Outcomes in Patients WithHyperglycemia Affected byCovid-19: Can We Do More onGlycemic Control?https://doi.org/10.2337/dc20-0723

OBJECTIVE

An important prognostic factor in any form of infection seems to be glucose controlinpatientswith type2diabetesmellitus. There isno informationabout theeffectsoftight glycemic control on Covid-19 outcomes in patients with hyperglycemia.Therefore, we examined the effects of optimal glycemic control in patients withhyperglycemia affected by Covid-19.

RESEARCH DESIGN AND METHODS

Fifty-nine patients with Covid-19 hospitalized with moderate disease were eval-uated.On thebasis of admission glycemia>7.77mmol/L, patientswere divided intohyperglycemic and normoglycemic groups. Interleukin 6 (IL-6) and D-dimer levelswere evaluated at admission andweekly during hospitalization. The composite endpoint was severe disease, admission to an intensive care unit, use of mechanicalventilation, or death.

RESULTS

Thirty-four (57.6%) patients were normoglycemic and 25 (42.4%) were hypergly-cemic. In the hyperglycemic group, 7 (28%) and 18 (72%) patients were diagnosedwith diabetes already before admission, and 10 (40%) and 15 (60%) were treatedwithout and with insulin infusion, respectively. The mean of glycemia duringhospitalizationwas 10.6560.84mmol/L in the no insulin infusion group and7.6961.85mmol/L in the insulin infusion group. At baseline, IL-6 and D-dimer levels weresignificantly higher in the hyperglycemic group than in the normoglycemic group(P < 0.001). Despite that all patients were on standard treatment for Covid-19infection, IL-6 and D-dimer levels persisted higher in patients with hyperglycemiaduring hospitalization. In a risk-adjusted Cox regression analysis, both patientswithhyperglycemia and patients with diabetes had a higher risk of severe disease thanthosewithout diabetes andwithnormoglycemia. Cox regression analysis evidencedthat patients with hyperglycemia treated with insulin infusion had a lower risk ofsevere disease than patients without insulin infusion.

CONCLUSIONS

Insulin infusion may be an effective method for achieving glycemic targets andimproving outcomes in patients with Covid-19.

1Department of Advanced Medical and SurgicalSciences, University of Campania “Luigi Vanvi-telli,” Naples, Italy2Department of PrecisionMedicine, University ofCampania “Luigi Vanvitelli,” Naples, Italy3Department of Infectious Diseases, Sant’AnnaHospital, Caserta, Italy4Department of Mental and Physical Health andPreventiveMedicine, University of Campania “LuigiVanvitelli,” Naples, Italy

Corresponding author: RaffaeleMarfella, raffaele.marfella@unicampania.it

Received 3 April 2020 and accepted 8 May 2020

This article contains supplementary material onlineat https://doi.org/10.2337/figshare.12275516.

This article is part of a special article collectionavailable at https://care.diabetesjournals.org/collection/diabetes-and-COVID19.

© 2020 by the American Diabetes Association.Readersmayuse this article as longas thework isproperly cited, the use is educational and not forprofit, and the work is not altered. More infor-mation is availableathttps://www.diabetesjournals.org/content/license.

Celestino Sardu,1 Nunzia D’Onofrio,2

Maria Luisa Balestrieri,2

Michelangela Barbieri,1

Maria Rosaria Rizzo,1 Vincenzo Messina,3

Paolo Maggi,3 Nicola Coppola,4

Giuseppe Paolisso,1 and Raffaele Marfella1

Diabetes Care 1

PATH

OPHYSIO

LOGY/CO

MPLIC

ATIO

NS

Diabetes Care Publish Ahead of Print, published online May 19, 2020

Severe acute respiratory syndrome co-ronavirus 2 (SARS-CoV-2) is the causeof the Covid-19 pandemic, which hasaffected.150,000 individuals and is thecause of;24,000 deaths in Italy as of thiswriting, representing almost 10% of in-fected patients. Furthermore, 5–10% ofpatients with Covid-19 require intensivecare unit (ICU) admission andmechanicalventilation (https://www.epicentro.iss.it).Hyperglycemia (defined as a blood glu-cose level .7.77 mmol/L) can occur inboth patients with and patients withoutdiabetes hospitalized for Covid-19 and iscommon among acute hospital admis-sions and critically ill patients, encom-passing those with no previous history ofhyperglycemia (1,2). However, precisenumbers on the prevalence and inci-denceof this stresshyperglycemia duringinfection are limited. In one study, pa-tients with infectious diseases withoutunderlying diabetes had average plasmaglucose values of 10.776 3.66mmol/L inthe absence of nutritional support, and.50% of all patients developed stress-induced hyperglycemia (3). Moreover,several studies demonstrated that ad-mission hyperglycemia was associatedwith an increase of poor outcomes andmortality in hospitalized patients pre-senting with an infectious disease (4).Possible mechanisms for this increasedmortality include hyperglycemia-inducedchanges in coagulation, worsening of en-dothelial function, and inflammatory cy-tokine overproduction. Indeed, our groupdemonstrated that healthy subjects andpatients with impaired glucose tolerancehave inflammatory cytokines interleukin(IL) 6, tumor necrosis factor-a (5), andD-dimer (6) overproduction following theappearance of hyperglycemia. Interest-ingly enough, Covid-19 infection is asso-ciated with severe pneumonia disease,disseminated intravascular coagulation,and septic shock with strong increasesin plasma IL-6 (7) and D-dimer (8) levels.Thus, elevated blood glucosemayworsenthe prognosis of patients with Covid-19,raising the risk formechanical ventilation,shock, and multiple organ failure neces-sitating ICU treatment. Thus, the absoluteburden of infection attributable to poorglycemic control in this population wouldbe substantial. In this context, a centralquestion is should patients with hyper-glycemia be treated with more attentionto glycemic control during the Covid-19infection? The Standards of Medical Care

in Diabetes recently developed by theAmerican Diabetes Association (ADA)recommend the range of 7.77–9.99mmol/Las a target level of blood glucose for themajority of critically ill patients (9). Inaddition, the recommendation from theSurviving Sepsis Campaign showed that#9.99 mmol/L should be targeted in themanagement of blood glucose (10). Fi-nally, the recommendations suggest thatit is reasonable to consider intensiveglucose control, with insulin infusion, inpatients with significant hyperglycemia(plasma glucose.9.99 mmol/L), regard-less of prior diabetes history. To date, it isunclear whether tight glycemic control(blood glucose range 7.77–9.99 mmol/L)is effective and warranted in patientswith Covid-19 with moderate disease.Therefore, our study evaluated whetherhyperglycemia is associated with a fur-ther increase in plasma inflammatorycytokine (IL-6) levels and coagulationactivation(asmonitoredbyplasmaD-dimerlevels) in hospitalized patients with Covid-19. Moreover, we evaluated whether poorglycemic control was associated with pooroutcomes and whether early optimal gly-cemic control along with hospitalizationreduces plasma IL-6 and D-dimer levels,thus improving outcomes for hospital-ized patients with Covid-19.

RESEARCH DESIGN AND METHODS

PatientsWe analyzed 187 patients positive forCovid-19 admitted to the Infection Dis-ease Departments of Vanvitelli Univer-sity and San Sebastiano Caserta Hospitalsince 20 February 2020. Among them,we selected 59 patients with moderatepneumonia disease (Fig. 1). Covid-19infectionwas categorized as follows (11):mild (patientswith fever andnoevidenceof pneumonia on imaging), moderate(patients with fever, respiratory tractsymptoms, and pneumonia on imagingwithout the need for invasive ventilation),and critical (occurrence of respiratoryfailure requiring mechanical ventilation,presence of shock, other organ failurethat requires monitoring, and treatmentin the ICU). Patients with previous in-flammatory disorders, malignancy, renaldiseases, or infections as well as patientswith critical Covid-19 infection at admis-sion were not eligible for the study. Allpatients were treated with standardprotocol, including noninvasive oxygentherapy, hydroxychloroquine 200 mg

(1 3 2/day), and lopinavir/ritonavir cps200/50 mg. Patients were categorizedas normoglycemic and hyperglycemic aswell as with or without diabetes on thebasis of a diagnosis preceding the currentillness (9). Hyperglycemia was definedas an admission plasma glucose levelof .7.7 mmol/L (9). Although intrave-nous infusion insulin is currently themosteffective method for controlling glucoseamong hospitalized patients, there isinsufficient evidence for recommendingor discouraging its early infusion (level ofevidence C). Therefore, after describingthe possible risks and benefits of insulininfusion therapy, patients voluntarily de-cided whether to receive insulin infusiontherapy. Continuous insulin infusion of50 IU Actrapid HM (Novo Nordisk) in50 mL NaCl (0.9% using a Perfusor fmpump) was started when blood glucoselevelswere.9.9mmol/L and adjusted tokeep blood glucose between 7.77 and9.99 mmol/L. When blood glucose fellto ,7.7 mmol/L, insulin infusion wastapered and eventually stopped. Afterthe start of the insulin infusion protocol,a glycemic control was provided everyhour to obtain three consecutive valuesthat were within the goal range. Theinfusion lasted until stable glycemic goaland at least for 24 h. Thereafter, sub-cutaneous insulin was initiated at thecessation of the infusion in the infusiongroup and at admission into the no in-sulin infusion group. Short-acting insulinwas given before meals, and intermedi-ate long-acting insulin was given in theevening. Patients with previous diabetesstopped at admission oral antidiabeticdrugs, such as metformin, sulfonylureas,dipeptidylpeptidase4 inhibitors, sodium–

glucose cotransporter 2 inhibitors, andglucagon-like peptide 1 agonists. The in-vestigation conformed with the principlesoutlined in the Declaration of Helsinki foruse ofhuman tissueor subjects. The studyprotocolwasapprovedbythe institutionalethics committees.Written informedcon-sent was obtained from all patients.

Laboratory and Imaging Evaluations

Real-Time RT-PCR Assay for SARS-CoV-2

Respiratory specimens were collected bythe local center for disease control andthen shipped to designated authoritativelaboratories to detect SARS-CoV-2. Thepresence of SARS-CoV-2 in respiratoryspecimens was detected by real-timeRT-PCR methods. Laboratory analyses

2 Glycemic Control in Patients With Covid-19 Diabetes Care

were obtained on admission before startingCovid-19 medical therapy and duringhospitalization.

Clinical Laboratory Measurements

Respiratory specimens, including nasaland pharyngeal swabs or sputum, weretested to exclude evidence of other viralinfections, including influenza, respiratorysyncytial virus, avian influenza, parain-fluenza, and adenovirus. Routine bacte-rial and fungal examinations were alsoperformed. Laboratory assessments con-sisted of a complete blood count, bloodchemical analysis, coagulation testing,assessment of liver and renal function,and measures of electrolytes, C-reactiveprotein, procalcitonin, lactate dehydro-genase, and creatine kinase. Venousblood for IL-6 (Human Quantikine ELISAKit; R&D Systems) and D-dimer (HumanELISA Kit; Invitrogen) levels were col-lected in EDTA-coated tubes immediatelyafter patients arrived at the departmentand weekly during hospitalization. Ra-diologic assessments included chest ra-diography or computed tomography (CT)at admission and weekly during hospi-talization, and all laboratory testing was

performed according to the clinical careneeds of the patient. We determined thepresence of a radiologic abnormality onthe basis of the documentation or de-scription in medical charts; if imagingscanswereavailable, theywere reviewedbyattendingphysicians in respiratorymedicinewho extracted the data. Major disagree-ment between two reviewerswas resolvedby consultation with a third reviewer.

Study OutcomesThe composite end point was admissionto an ICU, the use of mechanical venti-lation, or death. These outcomes wereused in a previous study to assess theseverity of Covid-19 infectious disease (12).

Statistical AnalysisContinuous variables were expressed asmedians and interquartile ranges (IQRs)or simple ranges, as appropriate. Cate-gorical variables were summarized ascounts and percentages. Because thecohort of patients in our study was notderived from random selection, all statisticsare deemed to be descriptive only. Risk-adjusted Cox regression analysis curvesshow survival from severe disease through

days of hospitalization. Cox models wereadjusted forage, sex,BMI,bloodpressure,heart rate, cholesterol, HDL cholesterol,LDL cholesterol, triglyceride levels, heartdisease, hypertension, dyslipidemia, cur-rent smoking, b-blockers, ACE inhibitors,calcium inhibitors, thiazide diuretics, andaspirin. Kaplan-Meier survival analysis wasperformed in patients divided into thefollowing groups: normoglycemiawith-out diabetes, normoglycemia with di-abetes, hyperglycemia without diabetes,andhyperglycemiawith diabetes.P,0.05was considered statistically significant. Allcalculations were performed using SPSSversion 23 statistical software.

RESULTS

Baseline Characteristics of Patients onAdmission and OutcomeAll patients had moderate Covid-19 dis-ease (fever, respiratory tract symptoms,pneumonia on imaging) without the needfor invasive ventilation. All were treatedwith the standard Covid-19 protocol, in-cluding oxygen therapy, hydroxychloro-quine, and antiviral treatment (Table 1).Of the 59 study patients, 34 (57.6%) werenormoglycemic and 25 (42.4%) were hy-perglycemic (glucose .7.7 mmol/L). Atadmission, glycemiawas 6.36 0.66mmol/Linpatientswithnormoglycemiaand11.0461.22 mmol/L in those with hypergly-cemia. Eight (23.5%)patientswithnormo-glycemia and 18 (72%) with hyperglycemiahad a diagnosis of diabetes before hospi-talization. There were no differences in themean age, sex, BMI, sex distribution,smoking habits, levels of plasma choles-terol, and triglycerides among the groups.The use of diuretics, ACE inhibitors, sta-tins, and calcium channel blocker ther-apy was similar in all study groups (Table1). b-Blocker use was almost twice asfrequent in patients with hyperglycemiacompared with those with normoglyce-mia, and the use of angiotensin receptorblockers tended to be greater in thosewith an elevated glucose (Table 1).Among patients with a diagnosis of di-abetes before hospitalization, therewere no differences in antidiabetic drugsbeing taken (eg, insulin, oral drugs) whenthey were categorized as hyperglycemicandnormoglycemic (Table 1). Themediantime from illness onset (before admission)todischargeordeathwas18days(IQR14–20) in patients with hyperglycemia and16 days (IQR 14–19) in patients withnormoglycemia. At admission, IL-6 and

Figure 1—Flowchart of study population.

care.diabetesjournals.org Sardu and Associates 3

D-dimer levels were higher in patientswith hyperglycemia than in those withnormoglycemia (P , 0.001) (Fig. 2).Moreover, both IL-6 and D-dimer levelswere correlated with admission bloodglucose levels (Fig. 2). All patients hadinterstitial lungabnormalitiesonchestCTscans once admitted. In all patients, thetypical findings of chest CT images ofCovid-19 on admission showed bilateralground glass opacity without subseg-mental areas of consolidation or massshadows. Subsequent chest CT images(7 days later) revealed that pneumoniadisease progressed with subsegmentalareas of consolidation and with massshadows of high density in both lungs in10 (40%) patients with hyperglycemia and

3 (8.8%) with normoglycemia (P , 0.01).The composite end point occurred in13 (52%) patients with hyperglycemiaand 5 (14.7%) with normoglycemia (P ,0.01). Further details regarding the in-dividual components are provided inSupplementary Table 1. In a risk-adjustedCox regression analysis, both patientswith diabetes and patients with hyper-glycemia had a higher risk of severedisease than patients without diabetesand normoglycemia (Fig. 3). Moreover, inKaplan-Meier survival analysis, small num-bers of patients with hyperglycemia withor without previous diabetes were freefrom severe disease compared withpatients with normoglycemia withoutprevious diabetes (P , 0.02) (Fig. 3).

Glucose-Lowering Treatment andOutcomeAmong the 25 (42.4%) patients withglycemic levels .7.7 mmol/L, 15 weretreated with insulin infusion. Eleven(44%) patients with diabetes who werehyperglycemic and 4 (11.8%) patientswithout diabetes who were hyperglyce-mic were treated with insulin infusion.There were no statistically significantdifferences in clinical and laboratory dataamong patients treated with or withoutinsulin infusion (Table 1). At admission,blood glucose levels were 12.32 6 1.48mmol/L in insulin infusion–treated pa-tients and 11.06 6 1.98 mmol/L in noinsulin infusion–treated patients. In theinsulin infusion group, the mean time

Table 1—Baseline clinical characteristics of patients with Covid-19

Patients withnormoglycemia

Patients withhyperglycemia P value

Hyperglycemia withoutinsulin infusion

Hyperglycemia withinsulin infusion P value

Patients 34 (57.6) 25 (42.4) 10 (40) 15 (60)

Age (years) 66.6 6 11.5 68.5 6 5.8 0.468 68.9 6 6.0 68.2 6 5.9 0.776

Sex (M/F) 28/6 20/5 8/2 12/3

BMI (kg/m2) 27.9 6 1.6 27.5 6 1.3 0.251 27.3 6 1.4 27.6 6 1.3 0.599

Systolic BP (mmHg) 122.4 6 8.5 116.2 6 5.4 0.002 115.9 6 5.8 116.3 6 5.3 0.849

Diastolic BP (mmHg) 79.3 6 6.6 79.9 6 7.1 0.769 81.7 6 5.6 78.7 6 7.8 0.301

Heart rate (beats/min) 86.4 6 6.0 88.4 6 11.6 0.413 93.1 6 11.6 85.2 6 10.9 0.097

Risk factorsDiabetes 8 (23.5) 18 (72) 0.001 7 (70) 11 (73.3) 0.601Heart disease 7 (20.6) 5 (20) ,0.129 2 (20) 3 (20) ,0.488Hypertension 26 (76.5) 18 (72) 0.462 8 (80) 10 (66.7) 0.399Hyperlipemia 9 (26.5) 6 (24.4) 0.538 2 (20) 4 (26.7) 0.545Cigarette smoking 6 (17.6) 5 (20) 0.539 2 (20) 3 (20) 0.687

Active treatmentsb-Blockers 12 (35.3) 17 (68) 0.013 6 (60) 11 (73.3) 0.393ACE inhibitors 13 (38.2) 10 (40) 0.551 6 (60) 4 (26.7) 0.106ARBs 13 (38.2) 15 (60) 0.082 6 (60) 9 (60) 0.663Calcium inhibitors 6 (17.6) 6 (24) 0.390 3 (30) 3 (20) 0.455Statins 19 (55.9) 9 (36) 0.106 6 (60) 3 (20) 0.053Thiazide diuretics 7 (20.6) 5 (20) 0.610 1 (10) 4 (26.7) 0.313Insulin 4 (11.8) 3 (12) 0.287 1 (10) 2 (20) 0.468Oral antidiabetic drugs 7 (20.6) 17 (68) 0.052 7 (70) 10 (66.7) 0.118Aspirin 29 (85.3) 23 (92) 0.359 9 (90) 14 (93.3) 0.650Low-molecular-weight

heparin 6 (17.6) 6 (24) 0.390 2 (20) 4 (26.7) 0.545

Laboratory analysesPlasmaglucose(mmol/L) 6.3 6 0.99 11.04 6 2.06 ,0.001 11.06 6 1.98 12.32 6 1.48 0.792Cholesterol (mg/dL) 209.2 6 16.9 203.2 6 22.1 0.244 200.6 6 15.6 205.0 6 25.9 0.637LDL-C (mg/dL) 136.1 6 16.7 129.5 6 20.9 0.185 126.5 6 15.1 131.5 6 24.4 0.562HDL-C (mg/dL) 36.1 6 2.9 37.1 6 4.1 0.257 39.6 6 4.4 37.1 6 4.0 0.892Triglycerides (mg/dL) 187.1 6 23.9 190.9 6 28.1 0.577 186.2 6 21.1 194.1 6 32.2 0.501Creatinine (mg/dL) 1.0 6 0.18 0.9 6 0.15 0.083 0.9 6 0.14 0.9 6 0.15 0.555

Covid-19 treatmentsAntiviral drugs 33 (97.1) 24 (96) 0.436 10 (100) 14 (93.3) 0.880Hydroxychloroquine 34 (100) 25 (100) d 10 (100) 15 (100) dAntibiotics 33 (97.1) 25 (100) 0.985 10 (100) 15 (100) d

Oxygen therapy 8 (23.5) 6 (24) 0.744 2 (20) 3 (20) 0.455

Data are mean 6 SD or n (%). ARB, angiotensin receptor blocker; BP, blood pressure; HDL-C, HDL cholesterol; LDL-C, LDL cholesterol.

4 Glycemic Control in Patients With Covid-19 Diabetes Care

Figure2—A: IL-6 levels at admission, 1week, and2weeksandafterhospitalization inpatientswithhyperglycemiaandnormoglycemia.B: D-dimer levelsat admission, 1week, and 2weeks, and after hospitalization in patientswith hyperglycemia and normoglycemia. C: IL-6 levels at admission, 1week, and2 weeks and after hospitalization in patients with hyperglycemia treated with insulin infusion and those not treated with insulin infusion. D: D-dimerlevels at admission, 1week, and2weeksandafterhospitalization inpatientswithhyperglycemic treatedwith insulin infusionand thosenot treatedwithinsulin infusion. For panels A–D, box plots display the median, 25th and 75th percentiles, and range. E: Regression analysis between admission bloodglucose levels and admission IL-6 levels. F: Regression analysis between admission blood glucose levels and admission D-dimer levels. *P, 0.05 vs.normoglycemia and vs. baseline values; §P , 0.05 vs. baseline.

care.diabetesjournals.org Sardu and Associates 5

required to achieve the blood glucosetarget was 8.7 6 2.7 h, and the sub-sequent mean duration of the insulininfusion was 32.7 6 4.9 h. After insulininfusion, multidose insulin (three ormoredaily doses) was used in all patients ofboth groups. The mean glycemia duringhospitalizationwas 10.656 0.84mmol/Lin the no insulin infusion group and7.696 1.85 mmol/L in the insulin infusiongroup (P , 0.001). After the treatmentperiod, plasma glucose reduction wasgreater in the insulin infusion group thanin the no insulin infusion group (4.57 61.09 vs. 1.966 1.06 mmol/L; P, 0.001).During hospitalization, IL-6 and D-dimerlevelswerehigher in theno insulin infusiongroup compared with insulin infusion

patients (P , 0.001) (Fig. 2). In patientswith hyperglycemia, chest CT imagesduring hospitalization revealed that pneu-monia disease progressed with subseg-mental areas of consolidation and withmass shadows of high density in bothlungs in 7 (70%) of 10 patients withhyperglycemia without insulin infusionand 3 (20%) patients with hyperglycemiawith insulin infusion (P , 0.01). Thecomposite end point occurred in five(33%) patients in the insulin infusiongroup and in eight (80%) patients in theno insulin infusion group (P , 0.01). Ina risk-adjusted Cox regression analysis,patients with hyperglycemia treatedwithout insulin infusion had a higher riskof severe disease than those treated with

insulin infusion (Fig. 3), which includes in-creasedmortality (Supplementary Tables 1).

CONCLUSIONS

Our keymessage is that optimal glycemiccontrol during hospitalization has beenassociated with reduction risk of severediseaseanddeath inpatientswithCovid-19.Moreover, among patients screened forthe study, more patients in the hyper-glycemic group were excluded from thisanalysis because of severe disease (38%vs. 27%). Still, hyperglycemia remaineda strongprognostic predictor of outcomein hospitalized patients with Covid-19.Furthermore, patients with Covid-19who were hyperglycemic versus normo-glycemic displayed a higher cumulative

Figure 3—A: Risk-adjusted Cox regression analysis curves showing survival from severe disease through 18 days for patients with Covid-19 stratified byhyperglycemia vs. normoglycemia. B: Risk-adjusted Cox regression analysis curves showing survival from severe disease through 18 days for patientswith Covid-19 stratified by with diabetes vs. no diabetes. C: Risk-adjusted Cox regression analysis curves showing survival from severe disease through18 days for patients with hyperglycemia and Covid-19 stratified by insulin infusion vs. no insulin infusion treatment.D: Kaplan-Meier analysis showingsurvival from severe disease through 18 days for patientswith hyperglycemia and no diabetes, hyperglycemia and diabetes, no hyperglycemia butwithdiabetes, and no hyperglycemia and no diabetes. HR, hazard ratio.

6 Glycemic Control in Patients With Covid-19 Diabetes Care

incidence of severe disease. Moreover,insulin infusion–mediated optimal bloodglucose control improves prognosis forhospitalized patients with Covid-19 andhyperglycemia. Previous studies evidencedthat hyperglycemia has been linked topoor outcomes in acutely ill hospitalizedpatients (13). Possible mechanisms for thisincreasedmortality include hyperglycemia-inducedchanges inthe immunesystemandincreases in inflammatory cytokines. It isrelatively clear from preclinical and clinicalstudies that several features associatedwith diabetes influence host response toinfection. Hyperglycemia affects differ-ent components of the host response,including the function of immune cellsand regulation of cytokines (14). Serumconcentrations of both proinflammatorycytokines and anti-inflammatory cytokines,including IL-2R, IL-6, tumor necrosis factor-a, and IL-10, increased in the majority ofpatients with severe disease and weremarkedly higher than those with mod-erate disease, suggesting that cytokinestorms might be associated with diseaseseverity (15). Similarly, SARS was alsocharacterized by exuberant inflammatoryresponses and lung damage. A previousstudy using a mouse model of SARSdemonstrated that rapid kinetics ofSARS-CoV replication and delay in IFN-Isignaling promoted inflammatory mono-cyte-macrophage accumulation, result-ing in elevated lung cytokine/chemokinelevels, vascular leakage, and suboptimalT-cell responses (16). Interestingly, weobserved that patients with hyperglyce-mia presented higher IL-6 levels com-pared with those with normoglycemia.Moreover, higher plasma blood glucoselevels on admission were associated withhigher plasma IL-6 levels. Despite fulltherapy for Covid-19 infection, patientswith hyperglycemia presented with higherlevels of IL-6 compared with those withnormoglycemia during hospitalization.Thus, elevated blood glucose may itselfcause an inflammatory response, leadingto severe Covid-19 disease and death.Onthe other hand, patients with Covid-19–associated pneumonia exhibit a numberof abnormal coagulation parameters (17),and coagulation abnormalities have beenassociated with a higher mortality rate(18,19). Patterns of disseminated intra-vascular coagulation were reported indeaths, and within this group, theD-dimer levels were higher (18,19). In-terestingly, we observed that patients

with hyperglycemia who progressed tosevere disease presented higher D-dimerlevels compared with those with normo-glycemia.Moreover, higher blood glucoselevels on admission were associatedwith higher D-dimer at admission. De-spite full therapy for Covid-19 infection,patients with hyperglycemia presentedwith higher levels of D-dimer comparedwith patients with normoglycemia duringhospitalization. Thus, elevated blood glu-cose may itself cause an inflammatoryresponse and an abnormal coagulationsystem, leading to severe Covid-19 diseaseand death.

The present findings mainly show aprotective effectof tight glycemic controlon outcomes of patients with hypergly-cemia with Covid-19 infection. Indeed,our observations evidence that a moresubstantial drop in glucose levels, ob-tained by insulin infusion, is associatedwith better outcomes in patients withCovid-19. As background for this associ-ation, we observed that insulin-treatedpatients with hyperglycemia reached op-timal glucose levels and low levels ofbothIL-6 and D-dimer, and thus, they had alowriskof severediseaseanddeathalongwith the hospitalization. Previous studiesevidenced that cytokine levels returnedto normal after insulin infusion and res-olution of the hyperglycemic crisis, re-ducing the risk of death (20,21). Afteradjustment for baseline glucose andother clinical predictors, we found thatfor every 0.56 mmol/L drop in glucoselevel between admission and 18 days,there was an 11% relative decrease insevere disease risk in patients with hy-perglycemia. However, this relation wasnot evidenced in patients with baselineglucose,7.7 mmol/L. Against these data,previous studies and a meta-analysis ofrandomized trials of intensive insulintherapy in critically ill patients failed tofind any benefit of tight glycemic controlfor all-cause mortality; moreover, tightversus mild glycemic control increasedthe frequency of mild and/or severehypoglycemia by about fivefold. All strat-ified analyses of mortality (by ICU type[medical, surgical, or mixed], time period[ICU stay,hospital stay, 28days, 3months,or 6months], or the presence of diabetes)did not identify any significant differencesamong the glycemic control groups (22).However, although the investigators didnot observe a significant reduction inoverallmortality in patients receiving the

insulin infusion, they suggested that theclustered ranking plot suggests that mildglycemic control (140 to ,180 mg/dL)achieves the best outcome in relation toall-cause mortality and hypoglycemia,which is consistent with the ADA (9) andthe American Association of Clinical En-docrinologists/ADA target glucose levels(23).

Our real-life study needs to extendour observations to a larger cohort ofrandomized patients. Because therewas no randomization of insulin infu-sion treatment, comparisonof thepatientsreceiving and not receiving insulin infu-sion cannot be assumed to be causal butcan be considered highly suggestive. Inthe small numbers available, lack ofsignificance between the differencesdoes not mean lack of important differ-ences. However, to the best of ourknowledge, this is the largest observationalstudy among patients with hyperglycemiawith Covid-19 who have experienced adefinite outcome. Our data evidencedthat optimal glucose control in the im-mediate postadmission period for almost18 days was associated with a significantreduction of inflammatory cytokines andprocoagulative status. Because inflamma-tory cytokines and procoagulative statushavebeenshown to inducepooroutcomein patients with Covid-19, we speculatethat optimal glycemic control, by reduc-ing IL-6 and D-dimer levels, may reducethe risk of progression of the infectiousdisease. Thus, in the critical care setting,insulin infusion may be an effectivemethod for achieving glycemic targetsand reducing mortality in patients withCovid-19.

Acknowledgments. The authors are grateful toLuigi Ruggiero (Medical Assistant to GeriatricUnit of Universita degli Studi della Campania“Luigi Vanvitelli”) for important work in theclinical evaluation of patients admitted at ourdepartment. The authors are also grateful toFedericaMiglietta (Medical Assistant toGeriatricUnit of Universita degli Studi della Campania“Luigi Vanvitelli”) for critical analysis of clinicaldata.Funded. Support for this study was by a PRIN2017 grant 2017FM74HK_002.Duality of Interest. No potential conflicts ofinterest relevant to this article were reported.Author Contributions. C.S. researched dataand contributed to the discussion: N.D., M.L.B.,and M.B. researched data. M.R.R. contributed tothe discussion. V.M., P.M., and N.C. researcheddata and contributed to the discussion. G.P.reviewed/edited themanuscript. R.M.wrote the

care.diabetesjournals.org Sardu and Associates 7

manuscript.R.M. is theguarantorof thisworkand,as such, had full access to all the data in the studyand takes responsibility for the integrity of thedata and the accuracy of the data analysis.

References1. van Vught LA, Wiewel MA, Klein KlouwenbergPM, et al. Admission hyperglycemia in critically illsepsispatients:associationwithoutcomeandhostresponse. Crit Care Med 2016;44:1338–13462. Khateeb J, Fuchs E, Khamaisi M. Diabetes andlung disease: a neglected relationship. Rev Dia-bet Stud 2019;15:1–153. Thompson BT. Glucose control in sepsis. ClinChest Med 2008;29:713–720, x4. Tiwari S, Pratyush DD, Gahlot A, Singh SK.Sepsis in diabetes: a bad duo. Diabetes MetabSyndr 2011;5:222–2275. Esposito K, Nappo F, Marfella R, et al. In-flammatory cytokine concentrations are acutelyincreased by hyperglycemia in humans: role ofoxidative stress. Circulation 2002;106:2067–20726. Nieuwdorp M, van Haeften TW, GouverneurMC, et al. Loss of endothelial glycocalyx duringacute hyperglycemia coincides with endothelialdysfunction and coagulation activation in vivo.Diabetes 2006;55:480–4867. Hendren NS, Drazner MH, Bozkurt B, CooperLT Jr. Description and proposed management ofthe acute COVID-19 cardiovascular syndrome.Circulation. 16 April 2020 [Epub ahead of print]DOI: 10.1161/CIRCULATIONAHA.120.0473498. Porfidia A, Pola R. Venous thromboembolisminCOVID-19patients. J ThrombHaemost.15April

2020 [Epub ahead of print] DOI: 10.1111/jth.148429. American Diabetes Association. 14. DiabetesCare in the Hospital. Standards ofmedical care indiabetes-2018. Diabetes Care 2018;41(Suppl. 1):144–15110. Rhodes A, Evans LE, Alhazzani W, et al.Surviving sepsis campaign: international guide-lines for management of sepsis and septic shock:2016. Intensive Care Med 2017;43:304–37711. Cheng ZJ, Shan J. 2019 Novel coronavirus:whereweare andwhatwe know. Infection 2020;48:155–16312. GuanWJ,Ni ZY, HuY, et al. Clinical character-istics of coronavirus disease 2019 in China. N EnglJ Med 2020;382:1708–172013. Dombrowski NC, Karounos DG. Pathophysiol-ogy andmanagement strategies for hyperglycemiafor patients with acute illness during and followinga hospital stay. Metabolism 2013;62:326–33614. Schuetz P, Castro P, ShapiroNI. Diabetes andsepsis: preclinical findings and clinical relevance.Diabetes Care 2011;34:771–77815. Chen G, Wu D, Guo W, et al. Clinical andimmunological features of severe and moderatecoronavirus disease 2019. J Clin Invest 2020;130:2620–262916. Channappanavar R, Fehr AR, Vijay R, et al.Dysregulated type I interferon and inflammatorymonocyte-macrophage responses cause lethalpneumonia in SARS-CoV-infectedmice. Cell HostMicrobe 2016;19:181–19317. Gao Y, Li T, HanM, et al. Diagnostic utility ofclinical laboratory data determinations for pa-tients with the severe COVID-19. J Med Virol. 17

March 2020 [Epub ahead of print] DOI: 10.1002/jmv.2577018. Tang N, Bai H, Chen X, Gong J, Li D, Sun Z.Anticoagulant treatment is associated withdecreased mortality in severe coronavirus disease2019 patients with coagulopathy. J Thromb Hae-most 2020;18:1094–109919. Toh CH, Alhamdi Y, Abrams ST. Currentpathological and laboratory considerations in thediagnosis of disseminated intravascular coagu-lation. Ann Lab Med 2016;36:505–51220. Chaudhuri A, Umpierrez GE. Oxidative stressand inflammation in hyperglycemic crises andresolution with insulin: implications for theacute and chronic complications of hypergly-cemia. J Diabetes Complications 2012;26:257–25821. Marfella R, Di Filippo C, PortogheseM, et al.Tight glycemic control reduces heart inflamma-tion and remodeling during acute myocardialinfarction in hyperglycemic patients. J Am CollCardiol 2009;53:1425–143622. Yamada T, Shojima N, Noma H, Yamauchi T,Kadowaki T. Glycemic control, mortality, andhypoglycemia incritically illpatients:asystematicreview and network meta-analysis of random-ized controlled trials. Intensive Care Med 2017;43:1–1523. Garber AJ, Abrahamson MJ, Barzilay JI,et al. Consensus statement by the AmericanAssociation of Clinical Endocrinologists andAmerican College of Endocrinology on thecomprehensive type 2 diabetes managementalgorithm – 2019 Executive summary. EndocrPract 2019;25:69–100

8 Glycemic Control in Patients With Covid-19 Diabetes Care