sign in sign up
<<
Home , Cardiovascular disease

Diabetes Care Volume 40, June 2017 793

Elsayed Z. Soliman,1 Jye-Yu C. Backlund,2 Electrocardiographic Ionut Bebu,2 Trevor J. Orchard,3 Bernard Zinman,4 John M. Lachin,2 and the Abnormalities and Cardiovascular DCCT/EDIC Research Group* Risk in Type 1 : The Epidemiology of Diabetes Interventions and Complications (EDIC) Study Diabetes Care 2017;40:793–799 | https://doi.org/10.2337/dc16-2050

OBJECTIVE We examined the association between the prevalence and incidence of electro- cardiographic (ECG) abnormalities and the development of cardiovascular disease (CVD) in patients with , among whom these ECG abnormalities are common. 1Epidemiological Research Center RESEARCH DESIGN AND METHODS (EPICARE), Department of Epidemiology and Pre- We conducted a longitudinal cohort study involving 1,306 patients with type 1 vention, and Department of , Section on 6 Cardiology, Wake Forest School of Medicine, RISK METABOLIC AND CARDIOVASCULAR diabetes (mean age 35.5 6.9 years; 47.7% female) from the Diabetes Control and Winston-Salem, NC Complications Trial/Epidemiology of Diabetes Interventions and Complications 2The Biostatistics Center, The George Washing- (DCCT/EDIC) Study. ECG abnormalities were defined by the Minnesota Code ton University, Rockville, MD 3 ECG classification as major, minor, or no abnormality. CVD events were defined University of Pittsburgh Graduate School of fi fi , Pittsburgh, PA as the rst occurrence of , , con rmed , coro- 4Lunenfeld Tanenbaum Research Institute, nary revascularization, congestive failure, or from any CVD. Mount Sinai Hospital, University of Toronto, Toronto, RESULTS Corresponding author: Elsayed Z. Soliman, esoliman@ During a median follow-up of 19 years, 155 participants (11.9%) developed CVD wakehealth.edu. events. In multivariable Cox proportional hazard models adjusted for demograph- Received 23 September 2016 and accepted 25 ics and potential confounders, the presence of any major ECG abnormalities as a February 2017. time-varying covariate was associated with a more than twofold increased risk of This article contains Supplementary Data online CVD events (hazard ratio [HR] 2.10 [95% CI 1.26, 3.48] vs. no abnormality/normal at http://care.diabetesjournals.org/lookup/ suppl/doi:10.2337/dc16-2050/-/DC1. ECG, and 2.19 [1.46, 3.29] vs. no major abnormality). Also, each visit (year) at *A complete list of participants in the DCCT/EDIC which the diagnosis of major ECG abnormality was retained was associated with a Research Group can be found at http://www 30% increased risk of CVD (HR 1.30 [95% CI 1.14, 1.48]). The presence of minor ECG .nejm.org/doi/suppl/10.1056/NEJMoa1409463/ abnormalities was not associated with a significant increase in CVD risk. suppl_file/nejmoa1409463_appendix.pdf. © 2017 by the American Diabetes Association. CONCLUSIONS Readers may use this article as long as the work The presence of major ECG abnormalities is associated with an increased risk of is properly cited, the use is educational and not for profit, and the work is not altered. More infor- CVD in patients with type 1 diabetes. This suggests a potential role for ECG screen- mation is available at http://www.diabetesjournals ing in patients with type 1 diabetes to identify individuals at risk for CVD. .org/content/license. 794 Abnormal ECG and CVD Risk in Type 1 Diabetes Diabetes Care Volume 40, June 2017

Type 1 diabetes is known to be associated years old; 726 participants were as- left [MC 7.1] and right [MC 7.2] bundle with a higher incidence and prevalence of signed to the primary prevention cohort branch block, major intraventricular cardiovascular disease (CVD) (1), the pri- (diabetes duration 1–5years,noreti- conduction delay [MC 7.4], and bifascic- mary in patients with nopathy, and urinary albumin excretion ular block [MC 7.8]), major Q-wave ab- type 1 diabetes (2–4). Increased CVD rate [AER] ,40 mg/day), and 715 to the normalities (MC 1.1.x, MC 1.2.x), minor risk is not, however, uniform in all pa- secondary intervention cohort (diabetes Q-wave abnormalities plus major ST/ tients with type 1 diabetes; it varies ac- duration 1–15 years, very mild to mod- T-wave abnormalities (MC 1.3.x plus cording to individual characteristics and erate nonproliferative retinopathy, and [MC4.1.x,MC4.2,MC5.1,MC5.2]), risk profile (1,5). Identifying these predic- AER #200 mg/day). Intensive therapy isolated major ST/T-wave abnormalities tive characteristics and risk markers is (n = 711) aimed to achieve levels of gly- (MC 4.1.x, MC 4.2, MC 5.1, MC 5.2), left necessary to improve our ability to iden- cemia as close to the nondiabetic range with strain pat- tify patients with type 1 diabetes who are as safely possible, whereas conventional tern (MC 3.1 and [MC 4.1.x, MC 4.2, MC at a higher risk. therapy (n = 730) aimed to maintain clin- 5.1, MC 5.2]), atrial fibrillation/flutter (ECG) is the most ical well-being, with no specificglucose (MC 8.3), major widely used noninvasive tool for cardiac targets. At the end of the DCCT (1993), (complete atrioventricular block [MC investigation. We recently showed that participants in the conventional treat- 6.1] and second-degree atrioventric- developing new ECG abnormalities is ment group were instructed in intensive ular block [MC 6.2]), major QT prolon- common in the course of type 1 diabe- diabetes therapy. In 1994, all surviving gation (QT index .116%), electronic tes; about three of every four partici- DCCT participants were invited to join pacemaker (MC 6.8), and others (atrio- pants in the Diabetes Control and the EDIC observational study. The study ventricular dissociation [MC 8.6], ven- Complications Trial (DCCT)/Epidemiol- was approved by the institutional re- tricular [MC 8.2], and ogy of Diabetes Interventions and Com- view board at each study site. All par- Wolf-Parkinson-White syndrome/ plications (EDIC) study developed at ticipants provided written informed pre-excitation [MC 6.4]). Minor ECG least one new ECG abnormality, and consent. abnormalities included minor isolated about one of every six developed at For the purpose of this analysis, we Q-wave abnormalities (MC 1.3.x), minor least one new major ECG abnormality included EDIC participants with a good- isolated ST/T-wave abnormalities (MC during 16 years of follow-up in EDIC quality ECG at the EDIC year 1 visit (re- 4.3, MC 4.4, MC 5.3, MC 5.4), high R (6). Prior reports have shown that the ferredtoas“baseline” in this article) waves/increased QRS voltage denoting presence of these ECG abnormalities in and at least one follow-up visit thereaf- left or right ventricular hypertrophy with- populations without diabetes is associ- ter. Patients who had CVD events before outstrainpattern(MC3.1,MC3.2,MC3.3, ated with an increased risk of CVD EDIC year 1 were eliminated from this MC 3.4), nonischemic ST segment eleva- events and all-cause mortality (7–11). analysis. Supplementary Fig. 1 shows tion (MC 9.2), incomplete bundle branch Similarly, in a small study with a relative- the inclusion and exclusion criteria ap- blocks (left [MC 7.6] and right [MC 7.3] ly short follow-up, the presence of ECG plied and how the analysis sample was bundle branch blocks and left anterior markers of myocardial in pa- achieved. hemiblock [MC 7.7]), minor QT prolonga- tients with type 1 diabetes was predic- tion (QT index .112% but ,116%), short tive of future coronary heart disease ECG PR interval (MC 6.5), axis deviation (left (12). However, no comprehensive re- EDIC participants underwent resting axis deviation [MC 2.1], ports have described the prognostic 12- ECG recording annually. ECG [MC 2.2]), ventricular premature beats significance of ECG abnormalities in pa- tracings were centrally read at an ECG (MC 8.1.2, MC 8.1.3), and others (low tients with type 1 diabetes, in whom core facility, initially (EDIC years 1–11) at QRS voltage [MC 9.1], premature atrial ec- CVD develops at least a decade sooner the University of Minnesota ECG Read- topic beats [MC 8.1.1], wandering atrial compared with the general population ing Center (Minneapolis, MN), then at pacemaker [MC 8.1.4], abnormal P-wave (1). Therefore, the purpose of this study the Epidemiological Cardiology Re- amplitude [MC 9.3], prolonged PR interval/ was to examine the association between search Center (EPICARE) of Wake Forest first-degree atrioventricular block [MC the presence of ECG abnormalities and School of Medicine (Winston-Salem, 6.3], marked sinus [MC 8.8], incident CVD events in patients with NC). The change in the ECG reading and marked [MC 8.7]) type 1 diabetes enrolled in the EDIC center did not affect the rate of devel- (14). Study, providing observational follow-up oping new ECG abnormalities (interac- of the DCCT cohort. tion P values between the reading Cardiovascular Events center and the DCCT treatment group All events were adjudicated by a mortal- RESEARCH DESIGN AND METHODS for any ECG abnormalities and major ity and morbidity review committee The EDIC observational component of ECG abnormalities were 0.98 and 0.65, whose members were blinded to the the DCCT/EDIC Study started in 1994, respectively). DCCT treatment group and level of gly- after completion of the DCCT, which ECG abnormalities were classified as cemia. CVD events used in this analysis consisted of a comparison of the effects major and minor abnormalities using occurred through December 2013 and of intensive versus conventional diabe- the standards of the Minnesota Code were defined using methods described tes therapy on long-term diabetes com- (MC) for ECG classification (14). Major elsewhere (15). These events included plications (13). During 1983–1989, DCCT ECG abnormalities included major ven- the first occurrence of either a nonfatal enrolled 1,441 individuals aged 13–39 tricular conduction defect (complete myocardial infarctiondincluding silent care.diabetesjournals.org Soliman and Associates 795

myocardial infarction, stroke, confirmed EDIC year 1 of participants who devel- use of -lowering and blood pres- angina, coronary artery revascularization, oped CVD events during follow-up and sure–lowering medications, composed and congestive heart failuredor death those who did not were compared using the fully adjusted model (model 2). Par- fromanyCVD.Aspartoftheeventsad- the Wilcoxon rank sum test for quanti- ticipants were censored at the time of an judication process, death from CVD was tative variables and the x2 test for cate- event, death, or 31 December 2013 (the further classified as sudden versus not gorical variables. end of follow-up), whichever occurred sudden based on timing. Silent myocardi- The association between ECG abnor- first. al infarction was defined as the presence malities and CVD abnormalities was ex- Additional analysis examined the ef- of significant serial Q-wave abnormalities amined using two approaches. First, Cox fect modification by sex and level of denoting a new myocardial infarction, ac- proportional hazards models were used HbA1c. The proportional hazards as- cording to the standards of the MC ECG to calculate the hazard ratios (HRs) and sumption was tested by adding time- classification, in the absence of adjudi- 95% CIs for the association between the varying interaction terms between the cated clinical myocardial infarction. presence (vs. absence) of ECG abnor- covariates and log(time) (17). A two- ECGs with major abnormalities that oc- malities (major and any, separately) as sided P value #0.05 was considered sta- curred on the same date as the silent time-varying covariates and the risk of tistically significant. All analyses were myocardial infarction that is part of the CVD events during EDIC follow-up. Sec- performed using SAS software (version CVD outcome were not included with ond, using similar models, we also ex- 9.3; SAS Institute, Cary, NC). the predictor ECGs. amined the CVD risk per number of visits (years), diagnosed separately as RESULTS Covariates major ECG abnormality and any abnor- This analysis included 1,306 participants During EDIC, demographic variables mality. Notably, when we initially exam- (mean age 35.5 6 6.9 years; 47.7% fe- (age and sex), smoking, use of lipid- ined the relationship between the male; 96.3% white) from EDIC, who lowering medications, and use of blood number of visits (years) at which an compose 92% of the surviving DCCT co- pressure–lowering medications were ECG abnormality was diagnosed and hort. Table 1 shows the characteristics self-reported. Fasting lipid profile and the CVD risk by fitting a smoothing of the analysis sample at EDIC year albumin excretion were assessed bien- splines model (16), we noticed the 1 (baseline) stratified by the occurrence nially, in alternate years, whereas HbA , 1c risk did not increase beyond five visits of CVD events during follow-up. As BMI, and were measured (years). Therefore, any number of visits shown, study participants who devel- annually. The mean values of BMI, blood (years) exceeding five was set to five. The oped CVD events during follow-up pressure, , and HbA over the com- 1c first approach investigates the associa- were more likely to be older, be current bined duration of DCCT and EDIC were tion between ECG abnormalities and smokers, have a longer duration of di- used as time-dependent covariates. The the risk of CVD per se, whereas the sec- abetes, and have higher HbA , systolic updated mean values were computed us- 1c ond approach investigates whether this blood pressure, diastolic blood pres- ing weights proportional to the time in- association further depends on the sure, non-HDL , total choles- terval between values because of the amount of time since the develop- terol, triglycerides, and albuminuria at different visit schedules during DCCT ment/diagnosis of ECG abnormality. baseline. and EDIC. A history of albuminuria was de- The minimally adjusted model (model During a median follow-up of 19 years fined as any sustained AER $30 mg/day 1) included age and sex at baseline (EDIC (mean 18 6 4 years), 155 participants during at least two consecutive visits dur- year 1), whereas the fully adjusted (11.9%) developed CVD events (inci- ing DCCT/EDIC. model (model 2) was further adjusted dence of 46.8 per 10,000 person-years). The presence of major ECG abnormal- for DCCT cohort (primary prevention co- The CVD events were distributed as fol- ities (yes vs. no) was defined as the pres- hort vs. secondary intervention cohort) lows: 148 were nonfatal (34 with clinical ence of such abnormalities on at least one and the most significant time-varying myocardial infarction, 38 with silent ECG up to the time of each annual ECG. traditional CVD risk factors. The tradi- myocardial infarction, 16 with con- Likewise, the presence of any ECG abnor- tional CVD risk factors that were initially firmed angina pectoris, 39 with coronary malities (yes vs. no) was defined as the considered included current smoking revascularization, 3 with congestive presence of such abnormalities on at least status and DCCT/EDIC weighted mean , and 18 with stroke) and one ECG up to the time of each annual of systolic blood pressure, diastolic 7 were fatal (2 sudden and 5 non- ECG. The presence of a major abnormality blood pressure, total cholesterol, HDL sudden deaths). In Cox proportional alone up to the most recent ECG (yes vs. cholesterol, LDL cholesterol, non-HDL hazard models adjusted for demo- no) was used as a time-dependent covar- cholesterol, triglycerides, HbA , and al- graphic characteristics (age and sex), iate at the time of a CVD event, as was 1c buminuria. Each of these CVD risk fac- the presence of any major ECG abnor- the presence of any abnormality (major tors was added one at a time to the mality as a time-varying covariate was or minor). The presence of a major abnor- minimally adjusted model. Four CVD associated with a more than 2.5-fold in- mality or minor abnormality versus no risk factors (HbA , albuminuria, systolic creased risk of a CVD event (HR 2.67 abnormality was used as a three-category 1c blood pressure, and non-HDL choles- [95% CI 1.62, 4.40] vs. no abnormality/ time-dependent covariate. terol) showed significant associations normal ECG, and 2.57 [1.72, 3.84] vs. no Statistical Analysis when added to the minimally adjusted major abnormality). This association Clinical characteristics at EDIC year 1 or model, and with the addition of DCCT was slightly attenuated after further the DCCT/EDIC weighted mean through cohort type (primary or secondary) and adjustment for potential confounders 796 Abnormal ECG and CVD Risk in Type 1 Diabetes Diabetes Care Volume 40, June 2017

Table 1—Participant characteristics at EDIC baseline stratified by occurrence of a cardiovascular event during follow-up Cardiovascular event Characteristics All participants (n =1,306) Yes (n =155) No(n =1,151) P value* Age (years) 35.5 6 6.9 39.1 6 6.1 35.1 6 6.9 ,0.001 Female sex 623 (47.7) 80 (51.6) 543 (47.2) 0.30 White race 1,257 (96.3) 149 (96.1) 1,108 (96.3) 0.91 Intensive group 655 (50.2) 71 (45.8) 584 (50.7) 0.25 Primary cohort 652 (49.9) 58 (37.4) 594 (51.6) ,0.001 Duration of diabetes (years) 13.5 6 4.9 14.5 6 5.2 13.3 6 4.8 0.013 Current smokers 245 (18.8) 41 (26.5) 204 (17.7) 0.009 Current use of blood pressure medications 118 (9.0) 24 (15.5) 94 (8.2) 0.003 Current use of lipid-lowering medications 29 (2.2) 7.7 (12) 1.5 (17) ,0.001 Albuminuria (ever) 226 (17.3) 43 (27.7) 183 (15.9) ,0.001 Weighted means from DCCT BMI (kg/m2)25.16 3.1 25.7 6 3.6 25.1 6 3.0 0.11 Systolic blood pressure (mmHg) 115 6 81186 91156 8 ,0.001 Diastolic blood pressure (mmHg) 74 6 6766 5746 5 ,0.001

HbA1c (%) 8.14 6 1.34 8.46 6 1.47 8.09 6 1.32 0.005 HDL cholesterol (mg/dL) 52 6 12 51 6 12 52 6 12 0.67 Non-HDL cholesterol (mg/dL) 130 6 30 144 6 30 128 6 29 ,0.001 LDL cholesterol (mg/dL) 114 6 26 126 6 26 112 6 26 ,0.001 Total cholesterol (mg/dL) 182 6 29 195 6 31 180 6 29 ,0.001 Triglyceride (mg/dL) 81 6 38 92 6 43 79 6 37 ,0.001 Data are n (%) for categorical variables and mean 6 SD for continuous variables. *P values are based on the x2 test for categorical variables and the Wilcoxon rank sum test for continuous variables.

(HR 2.10 [95% CI 1.26, 3.48] vs. no ab- ECG abnormalities with CVD events is as imbalance resulting from normality/normal ECG, and 2.19 [1.46, not necessarily driven by a certain group . 3.29] vs. no major ECG abnormality), of abnormalities. Our prior report from the EDIC study and did not differ by sex or HbA1c that showed a higher incidence of new (P for interaction = 0.63 and 0.16, re- CONCLUSIONS major ECG abnormalities in patients spectively). Figure 1 shows that the cu- In this analysis of the EDIC cohort, we with type 1 diabetes (6), and our find- mulative incidence of a major ECG examined the association between the ings from this analysis, which show that abnormality and of CVD events both in- presence and development of ECG ab- these abnormalities are predictive of creased in a parallel fashion throughout normalities with incident CVD in pa- poor CVD outcomes, emphasize the re- follow-up. Also, each visit (year) at tients with type 1 diabetes. We found ported high CVD risk in patients with which a diagnosis of a major ECG abnor- that the presence of major ECG abnor- type 1 diabetes (1). Nevertheless, pa- mality was retained was associated malities is associated with a more than tients with type 1 diabetes vary in their with a 30% increased risk of CVD (in twofold increased risk of CVD events af- individual risk profile and susceptibility the fully adjusted model: HR 1.30 [95% ter adjusting for potential confounders, to adverse outcomes (1,5). Hence, pa- CI 1.14, 1.48]) (Table 2). As shown in Fig. and that each additional visit (year) tients with type 1 diabetes must be 2, the log hazard rate for developing a during which a diagnosis of major ECG stratified according to their clinical char- CVD event was a strong linear function abnormality was retained is associated acteristics and risk factor profile for ef- of the cumulative number of visits with a 30% increased risk of a CVD fective patient management. In our (years) with a diagnosis of a major ECG event. study, the association of a major ECG abnormality. The potential mechanism by which abnormality with CVD events persisted The presence of any ECG abnormali- major ECG abnormalities are predictive after controlling for several key clinical ties and the presence of minor ECG ab- of outcome is not totally clear. However, characteristics and risk factors, suggest- normalities were not associated with a they are mostly markers of subclinical ing that ECG could be an additional tool significant increase in CVD risk in any of cardiac disease, including subclinical for examining CVD risk in patients with the models (Table 2). Supplementary myocardial injury/ischemia, myocardial type 1 diabetes. Table 1 shows the individual ECG abnor- dysfunction, and susceptibility to con- The prognostic significance of ECG ab- malities by CVD status. As shown, the duction defects and that normalities as predictors of adverse out- prevalence of individual major ECG ab- could be fatal by themselves or could comes in different populations is well normalities was generally higher in lead to myocardial dysfunction. Also, established (7–11). Notably, however, those who developed versus those ECG abnormalities could be an indica- the magnitude of risk associated with who did not develop CVD, suggesting tion of systemic disease negatively af- these ECG abnormalities varies across that the association between major fecting the cardiovascular system, such populations. These variations are care.diabetesjournals.org Soliman and Associates 797

Figure 1—Cumulative incidences of a major ECG abnormality and of CVD events. further emphasized by the different lev- those guidelines, using resting ECG for it is a class IIb recommendation. The fact els of recommendations by the American cardiovascular risk assessment in asymp- that the distribution and prognostic sig- Heart Association/American College of tomatic adults with and nificance of ECG abnormalities vary Cardiology Foundation regarding the uti- is a class IIa recommen- across populations makes it inappropri- lization of ECG as a screening tool in dation. On the other hand, for those ate to extend evidence developed in one asymptomatic adults (18). According to without hypertension or type 2 diabetes, population to another without testing.

Table 2—Association between the presence of ECG abnormalities as time-dependent covariates and cardiovascular risk Minimally adjusted model* Fully adjusted model† Major/any ECG abnormality (years 1–21)‡ HR (95% CI) P value HR (95% CI) P value No abnormality/normal ECG Reference Reference Minor abnormality 1.05 (0.71, 1.57) 0.80 0.94 (0.63, 1.41) 0.78 Major abnormality 2.67 (1.62, 4.40) 0.001 2.10 (1.26, 3.48) 0.004 No abnormality/normal ECG Reference Reference Any abnormality (major or minor) 1.23 (0.84, 1.80) 0.29 1.08 (0.73, 1.59) 0.70 No major abnormality Reference Reference Major abnormality 2.57 (1.72, 3.84) ,0.001 2.19 (1.46, 3.29) 0.001 Visits/years with a major abnormality (per visit)@ 1.37 (1.21, 1.55) ,0.001 1.30 (1.14, 1.48) ,0.001 Visits/years with any abnormality (per visit)@ 1.07 (0.98, 1.17) 0.12 1.04 (0.95, 1.14) 0.37 *Adjusted for sex and age at EDIC year 1. †Adjusted for sex and age at year 1, primary vs. secondary cohort, weighted mean of systolic blood pressure, use of blood pressure–lowering medications, use of lipid-lowering medications, non-HDL cholesterol, HbA1c, and albuminuria. The following covariates were entered into each model as time-varying covariates: systolic blood pressure, use of lipid-lowering medications, use of blood pressure–lowering medications, non-HDL cholesterol, HbA1c, and albuminuria. ‡Time-varying covariates. @Visits numbering more than five were set to five. 798 Abnormal ECG and CVD Risk in Type 1 Diabetes Diabetes Care Volume 40, June 2017

73.5 6 2.8 years) (8), and a 115% in- creased risk of CVD events (2.15 [1.56, 2.98]) in patients with who are older than 65 years (9). Needless to say, the differences in age, sex, and race among these popula- tions, as well as the definitions of out- comes, make it difficult to appropriately compare the magnitude of risk between major ECG abnormalities with CVD. Our study has some limitations. A sig- nificant association with an outcome is not always reflected as an improved pre- diction of that outcome. Hence, appro- priate statistical methods that show the usefulness of major ECG abnormalities in improving CVD risk prediction are needed. However, all concordance and predictive measures in the context of time-to-event data, such as the area un- der the receiver operating characteristic curve, the integrated discrimination im- provement, and the net reclassification improvement, are only defined for mod- els with fixed covariates (21). This is not the case in our study, where ECG status Figure 2—Number of visits with a major ECG abnormality and risk of CVD. The red line represents and most of the covariates were evalu- the log hazard (y-axis) of CVD risk associated with the number of years with any major ECG ated longitudinally over time for each abnormality. The yellow dashed lines represent 95% point-wise CIs for the log hazard (y-axis) of participant. Moreover, assessing the CVD risk associated with the number of years with any major ECG abnormality. *Visits number- predictive power of a model typically re- ing more than five were set to five. quires a larger number of events than we have in our cohort at this time. Never- theless, providing evidence for an associa- tion between major ECG abnormalities This highlights the importance of our the general population, supports this and CVD events in a comprehensive study study as the first comprehensive report approach and provides evidence that it like ours is a necessary firststeptojus- showing that ECG abnormalities are also is reasonable to presume that risk as- tify further efforts to examine the util- predictive of poor outcomes in type 1 sessment tools for use with the general ity of such an association in improving diabetes, which adds to the evidence population could also work for individu- prediction. of the potential benefitofECGasa als with type 1 diabetes. This does not, The majority of EDIC participants are screening tool in high-risk populations. of course, obviate the need to refine a Caucasian, which may limit the generaliz- However, the ideal frequency and cost- prediction score specifictotype1dia- ability of our results to other races/ effectiveness of using ECG to screen for betes,andECGmaybeofhelpinthis ethnicities. However, the ethnic makeup CVD during the routine care of patients regard, given the magnitude of the as- of the DCCT/EDIC cohort is similar to the with type 1 diabetes need to be as- sociation between major ECG abnormal- general population with type 1 diabetes, sessed. ities and CVD events we observed in which is largely Caucasian. Currently there are no widely used patients with type 1 diabetes, which is We used global classification of ECG CVD risk prediction algorithms specific stronger than the association observed abnormalities (major, minor) rather than to patients with type 1 diabetes, despite in other populations who are even older individual ECG abnormalities. Arguably, some efforts to develop such algorithms and have more . For exam- different individual ECG abnormalities (5,19,20). In the absence of data to the ple, the more than twofold association might have different associations with contrary, the current common approach (HR 2.19) between major ECG abnormal- CVD. However, our approach of using a to identifying CVD risk in patients with ities and CVD events in type 1 diabetes global classification for ECG abnormali- type 1 diabetes has been to apply the compares with only an 83% increased ties is common, and several previous re- same risk assessment and diagnostic risk of CVD events (HR 1.83 [95% CI ports have shown its usefulness for both strategies used in the general popula- 1.12, 2.97]) in an HIV-infected popula- the assessment (22–25) and prediction tion (1). Our conclusion that ECG abnor- tion (age 43.5 6 9.3 years) (10), a 47% (6–10) of CVD. The main reason to use a malities are predictive of poor outcomes increased risk of coronary heart disease global classification of minor/major ECG in type 1 diabetes, which is in accor- events (1.47 [1.16, 1.86]) in a general abnormalitiesdin addition to its simplic- dance with conclusions of studies of population of elderly patients (age ity, which enhances comparability across care.diabetesjournals.org Soliman and Associates 799

studiesdis that the approach overcomes Care, Becton Dickinson, Eli Lilly and Company, 12. Olson JC, Orchard TJ. Ischemic ECG changes statistical power concerns if ECG abnor- Extend , Insulet Corporation, LifeScan, predict in type 1 diabe- malities are used individually. Medtronic Diabetes, Nipro Home Diagnostics, tes. Int J Cardiol 2005;98:511 Nova Diabetes Care, Omron Corp., Perrigo Di- 13. Diabetes Control and Complications Trial Re- Finally, clinical practice and care for abetes Care, Roche Diabetes Care, and Sanofi. search Group, Nathan DM, Genuth S, et al. The type 1 diabetes have evolved since the Duality of Interest. No potential conflicts of effect of intensive treatment of diabetes on the start of the EDIC study in 1994. For exam- interest relevant to this article were reported. development and progression of long-term com- ple, HbA is monitored much more fre- Author Contributions. E.Z.S. wrote the man- plications in -dependent diabetes melli- 1c uscript. E.Z.S., J.-Y.C.B., I.B., and J.M.L. inter- tus. N Engl J Med 1993;329:977–986 quently in contemporary practice than preted the results. E.Z.S. and J.M.L. designed the 14. Prineas RJ, Crow R, Blackburn H. The Minne- before. If we had used time-fixed covari- study. J.-Y.C.B. and I.B. performed statistical sota Code Manual of Electrocardiographic Find- ates, this could have raised concerns analysis. J.-Y.C.B., I.B., T.J.O., B.Z., and J.M.L. ings: Standards and Procedures for Measurement about the conclusions of our study. How- critically revised the manuscript for content. and Classification. Boston, MA, John Wright, 1982 J.M.L. is the guarantor of this work and, as ever, as stated in RESEARCH DESIGN AND METH- 15. Diabetes Control and Complications Trial such, had full access to all the data in the study (DCCT)/Epidemiology of Diabetes Interventions ODS, we used time-dependent covariates, and takes responsibility for the integrity of the and Complications (EDIC) Study Research which enabled our statistical models to data and the accuracy of the data analysis. Group. Intensive diabetes treatment and car- capture any changes in key clinical factors diovascular outcomes in type 1 diabetes: the resulting from changes in either practice References DCCT/EDIC study 30-year follow-up. Diabetes – or participant behavior during follow-up. 1. de Ferranti SD, de Boer IH, Fonseca V, et al. Care 2016;39:686 693 Despite these limitations, this is, to Type 1 diabetes mellitus and cardiovascular dis- 16. Therneau TM, Grambsch PM. Modeling sur- ease: a scientific statement from the American vival data: Extending the Cox Model. Statistics for our knowledge, the first comprehensive Biology and Health. New York, Springer-Verlag, 2000 fi Heart Association and American Diabetes Asso- report of the prognostic signi cance of ciation. Diabetes Care 2014;37:2843–2863 17. Lachin JM. Biostatistical Methods: The As- ECG abnormalities in type 1 diabetes. 2. Khunti K, Davies M, Majeed A, Thorsted BL, sessment of Relative Risks. 2nd ed. New York, The uniform collection of data, including Wolden ML, Paul SK. and risk of John Wiley & Sons, 2011 centrally read ECG data, and the long- cardiovascular disease and all-cause mortality in 18. Greenland P, Alpert JS, Beller GA, et al.; American College of Cardiology Foundation; term follow-up of a cohort of patients insulin-treated people with type 1 and type 2 diabetes: a cohort study. Diabetes Care 2015; American Heart Association. 2010 ACCF/AHA with type 1 diabetes with extensive phe- 38:316–322 guideline for assessment of cardiovascular risk notyping are just a few of the many 3. Livingstone SJ, Looker HC, Hothersall EJ, et al. in asymptomatic adults: a report of the Ameri- strengths of the EDIC study. Risk of cardiovascular disease and total mortal- can College of Cardiology Foundation/American In conclusion, the presence of major ity in adults with type 1 diabetes: Scottish reg- Heart Association Task Force on Practice Guide- lines. J Am Coll Cardiol 2010;56:e50–e103 ECG abnormalities during the course of istry linkage study. PLoS Med 2012;9:e1001321 4. Secrest AM, Becker DJ, Kelsey SF, Laporte RE, 19. Zgibor JC, Ruppert K, Orchard TJ, et al. De- type 1 diabetes is associated with an in- Orchard TJ. Cause-specific mortality trends in a velopment of a coronary heart disease risk creased risk of CVD events. Identifying large population-based cohort with long- prediction model for type 1 diabetes: the Pitts- risk markers/predictors such as ECG ab- standing childhood-onset type 1 diabetes. Dia- burgh CHD in Type 1 Diabetes Risk Model. Di- abetes Res Clin Pract 2010;88:314–321 normalities in type 1 diabetes could help betes 2010;59:3216–3222 5. Vistisen D, Andersen GS, Hansen CS, et al. 20. Cederholm J, Eeg-Olofsson K, Eliasson B, guide future efforts toward the develop- Prediction of first cardiovascular disease event Zethelius B, Gudbjornsdottir¨ S; Swedish National ment of risk stratification tools to iden- in type 1 diabetes mellitus: the Steno Type 1 Diabetes Register. A new model for 5-year risk of tify those who may benefitfromcloser Risk Engine. Circulation 2016;133:1058–1066 cardiovascular disease in type 1 diabetes; from follow-up and earlier, more aggressive 6. Soliman EZ, Backlund JY, Bebu I, et al.; DCCT/ the Swedish National Diabetes Register (NDR). Diabet Med 2011;28:1213–1220 risk factor management. EDIC Research Group. Progression of electrocar- diographic abnormalities in type 1 diabetes dur- 21. Heagerty PJ, Zheng Y. Survival model pre- ing 16 years of follow-up: the Epidemiology of dictive accuracy and ROC curves. Biometrics Diabetes Interventions and Complications 2005;61:92–105 Acknowledgments. The authors acknowledge (EDIC) Study. J Am Heart Assoc 2016;5:e002882 22. Prineas RJ, Le A, Soliman EZ, et al.; Reasons the data processing and technical assistance of 7. Denes P, Larson JC, Lloyd-Jones DM, Prineas RJ, for Geographic and Racial Differences in Stroke Wanyu Hsu at the Biostatistics Center, The George Greenland P. Major and minor ECG abnormalities (REGARDS) Investigators. United States national WashingtonUniversity, Rockville,MD, andCharles in asymptomatic women and risk of cardiovascu- prevalence of electrocardiographic abnormalities C. Campbell at the EPICARE ECG Center, Wake lar events and mortality. JAMA 2007;297:978–985 in black and white middle-age (45- to 64-year) and Forest School of Medicine, Winston-Salem, NC. 8. Auer R, Bauer DC, Marques-Vidal P, et al.; older ($65-year) adults (from the Reasons for Funding. The DCCT/EDIC study has been sup- Health ABC Study. Association of major and mi- Geographic and Racial Differences in Stroke ported by U01 Cooperative Agreement grants nor ECG abnormalities with coronary heart dis- Study). Am J Cardiol 2012;109:1223–1228 (1982-93, 2011–2016) and contracts (1982– ease events. JAMA 2012;307:1497–1505 23. Vitelli LL, Crow RS, Shahar E, Hutchinson 2011) with the Division of Diabetes, Endocrinol- 9. Dobre M, Brateanu A, Rashidi A, Rahman M. RG, Rautaharju PM, Folsom AR; ogy, and Metabolic of the National Electrocardiogram abnormalities and cardiovas- Risk in Communities (ARIC) Study Investigators. Institute of Diabetes and Digestive and Kidney cular mortality in elderly patients with CKD. Clin Electrocardiographic findings in a healthy bira- Diseases (current grants U01 DK094176 and U01 J Am Soc Nephrol 2012;7:949–956 cial population. Am J Cardiol 1998;81:453–459 DK094157), and through support from the 10. Soliman EZ, Prineas RJ, Roediger MP, et al. 24. Walsh JA 3rd, Prineas R, Daviglus ML, et al. National Eye Institute, the National Institute of Prevalence and prognostic significance of ECG Prevalence of electrocardiographic abnormali- Neurological Disorders and Stroke, the Genetic abnormalities in HIV-infected patients: results ties in a middle-aged, biracial population: Coro- Clinical Research Centers Program (1993–2007), from the Strategies for Management of Antire- nary Artery Risk Development in Young Adults and the Clinical Translational Science Center Pro- troviral Therapy study. J Electrocardiol 2011;44: study. J Electrocardiol 2010;43:385.e1–385.e9 gram (2006–present), Bethesda, MD. 779–785 25. Denes P, Garside DB, Lloyd-Jones D, et al. Ma- The following industry contributors had no 11. Moyer VA; U.S. Preventive Services Task jor and minor electrocardiographic abnormalities role in the DCCT/EDIC study but provided free Force. Screening for coronary heart disease and their association with underlying cardiovascular or discounted supplies or equipment to support with electrocardiography: U.S. Preventive Ser- disease and risk factors in Hispanics/Latinos (from participants’ adherence to the study: Abbott Di- vices Task Force recommendation statement. the Hispanic Community Health Study/Study of abetes Care Division, Animas, Bayer Diabetes Ann Intern Med 2012;157:512–518 Latinos). Am J Cardiol 2013;112:1667–1675