Heart Rate-Corrected QT Interval Prolongation Predicts Risk of Coronary Heart Disease in Black and White Middle-Aged Men and Women the ARIC Study Jacqueline M

Home , QT interval

View metadata, citation and similar papers at core.ac.uk brought to you by CORE Journal of the American College of Cardiology providedVol. by Elsevier 43, No. - 4,Publisher 2004 Connector © 2004 by the American College of Cardiology Foundation ISSN 0735-1097/04/$30.00 Published by Elsevier Inc. doi:10.1016/j.jacc.2003.09.040 Risk Predictors and Coronary Disease Heart Rate-Corrected QT Interval Prolongation Predicts Risk of Coronary Heart Disease in Black and White Middle-Aged Men and Women The ARIC Study Jacqueline M. Dekker, PHD,*† Richard S. Crow, MD,‡ Peter J. Hannan, MSTAT,‡ Evert G. Schouten, MD, PHD,* Aaron R. Folsom, MD, MPH‡ Wageningen and Amsterdam, the Netherlands; and Minneapolis, Minnesota

OBJECTIVES We aimed to study the predictive value of heart rate-corrected QT interval (QTc) for incident coronary heart disease (CHD) and cardiovascular disease (CVD) mortality in the black and white general population, and to validate various QT measurements. BACKGROUND QTc prolongation is associated with higher risk of mortality in cardiac patients and in the general population. Little is known about the association with incident CHD. No previous studies included black populations. METHODS We studied the predictive value of QTc prolongation in a prospective population study of 14,548 black and white men and women, age 45 to 64 year. QT was determined by the NOVACODE program in the digital electrocardiogram recorded at baseline. RESULTS In quintiles of QTc, cardiovascular risk profile deteriorated with longer QTc, and risk of CHD and CVD mortality increased. The high risk in the upper quintile was mostly explained by the 10% with the longest QTc. The age-, gender-, and race-adjusted hazard ratios for CVD mortality and CHD in subjects with the longest 10% relative to the other 90% of the gender-specific QTc distribution were 5.13 (95% confidence interval 3.80 to 6.94) and 2.14 (95% confidence interval 1.71 to 2.69), respectively. The increased risk was partly, but not completely, attributable to other risk factors or the presence of chronic disease. The association was stronger in black than in white subjects. Manual- and machine-coded QT intervals were highly correlated, and the method of rate correction did not affect the observed associations. CONCLUSIONS Long QTc is associated with increased risk of CHD and CVD mortality in black and white healthy men and women. (J Am Coll Cardiol 2004;43:565–71) © 2004 by the American College of Cardiology Foundation

The heart rate-corrected QT (QTc) interval in the resting QT interval have been reported (11), there is no information 12-lead electrocardiogram (ECG) provides important prog- about the association of QTc with CVD mortality or incident nostic information in clinical practice. In post-myocardial CHD in the black population. Therefore, we studied QT in infarction patients (1), subjects referred for Holter monitor- relation to risk of CHD incidence and CVD mortality in a ing (2), and subjects with diabetes (3) QTc prolongation is large biracial U.S. population study, the Atherosclerosis Risk In Communities (ARIC) study. See page 572 Because measurement of QT is hampered by systematic differences between manual coders and between computerized associated with mortality risk. In inheritable syndromes, which QT measurements of different ECG systems (12), manually are characterized by QTc prolongation, affected patients have determined QT intervals were compared with machine- high risk of sudden death (4). Also, in several population derived QT intervals. In addition, to compare QT intervals studies (5–10), increased risk of cardiovascular disease (CVD) between subjects, adjustment for differences in heart rate is mortality was observed in subjects with QTc prolongation. essential. The widely used Bazett formula is known to over- Less is known about the association with incident coronary adjust at very high and low heart rates, and several alternative heart disease (CHD). Furthermore, though racial differences in methods have been proposed (13,14). We compared three commonly used formulae for the correction of heart rate.

From the *Department of Epidemiology and Public Health, Wageningen Univer- sity, Wageningen, The Netherlands, †Institute for Research in Extramural Medicine, METHODS VU University Medical Center, Amsterdam, The Netherlands; and the ‡Division of Epidemiology, School of Public Health, University of Minnesota, Minneapolis, Study population. The ARIC study is following a cohort Minnesota. Supported by a grant from the Netherlands Organization for Scientiﬁc of 15,792 middle-aged men and women (15). During 1987 Research (NWO) (Dr. Dekker). Manuscript received December 13, 2002; revised manuscript received August 15, to 1989, population samples were drawn from the 45- to 2003, accepted September 17, 2003. 64-year-old inhabitants of Forsyth County, North Carolina; 566 Dekker et al. JACC Vol. 43, No. 4, 2004 QT and CHD Risk in the General Population February 18, 2004:565–71

dardized B-mode ultrasonic technique (20). Use of cardiac Abbreviations and Acronyms medication was defined as self-reported use of medication ARIC ϭ Atherosclerosis Risk In Communities study for angina pectoris, heart failure, or rhythm disturbances CHD ϭ coronary heart disease during the last two weeks. ϭ CVD cardiovascular disease Methods of follow-up. The method of follow-up has been ECG ϭ electrocardiogram QTc ϭ heart rate-corrected QT interval described previously (21). Briefly, interviewers contacted participants annually by telephone to identify hospitaliza- tions and deaths. Death certificates and discharge lists from Jackson, Mississippi; the northwest suburbs of Minneapolis, local hospitals were surveyed by ARIC staff to detect Minnesota; and Washington County, Maryland; they were additional deaths and cardiovascular events. For hospital- invited for a home interview and a clinical examination. In ized patients with potential acute CHD events, trained Jackson, 46% of eligible subjects participated; approximately abstractors recorded the presenting signs and symptoms, 65% participated in the other three centers. For the present including chest pain, cardiac enzymes, and related clinical study, subjects with a self-reported history of physician- information. Up to three 12-lead ECGs were visually coded diagnosed heart attack, prior cardiovascular surgery or using the Minnesota Code, and wave-form evolution was coronary angioplasty, or with 12-lead Minnesota Code evaluated using side-by-side comparisons. Out-of-hospital evidence of prior myocardial infarction were excluded, deaths were investigated by means of the death certificate leaving 14,672 subjects. After exclusion of subjects with and, in most cases, an interview with next of kin and missing data on QT, the study population consists of 14,548 questionnaires completed by the patients’ physicians. Cor- men and women. oner reports and autopsy reports, when available, were used Baseline measurements. In the home interview, questions in validation. All potential in- and out-of-hospital clinical were asked about health behaviors, sociodemographics, and CHD events were reviewed and adjudicated by the ARIC disease history. The clinical examination included electro- Morbidity and Mortality Classification Committee using cardiography. After resting 5 to 10 min while electrodes published criteria (21). were placed, a standard 12-lead ECG and a 2-min paper In the present study, follow-up data through 1993 were recording of a three-lead (leads V , II, and V ) rhythm strip 1 5 used. Follow-up information on incident CHD and mor- were made in the supine position. The QT interval from the tality was complete for 95%. The CHD incidence was digital 12-lead ECG was determined by the NOVACODE defined as definite or probable myocardial infarction, cardiac program (16). The NOVACODE program generates an revascularization procedures (excluding thrombolytic ther- average waveform derived from all 12 simultaneously mea- apy), or definite CHD death. Cardiovascular disease mor- sured leads. This allows the system to determine the QT tality was based on the underlying cause from the death from the earliest QRS onset to the latest offset of the certificates: ICD-9 codes 390 to 459. T-wave. All NOVACODE measurements were visually verified. When measurements between computer and oper- Manual QT measurement. QT was manually determined ator differed by Ͼ5 ms, the operator’s reference points were in a case-cohort sample from the ARIC study (21), consist- used. Plasma total cholesterol and triglycerides were mea- ing of a random sample of 900 subjects, all incident CHD sured by enzymatic methods (17,18), and low-density li- cases, and all deaths. Measurements were performed at the poprotein cholesterol was calculated using the Friedewald University of Minnesota ECG Coding Center, using a formula. High-density lipoprotein cholesterol was measured digitizing tablet (Calcomp, Columbia, Maryland) and a after dextran-magnesium precipitation of non–high-density personal computer. Coders mounted the 3-lead two-min lipoproteins (19). Serum insulin was measured using a rhythm strip on the tablet, and used the mouse to mark QT radioimmunoassay (125I Insulin Kit, Cambridge Medical onset and T-wave offset and preceding RR intervals in three Diagnostics, Billerica, Massachusetts). Serum glucose was consecutive normal beats in all three leads. T-wave offset assessed by the hexokinase method. Prevalent diabetes was defined as the point of maximum change of slope as the mellitus was defined as a fasting glucose level of 126 mg/dl T-wave merges with the baseline (6). This accords with the (7.0 mmol/l) or more, nonfasting glucose level of 200 mg/dl method applied in the NOVACODE program (16). The (11.1 mmol/l) or more, and/or a history of, or treatment for, inter- and intracoder variability was evaluated by means of diabetes. Three seated blood pressure measurements were analysis of variance in a set of 10 ECGs, which were taken on the right arm, using a random-zero sphygmoma- measured three times by each coder. The QT interval was nometer. The mean of the last two measurements was used. used as the dependent variable. Between-subject variance 2 2 Hypertension was defined as systolic blood pressure of 140 was 356 ms , the between-coder variance 79 ms , and the mm Hg or more, diastolic pressure of 90 mm Hg or more, intercoder variance 15 ms2. or use of antihypertensive agents. The ratio of waist (um- Data analysis. The QT was corrected for heart rate using bilical level) and hip (maximum buttocks) circumferences three different methods: Bazett (22) ͑QTc ϭ QT/ was calculated as a measure of fat distribution. Average ͱRR [QT and RR in s]), the linear method proposed by carotid intima-media thickness was assessed using a stan- Hodges et al. (14) (QThodges ϭ QT ϩ 1.75 [heart rate JACC Vol. 43, No. 4, 2004 Dekker et al. 567 February 18, 2004:565–71 QT and CHD Risk in the General Population

Table 1. Age-, Gender-, and Race-Adjusted Baseline Characteristics in Gender-Speciﬁc Quintiles of QTc: The ARIC Study QTc Quintiles Characteristics* Q1 Q2 Q3 Q4 Q5 p for Trend* Age (yrs) 53.2 53.5 53.9 54.0 54.9 Ͻ0.0001 Race (% black) 30 23 25 25 30 0.42 Body mass index (kg/m2) 27.2 27.7 28.2 28.6 29.3 Ͻ0.0001 Waist-to-hip ratio 0.914 0.921 0.925 0.934 0.944 Ͻ0.0001 Heart rate (beats/min) 60 64 67 69 73 Ͻ0.0001 Current smoking (%) 26 27 26 27 31 0.0004 Smoking (cigarette yrs) 285 293 303 297 336 Ͻ0.0001 Systolic pressure (mm Hg) 120 121 122 125 129 Ͻ0.0001 Hypertension (%) 26 28 30 36 46 Ͻ0.0001 Triglycerides (mmol/l)‡ 1.13 1.17 1.20 1.26 1.35 Ͻ0.0001 Total cholesterol (mmol/l) 5.50 5.52 5.53 5.54 5.61 Ͻ0.0001 HDL cholesterol (mmol/l) 1.36 1.36 1.37 1.34 1.31 Ͻ0.0001 Insulin (pmol/l)†‡ 60 64 68 70 75 Ͻ0.0001 Glucose (mmol/l)† 5.46 5.48 5.49 5.50 5.55 Ͻ0.0001 Diabetes (%) 10 12 12 14 20 Ͻ0.0001 Intima-media thickness (mm) 0.735 0.741 0.745 0.748 0.767 Ͻ0.0001 ECG abnormalities (%)§ 21 21 22 25 41 Ͻ0.0001 Cardiac medication (%) 43346 0.003 Potassium (mmol/l) 4.42 4.39 4.37 4.34 4.26 Ͻ0.0001 Calcium (mg/dl) 9.82 9.80 9.79 9.79 9.78 Ͻ0.0001

Quintile cut points for QTc were 403, 415, 425, and 440 ms for men and 417, 429, 440, and 454 ms for women. *Test for trend: linear regression for continuous variables, and logistic regression for dichotomous variables, with quintiles as a continuous explanatory variable; †Subjects with diabetes or not in fasting state were excluded for these analyses; ‡Because insulin and triglycerides were not normally distributed, the geometric mean is presented; §ECG abnormalities: presence of Minnesota codes 1.1, 1.2 (major Q waves), any 4 or 5 codes (ST- and T-wave changes), or codes 7.1 to 7.6 (conduction disturbances). ARIC ϭ Atherosclerotic Risk in Communities study; ECG ϭ electrocardiographic; HDL ϭ high-density lipoprotein; QTc ϭ heart rate-corrected QT interval.

Ϫ 60] [QT in ms]) , and by the method proposed by Rautaharju creasing QTc. Longer QTc was also associated with lower (13) (QTindex ϭ QT · [heart rate ϩ 100]/656 [QT in ms]). serum potassium and calcium concentrations. Similar results Age-, gender-, and race-adjusted baseline characteristics were were observed in categories of QThodges and QTindex compared over categories of gender-specific quintiles of QTc, (data not shown). QThodges, and QTindex. Age-, race-, and gender-adjusted In Table 2, relative rates are shown in quintiles of QTc. baseline characteristics were derived from regression analyses, Risk of CVD mortality, incident CHD, and total mortality using the specific characteristic as the dependent, and age, gender, was greater with longer QTc in black men and women and and QT categories as explanatory variables. Relative risk of in white women. In white men, a U-shaped association was cardiovascular disease mortality, CHD incidence, and total mor- observed for CVD mortality. In the total cohort, the age-, tality in gender-specific quintiles and in the upper 10% of QTc, of gender-, and race-adjusted hazard ratios of CVD mortality, QThodges, and of QTindex were estimated using Cox propor- but not of incident CHD, in the lowest quintile were higher tional hazards analysis in the complete cohort. In the multivariable than in the second quintile. From the fourth quintile on, models, we forced in risk factors, which were previously suggested, relative risk for all end points increased. When adjusting for as possibly confounding or mediating in the literature. For analyses other cardiovascular risk factors one by one, none of these of the manually determined QT, Poisson regression for the explained a large part of the increased risk. After adjustment case-cohort design was used (23). for all other risk factors combined, only risk in the highest quintile was still significantly elevated. The large increase in RESULTS risk in the highest quintile was mostly explained by the 10% The correlation between the three QTc values was high: with the longest QT interval. The age-, gender-, and 0.81 between QTc and QThodges, 0.96 between QTc and race-adjusted hazard ratios for CVD mortality and CHD QTindex, and 0.96 between QThodges and QTindex. incidence were about 4 and 2, decreasing to about 2 and 1.5 Mean QTc was 436 Ϯ 25 ms in black women, 435 Ϯ 23 ms after adjustment for all other risk factors (Table 3). Results in white women, 421 Ϯ 27 ms in black men, and 422 Ϯ 22 were similar for QTc, QThodges, and QTindex. Stratifica- ms in white men. In Table 1, age-, gender-, and race- tion for race showed higher relative risks among the black adjusted baseline characteristics are shown across gender- than among the white population (Table 3). A test of specific quintiles of QTc. In all groups, men and women, statistical significance showed significant interaction be- and in blacks and whites, a more adverse cardiovascular risk tween QT prolongation and race. Using race- and gender- profile with increasing QTc was observed. Intima-media specific cutoff points did not change the observed difference thickness and the prevalence of hypertension, diabetes, and (data not shown). The reduction of the prognostic value of electrocardiographic abnormalities were greater with in- QTc prolongation in the multivariable model in the white 568 Dekker et al. JACC Vol. 43, No. 4, 2004 QT and CHD Risk in the General Population February 18, 2004:565–71

Table 2. Relative Rates (95% CI) of CVD Mortality, Incident CHD, and Total Mortality in Gender-Speciﬁc Quintiles of the Heart Rate-Adjusted QT Quintile (n) CVD Death CHD Incidence Total Mortality Age-adjusted Black women n ϭ 44 (4, 4, 1, 10, 25)* n ϭ 44 (6, 8, 3, 9, 18) n ϭ 104 (14, 13, 10, 23, 44) QTc Q1 (540) 1 1 1 Q2 (417) 1.30 (0.33–5.20) 1.73 (0.60–4.88) 1.21 (0.57–2.57) Q3 (435) 0.30 (0.03–2.69) 0.61 (0.15–2.42) 0.66 (0.38–1.94) Q4 (508) 2.57 (0.81–8.18) 1.52 (0.54–4.27) 1.69 (0.87–3.29) Q5 (553) 5.79 (2.01–16.64) 2.75 (1.09–6.94) 2.93 (1.61–5.35) White women n ϭ 27 (1, 2, 5, 9, 10) n ϭ 85 (6, 12, 14, 19, 34) n ϭ 120 (20, 16, 19, 33, 32) QTc Q1 (1,123) 1 1 1 Q2 (1,194) 1.86 (0.17–20.45) 1.86 (0.70–4.96) 0.74 (0.39–1.43) Q3 (1,132) 4.75 (0.56–40.64) 2.22 (0.85–5.77) 0.90 (0.48–1.69) Q4 (1,207) 7.90 (1.00–62.36) 2.71 (1.08–6.79) 1.44 (0.82–2.50) Q5 (1,121) 8.29 (1.06–64.95) 4.66 (1.95–11.15) 1.32 (0.75–2.31) Black men n ϭ 50 (1, 4, 4, 11, 30) n ϭ 64 (6, 7, 11, 12, 28) n ϭ 117 (12, 13, 12, 28, 52) QTc Q1 (355) 1 1 1 Q2 (254) 5.82 (0.65–52.08) 1.70 (0.57–5.06) 1.58 (0.72–3.47) Q3 (256) 5.02 (0.56–44.93) 2.30 (0.85–6.23) 1.21 (0.54–2.70) Q4 (296) 12.27 (1.58–95.15) 2.20 (0.83–5.88) 2.53 (1.29–4.99) Q5 (317) 31.47 (4.28–231) 4.96 (2.05–12.03) 4.35 (2.31–8.16) White men n ϭ 57 (14, 5, 6, 11, 21) n ϭ 252 (38, 53, 40, 54, 67) n ϭ 173 (25, 29, 28, 33, 58) QTc Q1 (882) 1 1 1 Q2 (1,001) 0.29 (0.11–0.82) 1.18 (0.78–1.79) 0.94 (0.55–1.60) Q3 (900) 0.39 (0.15–1.01) 0.98 (0.63–1.53) 0.99 (0.17–1.69) Q4 (1,161) 0.53 (0.23–1.16) 0.99 (0.65–1.50) 0.85 (0.51–1.43) Q5 (886) 1.22 (0.62–2.20) 1.54 (1.03–2.30) 1.76 (1.10–2.83) Total cohort Age-, gender-, and race-adjusted QTc Q1 1 1 1 Q2 0.83 (0.43–1.63) 1.39 (0.99–1.96) 1.06 (0.76–1.47) Q3 0.86 (0.45–1.66) 1.22 (0.86–1.74) 1.01 (0.72–1.40) Q4 1.85 (1.08–3.15) 1.36 (0.97–1.89) 1.42 (1.06–1.91) Q5 3.91 (2.40–6.37) 2.34 (1.72–3.19) 2.28 (1.73–3.00) Adjusted for cardiovascular risk factors† QTc Q1 1 1 1 Q2 0.93 (0.43–1.97) 1.31 (0.92–1.88) 1.04 (0.73–1.49) Q3 1.02 (0.49–2.13) 1.16 (0.79–1.70) 0.98 (0.68–1.41) Q4 1.50 (0.79–2.87) 1.20 (0.84–1.72) 1.10 (0.79–1.55) Q5 2.00 (1.07–3.75) 1.55 (1.08–2.23) 1.33 (0.95–1.87)

Quintile cut points for QTc were 403, 415, 425, and 440 ms for men and 417, 429, 440, and 454 ms for women. *Number of cases (number of cases in Q1 to Q5); †Adjusted for age, gender, race, heart rate, hypertension, systolic blood pressure, ECG abnormalities, body mass index, waist-hip ratio, current smoking, cigarette years, total cholesterol, HDL cholesterol, triglycerides, use of cardiac medication, intima-media thickness, and diabetes. CHD ϭ coronary heart disease; CI ϭ conﬁdence interval; CVD ϭ cardiovascular disease; HDL ϭ high-density lipoprotein; QTc ϭ heart rate-corrected QT interval. population was observed in the women only. This may be what lower than that of the machine QT. However, this was due to low event rates because none of the individual due to the higher correlation of the machine QT with heart variables explained this. The effect was observed only when rate; the Pearson correlations of manual- and machine- all variables were included, and it disappeared when a lower coded QTc with heart rate were 0.33 and 0.48, respectively. cutoff point was used. After adjustment for heart rate, they had similar prognostic The results did not differ over strata of heart rate, value (Table 5). diabetes, hypertension, or presence of ECG abnormalities. When stratifying for the presence of diabetes, hypertension, DISCUSSION use of cardiac medication, or ECG abnormalities combined, the predictive value (after adjustment for other risk factors) In this population-based cohort study of middle-aged men of QTc and QTindex was somewhat higher in the healthy and women, a longer QTc was associated with an adverse subpopulation (Table 4). cardiovascular risk proﬁle. Prolonged QTc was a strong The average machine-determined QT interval was 10 ms predictor of incident CHD and especially of CVD mortal- longer than the manually determined QT, and the correla- ity, independent of the other risk factors. The association tion was 0.81. In the age- gender-, and coder-adjusted was stronger in black than in white subjects. model, the predictive value of the manual QTc was some- QT was determined in the digital ECG, and manually in JACC Vol. 43, No. 4, 2004 Dekker et al. 569 February 18, 2004:565–71 QT and CHD Risk in the General Population

Table 3. Relative Rates (95% CI) of CVD Mortality, Incident CHD, and Total Mortality in the Highest Gender-Speciﬁc 10% Compared with the Other 90% of the QT Distribution* CVD Death CHD Incidence Total Mortality All Age-, gender-, and, race-adjusted QTc 5.13 (3.80–6.94) 2.14 (1.71–2.69) 2.35 (1.92–2.87) QThodges 3.15 (2.29–4.35) 1.76 (1.38–2.24) 1.81 (1.46–2.25) QTindex 3.57 (2.63–4.85) 2.03 (1.63–2.53) 2.06 (1.68–2.52) Adjusted for cardiovascular risk factors† QTc 2.53 (1.74–3.69) 1.51 (1.15–1.98) 1.37 (1.07–1.76) QThodges 2.45 (1.68–3.57) 1.36 (1.03–1.80) 1.59 (1.24–2.04) QTindex 2.13 (1.48–3.06) 1.47 (1.15–1.89) 1.43 (1.14–1.81) Blacks n ϭ 3,931 Age- and gender-adjusted Number of events/number at risk:‡ 46/3,451, 48/480 72/3,451, 36/480 150/3,451, 71/480 QTc 7.21 (4.80–10.82) 3.47 (2.32–5.18) 3.23 (2.43–4.28) QThodges 4.06 (2.68–6.15) 3.33 (2.23–4.96) 2.41 (1.80–3.24) QTindex 4.33 (2.87–6.53) 3.42 (2.31–5.05) 2.46 (1.85–3.28) Adjusted for cardiovascular risk factors† QTc 3.14 (1.93–5.12) 2.07 (1.24–3.46) 1.77 (1.23–2.55) QThodges 4.14 (2.49–6.90) 2.55 (1.55–4.20) 2.20 (1.54–3.14) QTindex 3.95 (2.35–6.66) 2.08 (1.30–3.34) 1.66 (1.18–2.33) Whites n ϭ 10,617 Age- and gender-adjusted Number of events/number at risk:‡ 59/9,601, 25/1,016 278/9,601, 59/1,016 239/9,601, 54/1,016 QTc 3.39 (2.12–5.44) 1.76 (1.33–2.16) 1.73 (1.28–2.33) QThodges 2.22 (1.31–3.74) 1.29 (0.94–1.76) 1.35 (0.97–1.86) QTindex 2.82 (1.76–4.53) 1.64 (1.25–2.16) 1.74 (1.31–2.31) Adjusted for cardiovascular risk factors† QTc 1.72 (0.94–3.16) 1.39 (1.00–1.92) 1.12 (0.79–1.60) QThodges 1.35 (0.71–2.54) 1.08 (0.77–1.54) 1.17 (0.81–1.69) QTindex 1.55 (0.88–2.74) 1.35 (1.01–1.82) 1.30 (0.94–1.80)

*Cut point for prolongation; QTc: 450 ms for men, 465 ms for women; QThodges: 441 ms for men, 451 ms for women; QTindex: 108 ms for men, 112 ms for women; †Adjusted for age, gender, (race in the combined analyses), heart rate, hypertension, systolic blood pressure, ECG abnormalities, body mass index, waist-hip ratio, current smoking, cigarette years, total cholesterol, high-density lipoprotein cholesterol, triglycerides, use of cardiac medication, intima-media thickness, and diabetes; ‡Number of events/number at risk in the lowest 90%, the upper 10%, respectively. CHD ϭ coronary heart disease; CI ϭ conﬁdence interval; CVD ϭ cardiovascular death; QTc ϭ heart rate-corrected QT interval. the three-lead rhythm strip, which were recorded at the The QT interval has to be corrected for heart rate to same visit. We observed a high agreement between compare individuals with different heart rates. There are machine- and manual-determined QT intervals, and both several methods, Bazett’s being the most commonly used. had similar prognostic value. Therefore, it can be concluded As illustrated by our data, most methods will be highly that NOVACODE-determined QT intervals can safely be correlated in a normal population. The Bazett correction used in population studies. This may not be true for patient has been criticized for the residual correlation with heart populations, because the normal pattern is easier for auto- rate. However, this does not explain the association with matic measurements to detect (14). CVD risk. In the present study adding heart rate to the age- Table 4. Relative Rates (95% CI) of CVD Mortality, Incident CHD, and Total Mortality in the Highest Gender-Speciﬁc 10% Compared with the Other 90% of the QT Distribution*, According to Health Status CVD Death CHD Incidence Total Mortality Adjusted for cardiovascular risk factors† Subjects with any diabetes, hypertension, ECG abnormalities,‡ or using cardiac medication Number of events/number at risk 120/5,963 255/5,963 284/5,963 QTc 2.78 (1.88–4.26) 1.36 (1.01–1.83) 1.48 (1.13–1.95) QThodges 2.91 (1.99–4.26) 1.55 (1.16–2.08) 1.75 (1.34–2.28) QTindex 2.45 (1.68–3.56) 1.46 (1.06–1.85) 1.53 (1.18–1.98) Subjects without diabetes, hypertension, ECG abnormalities, and not using cardiac medication Number of events/number at risk 26/7,140 138/7,140 155/7,140 QTc 6.38 (2.53–16.10) 2.40 (1.41–4.09) 1.71 (1.03–2.85) QThodges 3.72 (2.53–10.23) 0.93 (0.47–1.87) 1.54 (0.88–2.70) QTindex 4.39 (1.81–10.61) 1.97 (1.22–3.18) 1.86 (1.17–2.95)

*Cut point for prolongation: QTc: 450 ms for men, 465 ms for women; QThodges: 441 ms for men, 451 ms for women; QTindex: 108 ms for men, 112 ms for women; †Adjusted for age, gender, race, heart rate, systolic blood pressure, body mass index, waist-hip ratio, current smoking, cigarette years, total cholesterol, high-density lipoprotein cholesterol, triglycerides, and intima-media thickness; ‡ECG abnormalities: presence of Minnesota codes 1.1, 1.2 (major Q waves), any 4 or 5 codes (ST- and T-wave changes), or codes 7.1 to 7.6 (conduction disturbances). CHD ϭ coronary heart disease; CI ϭ conﬁdence interval; CVD ϭ cardiovascular disease; QTc ϭ heart rate-corrected QT interval. 570 Dekker et al. JACC Vol. 43, No. 4, 2004 QT and CHD Risk in the General Population February 18, 2004:565–71

Table 5. Comparison of Prognostic Value of Manual and Machine Measured QT in the Case Cohort Sample: Relative Rates (95% CI) of Highest Gender-Speciﬁc 10% Compared With the Other 90% of the QTc Distribution CHD CVD Death Incidence Total Mortality 514 ؍ n 445 ؍ n 178 ؍ Number of Events n Age-, gender- and coder-adjusted QTc NOVACODE 4.89 (3.26–7.35) 1.95 (1.36–2.79) 2.17 (1.56–3.01) QTc Manual 3.79 (2.44–5.88) 1.62 (1.12–2.36) 1.81 (1.28–2.55) Age-, coder-, gender-, and heart rate-adjusted QTc NOVACODE 2.24 (1.61–3.10) 1.48 (1.19–1.84) 1.49 (1.22–1.82) QTc Manual 2.54 (1.87–3.45) 1.45 (1.18–1.78) 1.52 (1.26–1.85)

*Cut points for prolongation in the sample of 900 subjects: QTc NOVACODE: 454 ms for men, 464 ms for women; QTc manual: 446 ms for men and 455 ms for women. CHD ϭ coronary heart disease; CI ϭ confidence interval; CVD ϭ cardiovascular death; QTc ϭ heart rate-corrected QT interval. and gender-adjusted model, the relative rates of CVD autonomic system imbalance and autonomic neuropathy mortality for QTc, QTindex, and QThodges were 4.37, (27), mutations of genes affecting cardiac ion channels 3.29, and 3.73, respectively. The prognostic value of QTc involved in cardiac repolarization (28), nonconducting scar was slightly better than QTindex or QThodges. tissue resulting from myocardial infarction (1), high glucose Certain drugs are known to lead to QTc prolongation, level (29,30), elevated insulin level (30–32), hypokalemia and may introduce bias. However, exclusion of subjects (32), obesity (33), and ventricular hypertrophy (16). All these using antihypertensive or cardioactive drugs did not change mechanisms seemed to contribute to QT prolongation in the the observed associations in our study, and the associations ARIC population. The QTc prolongation was associated were also present in subjects without signs of heart disease with the presence of ECG abnormalities, possibly resulting or chronic diseases. from small silent myocardial infarctions; QTc duration was This study is in line with previous reports on the also strongly related to all cardiovascular risk factors, includ- association of prolonged QTc with mortality risk (5–10) and ing fasting insulin. This has previously been observed in of CHD incidence (6,8). The finding of a gradually increas- Dutch elderly men (6) and in the nondiabetic subjects of the ing risk of CVD mortality with increasing QTc, and a Insulin Resistance and Atherosclerosis Study (IRAS) (31). nonlinear increase in the upper 10% of the population The presence of all these factors of the insulin resistance distribution was also observed in the Strong Heart study of syndrome is expected to lead to atherosclerosis, and, indeed, American Indians (10). We showed the QT-associated risk we confirmed the association between intima-media thick- may be even higher in the black population. This may be a ness and QTc, also reported in IRAS. Thus, the QTc may real phenomenon because there are known racial differences be viewed as a marker of subclinical atherosclerosis, which in the 12-lead ECG, and race-specific criteria for left indeed may contribute to the increased incidence of symp- ventricular hypertrophy have been proposed (11,24).In tomatic coronary heart disease associated with QTc prolon- healthy subjects without hypertension, diabetes, or his- gation. However, the simultaneous presence of other car- tory of cardiovascular disease in the ARIC study, black diovascular risk factors, ECG abnormalities, and intima- subjects had higher QRS voltage than white subjects, and media thickness did not completely explain the association QTc was shorter, especially after adjusting for other between QTc prolongation and either CHD incidence or cardiovascular risk factors (11). When we increased the CVD mortality. Gastadelli et al. (32) recently reported a cut point for prolongation for the white subjects in the direct effect of exogenous insulin on QTc, which was present study, the relative risks in white subjects in- mediated through lowering of serum potassium and an creased somewhat, but they were still considerably lower increase of noradrenaline levels. It was suggested that than in black subjects. Furthermore, racial differences insulin-stimulated cellular potassium uptake leads to have also been reported in the mechanism of CHD hyperpolarization of the cell membrane. In the present leading to mortality (25). The observed higher risk may study, adjustment for fasting insulin or potassium level contribute to the higher case fatality in the black popu- also did not explain the increased risk of QTc prolonga- lation (26). The interaction with race may also be a tion. But this may be only one of several possible chance finding because previous studies reported higher mechanisms of (acquired) suboptimal regulation of car- relative risks in white subjects with prolonged QTc than diac ion channels involved in ventricular repolarization. observed in the ARIC white population. In analogy with the findings in subjects with the long QT The QT interval reflects the time between the initial fast syndrome (28), this may directly be related to suscepti- depolarization of the ventricle and its repolarization. Several bility to both fatal and nonfatal arrhythmias. mechanisms have been reported to lead to QT prolongation: Some experimental work suggests subjects with QTc JACC Vol. 43, No. 4, 2004 Dekker et al. 571 February 18, 2004:565–71 QT and CHD Risk in the General Population prolongation may benefit from preventive treatment. In Correlates of QT prolongation in older adults (the Cardiovascular Health study). Am J Cardiol 1994;73:999–1002. obese subjects who lost weight, QT-interval duration nor- 14. McFarlane PW, Lawrie TDV. The normal electrocardiogram and malized (33). In addition, hypertensive rats treated with vectorcardiogram. In: McFarlane PW, Lawrie TDV, editors. Com- angiotensin-converting enzyme inhibitors showed a reduc- prehensive Electrocardiology: Theory and Practice in Health and Disease. New York, NY: Pergamon Press, 1989:407–57. tion of left ventricular mass and normalized QT (34). 15. The ARIC Investigators. The Atherosclerosis Risk in Communities In conclusion, apparently healthy men and women with (ARIC) study: design and objectives. Am J Epidemiol 1989;129:687– QT prolongation are at increased risk of incident CHD, and 702. 16. Rautaharju PM, Park LP, Chaitman BR, Rautaharju F, Zhang Z-M. especially of CVD mortality. Therefore, QTc is a cheap and The Novacode criteria for classification of ECG abnormalities and noninvasive tool to detect high-risk subjects who should their clinically significant progression and regression. J Electrocardiol receive special attention. 1998;31:157–87. 17. Siedel J, Hegele EO, Ziegenhorn J, Wahlefeld AW. Reagent for the enzymatic determination of total serum cholesterol with improved Reprint requests and correspondence: Dr. Jacqueline M. Dek- lipolytic efficiency. Clin Chem 1983;29:1075–80. ker, Institute for Research in Extramural Medicine, VU University 18. Nagele U, Hagele EO, Sauer G, et al. Reagent for the enzymatic Medical Center, Van der Boechorststraat 7, 1081 BT Amsterdam, determination of serum total triglycerides with improved lipolytic The Netherlands. E-mail: [email protected]. efficiency. J Clin Chem Clin Biochem 1984;22:165–74. 19. Warnick GR, Benderson JM, Albers JJ. Quantification of high-density lipoprotein subclasses after separation by dextran sulfate and MG2ϩ precipitation (abstr). Clin Chem 1982;28:1574. REFERENCES 20. Bond MG, Barnes R, Riley WA, et al. High resolution B-mode ultrasound scanning methods in the Atherosclerosis Risk In Commu- 1. Peters RW, Byington RP, Barker A, Yusuf S, for the BHAT Study nities study (ARIC). J Neuroimaging 1991;1:68–73. Group. Prognostic value of prolonged ventricular repolarization fol- 21. White AD, Folsom AR, Chambless LE, et al. Community surveil- lowing myocardial infarction: the BHAT experience. J Clin Epidemiol lance of coronary heart disease in the Atherosclerosis Risk In Com- 1990;43:167–72. munities (ARIC) study: methods and initial two years’ experience. 2. Algra A, Tijssen JPG, Roelandt JRTC, Pool J, Lubsen J. QT J Clin Epidemiol 1996;49:223–33. prolongation measured by standard 12-lead electrocardiography is an 22. Bazett HC. An analysis of time relations of the electrocardiogram. independent risk factor for sudden death due to cardiac arrest. Heart 1920;7:53–70. Circulation 1991;83:1888–94. 23. Schouten EG, Dekker JM, Kok FJ, et al. Risk ratio and rate ratio 3. Naas AAO, Davidson NC, Thompson C, et al. QT and QTc estimation in case-cohort designs: hypertension and cardiovascular dispersion are accurate predictors of cardiac death in newly diagnosed mortality. Stat Med 1993;12:1733–45. non-insulin dependent diabetes: cohort study. BMJ 1998;316:745–6. 24. Okin PM, Wright JT, Nieminen MS, et al. Ethnic differences in 4. Vincent GM, Timothy KW, Leppert M, Keating M. The spectrum of electrocardiographic criteria for left ventricular hypertrophy: the LIFE symptoms and QT intervals in carriers of the long QT syndrome: an study. Am J Hypertens 2002;15:663–71. update. N Engl J Med 1992;327:846–52. 25. Stein CM, Lang CC, Singh I, He HB, Wood AJJ. Increased vascular 5. Schouten EG, Dekker JM, Meppelink P, Kok FJ, Vandenbroucke JP, adrenergic vasoconstriction and decreased vasodilation in blacks: ad- Pool J. QT-interval prolongation predicts cardiovascular mortality in ditive mechanism leading to enhanced vascular reactivity. Hyperten- an apparently healthy population. Circulation 1991;84:1516–23. sion 2000;36:945. 6. Dekker JM, Schouten EG, Klootwijk P, Pool J, Kromhout D. 26. Williams JE, Massing M, Rosamond WD, Sorlie PD, Tyroler HA. Association between QT interval and coronary heart disease in Racial disparities in CHD mortality from 1968 to 1992 in the state middle-aged and elderly men: the Zutphen study. Circulation 1994; economic areas surrounding the ARIC study communities. Ann 90:779–85. Epidemiol 1999;9:472–80. 7. De Bruyne MC, Hoes AW, Kors JA, Hofman A, van Bemmel JH, 27. Zipes DP, Miyaki T. The autonomic nervous system and the heart: Grobbee DE. Prolonged QT interval predicts cardiac and all-cause basis for understanding interactions and effects on arrhythmias devel- mortality in the elderly: the Rotterdam study. Eur Heart J 1999;20: opment. In: Zipes DP, Jalife J, editors. Cardiac Electrophysiology: 278–84. From Cell to Bedside. Philadelphia, PA: WB Saunders, 1990:312–30. 8. Elming H, Holm E, Jun L, et al. The prognostic value of the QT 28. Roden DM, Spooner PM. Inherited long QT syndromes: a paradigm interval and QT interval dispersion in all-cause and cardiac mortality for understanding arrhythmogenesis. J Cardiovasc Electrophysiol and morbidity in a population of Danish citizens. Eur Heart J 1999;10:1664–83. 1998;19:1391–400. 29. Marfella R, Nappo F, De Angelis L, Siniscalchi M, Rossi F, Giugliano 9. Sharp DS, Masaki K, Burchfiel CM, Yano K, Schatz IJ. Prolonged D. The effect of acute hyperglycaemia on QTc duration in healthy QTc interval, impaired pulmonary function, and a very lean body mass men. Diabetologia 2000;43:571–5. jointly predict all-cause mortality in elderly men. Ann Epidemiol 30. Dekker JM, Feskens EJM, Schouten EG, Klootwijk P, Pool J, 1998;8:99–106. Kromhout D. The Zutphen Elderly study: QTc duration is associated 10. Okin PM, Devereux RB, Howard BV, Fabsitz RR, Lee ET, Welty with levels of insulin and glucose tolerance. Diabetes 1996;45:376–80. TK. Assessment of QT interval and QT dispersion for prediction of 31. Festa A, D’Agostino R, Rautaharju P, et al. Is QT interval a marker all-cause and cardiovascular mortality in American Indians: the Strong of subclinical atherosclerosis in non-diabetic subjects? The Insulin Heart study. Circulation 2000;101:61–6. Resistance Athersoclerosis Study (IRAS). Stroke 1999;30:1566–71. 11. Vitelli LL, Crow RS, Shahar E, Hutchinson RG, Rautaharju PM, 32. Gastaldelli A, Emdin M, Conforti F, Camastra S, Ferrannini E. Folsom AR, for the Atherosclerosis Risk in Communities (ARIC) Insulin prolongs the QTc interval in humans. Am J Physiol Regul Study Investigators. Electrocardiographic findings in a healthy biracial Integr Comp Physiol 2000;279:R2022–5. population. Am J Cardiol 1998;81:453–9. 33. Carella MJ, Mantz SL, Rovner DR, et al. Obesity, adiposity, and 12. McLaughlin NB, Campbell RW, Murray A. Accuracy of four auto- lengthening of the QT interval: improvement after weight loss. Int J matic QT measurement techniques in cardiac patients and healthy Obes Relat Metab Disord 1996;20:938–42. subjects. Heart 1996;76:422–6. 34. Bailard C, Mansier P, Ennezat PV, et al. Converting enzyme 13. Rautaharju PM, Manolio TA, Psaty BM, Borhani NO, Furberg CD, inhibition normalizes QT interval in spontaneously hypertensive rats. for the Cardiovascular Health Study Collaborative Research Group. Hypertension 2000;36:350–4.