er,thelatter
patients
wereyo
unge
r
Zhu
etal.(3)
Hub
eiProvinc
e,China
7,33
7,of
who
m95
2with
T2D
T2D
patientsha
dhigh
ermortality:
7.8%
vs.
2.7%
overall,ad
justed
HR1.49
(95%
CI1.13–
1.96
;P5
0.00
5);well-co
ntrolledblood
gluc
oseco
nferslower
all-ca
usemortality,
adjusted
HR
0.14
(95%
CI0.03
–0.60
;P5
0.00
8)
Blood
gluc
oseco
rrelated
with
comorbid
CHD,
hypertens
ion
T2D
patientsha
delev
ated
WBC,ne
utrophils,
Cr,CRP,D-dim
er,IL-6,LD
H,PCT,
and
lympho
pen
iavs.patientswith
outdiabetes
;T2
Dpatientswith
well-co
ntrolledvs
.poo
rlyco
ntrolledblood
gluc
oseha
dsign
ifica
ntly
fewer
inciden
cesof
elev
ated
WBC,
neutrophils,ALT
,AST,
Cr,CRP,D-dim
er,
PCT,
andlympho
pen
ia;no
Pva
lues
Iaco
bellis
etal.(29)
Miami,FL
85Admission
hyperglyc
emia
bes
tpredictedpoo
rch
estradiologica
loutco
mes
BMIco
rrelated
with
poo
rch
estradiologica
lou
tcom
esNot
exam
ined
Liet
al.(30)
Wuh
an,China
132,
ofwho
m13
0with
T2D
Patientswith
diabetes
stratifi
edbyad
mission
gluc
ose:
grou
p1(#
11mmol/L)vs
.grou
p2(.
11mmol/L);grou
p2ha
dlong
erdiabetes
duration,
morelikelyto
suffer
ACI,ICU
admission
,dea
th
Nodifferen
cein
comorbidities
ingrou
p1vs.
grou
p2
Eleva
tedWBC,CRP,D-dim
er,ESR,IL-6,an
dlympho
pen
iain
grou
p2vs.grou
p1;
WBC
(.10
9/L),Cr(,
57/0
mmol/L),
D-dim
er($
1.5
mg/L),hs
-TnI
(.26
.2pg/mL),LD
H(.
245
units
/L),PCTun
ivariate
OR.1forin-ho
spita
lco
mplications
Cha
oet
al.
(31)
Taiwan
452
Highgluc
oseva
riability
with
inthefirstday
ofICU
admission
correlated
with
30-day
mortality,
partic
ularly
inpatientswith
out
diabetes
.Highgluc
oseva
riability
was
more
freq
uent
inpatientswith
diabetes
Exc
eptfordiabetes
,no
differen
cein
othe
rco
morbidities
(e.g.,CKD,CHD,
cerebrova
scular
disea
se)inpatientswith
high
vs.low
gluc
oseva
riability;APACHEIIsc
ore
indep
enden
tlyco
rrelated
with
high
er30
-day
mortality
Nodifferen
cesin
Cr,CRP,a
ndPCTin
patients
with
high
vs.low
gluc
oseva
riability
Con
tinue
don
p.25
54
diabetes.diabetesjournals.org Feldman and Associates 2553
Tab
le1—
Continue
dStudy
Loca
tion
Partic
ipan
ts(n)
Diabetes
find
ings
Com
orbidities
find
ings
*Selec
tlaboratoryfind
ings
**
Bod
eet
al.
(32)
U.S.
1,12
2Diabetes
and/orun
controlledhy
perglyc
emia
increa
sedho
spita
llen
gthof
stay
and
mortality
Kidne
yfunc
tion,
asas
sessed
byeG
FR,was
lower
inpatientswith
diabetes
and/or
unco
ntrolledhy
perglyc
emia
atad
mission
Eleva
tedCrin
patientswith
diabetes
and/or
unco
ntrolledhy
perglyc
emia
vs.patients
with
outdiabetes
orwith
controlledblood
gluc
osepatients
Williamso
net
al.(4)
U.K.
10,926
COVID-
19dea
thsvs
.17
,278
,392
controls
ubjects
Diabetes
with
HbA1c,7.5%
(58mmol/m
ol),HR
1.31
(95%
CI1.24
–1.37
),an
dwith
HbA1c$7.5%
(58mmol/m
ol),HR1.95
(95%
CI1.83
–2.07
),fordea
th,ad
justed
forag
e,se
x,co
morbidities
,sm
oking,
socioe
cono
mic
status
.Mixed
race
,HR
1.43
(95%
CI1.11
–
1.85
);Sou
thAsian
,HR
1.44
(95%
CI1.32
–
1.58
);an
dBlack
,HR1.48
(95%
CI1
.30–
1.69
);ris
ksfordea
thafterad
justmen
tforthesa
me
varia
bles
BMI30
–34
.9kg
/m2(obes
eclas
sI)
nons
ignifica
ntHR
1.05
(95%
CI1.00
–1.11
),BMI35
–39
.9kg
/m2(obes
eclas
sII)
HR
1.40
(95%
CI1.30
–1.52
),BMI$40
kg/m
2(obes
eclas
sIII)HR
1.92
(95%
CI1.72
–2.13
),hy
pertens
ionHR
0.89
(95%
CI0.85
–0.93
),CHD
HR
1.17
(95%
CI1.12
–1.22
),reduc
edkidne
yfunc
tioneG
FR30
–60
mL/min/1.73m
2
HR1.33
(95%
CI1
.28–
1.40
),eG
FR,30
mL/
min/1.73m
2HR
2.52
(95%
CI2.33
–2.72
),stroke
/dem
entia
HR2.16
(95%
CI2
.06–
2.27
),fordea
th,ad
justed
forthesa
meparam
eters
asdiabetes
Not
exam
ined
Holman
etal.
(33)
U.K.
464T1
DCOVID-19
dea
ths,
10,525
T2D
COVID-19
dea
ths
T1D:HbA1c$10
.0%
(86mmol/m
ol)HR
2.23
,T2
D:HbA1c7.5–
8.9%
(59–
74mmol/m
ol)HR
1.22
(95%
CI1.15
–1.30
),HbA1c9.0–
9.9%
(75–
85mmol/m
ol)HR
1.36
(95%
CI1.24
–
1.50
),HbA1c$10
.0%
(86mmol/m
ol)H
R1.61
(95%
CI1
.47–
1.77
);allP
,0.00
01,a
djusted
forag
e,se
x,dep
rivation,
ethn
icity
,clinical,
CVD,CKD,am
ongothe
rs
T1D:inv
erse
relatio
nof
eGFR
with
HR;U
-sha
pe
relatio
nof
BMIwith
HR,referenc
eto
overweigh
tca
tego
ry(BMI25
.0–29
.9kg
/m2);
CVD
HR.1,
nosign
ifica
nceof
hypertens
ion
andch
oles
terol.T2
Dha
dthesa
meris
ks,p
lus
hypertens
ionHR
,1
Not
exam
ined
Zha
nget
al.
(34)
Wuh
an,China
166
Diabetes
andhy
perglyc
emia
seco
ndaryto
COVID-19increa
setheris
kof
critica
ldisea
se(32.8%
and38
.1%,resp
ectiv
ely,
vs.9.5%
overall,P,
0.05
forboth)
andco
mpos
iteou
tcom
e(IC
U,mec
hanica
lven
tilationus
e,dea
th)
Hyp
ertens
ionwas
freq
uent
inpatientswith
diabetes
andse
cond
aryhy
perglyc
emia
(P5
0.02
9)
Eleva
tedWBC,ne
utrophils,ALT
,AST,
CRP, D
-dim
er,E
SR,ferritin,IL-8,
LDH,a
ndN-terminal
pro-B
NPin
COVID-19patientswith
diabetes
andhy
perglyc
emia
seco
ndaryvs
.with
out
diabetes
andwith
norm
oglyce
mia
Wan
get
al.
(35)
Wuh
an,China
605
Admission
FBG
$7.0mmol/L
multiv
ariate
HR
2.30
(95%
CI1.49
–3.55
;P5
0.00
02)for
28-day
mortality;
admission
FBG
$7.0an
d6.1–
6.9vs.,6.1mmol/L
OR
3.99
(95%
CI
2.71
–5.88
)an
d2.61
(95%
CI1.64
–4.41
),resp
ectiv
ely,
for28
-day
in-hos
pita
lco
mplications
Hyp
ertens
ionan
dCHD
hadno
sign
ifica
nteffect
on28
-day
mortality;
CKD
and
cerebrova
scular
disea
seha
dun
ivariate
HR
.1for28
-day
mortality
Not
exam
ined
Smith
etal.
(36)
NJ
184
Mos
tpatientsha
ddiabetes
(62.0%
)or
prediabetes
(23.9%
);intubated
patientsha
dhigh
erFB
G(P
50.01
3)an
dHbA1c(P
50.03
4)vs.no
nintub
ated
Mos
tco
mmon
preex
istin
gco
ndition
s:hy
pertens
ion(60.3%
),hy
perlip
idem
ia(33.7%
),dem
entia
(13.0%
),CKD
(13.0%
),CAD
(12.0%
),an
dCHD
(10.9%
);intubated
patientsha
dhigh
erBMI(P
50.03
0)vs.
nonintub
ated
Not
exam
ined
Con
tinue
don
p.25
55
2554 COVID-19 and Diabetes: Collision and Collusion Diabetes Volume 69, December 2020
Tab
le1—
Continue
dStudy
Loca
tion
Partic
ipan
ts(n)
Diabetes
find
ings
Com
orbidities
find
ings
*Selec
tlaboratoryfind
ings
**
Sim
onne
tet
al.(39)
Lille,Fran
ce12
4Diabetes
was
notaris
kfactor
inun
ivariate
logistic
regres
sion
analysis
Obes
ity($
35kg
/m2BMI)un
ivariate
OR
6.75
(95%
CI1
.76–
25.85;
P5
0.01
5),m
ultiv
ariate
OR
7.36
(95%
CI1.63
–33
.14;
P5
0.02
1);
hypertens
ionun
ivariate
OR
2.81
(95%
CI
1.25
–6.3;
P5
0.01
2)but
notsign
ifica
ntin
multiv
ariate
analysis;dyslip
idem
iawas
not
aris
kfactor
inun
ivariate
logistic
regres
sion
analysis
Not
exam
ined
Gao
etal.(41)
Wen
zhou
,China
150
Diabetes
morepreva
lent
inob
ese(24.0%
)vs.
nono
bes
e(14.7%
)COVID-19patients
Obes
ityha
dOR
3.00
(95%
CI1.22
–7.38
)after
adjustmen
tforag
e,se
x,sm
okingstatus
,hy
pertens
ion,
diabetes
,dyslip
idem
ia
Eleva
tedCRPan
dlympho
pen
iain
obes
evs.
nono
bes
eCOVID-19patients
Shi
etal.(43)
Wuh
an,China
1,56
1,of
who
m15
3with
diabetes
analyzed
vs.
153ag
e-an
dse
x-match
ed15
3patients
with
out
diabetes
Diabetes
(multiv
ariate
HR
1.58
[95%
CI0.84
–
2.99
])no
tan
indep
enden
tris
kforin-hos
pita
ldea
th;patientswith
diabetes
likelierto
be
admitted
toICU
andex
perienc
eco
mplications
(ACI,AKI,ARDS,etc.)an
ddea
th;no
nsurvivo
rpatientswith
diabetes
likelierto
have
hypertens
ionan
dCVD
(P,
0.05
);hy
pertens
ionmultiv
ariate
HR
3.10
(95%
CI1.14
–8.44
)forin-hos
pita
ldea
thof
patientswith
diabetes
Hyp
ertens
ionmultiv
ariate
HR
2.50
(95%
CI
1.30
–4.78
)an
dCVD
multiv
ariate
HR
2.24
(95%
CI1.19
–4.23
)as
sociated
with
in-hos
pita
ldea
th
Eleva
tedPCTan
dlower
CD81
Tce
llsinpatients
with
diabetes
vs.patientswith
outdiabetes
;elev
ated
gluc
ose,
HbA1c,WBC,ne
utrophils,
Cr,CRP, D
-dim
er,P
CT,
PT,
andlympho
pen
iaan
dlower
eGFR
,CD31
,CD41
,CD81
,CD19
1,an
dCD16
156
1ce
llsin
nons
urvivo
rvs.su
rvivor
patientswith
diabetes
Lassaleet
al.
(40)
U.K.
640COVID-19
hosp
italizations
from
340,96
6registrantsin
UK
Bioban
ksu
bse
tfrom
900COVID-19
hosp
italizations
and42
8,49
4registrants
Diabetes
morepreva
lent
andHbA1chigh
erin
hosp
italized
vs.n
onho
spita
lized
patients(fu
lldatase
t),P,
0.00
1;Lo
gHbA1cremaine
das
sociated
inmultiv
ariate
analysis
(OR
1.60
[95%
CI1
.02–
2.52
];P5
0.04
3;su
b–datase
t);diabetes
morepreva
lent
inBlack
andAsian
patients(fu
lldatase
t)
CVD,hy
pertens
ion,
BMI,WHR
high
eran
dch
oles
terol,HDL-clower
inho
spita
lized
vs.
nonh
ospita
lized
patients(fu
lldatase
t),P,
0.00
1;BMI,WHR,ch
oles
terolrem
aine
dsign
ifica
ntin
multiv
ariate
analysis;Black
patients(OR
2.66
[95%
CI1.82
–3.91
];P,
0.00
1)moresu
scep
tible
toho
spita
lization,
with
adjustmen
tforag
e,se
x,co
morbidities
,an
dso
cioe
cono
mic
factors
Eleva
tedCRPin
hosp
italized
vs.
nonh
ospita
lized
COVID-19patientsbut
did
notremainsign
ifica
ntin
multiv
ariate
analysis
Pric
e-Hay
woo
det
al.(45)
LA3,48
118
.5%
ofBlack
patientsha
ddiabetes
vs.1
0.9%
White.Noan
alysis
perform
edto
disea
sese
verity.
Black
race
was
aho
spita
lizationris
kbut
notan
indep
enden
tin-hos
pita
lmortality
risk
Cha
rlson
Com
orbidity
Index
scoreOR
1.05
(95%
CI1.00
–1.10
)forho
spita
lization
(acc
ountingforrace
,ag
e,se
x,low-inc
ome
area
ofresiden
ce,insu
ranc
eplan,
obes
ity)
but
HR
0.99
(95%
CI0.94
–1.03
)for
in-hos
pita
ldea
th;hy
pertens
ionan
dCKD
morepreva
lent
inBlack
vs.White
patients
Asidefrom
othe
ralte
redmarke
rs,A
ST,
Cr,CRP,
PCT,
andlympho
pen
iaha
dsign
ifica
ntHR.1
forin-hos
pita
ldea
th,afterad
justmen
tfor
race
,ag
e,se
x,co
morbidities
,low-inc
ome
area
ofresiden
ce,an
dlaboratorymea
sures
ALT
,alanine
aminotrans
ferase
;APACHEII,
Acu
tePhy
siolog
yan
dChron
icHea
lthEva
luationII;
BNP,b
rain
natriuretic
peptid
e;CAD,c
oron
aryartery
dise
ase;
CCD,c
hron
icca
rdiacdisea
se;
CFR
,cas
efatalityrate;C
HD,c
oron
aryhe
artd
isea
se;C
K,c
reatinekina
se;C
K-M
B,c
reatinekina
se,m
usclean
dbraintype;
CRP,C
-rea
ctiveprotein;e
GFR
,estim
ated
glom
erular
filtrationrate;
FBG,fas
tingblood
gluc
ose;
HDL-c,
high
-den
sity
lipop
roteinch
oles
terol;IL-6,interleuk
in6;
PT,
prothrombintim
e;T1
D,typ
e1diabetes
;T2D
,typ
e2diabetes
;TT,
thrombintim
e.*C
ondition
sco
morbid
with
diabetes
cons
idered
.**Selec
tlab
oratoryfind
ings
fors
ignifica
ntdifferen
cesreportedin
immun
ece
llpop
ulations
,cytok
ines
,and
biomarke
rsof
infectionan
dkidne
y,liver,a
ndca
rdiacdam
age.
Cha
nges
wererepo
rted
ifthereweresign
ifica
ntdifferen
cesin
either
mea
nva
lues
orin
thenu
mber
ofpa
tientsab
oveacu
toffva
lue.
diabetes.diabetesjournals.org Feldman and Associates 2555
study (n 5 1,317 participants with diabetes, 88.5% ofwhom had type 2 diabetes) observed that diabetes typeand glycated hemoglobin (HbA1c) level did not affect theprimary outcome in univariate analysis, i.e., tracheal intuba-tion for mechanical ventilation and/or death within 7 days ofadmission (25). Another large study, led by the NationalHealth Service (NHS) England, also focused on both type1 (n 5 364) and type 2 (n 5 7,434) diabetes–associatedCOVID-19 deaths and determined multivariate ORs of2.86 and 1.80, respectively, with adjustment for age, sex,ethnicity, deprivation, CVD, and cerebrovascular disease,though they could not adjust for other frequent comor-bidities, hypertension, chronic kidney disease (CKD), andBMI, due to data set limitations (26). Notably, moststudies have not differentiated diabetes type; CORO-NADO found no differences between type 1 and type2 diabetes in COVID-19 outcomes, but there were only39 patients with type 1 diabetes. In contrast, the NHSEngland study might suggest that patients with type1 diabetes are at greater risk, though this remains tobe validated by additional studies (Fig. 1).
A study from China with 258 COVID-19 patients, of whom63 had diabetes, reported diabetes had a multivariate HR of3.64 for death, with adjustment for age, comorbidities, andinflammatory markers (27). Guo et al. (28) accounted forcomorbidities by comparing mortality in patients without di-abetes (0%) versus with diabetes (16.5%) without comorbid-ities; however, they failed to consider age, which significantlydiffered between groups. In a study of COVID-19 patients withtype 2 diabetes, diabetes led to a higher all-cause mortality of7.8% (vs. 2.7%), with HR 1.49, with adjustment for age, sex,and infection severity (3). These studies of cohorts with
diabetes confirm the concept that persons with diabeteswho contract COVID-19 disease have poorer outcomes.
Glycemic Control and Elevated Fasting Blood GlucoseWell-controlled blood glucose has emerged as an importantoutcome parameter and conferred lower mortality (HR 0.14)in a propensity score–matching model that accounted for age,sex, comorbidities, and several additional parameters (3). Thisfinding agrees with other studies that identified diabetes and/or uncontrolled or variable hyperglycemia at admission(29,30), ICU admission (31), or during in-hospital stay (32)as a severe disease or mortality risk. In the large U.K. Open-SAFELY study of 10,926 COVID-19 deaths in comparisonwitha database of 17,278,392 adults, greater mortality occurredwith poorer glycemic control (stratified by HbA1c) (4). Patientswith diabetes with HbA1c ,7.5% had a fully adjusted HR of1.31 for death, whereas HR was 1.95 with HbA1c $7.5%.These findings were mirrored by the NHS England study inboth patients with type 1 diabetes (HbA1c $10.0%, HR 2.23)and patients with type 2 diabetes (HbA1c 7.5–8.9%, HR 1.22;HbA1c 9.0–9.9%, HR 1.36; and HbA1c$10.0%, HR 1.61) (33).
COVID-19 can also induce hyperglycemia in patientswithout diabetes, secondary to infection, which increasesthe risk of critical disease (34,35). Finally, prediabetes,characterized by elevated fasting blood glucose or impairedinsulin sensitivity, has been mostly overlooked in COVID-19 studies but could nevertheless pose a threat to clinicaloutcomes (Fig. 1). In a U.S. study of 184 patients, most haddiabetes (62.0%) or prediabetes (23.9%), and stratifyingpatients solely by elevated fasting blood glucose or HbA1cincreased the risk of intubation (36). A China study alsofound that elevated fasting blood glucose (.7.54 mmolcutoff) independently predicted mortality (HR 1.19) (27).
Figure 2—Illustration of parallels in acute COVID-19 pathology versus chronic diabetes pathology. COVID-19 infection induces acuteinflammatory cytokine storm, hyperglycemic surges, and acute organ damage. Diabetes is characterized by chronic, low-grade inflammation,glucose variability, and slowly progressing tissue damage in microvascular (CKD, neuropathy, brain) and macrovascular (CVD) complications.Additional shared detrimental mechanisms include hypercoagulation, endothelial dysfunction, and fibrosis. Drawn in part with BioRender.
2556 COVID-19 and Diabetes: Collision and Collusion Diabetes Volume 69, December 2020
Overall, there is a consensus from clinical studies andmeta-analyses (36 and reviewed in 37) that diabetes isa risk factor for serious COVID-19 infection and mortality,though this dependency may be less significant by multi-variate analysis in some studies. Varying study results arelikely due to the fact that many, but not all, patients withdiabetes suffer from additional comorbidities, such asobesity, hypertension, and CVD, which are independentrisk factors (Fig. 1).
Comorbidities and COVID-19
Comorbidities in General COVID-19 Patient CohortsObesity (19,20,25,39–41), CKD (19,20), CVD (5,20), andhypertension (20) persist as risk factors for hospitalizationor serious COVID-19 disease in multivariate analysis insome studies, after adjustment for various clinical varia-bles (Table 1 and Fig. 1), and in meta-analyses (37). Ina French cohort (n 5 124), obesity (BMI $35 kg/m2), butnot diabetes, was a strong predictor for mechanical ven-tilation use, with multivariate OR 7.36, after adjustmentfor age, sex, diabetes, and hypertension (39). The Open-SAFELY study reported that mortality risk increased withBMI, with HR 1.40 for class II obesity (BMI 35–39.9 kg/m2) and HR 1.92 for class III obesity (BMI$40 kg/m2) (4).This was similar to a NYC study, where BMI proportion-ately increased hospitalization risk (20). In a China cohort(n5 150), obesity was an independent predictor of seriousinfection (multivariate OR 3.0) and obese patients werelikelier to have diabetes versus other age- and sex-matchedCOVID-19 patients, underscoring the frequent occurrenceof comorbidities in patients with diabetes (41). Surpris-ingly, obesity with BMI $40 kg/m2 was not a risk forin-hospital mortality in a NYC cohort (5).
There are fewer reports on comorbid dyslipidemia. Themost comprehensive analysis leveraged data from the UKBiobank as a control population (n 5 428,494) versushospitalized COVID-19 patients (n 5 900) (40). Diabetes,HbA1c, CVD, hypertension, BMI, and waist-hip-ratio (WHR)were higher and cholesterol and HDL cholesterol lower inCOVID-19 patients. Log(HbA1c), BMI, and WHR (OR . 1)and total cholesterol (OR , 1) remained significant inmultivariate analysis in a subset of 340,966 UK Biobankregistrants vs. 640 COVID-19 hospitalized patients. Finally,LDL did not vary significantly between patients with diabeteswith poorly or well-controlled glucose (3) and was protectivefrom ARDS (HR 0.63) but not death (22).
Comorbidities in Cohorts of Patients With COVID-19 andDiabetesPatients with diabetes frequently suffer from comorbid-ities, e.g., obesity, dyslipidemia, hypertension, CVD, andCKD (42), which would predispose them to poorer COVID-19 outcomes. In mostly CORONADO participants withtype 2 diabetes, obesity by BMI positively predicted thestudy primary outcome, with OR 1.28 (i.e., tracheal in-tubation and/or death within 7 days of admission) (25).Dyslipidemia, although present in 51.0% of patients, did
not significantly increase risk of the composite primaryoutcome (25). In a second NHS England study, those whodied from COVID-19 (type 1 diabetes, n 5 464; type2 diabetes, n 5 10,525) were compared with individualswith diabetes registered to a practice (type 1, n5 264,390;type 2, n 5 2,874,020) to identify mortality risk factors(33). Type 1 diabetes shared the same risks as type 2 di-abetes for COVID-19 mortality, with preexisting CVD,CKD, and obesity identified as independent factors. Onestudy, with COVID-19 patients with diabetes (n5 153) ageand sex matched to 153 COVID-19 patients without di-abetes reported that CVD and hypertension were inde-pendent risk factors for mortality risks among all patients(43). These studies support the idea that comorbidities inpatients with diabetes, independent of diabetes itself,increase adverse COVID-19 disease outcomes.
Cumulative Comorbidities EffectFurthermore, COVID-19 patients with more than one comor-bidity may be especially vulnerable. In NYC, COVID-19patients were far likelier to have two or more comorbidities,constituting 88% of hospital admissions versus admissions ofpatientswith only one comorbidity (6.3%) or no comorbidities(6.1%) (17). In a nationwide study in China (n 5 1,590), theHR was 1.79 for one comorbidity and as high as 2.59 for twoor more comorbidities after adjustment for age and smokingstatus (44). When the data from this cohort were used todevelop a scoring system to predict serious clinical trajectoriesfrom admission status, the number of comorbidities (OR1.60) emerged as 1 of 10 variables (24). The CharlsonComorbidity Index, a score based on the presence of comor-bidities from a list that includes diabetes and kidney andcardiac diseases, had a multivariate OR of 1.05 for hospital-ization but an HR of only 0.99 for in-hospital death (45).
Overall, in assessment of risk for a COVID-19 patientwith diabetes at admission, overall comorbidities, includ-ing degree of glucose control (assessed by HbA1c [36,40]),fasting blood glucose (36), obesity (19,25,39,40), and thenumber of additional comorbid conditions, will be impor-tant clinical parameters to consider (Fig. 1).
Pediatric Diabetes and Comorbidities in COVID-19Fortunately, there is agreement to date that most pediatricCOVID-19 patients present with asymptotic or mild dis-ease (46). Nevertheless, some children suffer from moreserious COVID-19 infection, requiring hospitalization andeven pediatric ICU (PICU) (Table 2). The reasons forserious illness remain incompletely understood; however,drawing a parallel to adults, the presence of comorbidities,which are less frequent in young patients, may be onereason fewer children are vulnerable to COVID-19 but whysome still fall critically ill. Given the recent rise in type2 diabetes and obesity in youth, there could be a significantnumber of children at risk. Unfortunately, the few studiesthat have examined diabetes and other comorbidities inchildren with COVID-19 are relatively small, making ithard to draw conclusions.
diabetes.diabetesjournals.org Feldman and Associates 2557
Tab
le2—
Ove
rview
ofped
iatric
andpregna
ncyCOVID
-19clinical
coho
rts
Study
Loca
tion
Partic
ipan
ts(n)
Diabetes
find
ings
Com
orbidities
find
ings*
Selec
tlaboratoryfind
ings
**
She
kerdem
ian
etal.(47)
U.S.,Can
ada
48ped
iatric
patientsad
mitted
toPICUs
8%ha
ddiab
etes
83%
hadsign
ifica
ntco
morbidities
:15
%wereob
ese,
6%ha
dco
ngen
italh
eart
dise
ase
Not
exam
ined
Cha
oet
al.(48)
NYC
21ped
iatric
outpatients,
33to
GPMU,13
toPICU
Diabe
tesno
tedin
critica
lillnes
s(3
of13
)but
notsign
ifica
ntly
correlated
toPICU
Obes
itypreva
lent
incritica
lillnes
s(3
of13
)but
notsign
ifica
ntly
correlated
toPICU
Lower
ASTan
delev
ated
CRP,
PCT,
andpro-B
NPin
PICU
vs.
non-PICU
patie
nts
Zac
haria
het
al.(49)
NYC
50ho
spita
lized
ped
iatric
patients
Diabe
tesdid
notraisetheris
kof
seve
redise
ase,
but
fewpatients
haddiabetes
(n5
3)
Significa
ntly
morepatientswereob
ese
with
seve
re(67%
)vs.
nons
evere(20%
)COVID-19(P
50.03
)
Eleva
tedCRPan
dPCTin
seve
revs
.no
nsev
ereCOVID-19
Ottoet
al.(50)
U.S.
424patientsag
e0–
21ye
ars,of
who
m77
wereho
spita
lized
Diabe
tesno
tedinfreq
uently
13%
ofallp
atientswereob
ese
Not
exam
ined
Ebek
ozienet
al.(51)
U.S.
33COVID-19pos
itive
,31
COVID-19–
likeT1
Dped
iatric
andad
ultpa
tients
Hyp
erglyc
emia
andDKAwere
common
adve
rseou
tcom
esObes
itywas
preva
lent;CVD,
hypertens
ion,
hyperlip
idem
iaalso
pres
ent
Not
exam
ined
Sen
tilhe
set
al.(52)
Fran
ce54
pregn
antfemales
Onlyfour
hadge
stationa
ldiabetes
mellitus
;sa
mple
size
toosm
all
foran
ypoten
tiallinkto
COVID-
19
Obes
itymay
bearis
k;on
lytw
oha
dge
stationa
lhyp
ertens
ion;
samplesize
toosm
allfor
anypoten
tiallinkto
COVID-19
Eleva
tedALT
,AST,
CRP,an
dlympho
pen
iain
hosp
italized
COVID-19pa
tients
Lokk
enet
al.(53)
WA
46pregn
antfemales
Onlyon
eha
dge
stationa
ldiabetes
mellitus
;sa
mple
size
toosm
all
foran
ypoten
tiallinkto
COVID-
19
26.1%
hadan
underlyingco
ndition
;two-
third
swereov
erweigh
t(28
.6%
,n5
12)
orob
ese(35.7%
,n5
15)by
prep
regn
ancy
BMI;15
%dev
elop
edse
vere
disea
se,of
who
m80
%were
overweigh
tor
obes
eby
prepregn
ancy
BMI;on
lytw
oha
dge
stationa
lhy
pertens
ion;
samplesize
toosm
allfor
anypoten
tiallinkto
COVID-19
Not
exam
ined
atpo
pulationleve
l
Knigh
tet
al.(54)
U.K.
427pregn
antfemales
3%ha
ddiab
etes
,12
%ha
dge
stationa
ldiabetes
mellitus
,;no
analysis
performed
for
disea
sese
verity
35%
overweigh
t,34
%ob
ese,
34%
pree
xistingco
morbidities
;no
analysis
performed
fordisea
sese
verity
Not
exam
ined
Kay
emet
al.(55)
Fran
ce61
7pregn
antfemales
Preex
istin
gdiabetes
(2.3%
preva
lenc
ein
totalp
opulation)
raised
theris
kof
seve
redise
ase,
RR
3.8(95%
CI1.4–
10.7),but
notg
estatio
naldiabe
tesmellitus
(11.5%
prev
alen
ce)
BMI(RR
1.9[95%
CI1.4–
2.5]),
hypertens
ion,
gestationa
lhyp
ertens
ion
orpreec
lamps
iaweremoreco
mmon
inse
vere
disea
se
Not
exam
ined
ALT
,alanine
aminotrans
ferase
;BNP,b
rainna
triuretic
pep
tide;
CRP,C
-rea
ctiveprotein;D
KA,d
iabeticke
toac
idos
is;G
PMU,g
eneralped
iatricmed
icalun
it;T1
D,typ
e1diabe
tes.*C
ondition
sco
morbid
with
diabetes
cons
idered
.**Selec
tlab
oratoryfind
ings
fors
ignifica
ntdifferen
cesreportedin
immun
ece
llpop
ulations
,cytok
ines
,and
biomarke
rsof
infectionan
dkidne
y,liver,a
ndca
rdiacdam
age.
Cha
nges
wererepo
rted
ifthereweresign
ifica
ntdifferen
cesin
either
mea
nva
lues
orin
thenu
mber
ofpa
tientsab
oveacu
toffva
lue.
2558 COVID-19 and Diabetes: Collision and Collusion Diabetes Volume 69, December 2020
A cross-sectional study of 48 pediatric patients (0–21years old), admitted to PICUs across the U.S. and Canada,found 83% had significant comorbidities: 15% were obese,8% had diabetes, and 6% had congenital heart disease (47). Achildren’s hospital in NYC (n 5 67, aged 1 month–21 years)admitted 13 patients to PICU, noting the presence of bothdiabetes (3 of 13) and obesity (3 of 13) but not to significance;however, the cohort was small (48). Another study (n 5 50,aged 6 days–21 years) at a different NYC children’s tertiarycare center found significantly more obesity in severe (67%)versus nonsevere (20%) COVID-19, but not diabetes, possiblydue to the small number of patients with diabetes (n 5 3)(49). Obesity is a recurrent theme and was relatively prev-alent in other pediatric studies also (50,51).
The cumulative evidence from pediatric studies sug-gests that comorbidities may be a predisposing factor forserious COVID-19 infection in children, particularly obe-sity. The impact of diabetes remains unclear due to rela-tively low study participant numbers (Fig. 1).
Pregnancy, Diabetes, and Comorbidities in COVID-19Pregnancy is a vulnerable period, particularly since gesta-tional diabetes mellitus may develop; yet, few studies haveexamined pregnant women admitted for COVID-19 infec-tion (Table 2). A French cohort of 54 pregnant women withsuspected or confirmed COVID-19 included four patientswith gestational diabetes mellitus and two with gestationalhypertension, which were too few to analyze for a potentiallink to infection severity (52). However, prepregnancy over-weight or obese BMI were relatively prevalent, which theauthors concluded could be a risk factor for COVID-19disease. Another small study (n 5 46), in the U.S., alsofound a high prevalence of elevated prepregnancy BMI(28.6%, overweight, and 35.7%, obese) (53). Moreover,15% of pregnant patients developed severe infection, ofwhom 80% were overweight or obese. A U.K. study of427 pregnant women with confirmed COVID-19 drewsimilar observations, finding that 35% of patients wereoverweight and 34% were obese (54). The diabetes preva-lence was 3%, whereas it was 12% for gestational diabetesmellitus, but no analysis of disease severity was performed.
The largest study to date was in 617 pregnant Frenchwomen (55). Preexisting diabetes was present in 2.3% of thetotal population and raised the chance of severe disease, witha risk ratio (RR) of 3.8. In contrast, gestational diabetesmellitus, at 11.5% prevalence, did not affect outcomes forinfection severity. The investigators did not discuss reasons forthe difference in risk from preexisting diabetes versus gesta-tional diabetes mellitus, but it raises the question of whethergestational diabetes mellitus interacts distinctly with COVID-19 pathophysiology (Fig. 1). Diabetes complications, for in-stance, from preexisting diabetes, could be a factor for seriousinfection, which draws parallels to studies of general popula-tions with diabetes (25). The study also found that BMI has anRR of 1.9, hypertension an RR of 2.4, and gestational hyper-tension or preeclampsia an RR of 2.4 for severe COVID-19,though the latter two did not reach significance.
Collectively, the data from pregnancy cohorts echofindings from adult studies, with diabetes, obesity, andcomorbidities likely predisposing to poorer outcomes.However, it is possible that gestational diabetes mellitusmay not be a factor, though larger studies are needed for usto definitively conclude this.
Race, Diabetes, and Comorbidities in COVID-19Race disparities are an emergent theme during the COVID-19 pandemic (Table 3). The precise reasons to date remainunclear, though the prevalence of comorbidities, includingobesity, (56) and socioeconomic factors (57) have beensuggested. Of the U.S. population, 18% are Hispanic, 13%Black, and 0.7% American Indian or Alaska Native; yet,these groups have disproportionately constituted 33%,22%, and 1.3%, respectively, of adult U.S. COVID-19 cases(58) and are also highly represented in hospitalized pedi-atric patients (50).
Several observational studies have taken a more de-tailed look to understand these racial disparities. InDetroit cohorts, Black race did not increase risk of severeinfection (19,59); however, diabetes or comorbidities prev-alence by race was not examined (19). These findings partlyagree with those of a Georgia study (n5 297), which foundthat although hospitalizations among Black patients(83.2%) were disproportionate to numbers among otherraces, indicating greater disease severity, Black patients didnot have higher mechanical ventilation use or mortality (60).This study also reported the prevalence of comorbidities,which did not differ significantly for diabetes in Black versusother races but did differ for hypertension and mean BMI. Alarger Louisiana cohort (n 5 3,481) similarly concluded thatBlack race was a hospitalization risk but not an independentin-hospital mortality risk (45). Although the investigatorsfound diabetes, hypertension, and CKD prevalence to behigher in Black versus White patients, they did not performan analysis for disease severity. A California study (n 51,052) analyzed hospitalization risk for Black, Asian, andHispanic race relative toWhite, but only Black race had anOR2.7, after adjustment for sex, age, comorbidities, and socio-economic factors (57). U.K. studies have also noted greatersusceptibility of Black patients, and other race minorities, toCOVID-19 disease (61) and hospitalization (40), after ad-justment for several cardiometabolic and socioeconomicfactors. Strikingly, a NYC study found that Black race wasprotective for critical illness and death, whereas Hispanic racewas a risk for hospitalization (20).
Importantly, some studies have reported increased mor-tality risk for Black race and other minorities. Analysis of NYCdemographics and COVID-19 deaths (n5 4,260) revealed thatHispanic (22.8%) and Black (19.8%) patients had the highestage-adjusted mortality per 100,000, which corresponded tothe highest obesity rates: 25.7% and 35.4%, respectively (56).However, the study did not adjust for other importantvariables. Lacking complete U.S. nationwide disaggregateddata by race, Millett et al. (62) analyzed county-level de-mographics and COVID-19 deaths. Counties with a greater
diabetes.diabetesjournals.org Feldman and Associates 2559
Tab
le3—
Ove
rview
ofCOVID
-19clinical
coho
rtswithinve
stigationofsu
scep
tibility
byrace
andethn
icity
Study
Loca
tion
Partic
ipan
ts(n)
Diabetes
find
ings
Com
orbidities
find
ings
*Selec
tlaboratoryfind
ings
**
Stoke
set
al.(58)
U.S.
599,63
6of
know
nrace
Noco
rrelationstud
yof
diabetes
torace
perform
ed33
%Hispan
ic,2
2%Black
,1.3%
American
Indianor
Alask
aNative,
which
acco
unt
for18
%,13
%,an
d0.7%
oftheU.S.
population,
resp
ectiv
ely,
sugg
estin
gthey
weredisprop
ortio
natelyaffected
byCOVID-19
Not
exam
ined
Bha
rgav
aet
al.(59)
Detroit,
MI
197
Diabetes
morefreq
uent
inpatientswith
seve
re(48.6%
)vs
.no
nsev
ereinfection
(30.1%
),OR
2.20
(95%
CI1
.21–
4.0;
P5
0.00
9)in
univariate
analysis
butno
tmultiv
ariate;no
correlationstud
yof
diabetes
torace
performed
Obes
ity,hy
pertens
ion,
cong
estiv
ehe
art
failure,ce
rebrova
scular
disea
sedid
not
increa
seun
ivariate
ORof
seve
redisea
se,
thou
ghCKDdid;8
2.1%
wereBlack
,and
Black
race
was
notaris
kforse
vere
infection
Eleva
tedCran
dPCTha
dsign
ifica
ntun
ivariate
OR.1for
seve
redisea
se;elev
ated
Cr
from
bas
elinean
dinitial
CRP
hadsign
ifica
ntmultiv
ariate
OR
.1forse
vere
infection
Goldet
al.(60)
GA
297;
Black
hosp
italizations
(83.2%
)were
disproportio
nate
toothe
rrace
s,indicatinggrea
ter
disea
sese
verity
Diabetes
prev
alen
cedid
notdiffer
sign
ifica
ntly
inBlack
patients(41.7%
)vs
.inpatientsof
othe
rrac
es(32.0%
)P5
0.21
Hyp
ertens
ionmoreco
mmon
inBlack
patie
nts(69.6%
)vs
.patientsof
othe
rrace
s(54.0%
),P5
0.04
7;mea
nBMI
high
erin
Black
(31.4%
)patientsvs
.pa
tientsof
othe
rrace
s(29.6%
),P5
0.00
3;Black
patientsdid
noth
avehigh
ermec
hanica
lven
tilationus
eor
mortality
Not
exam
ined
Aza
ret
al.(57)
CA
1,05
2co
nfirm
edca
ses
Diabetes
hadOR2.2,
P,
0.01
,forho
spita
lad
mission
,in
multiv
ariate
analysis
with
adjustmen
tforse
x,ag
e,co
morbidities
,so
cioe
cono
mic
factors;
noco
rrelation
stud
yof
diabetes
torace
perform
ed
Non
-Hispan
icAfrican
American
sha
dOR
2.7,
P5
0.00
7,forh
ospita
ladmission
vs.
non-Hispan
icWhites,
afterad
justmen
tforthesa
meva
riablesas
listedfor
diab
etes
find
ings
Not
exam
ined
Raisi-Estab
ragh
etal.(61)
U.K.
1,32
6pos
itive
,3,18
4ne
gativ
eCOVID-19tests
from
UKBioban
k
Diabetes
notaris
kforsu
scep
tibility
topos
itive
vs.ne
gativ
eCOVID-19test;no
correlationstud
yof
diabetes
torace
perform
ed
Hyp
ertens
ion,
high
choles
teroln
otris
ksfor
susc
eptib
ility
topos
itive
vs.ne
gativ
eCOVID-19test;Black
,Asian
,an
dminority
ethn
icgrou
pmoresu
scep
tible
topo
sitiv
evs
.ne
gativ
eCOVID-19test,
with
adjustmen
tforag
e,se
x,BMI,
diab
etes
,hy
pertens
ion,
choles
terol,an
dso
cioe
cono
mic
factors
Not
exam
ined
ElC
haar
etal.(56)
NYC
4,26
0de
aths
Diabetes
notinve
stigated
Hispa
nican
dBlack
patie
ntsha
dhigh
est
age-ad
justed
mortalityratesper
100,00
0(22.8%
and19
.8%
,resp
ectiv
ely,
vs.
othe
rethn
icgrou
ps)
corres
pon
dingto
thegrou
pswith
thehigh
esto
bes
ityrates,
25.7%
and35
.4%
,resp
ectiv
ely,
P,
0.05
;the
twoNYCborou
ghswith
high
est
mortalityrates,
Bronx
(6%
)and
Brook
lyn
(5.4%
),also
hadthehigh
estob
esity
rates,
32%
and27
%,resp
ectiv
ely
Not
exam
ined
Con
tinue
don
p.25
61
2560 COVID-19 and Diabetes: Collision and Collusion Diabetes Volume 69, December 2020
proportion of Black residents (i.e., above national average,$13%) hadmore COVID-19 cases (rate ratio 1.24) and deaths(rate ratio 1.18), after adjustment for county-level traits, e.g.,age, comorbidities, poverty, and pandemic duration. Diabetesprevalence was also higher (13.9% vs. 11.1%) in counties withhigh ($13%) and low (,13%) proportion of Black residentsbut did not correlate with COVID-19 cases (rate ratio 0.97) ordeaths (nonsignificant rate ratio 1.01), after adjustment fordemographics, comorbidities, and socioeconomic factors.Thus, diabetes, or other cardiometabolic effects, may not besolely attributable to COVID-19 risk in Black patients. Finally,large population-based studies, OpenSAFELY and NHSEngland, found higher mortality risk for Asian and Blackraces, after adjustment for age, sex, comorbidities, andsocioeconomic status (4,26,33).
Overall, Black, Hispanic, and possibly other races may berisk factors for serious COVID-19 infection or death, but thefactors driving this disparity are presently unclear (Fig. 1).
COVID-19 and Diabetes Pathology: Collision andCollusionGiven the relatively short time that has elapsed since theSARS-CoV-2 pandemic broke out, its pathophysiology remainsincompletely understood. However, like its predecessorsSARS-CoV and Middle East respiratory syndrome corona-virus (MERS-CoV), SARS-CoV-2 gains cellular entry byleveraging the ACE2 receptor, a master regulator of therenin-angiotensin system. The major viral spike glycopro-tein (S1) binds to ACE2 (63), while proximal serine pro-teases, like the transmembrane serine protease 2, cleave thevirus spike protein and ACE2, promoting viral internalization(64). Infection induces cell death, which triggers inflamma-tory cytokine production and inflammatory immune cellrecruitment (65). SARS-CoV-2 also infects circulating im-mune cells, stimulating lymphocyte apoptosis and inflam-matory cytokine secretion, known as “cytokine storm” (6).High circulating cytokine levels contribute to SARS-CoV-2–driven multiorgan failure and disrupted endocrine signalingand hyperglycemia surges (66). Widespreadmultitissue ACE2expression, e.g., lung, heart, kidney, and nerve (67), leads totropism, as validated by viral detection within multipletissues (7,68). Tropism potentially constitutes another path-way to multiorgan damage in COVID-19 patients, e.g., acutecardiac injury (ACI) and acute kidney injury (AKI) (13,14).
Although the inflammatory, hyperglycemic, and tissuedamage response is intensely acute in COVID-19 infection,it is mirrored by diabetes pathology (Fig. 2), which ischaracterized by chronic, low-grade inflammation, im-paired glycemic control, and slowly progressive multitissueinjury, e.g., diabetic microvascular (CKD, neuropathy,brain) and macrovascular (CVD) complications (8,69).Although the underlying reasons for the susceptibilityof patients with diabetes to COVID-19 remain unclear,commonalities in pathology suggest that acute COVID-19–induced adverse reactions may superimpose on preexistinginflammation, glucose variability, and multitissue injury inpatients with diabetes to aggravate outcomes (Fig. 1).
Tab
le3—
Continue
d
Study
Loca
tion
Partic
ipan
ts(n)
Diabetes
find
ings
Com
orbidities
find
ings
*Selec
tlaboratoryfind
ings
**
Millet
etal.(62)
U.S.
Nationw
idepop
ulation
dem
ographics
andCOVID-
19dea
ths
Diabetes
preva
lenc
ein
coun
tieswith
,13
%Black
residen
tswas
11.1%
vs.13
.9%
inco
untie
swith
$13
%Black
residen
ts,P
valueno
tstated
;diabetes
preva
lenc
edid
notco
rrelatewith
coun
tiesby
COVID-19
case
s(ra
teratio
0.97
[95%
CI0
.92–
1.03
])or
dea
ths(ra
teratio
1.01
[95%
CI0.88
–
1.16
]),afterad
justmen
tfordem
ographics
,co
morbidities
,and
socioe
cono
mic
factors
Cou
ntieswith
high
erBlack
residen
tprop
ortio
ns($
13%
)had
moreCOVID-19
case
s(ra
teratio
1.24
,95%
CI1
.17–
1.33
)an
ddea
ths(ra
teratio
1.18
,95%
CI1
.00–
1.40
),afterad
justingforco
unty-lev
eltraits,e.g.,ag
e,co
morbidities
,po
verty,
andep
idem
icdu
ratio
n
Not
exam
ined
*Con
ditio
nsco
morbidwith
diabe
tesco
nsidered
.**Selec
tlab
oratoryfind
ings
fors
ignifica
ntdifferen
cesreportedinim
mun
ece
llpop
ulations
,cytok
ines
,and
biomarke
rsof
infectionan
dkidne
y,liver,an
dca
rdiacdam
age.
Cha
nges
werereportedifthereweresign
ifica
ntdifferen
cesin
either
mea
nva
lues
orin
thenu
mber
ofpatientsab
oveacu
toffva
lue.
diabetes.diabetesjournals.org Feldman and Associates 2561
Do Preexisting Diabetes Complications PredisposePatients to Acute COVID-19–Induced Organ Damage?Few studies have stratified COVID-19 patients by diabetesstatus to examine the possibility that preexisting micro-and macrovascular complications render patients suscep-tible to acute organ injury (Fig. 1). CORONADO (n 5 1,317)demonstrated that preexisting microvascular (OR 2.14) andmacrovascular (OR 2.54) complications independently associ-ated with 7-day mortality (25), suggesting that the presence ofdiabetes complications may set patients on poorer clinicaltrajectories. In a NYC study of 5,449 severe COVID-19patients, of whom 1,993 developed AKI, diabetes was a riskfor renal damage, with 41.6% developing AKI vs. 28.0% whodid not (70). Diabetes also correlated with progressive damagein AKI stage 1 (39.7%), stage 2 (43.2%), and stage 3 (43.5%) byKidney Disease: Improving Global Outcomes (KDIGO) criteria.After adjustment for age, sex, and race, diabetes had an OR of1.76 for AKI. However, the study did not state whether AKIcorrelated with preexisting CKD, since baseline CKD data werenot available, although associations with preexisting CKD andAKI have been noted in meta-analysis (71).
Although diabetes was not an independent risk for COVID-19 death in a cohort of 153 patients with diabetes comparedwith age- and sex-matched individuals without diabetes,patients with diabetes were more likely to have preexistingCVD and be admitted to ICUs and experience acute complica-tions (ACI, AKI, ARDS) (43). Nonsurvivor patients with diabeteshad higher blood glucose levels and a greater chance of ACI orAKI, in addition to an altered inflammatory and immune systemprofile (see Are Patients With Diabetes Predisposed to AcuteCOVID-19–Induced Inflammatory Response?).Within a cohortwith diabetes (n5 952), patients with well-controlled glucosewere also less likely to suffer from hypertension and CVD.They were also at lowered risk of AKI (HR 0.12) and ACI(HR 0.24), after adjustment for comorbidities (3), indicat-ing that even if preexisting microvascular complicationscontribute to acute organ injury, additional factors, suchas glucose control or inflammation, may also participate.
Additional Aspects of COVID-19 Tropism Relevant toDiabetesOne particular aspect of COVID-19 tropism meriting closeattention from a diabetes perspective is the possibility ofincreasing the incidence of b-islet damage–induced type1 diabetes. Drawing parallels, SARS-CoV may have beenresponsible for acute type 1 diabetes onset by leveragingb-islet ACE2 expression to induce loss of islets (72). It ispossible that COVID-19 might also trigger acute-onset type1 diabetes in individuals predisposed to autoimmunity (73).Indeed, the multicenter regional data from North WestLondon just reported an 80% increase in new-onset type1 diabetes cases and diabetic ketoacidosis in children up tothe age of 16 years during the COVID-19 pandemic peak (74).Moreover, COVID-19 tropism through ACE2 expression inadipose tissue may underlie the link to obesity as a seriousinfection risk, since adipose tissue could potentially serve asa reservoir of viral shedding (75).
Are Patients With Diabetes Predisposed to AcuteCOVID-19–Induced Inflammatory Response?Although the full cytokine storm profile in COVID-19 is notfully characterized yet, hyperinflammation predicts seriousdisease (Fig. 1). Lymphopenia along with elevation in whiteblood cells (WBC), neutrophils, C-reactive protein (CRP),erythrocyte sedimentation (ESR), ferritin, IL-6, and procalci-tonin (PCT) associates with poorer COVID-19 clinical course,defined as serious infection, ARDS, ICU admission, or death,in studies in multiple countries (Table 1). COVID-19 patientsexperience, in parallel to inflammation, elevated AST, brainnatriuretic peptide, hypersensitive troponin I (hs-TnI), cre-atine kinase (muscle and brain type), lactate dehydrogenase(LDH), and creatinine (Cr), indicative of tissue damage.Clotting homeostasis is similarly compromised, e.g., withelevated D-dimer with longer thrombin or prothrombintime, which also correlate with clinical progression. Ameta-analysis found higher AST (.40 units/L), Cr ($133mmol/L), D-dimer (.0.5 mg/L), hs-TnI (.28 pg/mL), LDH(.245 units/L), and PCT (.0.5 ng/mL) and lower WBC(,4 3 109 per L) defines an OR .1 for critical illness (76).
Diabetes is also characterized by chronic, low-gradeinflammation, which is also a prominent feature of itscomplications, diabetic CKD, CVD, and neuropathy(8,77,78). Several proinflammatory molecules from theCOVID-19 cytokine storm cascade are shared with type2 diabetes pathophysiology, such as CRP, IL-6 (77), andPCT (79). The underlying chronic inflammatory state indiabetes may be “locked and loaded” for virus-induceddamage, promoting a vicious cycle of cytokine releaseand hyperglycemic surges, leading to more widespreadmultiorgan damage, including injury to tissues alreadyweakened by preexisting diabetes complications.
Worryingly for patients with diabetes, and as an addedlayer of risk, they are more prone to cytokine storm, whichpredicts poorer outcomes (Table 1). Admission CRP (OR1.93) and AST (OR 2.23) independently predicted 7-daymortality in the CORONADO COVID-19 patients withdiabetes (25). In Chinese cohorts, patients with diabeteshad a more inflammatory profile than patients withoutdiabetes (3,27). More favorable inflammatory and tissuebiomarker profiles were also evident in patients with type2 diabetes with well-controlled versus poorly controlledblood glucose (3,30). Another study found differences innumerous inflammation and organ damage biomarkers innonsurviving versus surviving patients with diabetes,which also correlated with glucose and HbA1c levels(43). Moreover, elevated inflammation and organ damagebiomarkers were present in COVID-19 patients with di-abetes and hyperglycemia secondary versus without di-abetes and with normoglycemia (34).
One inflammatory biomarker, with deep roots in di-abetes pathophysiology, not widely investigated in COVID-19, is soluble urokinase-type plasminogen activator re-ceptor (suPAR). In Greek (n 5 57) and U.S. (n 5 21)COVID-19 cohorts, we found that admission suPAR pre-dicted severe respiratory failure (80). suPAR correlates
2562 COVID-19 and Diabetes: Collision and Collusion Diabetes Volume 69, December 2020
with diabetes risk (81) and reflects the underlying chronicinflammatory process of its micro- (82) and macrovascularcomplications (83).
The reasons for the susceptibility of patients withdiabetes to COVID-19 are multifaceted and reflect thecomplex pathophysiology of both diabetes and COVID-19 infection. Diabetes and its comorbidities, inflamma-tion, glucose variability, and other factors, may “collide andcollude” to disproportionally set COVID-19 patients withdiabetes on poorer clinical trajectories (Fig. 2).
Diabetes and COVID-19 SequelaeIt is becoming clear that COVID-19 survivors suffer frompersistent symptoms (84) and may also face a lifetime ofsequelae, which draws parallels to SARS-CoV and MERS-CoV(10,85). Although the pandemic has not yet lasted longenough to measure long-term outcomes, the evidence todate suggests a significant burden of possibly irreversiblenew complications. For instance, COVID-19, like SARS-CoVand MERS-CoV, may aggravate preexisting CVD or eveninduce new cardiac pathology (86), including in patientswith type 2 diabetes (87). COVID-19 patients with preexist-ing CKD are likelier to suffer AKI (71). COVID-19 also elicitsneurological manifestations (88) and cognitive impairment(89), which exhibit shared pathology with diabetes throughcytokine storm, hypercoagulability, and endothelial d