Effect of A1C and Glucose onPostoperative Mortality inNoncardiac and Cardiac SurgeriesDiabetes Care 2018;41:782–788 | https://doi.org/10.2337/dc17-2232

OBJECTIVE

HemoglobinA1c (A1C) is used in assessment of patients for elective surgeries becausehyperglycemia increases risk of adverse events. However, the interplay of A1C,glucose, and surgical outcomes remains unclarified, with often only two of these threefactors considered simultaneously. We assessed the association of preoperative A1Cwith perioperative glucose control and their relationship with 30-day mortality.

RESEARCH DESIGN AND METHODS

Retrospective analysis on 431,480 surgeries within the Duke University Health Sys-tem determined the association of preoperative A1C with perioperative glucose(averaged over the first 3 postoperative days) and 30-day mortality among 6,684noncardiac and 6,393 cardiac surgeries with A1C and glucose measurements. Ageneralized additive model was used, enabling nonlinear relationships.

RESULTS

A1C and glucose were strongly associated. Glucose and mortality were positivelyassociated for noncardiac cases: 1.0% mortality at mean glucose of 100 mg/dL and1.6% at mean glucose of 200 mg/dL. For cardiac procedures, there was a strikingU-shaped relationship between glucose andmortality, ranging from4.5%at 100mg/dLto a nadir of 1.5% at 140mg/dL and rising again to 6.9% at 200mg/dL. A1C and 30-daymortality were not associated when controlling for glucose in noncardiac or cardiacprocedures.

CONCLUSIONS

Although A1C is positively associated with perioperative glucose, it is not associatedwith increased 30-day mortality after controlling for glucose. Perioperative glucosepredicts 30-day mortality, linearly in noncardiac and nonlinearly in cardiac proce-dures. This confirms that perioperative glucose control is related to surgical out-comes but that A1C, reflecting antecedent glycemia, is a less useful predictor.

Hyperglycemia in the perioperative period is associated with adverse outcomes. Therelationship between hyperglycemia and stress is well established, with higher rates of1) infection, 2) delayed wound healing, 3) neurologic injuries, and 4) postoperativemortality (1). Increased insulin resistance induced by surgical stress and nociceptivesignals during surgery are significant contributing factors (2,3). Gandhi et al. (4) found a30% increase in rate of adverse postoperative events for every 20-mg/dL increase inintraoperative glucose level. Attempts to study the problem are frustrated by the factthat the definition of intraoperative hyperglycemia has been highly variable in bothrange and format (5).

1Department of Statistical Science, Duke Univer-sity, Durham, NC2Department of Anesthesiology, Duke UniversitySchool of Medicine, Durham, NC3Division of Endocrinology, Metabolism, and Nu-trition, Department of Medicine, Duke UniversitySchool of Medicine, Durham, NC4Department of Neurobiology, Duke UniversitySchool of Medicine, Durham, NC

Corresponding author: Rebecca A. Schroeder,rebecca.schroeder@duke.edu.

Received 24 October 2017 and accepted 16January 2018.

This article contains Supplementary Data onlineat http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc17-2232/-/DC1.

This article is featured in a podcast available athttp://www.diabetesjournals.org/content/diabetes-core-update-podcasts.

© 2018 by the American Diabetes Association.Readers may use this article as long as the workis properly cited, the use is educational and notfor profit, and the work is not altered. More infor-mation is available at http://www.diabetesjournals.org/content/license.

Willem van den Boom,1

Rebecca A. Schroeder,2

Michael W. Manning,2 Tracy L. Setji,3

Gic-Owens Fiestan,4 and David B. Dunson1

782 Diabetes Care Volume 41, April 2018

EPIDEM

IOLO

GY/HEA

LTHSERVICES

RESEA

RCH

Hemoglobin A1c (A1C) level gives an in-directmeasurement of howeffectively anindividual’s blood glucose is controlled(6,7) and broadly predicts serum glucose(8). It is therefore often used to determineif a patient should be allowed to proceedwith elective surgical procedures (9), eventhough its relationwith poor postsurgical out-comes is less clear (6,10–13). Subramaniametal. (14) contend that glycemic variabilitywith elevated preoperative A1C predictsadverse outcomes in surgery, and indeed,glycemic variability does appear detri-mental (15,16). Conversely, Ambiru et al.(17) find that A1C is not an independentpredictor of infection when jointly ana-lyzed with blood glucose level. Interest-ingly, both high and low A1C levels areassociated with poor outcomes followingsurgery (18–20), suggesting that the re-lationship follows some nonlinear shapesuch as a parabola.Despite numerous studies into the ef-

fect of A1C and hyperglycemia on postop-erative outcomes (21,22), most considerhyperglycemia alone (1,23–26) or A1Calone (13,20,27). When studied together,they aremost often not in amultivariableanalysis (28,29) in which one can assessthe effect of an indicatorwhile controllingfor another. Such analysis is pertinentwhen considering A1C and glucose be-cause they are correlated. A thoroughunderstanding of the interplay of pre-operative A1C, perioperative glucose,andpostoperativeoutcomes should informthe practice of A1C screening for electivesurgery. Therefore, we investigate the as-sociation of A1C with postoperative mor-tality while controlling for perioperativeglucose level.Weused a large, single-centerdatabase including a wide variety of sur-gical procedures to examine these rela-tionships. Patients undergoing cardiacand noncardiac procedures were analyzedseparately.

RESEARCH DESIGN AND METHODS

Data DescriptionFollowing institutional review board ap-proval, de-identified data from 246,650patients aged$18 years who underwenta total of 431,480 surgical procedures atDuke University Health System from 1999through 2013 were obtained from a data-base of electronic medical records for ret-rospective analysis. Figure 1 describes theselection of the final 13,077 procedures.The primary clinical outcomeswere av-

erage perioperative blood glucose level

and30-daypostoperativemortality. Aver-age perioperative blood glucose level wascomputed as the mean of glucose meas-urements from the day of surgery untilpostoperative day 3. Using the medianpostoperative glucose instead to obviatethe effect of outlier values does not ma-terially change our findings. Thirty-daypostoperative mortality was defined tooccur whenever the postoperative statusin the enterprise datawarehousewas anyoption other than “alive” with a datewithin 30 days of surgery. Statuses otherthan “alive”were all labels correspondingto death such as “presumeddead” or “ex-pired at Duke.” Analogously, we consid-ered 3-year and time-unlimited mortalitybased on medical records up to 2015.

We chose mortality as the measure ofadverse postoperative outcome becauseit was, unlike other outcomes, reliablycaptured in the data. Our large databaseallowed the use of mortality, whereas

smaller studies often lack statistical powerto get results for such a rare outcome. Asan outcome variable, mortality is very in-formative. Most interventions in medicineare focused on delaying mortality, and ithas virtually no diagnostic error, unlikeother outcomes. For instance, length ofstay is more subjective as it is affected bymany nonphysiological determinates in-cluding bed availability, transportation, ef-ficiency of the discharge process, andvariability in human decision-making.

The database did not contain reliableinformation on important confounders ofglycemia or mortality such as comorbid-ities. To reduce confounding within theseconfines, the analysis controlled for typeof procedure as provided by primary ICD-9 procedure codes. There were 680 non-cardiac and 61 cardiac distinct ICD-9 codespresent. To minimize the number of dis-tinct types of surgery in the analysis, onlythe most common ICD-9 codes were

Figure 1—Case inclusion flow chart. Visual representation of how the 6,684 noncardiac and 6,393cardiac cases analyzed were selected from the 431,480 surgical procedures available.

care.diabetesjournals.org van den Boom and Associates 783

included, comprising distinct ICD-9 codesfor199noncardiacand9cardiacproceduresand accounting for 80%of the total numberof operations. Age, BMI, and sex served asadditional controls. Surgeries were groupedas noncardiac and cardiac for analysis be-cause of the prevalence of cardiopulmonarybypass and insulin use in cardiac surgeries.Diabetes is an important confounder of

both preoperative A1C and perioperativeglucose. Unfortunately, our data capturedthis via ICD coding with an implausibly lowdiagnosis rate.Therefore,wedidnot includeit in ourmain analysis. Results for the subsetwith a diabetes diagnosis are mentionedand provided in the Supplementary Data.The primary independent variable in

the analysis was preoperative A1C, mea-sured closest to but not.90 days prior tothe day of surgery. Eliminating cases lack-ing A1C measurements within 90 days ofsurgery or glucose values in the 3 daysstarting with the surgery date left 6,684noncardiac and 6,393 cardiac procedures;missingA1Cmeasurementswere themostcommon cause of culling cases (Fig. 1).During the period examined, measure-ment of A1C was not used as a standardmethod for screening patients without di-abetes for surgery. Supplementary Fig. 1shows that the A1C measurements wereclosertothesurgeryforcardiacthanfornon-cardiac procedures. Table 1 shows sum-mary statistics for various data subsets.

Statistical AnalysisPrevious studies (18,19) suggest a nonlin-ear relationship betweenA1C andmortal-ity. Generalized additive models (30), atype of multivariable regression, allowfor such nonlinearity. We include proce-dure type as a random effect to capturethe correlation between similar caseswhile mitigating biases due to differencesbetween procedure types. Generalizedadditive models with random effects arereferred to as generalized additivemixed-effects models. We use such models toestimate the relation between preopera-tive A1C and both average perioperativeglucose and postoperative mortality whilecontrolling for age, BMI, sex, proceduretype, and, for mortality, average perioper-ative glucose. The results when not con-trolling for procedure type are similar tothose presented in this study with all con-clusions being the same.Generalized additive models were fit

using R version 3.2.3 with the package“mgcv” version 1.8–9. An identity link

withGaussian errorswasused for averageperioperative glucose, whereas a logistic-binomial linkwasused formortality.All con-tinuous predictors were treated as havingpotentially nonlinear effects. Note that ouranalysis did not formally test for differencesbetween noncardiac and cardiac proce-dures.

RESULTS

Average Perioperative Blood GlucoseLevelFigure 2 shows the association betweenaverage perioperative glucose and A1C

from the generalized additive model.There is a significant, positive, nonlinearrelationship between A1C and averageperioperative glucose (P , 0.001). Therelationship is virtually linear betweenan A1C of 6% (42 mmol/mol) and 7.5%(59 mmol/mol), with flattening of thecurve at higher A1C values. The relation-ship of A1C with average glucose is stron-ger for patients undergoing noncardiacthan cardiac procedures, with significantlygreater variation in average perioperativeglucose (Fig. 2) as previously shown inTable 1.

Table 1—Demographic and clinical characteristics

Characteristic Mean or n (%) SD

Quartiles

Q1 Median Q3

Data set: 167,376 noncardiac proceduresnot constrained by havingpreoperative A1C and perioperativeglucose measured or by proceduretype

Age (years) 53 18 39 55 67BMI 29 6.6 24 28 3230-day postoperative mortality 1,456 (0.9%)Female 96,873 (58%)

Data set: 6,684 noncardiac proceduresconstrained by having preoperativeA1C and perioperative glucosemeasured and being among199 types considered

Age (years) 60 15 50 62 71BMI 32 7.7 26 30 3630-day postoperative mortality 156 (2.3%)Female 3,518 (53%)Average perioperative glucose (mg/dL) 149 42 118 142 172No. of perioperative glucose

measurements 11 11 2 8 15A1C (%|mmol/mol) 7.1|54 1.7|19 5.9|41 6.6|49 7.8|62

Data set: 12,199 cardiac procedures notconstrained by having preoperativeA1C and perioperative glucosemeasured or by procedure type

Age 62 14 54 63 72BMI 29 6.0 25 28 3230-day postoperative mortality 273 (2.2%)Female 4,557 (37%)

Data set: 6,393 cardiac proceduresconstrained by having preoperativeA1C and perioperative glucosemeasured and being among 9 typesconsidered

Age 64 12 56 65 73BMI 29 6.0 25 28 3230-day postoperative mortality 116 (1.8%)Female 2,257 (35%)Average perioperative glucose (mg/dL) 146 19 134 145 156No. of perioperative glucose

measurements 32 13 24 33 42A1C (%|mmol/mol) 6.4|46 1.4|15 5.6|38 6.0|42 6.7|50

Summary statistics of demographic and clinical characteristics for the cardiac and noncardiac setsof surgeries not constrained and constrained by having both preoperative A1C and perioperativeglucose measured and by being one of the procedure types considered.

784 Effect of A1C and Glucose on Mortality Diabetes Care Volume 41, April 2018

Supplementary Fig. 2 illustrates thesame analysis but only includes caseswith a diabetes diagnosis as documentedthrough ICD codes in the data. The maindifference with Fig. 2 is that the perioper-ative glucose is higher at low levels ofpreoperative A1C.

30-Day Postoperative MortalityFigure 3 shows the associations betweenpostoperativemortality and A1C, averageperioperative glucose, age, and BMI.Supplementary Tables 1 and 2 providethe results in tabular format, also whennot controlling for glucose. For noncar-diac surgeries, increased mortality wasassociated with female sex (odds ratio1.8 [95% CI 1.2–2.7]; P = 0.004), very lowBMI (P = 0.001), lower A1C (P = 0.01), in-creased age (P , 0.001), and increasedaverage perioperative glucose (P = 0.04)after controlling for other predictors. Incardiac surgeries, increased mortalitywas associated with increased age (P ,0.001), BMI (P = 0.001), and both low andhigh average glucose (P, 0.001). Of note,A1C was not a significant (P = 0.88) predic-tor of postoperative mortality in cardiacsurgeries.The results demonstrate a clear effect

of glycemic control on postoperative mor-tality. Furthermore, they illustrate a nota-bly different relationship in noncardiacversus cardiac procedures. The associationbetween average perioperative glucose

and 30-day mortality was linear for noncar-diac procedures, whereas it was U-shapedfor cardiac procedures and was stronglyassociated with 30-day mortality outsideof 120–160 mg/dL. The mortality levelscorresponding to a glucose of 100, 140,and 200 mg/dL were 1.0% (95% CI 0.7–1.6%), 1.3% (95% CI 0.9–1.8%), and 1.6%(95% CI 1.1–2.3%) for noncardiac and4.5% (95% CI 2.3–8.7%), 1.5% (95% CI1.0–2.1%), and 6.9% (95% CI 3.3–14.1%)for cardiac cases, respectively. Controllingfor blood glucose levels, A1C did not pre-dict increased 30-day mortality despitethe strong relationship between A1Cand glucose in Fig. 2. In fact, A1Cexhibiteda negative association with 30-day mor-tality in noncardiac procedures after ad-justment for glucose and other factors.

Supplementary Fig. 3 shows that therewas no evidence of a difference in thecases studied in this study versus thosewith a diabetes diagnoses as documentedvia ICD codes in the data.

The Effect of Preoperative A1C,Different Time Windows, andPostoperative MortalityA1C was negatively associated with30-day mortality in noncardiac proce-dures after controlling for glucose, age,BMI, and sex. This changed for longertime windows for mortality. Supplemen-tary Fig. 4 contains similar A1C plots asFig. 3 with probability of 3-year and time-

unlimitedmortalityonthey-axis. As the timewindow increased, the association of A1Cwithmortality becamemore positive. Thiswasdespiteusing the conservative approachof treating right-censored individuals as sur-vivors: mortality records continued up to2015, whereas the last procedure occurred,3years prior, in 2013.Whennot controllingfor perioperative blood glucose, the trend ofmortality versus A1C, as shown in Fig. 3 andSupplementary Fig. 2, remained largelythe same for noncardiac surgeries. How-ever, the association became more posi-tive for cardiac procedures than whencontrolling for perioperative glucose.

CONCLUSIONS

A retrospective analysis of electronicmedical records from the Duke UniversityHealth System demonstrates that preop-erative A1C predicts average periopera-tive blood glucose (3 postoperative dayaverage) and that the average periopera-tive glucose predicts 30-day mortality.However, after controlling for age, BMI,sex, and perioperative glucose, elevatedpreoperative A1Cwas not positively asso-ciated with 30-daymortality. These state-ments held across both noncardiac andcardiac procedures, although the natureof the relationships differed betweennoncardiac and cardiac procedures.

The expected positive association be-tween A1C and average perioperativeglucose was present but not linear. Thisdeviates from the common linear trans-lation (8) of A1C to average glucose andwas different between cardiac andnoncar-diac surgical populations. These differen-ces are likely due to aggressive treatmentof hyperglycemia with intravenous insulinin the perioperative setting in patients un-dergoing cardiac surgery.

Our results clearly demonstrate an as-sociation between average perioperativeblood glucose levels and 30-day mortal-ity. In noncardiac surgeries, higher aver-age blood glucose was associated withhigher 30-day mortality. The relationshipbetween average glucose and mortalitywas notably different in patients under-going cardiac surgery. We found a strik-ing U-shaped curve in which averageperioperative blood glucose of ,120 or.160mg/dLwasassociatedwithamarkedincrease in 30-day mortality in patientsundergoing cardiac surgery. The distinctglucose-mortality relationships betweencardiac and noncardiac surgery suggest

Figure 2—Averageperioperative glucose level versus preoperativeA1Cpercentage. Visual summaryof the association between preoperative A1C and the average of glucose from the day of surgerythrough postoperative day 2 from the generalized additive model fit on the 6,684 noncardiac (left)and 6,393 cardiac surgeries (right). The solid line is the mean prediction and the dashed lines the95% CIs.

care.diabetesjournals.org van den Boom and Associates 785

a complex, variable interaction betweenglucose metabolism and postoperativerecovery.The U-shaped curve in cardiac surgery

confirms the previously established harmof both perioperative hypoglycemia andhyperglycemia. Current American Diabe-tes Association guidelines recommend atarget glucose range of 80–180 mg/dL inthe perioperative period, with consider-ation of a lower target (,140 mg/dL) inselect patients such as those who under-went cardiac surgery if the goal can beobtained without hypoglycemia (31).The Canadian Diabetes Association men-tions a target range of 90–180mg/dL (32).Further, the Society of Thoracic Surgeonsrecommends maintaining blood glucosein patients undergoing cardiac surgery,180 mg/dL irrespective of a diagnosisof diabetes (33). These targets are slightly

more liberal than our results would sug-gest for patients who have had cardiacsurgery, particularly regarding values inthe lower end of the target range.

Our finding of increased mortality withtightly controlled blood glucoses is similarto the findings demonstrated in the Nor-moglycemia in Intensive Care Evaluation-Survival Using Glucose Algorithm Regulation(NICE-SUGAR) (34) trial that showedhigher mortality in critically ill patientsrandomized to intense blood glucose con-trol (81–108 mg/dL) compared with amore moderate target of ,180 mg/dL.Hypoglycemia itself increases risk of mor-tality. Themechanism for increasedmortal-ity in patients with tightly controlled bloodglucoses may be related to adverse car-diovascular effects from lowering glucosewith insulin (34). Experimental studieshave shown that insulin and hypoglycemia

can cause vasodilatation, hypotension,and exhaustion of the sympathetic systemresponse. Further, the risk of frank hypo-glycemia is higher when glucoses aretightly controlled, and recent hypoglyce-mia reduces the physiological responseto subsequent hypoglycemia (35). Furtherresearch is needed to understand thiscomplex interplay of insulin, hypoglyce-mia, and cardiovascular effects.

There have been conflicting resultswhen looking at the association betweenincreased A1C and mortality, with someinvestigators demonstrating increasedmortality (6,12–14), whereas others(10,11,36) have not. Unfortunately, thesestudies examined small groupsof patientsundergoing a specific type of surgery orconsidered one rare outcome, whereasour data included many patients under-going a broad variety of procedures. In ourstudy, an elevated preoperative A1C wasnot predictive of increased 30-day mortal-ity in either patients who had noncardiacsurgery or patients who had cardiac sur-gery after controlling for blood glucose,despite the strong positive associationbetween A1C and average perioperativeglucose. These findings suggest that peri-operative glycemic control may neutralizethe effect of high A1C on postoperativemortality. It is possible that a high A1C,signifying uncontrolled or more difficultto control disease,might promptmore vig-ilance and aggressive treatment from peri-operative clinicians.

Interestingly, a negative associationwas found between preoperative A1Cand 30-day mortality in noncardiac sur-geries. This is probably driven by verylow A1C corresponding with high postop-erative mortality, as found previously(18,19). Further, low BMI correspondedwith higher mortality in noncardiac pro-cedures similar to the trend noted in amultivariate analysis in cancer survival(37). Baseline nutritional status or de-creased physiological reserve for anotherreason may be contributors. Additionally,the negative association between A1Cand mortality in noncardiac surgeries maybe partially confounded by acute physicianinterventiondfor instance, if those withhigher A1C are treated more aggressivelyin the perioperative and close postopera-tive periods. To determine the long-termassociation of A1C and mortality in lightof no or even negative short-term effect,we extended the time frame for mortal-ity. Over subsequent years, an elevated

Figure 3—Probability of postoperativemortality (Pr. of postop. mort.) versus A1C, glucose, age, andBMI. Visual summary of the associations between 30-day postoperative mortality and preoperativeA1C, average of glucosemeasurements from day of surgery up to postoperative day 2, age, and BMIfrom the generalized additive logistic model fit on the 6,684 noncardiac (left) and 6,393 cardiacsurgeries (right). The solid line is the estimated probability and the dashed lines the 95% CIs.

786 Effect of A1C and Glucose on Mortality Diabetes Care Volume 41, April 2018

preoperative A1Cwas associatedwith long-term mortality in our patients, which is con-sistentwithexisting literature (18), providingfurther validation of our findings.There are limitations toour study. These

are retrospective data extracted fromelec-tronic medical records and, as such, sub-ject to knownweaknesses and biases. Lessthan one-tenth of the patients who hadnoncardiac procedures had preoperativeA1C. Table 1 shows that those who didare older, more overweight, have highermortality, and are more often male. Exis-tence of an A1Cmeasurement may reflectthe diagnosis of diabetes or concern forthe health status of a patient. We are,however, most interested in this subsetof the population as we believe that thiswill better inform practice within thisgroup. This relates to the limitation thatwe were unable to definitely say whichof these patients carried a preoperativediagnosis of diabetes orwere receiving gly-cemic therapy, both known to influencethe relation between hyperglycemia andpostoperative outcomes (1,38). Also, asother important adverse outcomes werenot reliably captured, we limited our-selves to mortality.Similarly, observational studies like this

cannot prove causation. We are unableto state that improved perioperative glu-cose control is directly responsible fordecreased mortality, only that they areassociated. Interestingly, a patient who isdeferred for elective surgery due to an el-evated A1C may well be accepted afterachieving a more favorable A1C target.Whether this actually improves their riskprofile or the odds of achieving successfulperioperative glucose control is unknown.Although previous randomized studies(39) have established the effect of specificblood glucose management strategies onglucose and A1C, they did not focus onmortality as an outcome. Such work isnecessary to determine the effectivenessof using A1C as a screening tool in theelective surgery population.We also recognize that the measure-

ment of an A1C level has inherent limita-tions. These include possible racial andethnic differences, particular hemoglo-binopathies, and other conditions thatshorten the life span of the erythrocyteor otherwise falsely lower or elevate A1C(40). Also, there are many factors thatimpact perioperative glucose controlthat were not examined in this analysisincluding but not limited to the use of

cardiopulmonary bypass; choice of phar-macologic agent, dose, and strategy to treathyperglycemia; intensity of nursing care;and patient comorbidities.

In conclusion, we sought to clarify therelationships among preoperative A1C,perioperative glucose control, and post-operative mortality as they inform thecurrent practice of using preoperativeA1C as a screen in offering elective sur-gery. PreoperativeA1Cwas indeed a goodpredictor of a patient’s average perioper-ative glucose level. Further, perioperativeglucose was associated with 30-day mor-tality, linearly in noncardiac and stronglyU-shaped in cardiac procedures. However,taking age, BMI, andperioperative glucoseinto account, A1C did not predict 30-daymortalityother thanatextremely low levelsand, even then, only in patients who hadnoncardiac surgery. Thus, our results sug-gest that perioperative glucose controlmay bemore important than preoperativeA1C in predicting 30-day postoperativemortality.

Funding.W.v.d.B. andD.B.D.weresupportedbyAccenture PLC for this research.Duality of Interest. No potential conflicts of in-terest relevant to this article were reported.Author Contributions. W.v.d.B. and G.-O.F.conducted the analysis of data and produced aninitialdraftof themanuscript.R.A.S. conceived thestudy. R.A.S., M.W.M., T.L.S., and D.B.D. contrib-uted to the interpretation of the results andeditorial review and revision of the manuscript.W.v.d.B. is the guarantor of thiswork and, as such,had full access to all of the data in the study andtakesresponsibility for the integrityof thedataandthe accuracy of the data analysis.Prior Presentation. Parts of this study werepresented at the 77th Scientific Sessions of theAmerican Diabetes Association, San Diego, CA,9–13 June 2017.

References1. Kwon S, Thompson R, Dellinger P, Yanez D,Farrohki E, Flum D. Importance of perioperativeglycemic control in general surgery: a report fromthe Surgical Care and Outcomes Assessment Pro-gram. Ann Surg 2013;257:8–142. Collier B, Dossett LA, May AK, Diaz JJ. Glucosecontrol and the inflammatory response. Nutr ClinPract 2008;23:3–153. Thorell A, Nygren J, Ljungqvist O. Insulin resis-tance: a marker of surgical stress. Curr Opin ClinNutr Metab Care 1999;2:69–784. Gandhi GY, Nuttall GA, Abel MD, et al. Intra-operative hyperglycemia and perioperative out-comes in cardiac surgery patients. Mayo ClinProc 2005;80:862–8665. Puskas F, Grocott HP, White WD, Mathew JP,Newman MF, Bar-Yosef S. Intraoperative hyper-glycemia and cognitive decline after CABG. AnnThorac Surg 2007;84:1467–1473

6. Jupiter DC, Humphers JM, ShibuyaN. Trends inpostoperative infection rates and their relation-ship to glycosylated hemoglobin levels in diabeticpatients undergoing foot and ankle surgery. J FootAnkle Surg 2014;53:307–3117. Sacks DB. Correlation between hemoglobinA1c (HbA1c) and average blood glucose: canHbA1c be reported as estimated blood glucoseconcentration? J Diabetes Sci Technol 2007;1:801–8038. Nathan DM, Kuenen J, Borg R, Zheng H,Schoenfeld D, Heine RJ; A1c-Derived AverageGlucose Study Group. Translating the A1C assayinto estimated average glucose values. DiabetesCare 2008;31:1473–14789. Dhatariya KK, Wiles MD. Pre-operative testingguidelines: aNICE try but not enough. Anaesthesia2016;71:1403–140710. Faritous Z, Ardeshiri M, Yazdanian F, Jalali A,Totonchi Z, Azarfarin R. Hyperglycemia or highhemoglobin A1C: Which one is more associatedwithmorbidity andmortality after coronary arterybypass graft surgery? Ann Thorac Cardiovasc Surg2014;20:223–22811. Iorio R, Williams KM, Marcantonio AJ, SpechtLM, Tilzey JF, Healy WL. Diabetes mellitus, he-moglobin A1C, and the incidence of total jointarthroplasty infection. J Arthroplasty 2012;27:726–729.e112. Kuhl J, Sartipy U, Eliasson B, Nystrom T,Holzmann MJ. Relationship between preopera-tive hemoglobin A1c levels and long-termmortal-ity after coronary artery bypass grafting inpatientswith type 2 diabetesmellitus. Int J Cardiol2016;202:291–29613. O’Sullivan CJ, Hynes N, Mahendran B, et al.Haemoglobin A1c (HbA1C) in non-diabetic and di-abetic vascular patients. Is HbA1C an independentrisk factor and predictor of adverse outcome? EurJ Vasc Endovasc Surg 2006;32:188–19714. Subramaniam B, Lerner A, Novack V, et al.Increased glycemic variability in patients withelevated preoperative HbA1C predicts adverseoutcomes following coronary artery bypassgrafting surgery. Anesth Analg 2014;118:277–28715. Duncan AE, Abd-ElsayedA,Maheshwari A, XuM, Soltesz E, Koch CG. Role of intraoperativeand postoperative blood glucose concentrationsin predicting outcomes after cardiac surgery.Anesthesiology 2010;112:860–87116. Egi M, Bellomo R, Stachowski E, French CJ,Hart G. Variability of blood glucose concentrationand short-term mortality in critically ill patients.Anesthesiology 2006;105:244–25217. Ambiru S, Kato A, Kimura F, et al. Poor post-operative blood glucose control increases surgicalsite infections after surgery for hepato-biliary-pancreatic cancer: a prospective study in a high-volume institute in Japan. J Hosp Infect 2008;68:230–23318. Abdelhafiz AH, Sinclair AJ. Low HbA1c andincreased mortality risk-is frailty a confoundingfactor? Aging Dis 2015;6:262–27019. Engoren M, Schwann TA, Arslanian-EngorenC, Maile M, Habib RH. U-shape association be-tween hemoglobin A1c and late mortality in pa-tientswithheart failure after cardiac surgery. Am JCardiol 2013;111:1209–121320. Underwood P, Askari R, Hurwitz S, ChamarthiB, Garg R. Preoperative A1C and clinical outcomesin patients with diabetes undergoing major

care.diabetesjournals.org van den Boom and Associates 787

noncardiac surgical procedures. Diabetes Care2014;37:611–61621. Bock M, Johansson T, Fritsch G, et al. Theimpact of preoperative testing for blood glu-cose concentration and haemoglobin A1c onmortality, changes in management and compli-cations in noncardiac elective surgery: a system-atic review. Eur J Anaesthesiol 2015;32:152–15922. Duggan EW, Carlson K, Umpierrez GE.Perioperative hyperglycemia management: anupdate. Anesthesiology 2017;126:547–56023. Frisch A, Chandra P, Smiley D, et al. Preva-lence and clinical outcome of hyperglycemia inthe perioperative period in noncardiac surgery.Diabetes Care 2010;33:1783–178824. Gandhi GY, Nuttall GA, Abel MD, et al. Inten-sive intraoperative insulin therapy versus conven-tional glucosemanagement during cardiac surgery:a randomized trial. Ann Intern Med 2007;146:233–24325. Vilar-Compte D, Alvarez de Iturbe I, Martın-Onraet A, Perez-Amador M, Sanchez-HernandezC, Volkow P. Hyperglycemia as a risk factor forsurgical site infections in patients undergoingmastectomy. Am J Infect Control 2008;36:192–19826. Wallia A, Schmidt K, Oakes DJ, et al. Glycemiccontrol reduces infections in post-liver transplantpatients: results of a prospective, randomizedstudy. J Clin Endocrinol Metab 2017;102:451–459

27. Domek N, Dux K, Pinzur M, Weaver F, RogersT. Association between hemoglobin A1c and sur-gical morbidity in elective foot and ankle surgery.J Foot Ankle Surg 2016;55:939–94328. Sadoskas D, Suder NC, Wukich DK. Perioper-ative glycemic control and theeffect on surgical siteinfections in diabetic patients undergoing foot andankle surgery. Foot Ankle Spec 2016;9:24–3029. Shibuya N, Humphers JM, Fluhman BL,Jupiter DC. Factors associated with nonunion, de-layed union, and malunion in foot and ankle sur-gery in diabetic patients. J Foot Ankle Surg 2013;52:207–21130. Hastie T, Tibshirani R. Generalized AdditiveModels. London, England, ChapmanandHall, 199031. American Diabetes Association. Diabetescare in the hospital. Sec. 14. In Standards of Med-ical Care in Diabetesd2017 [published correctionappears in Diabetes Care 2017;40:986]. DiabetesCare 2017;40(Suppl. 1):S120–S12732. Canadian Diabetes Association Clinical Prac-tice Guidelines Expert Committee; Cheng AY. Ca-nadian Diabetes Association 2013 clinical practiceguidelines for the prevention andmanagement ofdiabetes in Canada. Introduction. Can J Diabetes2013;37(Suppl. 1):S1–S333. Lazar HL, McDonnell M, Chipkin SR, et al.;Society of Thoracic Surgeons Blood GlucoseGuideline Task Force. The Society of Thoracic Sur-geons practice guideline series: blood glucose

management during adult cardiac surgery. AnnThorac Surg 2009;87:663–66934. Finfer S, Chittock DR, Su SY, et al.; NICE-SUGAR Study Investigators. Intensive versus con-ventional glucose control in critically ill patients.N Engl J Med 2009;360:1283–129735. Bellomo R, Egi M. What is a NICE-SUGAR forpatients in the intensive care unit?Mayo Clin Proc2009;84:400–40236. Rollins KE, Varadhan KK, Dhatariya K, LoboDN. Systematic review of the impact of HbA1con outcomes following surgery in patients withdiabetes mellitus. Clin Nutr 2016;35:308–31637. Zhang SS, Yang H, Luo KJ, et al. The impact ofbody mass index on complication and survival inresected oesophageal cancer: a clinical-based co-hort and meta-analysis. Br J Cancer 2013;109:2894–290338. Blaha J, Mraz M, Kopecky P, et al. Perioper-ative tight glucose control reduces postoperativeadverse events in nondiabetic cardiac surgery pa-tients. J Clin Endocrinol Metab 2015;100:3081–308939. Umpierrez GE, Smiley D, Zisman A, et al. Ran-domized study of basal-bolus insulin therapy inthe inpatient management of patients with type 2diabetes (RABBIT 2 trial). Diabetes Care 2007;30:2181–218640. Sacks DB. A1C versus glucose testing: a com-parison. Diabetes Care 2011;34:518–523

788 Effect of A1C and Glucose on Mortality Diabetes Care Volume 41, April 2018