e(uhertaarmidoprbetapoT(p

d

RA

The Journal of Emergency Medicine, Vol. 38, No. 2, pp. 264–270, 2010Copyright © 2010 Elsevier Inc.

Printed in the USA. All rights reserved0736-4679/08 $–see front matter

doi:10.1016/j.jemermed.2007.11.088

Public Health inEmergency Medicine

MULTICENTER SURVEY OF EMERGENCY PHYSICIAN MANAGEMENT ANDREFERRAL FOR HYPERGLYCEMIA

Adit A. Ginde, MD, MPH,*† Kate E. Delaney, BA,‡ Daniel J. Pallin, MD, MPH,§and Carlos A. Camargo Jr, MD, DRPH‡

*Department of Emergency Medicine, University of Colorado Denver School of Medicine, Aurora, Colorado, †Department ofEmergency Medicine, Beth Israel Deaconess Medical Center, Boston, Massachusetts, ‡Department of Emergency Medicine,

Massachusetts General Hospital, Boston, Massachusetts, and §Department of Emergency Medicine, Brigham andWomen’s Hospital, Boston, Massachusetts

Corresponding Address: Adit A. Ginde, MD, MPH, Department of Emergency Medicine, University of Colorado Denver School of

Medicine, 12401 E. 17th Avenue, B-215, Aurora, CO 80045; E-mail: adit.ginde@ucdenver.edu

ibptE

es

AaTonn

tgdgttd

Abstract—Background: The Emergency DepartmentED), with its high-risk and often disenfranchised patient pop-lation, presents a novel opportunity to identify patients asaving undiagnosed or uncontrolled diabetes. Objective: Tovaluate Emergency Physician opinion on management andeferral for incidental hyperglycemia and on ED-based diabe-es screening. Methods: We conducted a web-based survey ofll attending and resident Emergency Physicians at threecademic EDs. We asked for glucose thresholds to treat andefer non-diabetic and diabetic ED patients for hyperglyce-ia, comparing physicians’ ideal and actual practices. We also

nquired about interest in and barriers for active ED-basediabetes screening compared to use of blood glucose valuesbtained during usual ED care. Results: We contacted 185hysicians, and 152 (85%) completed the survey; 75% ofespondents reported routine outpatient referral of non-dia-etic patients for random glucose values > 200 mg/dL. How-ver, a majority (71%) believed that they should use a lowerhreshold to refer than they currently use. Nearly all (92%)greed that Emergency Physicians should inform non-diabeticatients of elevated glucose values; 53% supported and 21%pposed active ED-based screening of asymptomatic patients.he most commonly cited barriers were limited follow-up

69%), insufficient time/resources (51%), and outside scope ofractice (36%). Conclusion: Emergency Physicians support

Dr. Ginde was supported by the Emergency Medicine Foun-ation Research Fellowship Grant.

ECEIVED: 1 September 2007; FINAL SUBMISSION RECEIVED

CCEPTED: 16 November 2007

264

mproved recognition of and referral for hyperglycemia,ased on glucose values collected during usual ED care. Welan to develop tools to interpret random ED glucose values inhe context of undiagnosed and uncontrolled diabetes. © 2010lsevier Inc.

Keywords—diabetes mellitus; hyperglycemia; survey;creening; emergency medicine

INTRODUCTION

lthough approximately 21 million Americans have di-betes, an estimated 6 million remain undiagnosed (1).he Emergency Department (ED), with its high-risk andften disenfranchised patient population, presents aovel opportunity to identify patients as having undiag-osed or uncontrolled diabetes.

Serum glucose tests are ordered at 18% of ED visits inhe United States, and many others require capillarylucose (2). Prior studies have demonstrated high rates ofiabetes risk factors, hyperglycemia, and elevated hemo-lobin A1C among non-diabetic ED patients (3–5). Al-hough stressors such as infections or pain may contributeo ED hyperglycemia, it may also indicate undiagnosediabetes. ED patients want to be informed of and have

vember 2007;

: 8 No

fdtri

Pktaaida

S

WtrMwsdeNst

S

Wuarturb

mtadtfEAiTm

ar5w

D

W(dcrAs

Ossr5y

ttaEg(mmrwtf1mt

ipptow�foC(p

Survey on ED Hyperglycemia 265

ollow-up for incidental hyperglycemia, but preliminaryata suggest that Emergency Physicians’ discharge instruc-ions rarely include this information (3,6). Identification andeferral of patients with unexplained hyperglycemia maymprove outpatient diagnosis and control of diabetes.

Current standard of care and opinions of Emergencyhysicians toward incidental hyperglycemia are notnown. In this multi-center survey, we hypothesized thathere would be a discrepancy between physicians’ idealnd actual practices in recognition of hyperglycemia. Welso assessed ED management of hyperglycemia, opin-on toward active ED-based screening for undiagnosediabetes, and differences in opinion between attendingnd resident physicians.

MATERIALS AND METHODS

tudy Design and Population

e conducted a cross-sectional survey of physicians athree academic EDs in April/May 2007. All attending andesident physicians in the three Departments of Emergency

edicine were eligible. Attending staff at the three sitesere part of separate physician groups, whereas two sites

hared a common group of resident physicians. Respon-ents with incomplete surveys (� 80% completed) werexcluded. All sites participate in the Emergency Medicineetwork (www.emnet-usa.org), which coordinated the

tudy. The Institutional Review Boards at all sites approvedhe protocol with a waiver of written informed consent.

urvey Content and Administration

e administered the survey in web-based format only,sing EZSurvey Software (Raosoft, Seattle, WA). We sentn e-mail to eligible physicians to introduce the study andequest voluntary participation. We tracked study participa-ion and prevented multiple responses by the same individ-al, but actual survey responses were anonymous. Non-espondents were sent a total of three reminder emailsefore the target response rate of � 80% was achieved.

The survey consisted of three sections: general infor-ation, non-diabetic ED patients, and diabetic ED pa-

ients (Appendix). Participants were asked about theirge, sex, level of training, and self-rated knowledge ofiabetes and hyperglycemia. The next section pertainedo hyperglycemia in non-diabetic ED patients. We askedor glucose thresholds for management and referral ofD hyperglycemia, comparing ideal and actual practice.dditionally, we inquired about the participants’ interest

n and barriers to active ED-based screening for diabetes.he final section assessed the practice of referral and

anagement for hyperglycemia in diabetic patients. For l

ll items requesting glucose thresholds, we presentedesponse choices as �: 125, 160, 200, 250, 300, 400, and00 mg/dL, or other value (with respondent asked torite-in their preferred threshold).

ata Analysis

e performed statistical analyses using Stata 9.0StataCorp LP, College Station, TX). We used basicescriptive statistics with 95% confidence intervals andompared proportions for categorical responses betweenesident and attending physicians using chi-squared test.ll p-values are two-tailed, with p � 0.05 considered

tatistically significant.

RESULTS

f 185 eligible attending and resident Emergency Phy-icians, we received 152 complete surveys (82% re-ponse rate). The response rate was 90% (78 of 87) foresidents and 76% (74 of 98) for attendings. There were5 (36%) female respondents and the median age was 32ears (inter-quartile range 29–39 years).

Emergency Physicians’ current practice and opinionoward hyperglycemia in non-diabetic and diabetic pa-ients are presented in Table 1, stratified by resident orttending status. The threshold glucose value triggeringD treatment (e.g., intravenous fluids, insulin) for hyper-lycemia in patients without known diabetes varied: 3322%) reported initiation of treatment for glucose � 200g/dL, 67 (44%) for � 250 mg/dL, 128 (84%) for � 300g/dL and 144 (95%) for � 400 mg/dL. Eight (5%)

espondents were unsure or expressed that their thresholdas situation dependent. For patients with known diabe-

es, 49 (32%) report that they would institute treatmentor glucose � 200 mg/dL, 90 (59%) for � 250 mg/dL,38 (91%) for � 300 mg/dL, and 148 (97%) for � 400g/dL. Four (3%) respondents were unsure or expressed

hat their threshold was situation dependent.When asked what threshold glucose would lead to

nitiation of outpatient treatment (e.g., insulin, oral hy-oglycemic agents) in consultation with the primary carehysician, 79 (52%) responded that they would deferreatment initiation to the outpatient providers, regardlessf glucose level. Of the remaining respondents, 17 (11%)ould initiate for glucose � 200 mg/dL, 35 (23%) for250 mg/dL, 61 (40%) for � 300 mg/dL, and 70 (46%)

or � 400 mg/dL. Three (2%) respondents were unsuref their threshold for initiation of outpatient treatment.onsidering diabetes screening based on symptoms, 127

84%) would send glucose for a patient with polyuria/olydipsia, 83 (55%) for unexplained infection (i.e., cel-

ulitis), 68 (45%) for blurred vision, 32 (21%) for diabe-

t2s

iit(Ob1

Ed

dmuEdc

ftbsnrwEg

T

S

N

D

*E

266 A. A. Ginde et al.

es risk factors, and 57 (38%) for patient request. Of note,1 (14%) reported that they do not order glucose as acreening test for diabetes under any circumstances.

When asked about barriers to active ED-based screen-ng for diabetes, the most commonly cited obstacles werenability to secure follow-up (n � 105, 69%), insufficientime/resources (n � 78, 51%), outside scope of practicen � 55, 36%), and lack of familiarity (n � 40, 26%).nly 16 (11%) respondents believed that active ED-ased screening for diabetes would be low yield, and8 (12%) reported that there are no significant barriers.

DISCUSSION

mergency Physicians are trained to respond to clinical

able 1. Emergency Physicians’ Opinion toward ManagemeDiagnosed Diabetes

Survey Item

elf-rated knowledge of hyperglycemia*Poor/fairAverageGood/excellenton-diabetic patientsIn actual practice, threshold ED glucose value to refer for

outpatient diabetes testing*125 mg/dL160 mg/dL200 mg/dL250 mg/dL300 mg/dL or higher

Ideal threshold ED glucose for outpatient referral lower thanactual practice

Important to inform ED patients of incidental hyperglycemia*Strongly disagree/disagreeNeutralAgreeStrongly agree

Worthwhile to screen asymptomatic, high-risk ED patientsfor diabetes*

Strongly disagreeDisagreeNeutralAgreeStrongly agree

iabetic patientsIn practice, threshold ED glucose to refer for adjustment of

diabetes medications160 mg/dL or lower200 mg/dL250 mg/dL300 mg/dL350 mg/dL or higher

Ideal threshold ED glucose value for med adjustment lowerthan actual practice

p � 0.05 using chi-squared test.D � Emergency Department; CI � confidence interval.

ata relevant to acute care. However, in the process of this m

ata collection, values sufficient to motivate long-termanagement are sometimes uncovered, and when not acted

pon, may be viewed as “missed opportunities.” The role ofmergency Physicians in recognizing and acting on theseata is unclear, however, prior studies have evaluated thisoncept of ED-based preventive medicine (7–9).

Although Emergency Physicians’ thresholds and re-erral for elevated blood pressure have been evaluated,hreshold values for referral and treatment of elevatedlood glucose are not well characterized (8). In ourtudy, 75% of Emergency Physicians reported referral ofon-diabetic patients for outpatient diabetes testing whenandom glucose values are � 200 mg/dL. These valuesere lower than the reported threshold for initiation ofD or outpatient treatment, which indicates that Emer-ency Physicians may recognize incidental hyperglyce-

Referral for ED Hyperglycemia in Patients with or without

esident (n � 78) Attending (n � 74) Total (n � 152)

% (95% CI) % (95% CI) n (%)

19% (10–28) 7% (1–13) 20 (13%)54% (43–65) 48% (36–60) 77 (51%)27% (17–37) 45% (34–57) 54 (36%)

0 9% (3–16) 7 (5%)19% (10–28) 31% (20–42) 38 (25%)47% (36–59) 41% (30–53) 68 (45%)15% (7–24) 5% (0–11) 16 (10%)18% (9–27) 12% (5–20) 23 (15%)77% (67–86) 64% (53–76) 106 (71%)

5% (0–10) 4% (0–9) 7 (5%)1% (0–4) 5% (0–11) 5 (3%)

57% (46–68) 27% (17–38) 64 (43%)36% (25–47) 63% (52–74) 74 (49%)

4% (0–8) 5% (0–11) 7 (5%)6% (1–12) 26% (15–36) 24 (16%)

33% (23–44) 19% (10–28) 40 (26%)50% (39–61) 34% (23–45) 64 (42%)6% (1–12) 16% (8–25) 17 (11%)

5% (0–10) 10% (3–17) 11 (7%)42% (31–54) 40% (28–51) 62 (41%)23% (14–33) 26% (16–36) 37 (25%)24% (15–34) 22% (12–32) 35 (23%)5% (0–10) 3% (0–7) 6 (4%)

83% (75–92) 75% (65–85) 120 (79%)

nt and

R

ia at levels lower than those necessitating ED treat-

mstcl

geftEisrcppwwostsnoit

eadEofaaosiwrl

dtficiinavun

TwrptreHih

MngratEu

1

Survey on ED Hyperglycemia 267

ent. However, a majority (71%) still believed that theyhould ideally refer for glucose thresholds lower thanhose used in actual practice. Indeed, Rolka et al. suggestonsideration of casual glucose screening thresholds asow as 120 mg/dL (10).

Nearly all (92%) agreed or strongly agreed that Emer-ency Physicians should inform non-diabetic patients oflevated glucose values during their ED visit and referor outpatient follow-up, although a majority believedhat outpatient treatment should not be initiated in theD. A minority (8%) were neutral or opposed to inform-

ng patients of elevated glucose values. Although notpecifically addressed, other survey comments by theseespondents suggest a belief that this identification andommunication of non-critical hyperglycemia wouldlace undue burden on the Emergency Physician. Ourrior work suggests that nearly all (95%) ED patientsant to be informed of elevated blood glucose and areilling to follow-up if instructed (3). Because prior datan random glucose and opinions expressed in this surveyuggest that Emergency Physicians should inform pa-ients of hyperglycemia at lower glucose thresholds, aignificant opportunity for improved care exists (6). Thiseed may be addressed by physician education, flaggingf non-critical glucose values, especially in non-diabeticndividuals, and hyperglycemia discharge instructionsriggered by glucose values.

Our results suggest that Emergency Physicians gen-rally support opportunistic screening for undiagnosednd uncontrolled diabetes using glucose values obtaineduring usual ED care. Fewer respondents endorsed activeD-based diabetes screening in patients with symptomsf diabetes (e.g., polyuria/polydipsia or unexplained in-ection), and only half supported ED screening forsymptomatic individuals. Barriers to active screening ofsymptomatic patients were related more to circumstancesf ED practice (lack of follow-up/time/resources and out-ide scope of practice), than to physician knowledge/will-ngness or patient need. Compared to residents, attendingsere more likely to have a lower glucose threshold for

eferral of non-diabetic patients, but they also were moreikely to resist active ED-based diabetes screening.

Although there are some data on random glucose foriabetes screening, the optimal glucose thresholds andiming for ED treatment and referral need further clari-cation (10). Active ED-based diabetes screening wouldlearly require additional time and resources, and accord-ngly, Emergency Physician support for diabetes screenings limited more by infrastructure rather than perceivedeeds of ED patients. However, enhanced recognitionnd referral for incidental hyperglycemia using glucosealues obtained during usual ED care may provide annobtrusive method to improve identification of undiag-

osed and uncontrolled diabetes in the ED.

LIMITATIONS

his study has several potential limitations. Our sitesere all academic centers and from the same geographic

egion, which may limit generalizability. We relied onhysician self-report, which may not reflect actual prac-ice. Because there are limited data for interpretation ofandom glucose in the ED setting, we were unable tovaluate the validity of glucose threshold responses.owever, differences between ideal and actual practice

llustrate potential for improved recognition of incidentalyperglycemia.

CONCLUSIONS

ost Emergency Physicians support improved recog-ition of and referral for hyperglycemia based onlucose values collected during usual ED care, but areeticent about active diabetes screening that requiresdditional testing, especially in asymptomatic pa-ients. We plan to develop tools to interpret randomD glucose values in the context of undiagnosed andncontrolled diabetes.

REFERENCES

1. Cowie CC, Rust KF, Byrd-Holt DD, et al. Prevalence of diabetesand impaired fasting glucose in adults in the US population.Diabetes Care 2006;29:1263–8.

2. McCaig LF, Nawar EN. National Hospital Ambulatory MedicalCare Survey: 2004 emergency department summary. Adv Data2006;372:1–29.

3. Ginde AA, Delaney KE, Lieberman RM, Vanderweil SG, Ca-margo CA Jr. Estimated risk for undiagnosed diabetes in theemergency department: a multicenter survey. Acad Emerg Med2007;14:492–5.

4. Charfen MA, Ipp E, Qazi MF, Shin EC, Lewis RJ. The yield ofscreening for diabetes in high-risk emergency department patients(abstract). Acad Emerg Med 2006;13:S121–2.

5. Silverman RA, Pahk R, Carbone M, et al. The relationship ofplasma glucose and HbA1c levels among emergency departmentpatients with no prior history of diabetes mellitus. Acad EmergMed 2006;13:722–6.

6. Graffeo CS, Holland CK. Unexplained hyperglycemia in emergencydepartment patients (abstract). Acad Emerg Med 2001;8:529–30.

7. Irvin CB. Public health preventive services, surveillance, andscreening: the emergency department’s potential. Acad EmergMed 2000;7:1421–3.

8. Backer HD, Decker L, Ackerson L. Reproducibility of increasedblood pressure during an emergency department or urgent carevisit. Ann Emerg Med 2003;41:507–12.

9. Pallin DJ, Muennig PA, Emond JA, Kim S, Camargo CA Jr.Vaccination practices in US emergency departments, 1992–2000.Vaccine 2005;23:1048–52.

0. Rolka DB, Narayan KM, Thompson TJ, et al. Performance ofrecommended screening tests for undiagnosed diabetes and dysg-

lycemia. Diabetes Care 2001;24:1899–903.

S

Ihoa

G

N

1

268 A. A. Ginde et al.

APPENDIX

ite __ __ __ Survey __ __ __Hyperglycemia and Diabetes

nstructions: We appreciate your time in completing this simple but important survey. We know how busy you are, andow many requests of this type you receive. Your responses will help us to better understand the knowledge and practicef emergency physicians as they relate to ED hyperglycemia and diabetes. Your name will not appear on the survey,nd only aggregate responses will be shared and published.

eneral Information

1. What is your level of training?□ 1 Resident PGY-1 □ 4 Attending 5 years or less□ 2 Resident PGY-2 □ 5 Attending 6–10 years□ 3 Resident PGY-3 or greater □ 6 Attending 11 years or greater

2. What is your sex? □ 1 Male □ 2 Female 3. What is your age? __ __ __

4. Please rate your knowledge and familiarity with treatment and referral for diabetic and non-diabetic patients presenting withhyperglycemia.□ 1 Poor □ 2 Fair □ 3 Average □ 4 Good □ 5 Excellent

on Diabetic ED patients

5. In practice, I usually refer asymptomatic, non-diabetic patients for outpatient diabetes testing when random (non-fasting) EDglucose values are greater than or equal to:□ 1 125 mg/dl □ 5 300 mg/dl□ 2 160 mg/dl □ 6 400 mg/dl□ 3 200 mg/dl □ 7 500 mg/dl□ 4 250 mg/dl □ 8 Other: __ __ __ mg/dl

6. In theory, I should refer an asymptomatic, non-diabetic patient for outpatient diabetes testing for a random ED glucose valuegreater than or equal to:□ 1 125 mg/dl □ 5 300 mg/dl□ 2 160 mg/dl □ 6 400 mg/dl□ 3 200 mg/dl □ 7 500 mg/dl□ 4 250 mg/dl □ 8 Other: __ __ __ mg/dl

7. I usually institute ED treatment (i.e., IV fluids, insulin) for non-diabetic patients with random ED glucose values greater than orequal to:□ 1 125 mg/dl □ 5 300 mg/dl□ 2 160 mg/dl □ 6 400 mg/dl□ 3 200 mg/dl □ 7 500 mg/dl□ 4 250 mg/dl □ 8 Other: __ __ __ mg/dl

8. If a non-diabetic patient is stable for discharge, I usually initiate outpatient treatment (i.e. insulin, oral hypoglycemic), inconsultation with the PCP, for a random ED glucose value greater than or equal to:□ 1 125 mg/dl □ 6 400 mg/dl□ 2 160 mg/dl □ 7 500 mg/dl□ 3 200 mg/dl □ 8 Other: __ __ __ mg/dl□ 4 250 mg/dl □ 9 I do not initiate outpatient diabetes treatment and defer

treatment initiation to outpatient providers□ 5 300 mg/dl

9. It is important for emergency physicians to inform non-diabetic patients of elevated glucose levels during their ED visit and referfor outpatient follow-up.□ 1 Strongly disagree □ 2 Disagree □ 3 Neutral □ 4 Agree □ 5 Strongly Agree

0. If I am not sending labs for other reasons, I routinely check glucose to screen for diabetes in non-diabetic ED patients with:(check all that apply)□ 1 Unexplained infection (i.e. cellulitis) □ 5 Polyuria/polydipsia□ 2 Blurred vision □ 6 Patient request□ 3 Patients with diabetes risk factors □ 7 I do not screen for diabetes

□ 4 Other: _______________________

S

1

1

D

1

1

1

Survey on ED Hyperglycemia 269

ite __ __ __ Survey __ __ __

1. It is worthwhile to screen asymptomatic, high-risk patients for diabetes during their ED visit and refer those with abnormalscreens for follow-up.□ 1 Strongly disagree □ 2 Disagree □ 3 Neutral □ 4 Agree □ 5 Strongly Agree

2. What are significant barriers to your support of ED screening for diabetes in high-risk patients? (check all that apply)□ 1 Insufficient time/resources □ 6 Inability to ensure adequate follow-up□ 2 Will be low yield □ 7 Patient acuity/unwillingness□ 3 Outside the scope of practice □ 8 Lack of familiarity with screening□ 4 Lack of data for screening □ 9 Other: ____________________□ 5 No significant barriers

iabetic ED patients

3. In practice, I usually advise diabetic patients to discuss adjustment of diabetes medications with their PCP when random EDglucose values are greater than or equal to:□ 1 125 mg/dl □ 5 300 mg/dl□ 2 160 mg/dl □ 6 400 mg/dl□ 3 200 mg/dl □ 7 500 mg/dl□ 4 250 mg/dl □ 8 Other: __ __ __ mg/dl

4. In theory, I should advise a diabetic patient to discuss adjustment of diabetes medications with their PCP when for a randomED glucose value greater than or equal to:□ 1 125 mg/dl □ 5 300 mg/dl□ 2 160 mg/dl □ 6 400 mg/dl□ 3 200 mg/dl □ 7 500 mg/dl□ 4 250 mg/dl □ 8 Other: __ __ __ mg/dl

5. I usually institute ED treatment (i.e. insulin, oral hypoglycemic, IV fluids) for diabetic patients with random ED glucose valuesgreater than or equal to:□ 1 125 mg/dl □ 5 300 mg/dl□ 2 160 mg/dl □ 6 400 mg/dl□ 3 200 mg/dl □ 7 500 mg/dl□ 4 250 mg/dl □ 8 Other: __ __ __ mg/dl

THANK YOU for your participation!

270 A. A. Ginde et al.

ARTICLE SUMMARY1. Why is this topic important?

Although approximately 21 million Americans havediabetes, an estimated 6 million remain undiagnosed.Identification and referral of patients with unexplainedhyperglycemia may improve outpatient diagnosis andcontrol of diabetes.2. What does this study attempt to show?

In this multi-center survey, we assessed EmergencyPhysician opinion on management and referral for inci-dental hyperglycemia and on ED-based diabetes screen-ing. In particular, we evaluated for discrepancy betweenEmergency Physicians’ ideal and actual practices in rec-ognition of hyperglycemia.3. What are the key findings?

Most Emergency Physicians support improved recog-nition of and referral for hyperglycemia based on glucosevalues collected during usual ED care. A majority ofrespondents believed that they should use a lower glucosethreshold to refer non-diabetic patients for outpatienttesting than they currently use in actual practice. Emer-gency Physicians are reticent about active diabetesscreening that requires additional testing, especially inasymptomatic patients.4. How is patient care impacted?

Although patient care is not directly impacted by theresults of this study, the findings will improve awarenessand referral patterns for undiagnosed or uncontrolleddiabetes using glucose values obtained as part of usualED care. Additionally, development of better tools tointerpret random ED glucose values in the context ofundiagnosed and uncontrolled diabetes is indicated.