Relationship of QT interval duration with carotid intima media thickness in a clinically healthy population undergoing cardiovascular risk screening


Dr Bernhard Strohmer, Paracelsus Private Medical University, Salzburger Landeskliniken, Department of Cardiology, Muellner Hauptstrasse 48, A-5020 Salzburg, Austria.
(fax: ++43 662 4482 3486; e-mail:


Objectives.  To investigate the relationship between cardiac repolarization (QT interval duration) and intima media thickness (IMT) of the carotid arteries as surrogate measures of subclinical atherosclerosis.

Design.  Prospective study with consecutive subjects enrolled in the SAPHIR program (Salzburg Atherosclerosis Prevention Program in Subjects at High Individual Risk).

Setting.  The analysis of the material was performed at the departments of medicine and neurology of a university hospital.

Subjects.  The study cohort comprises a population-based sample of 1199 clinically healthy subjects (851 men and 348 women; age 39–66 years). Exclusion criteria were cardiovascular disease, diabetes, atrial fibrillation, bundle branch block and use of medication affecting QT interval duration.

Main outcome measures.  IMT of common (CCA) and internal carotid arteries (ICA) was measured by B-mode ultrasound. QT interval duration was determined in the resting 12-lead electrocardiogram by an automatic analysis program. The QT intervals were corrected for heart rate with five standard equations (QTc-Bazett, -Fridericia, -Framingham, -Hodges and -Rautaharju) and tested for their relationship with carotid IMT after adjustment for clinical and metabolic variables.

Results.  Females had higher heart rates than males (64 ± 10 b min−1 vs. 60 ± 9 b min−1, P < 0.0005), with longer mean QT (410 ± 28 ms vs. 404 ± 28 ms, P = 0.003) and QTc intervals in all correction formulae (P < 0.0005). Significant correlations between QT/QTc and ICA IMT (r =0.14–0.16) were found in males. In the general linear model the association between QTc (except for Bazett) and ICA IMT remained significant after adjusting for age, BMI and further cardiovascular risk factors. In females the crude correlations between QT/QTc and ICA IMT were lower than those with CCA IMT. Only the correlation between uncorrected QT and CCA IMT (r = 0.15, P = 0.006) remained significant after adjustment for covariates.

Conclusions.  The results of the present study demonstrate that QT and QTc prolongation are in part associated with IMT of carotid arteries, which is an established risk marker of subclinical atherosclerosis. In men the data support the hypothesis of an association between QTc and ICA IMT. In women a statistically significant relationship was found between the uncorrected QT interval and CCA IMT. These findings suggest that differences in carotid IMT and ventricular repolarization between genders might be related to hormonal and nonhormonal effects.


Prolonged QT interval and increased carotid artery intima media thickness (IMT) have both been associated with cardiovascular morbidity and mortality in epidemiological studies. However, the association between the duration of the QT interval and the risk of total and cardiovascular mortality remains controversial [1].

A bulk of published data are available linking QT interval duration to an increased risk of cardiovascular morbidity [2–4]. The QT interval has also been related to various components of the insulin resistance syndrome, such as body mass index (BMI), obesity, blood pressure and fasting insulin [5–7]. The association of the heart rate (HR)-corrected QT interval duration with subclinical atherosclerotic disease, determined by ultrasonographic measurement of carotid IMT is of particular clinical interest for early detection and prevention of cardiovascular diseases [8]. Several studies demonstrated an association between carotid artery IMT and unfavourable cardiovascular risk factor levels [9–13], prevalent and incident cardiovascular disease [14–17], and atherosclerosis in other parts of the arterial system [18, 19]. Thus, carotid artery IMT is considered to be a reliable indicator of generalized atherosclerosis [20]. Moreover, subclinical disease of the carotid wall has been reported to be an independent risk factor for cardiovascular mortality [21].

We hypothesized that the interaction between QT interval duration – with and without correction for HR by different formulae – and carotid artery IMT might be used as early marker for cardiovascular risk stratification in apparently healthy subjects.

Materials and methods


A total of 1799 subjects were participants of the Salzburg Atherosclerosis Prevention Program in Subjects at High Individual Risk (SAPHIR). This population-based prospective study was conducted to investigate the role of various genetic and metabolic factors for progression of atherosclerotic vascular disease and to initiate appropriate interventions in subjects at high cardiovascular risk. All individuals were Caucasians and the population was homogeneous with regard to ethnic background. The local ethics committee approved the study. All subjects gave written, informed consent before entering the programme. A total of 1199 subjects of the SAPHIR programme were included in the present study after exclusion of patients with conditions which are known to prolong QT interval duration: history or presence of cardiovascular, cerebrovascular or peripheral arterial diseases or interventions, diabetes (patients on hypoglycaemic medication or fasting plasma glucose concentrations ≥126 mg dL−1), left or right bundle branch block, significant Q waves on surface electrocardiogram (ECG), atrial fibrillation and use of medication affecting QT interval duration (i.e. beta-blocker, digitalis, anti-dysrhythmic or antidepressant medication). The study cohort consisted of 851 males aged between 39 and 66 years and 348 females aged between 39 and 65 years.

Cardiovascular risk indicators

At baseline, all study participants were subjected to a thorough screening programme, which included assessment of a detailed personal and family history, physical examination, determination of anthropomorphic parameters, and measurement of various biochemical parameters. Venous blood samples were collected in the fasting state for determination of total cholesterol, triglycerides, HDL cholesterol, LDL cholesterol, plasma glucose, fasting insulin and plasminogen activator inhibitor type 1 (PAI-1). These laboratory analyses were determined using commercially available kits. Insulin sensitivity (k-ITT) was estimated using the short insulin tolerance test as described previously. Subjects with a systolic blood pressure ≥140 mmHg, a diastolic blood pressure ≥90 mmHg, and/or on antihypertensive medication were regarded as hypertensive. BMI was defined as weight divided by the square of the height (kg m−2). Smoking status was categorized in recent smoker or nonsmoker. Physical activity was defined as ‘yes’ if the subject's work involved heavy labour or if the subject was involved in some physical fitness training for 2 h or more per week.

Assessment of carotid atherosclerosis and cardiac repolarization

The IMT (in mm) of common (CCA) and internal (ICA) carotid arteries was measured by high-resolution B-mode ultrasound (HDI 3000 CV, ATL, Munich, Germany) according to a standard protocol [22]. IMT measurements were available in 1124 subjects, 785 males and 339 females. HR (b min−1) and QT intervals (ms) were determined automatically from resting ECGs at the same diurnal period by the GE Marquette 12SLTM ECG Analysis Program (GE Medical Systems, Menomonee Falls, WI, USA). QT interval was corrected for HR [RR interval (s) =60/HR] by calculating QTc (ms) according to five previously published formulae [23–27]: the log-linear equations Bazett [QTc = QT/RR1/2] and Fridericia [QTc = QT/RR1/3], the linear regression equations Framingham [QTc = QT + 154(1 − RR)] and Hodges [QTc = QT + 1.75(HR − 60)] and the equation QT60 = QT + [410–656/(1 + HR/100)], proposed by Rautaharju.

Statistical analysis

Statistical analyses were performed using the SPSS 11.0 (Statistical Package for the Social Sciences, Inc., Chicago, IL, USA) software system. The statistical methods included Pearson and Spearman rank correlations, t-tests, simple linear regression and general linear model analysis. The dependent variables were the baseline QT intervals as well as the HR-corrected QT intervals, here referred to as QTc-Bazett, QTc-Fridericia, QTc-Framingham, QTc-Hodges and QTc-Rautaharju. The independent variables were CCA IMT and ICA IMT respectively. We examined the crude associations between these variables as well as the associations between them after adjusting for various covariates/confounders. Analysis was carried out separately for male and female subjects. P-values ≤0.05 are considered as significant, P-values between 0.05 and 0.10 are referred to as weakly significant.


The descriptive statistics for the dependent and independent variables are given in Table 1. Females had significantly higher HRs than males (64 ± 10 b min−1 vs. 60 ± 9 b min−1, P < 0.0005), with longer mean QT (410 ± 28 ms vs. 404 ± 28 ms, P = 0.003) and QTc intervals as calculated by the five different correction formulae (P < 0.0005). As far as the IMT was concerned, we observed site-specific differences between the genders. CCA IMT was slightly higher in females than in males, whereas for the ICA a higher mean IMT was detected in males compared with females (P < 0.04 respectively).

Table 1.  Baseline clinical characteristics of male and female subjects including QT/QTc intervals and carotid intima media thickness (IMT)
 Males (n = 851)Females (n = 348)P-value
Mean ± SDRangeMean ± SDRange
  1. BMI, body mass index; CCA, common carotid artery; ICA, internal carotid artery.

Age (years)48.8 ± 5.439–6655.6 ± 4.239–65<0.0005
BMI (kg m−2)26.7 ± 3.616.5–43.426.4 ± 4.817.2–41.8 n.s.
Hypertension (%)53.444.00.004
Physical activity (%)26.819.50.008
Heart rate (b min−1)60.2 ± 9.434–11763.7 ± 9.641–106<0.0005
QT (ms)404.2 ± 28.4322–502409.5 ± 28.2322–5100.003
QTc-Bazett (ms)401.9 ± 22.2307–476419.1 ± 22.8352–479<0.0005
QTc-Fridericia (ms)402.4 ± 19.5313–473415.6 ± 20.1357–487<0.0005
QTc-Framingham (ms)400.9 ± 20.0307–473415.2 ± 20.0354–488<0.0005
QTc-Hodges (ms)404.4 ± 20.4314–476415.9 ± 20.5359–496<0.0005
QT60-Rautaharju (ms)403.2 ± 20.0308–473417.3 ± 20.3355–489<0.0005
IMT-CCA (mm)0.753 ± 0.1210.48–1.350.760 ± 0.1220.48–1.27<0.039
IMT-ICA (mm)0.822 ± 0.1280.50–1.520.815 ± 0.1230.46–1.38<0.036

Crude analysis

The crude associations between QT/QTc and IMT of CCA and ICA were measured by correlation coefficients (Table 2) and, equivalently, by examining the coefficient B in the simple linear regression model (data not shown). In males (Table 2), all correlations between QT/QTc and ICA IMT as well as all correlations between QTc and CCA IMT were significant. The correlations between QT/QTc and ICA IMT were higher than those between QT/QTc and CCA IMT. In females, the opposite is the case as the correlations between QT/QTc and CCA IMT were higher than those between QT/QTc and ICA IMT. However, in females none of the correlations with ICA IMT reached statistical significance. For CCA IMT, only the correlations with the uncorrected QT interval and with QTc-Hodges were significant. In females, the correlations between CCA (ICA) IMT and the uncorrected QT interval were higher than those with HR-corrected QT intervals, whereas in males the correlations with HR-corrected QT intervals were higher than those with uncorrected QT intervals.

Table 2.  Crude correlations (rho) between QT/QTc intervals and carotid intima media thickness (IMT)
 Males (n = 785)Females (n = 339)
  1. CCA, common carotid artery; ICA, internal carotid artery.


Adjusted analysis

The associations between QT/QTc and IMT of CCA and ICA were analysed taking potential confounders of the QT interval as well as of the IMT into account. The association was examined after adjusting for age and BMI using the general linear model analysis. Subsequently, a further pool of clinically important covariates was considered, which included the following variables in males: CCA or ICA IMT, age, BMI, physical activity, hypertension, insulin resistance (k-ITT), LDL cholesterol and HDL cholesterol. In females, the variables CCA or ICA IMT, age, BMI, physical activity, hypertension, and PAI-1 were determined by a combination of statistical procedures, such as stepwise regression and clinical considerations. The descriptive statistics of the covariates are given in Table 1. Correlations of these variables with QT/QTc and CCA IMT as well as ICA IMT are outlined for both sexes in Table 3.

Table 3.  Correlation analysis (Pearson's rho) between QT/QTc intervals and important clinical and biochemical covariates
AgeBMIPhysical activityHypertensionk-ITTLDLHDLAgeBMIPhysical activityHypertensionPAI-1
  1. +P ≤ 0.10; *P ≤ 0.05; **P ≤ 0.01.

  2. Biochemical parameters (mean ± SD) in males: k-ITT (% min−1) 4.23 ± 1.35, LDL cholesterol (mg %) 146.08 ± 36.32 and HDL cholesterol (mg %) 55.80 ± 13.36; in females: plasminogen activator inhibitor type 1, PAI-1 (ng mL−1) 2.12 ± 1.34.


The regression coefficients B and their standard errors (SE) are given for CCA and ICA after adjustment for age, BMI and further relevant risk factors in Table 4. The coefficient B can be interpreted as the mean expected increase in the dependent variable QT/QTc corresponding to an increase of 1 mm in the respective IMT parameter and all other covariates remain unchanged. A statistically significant association was found in males between corrected QT intervals (except for QTc-Bazett) and ICA IMT (Fig. 1) and in females between uncorrected QT interval and CCA IMT (Fig. 2).

Table 4.  Linear model analysis for QT/QTc intervals and intima media thickness (IMT) of common and internal carotid artery
 Common carotid artery (CCA)Internal carotid artery (ICA)
  1. Coefficient B of CCA, respectively ICA gives the mean expected rise in the dependent variable QT/QTc when the IMT variable increases by 1 mm and all other covariates remain unchanged. SE =the standard error of B. All data are adjusted for age, body mass index and further risk factors, i.e. in males: physical activity, hypertension, insulin resistance (k-ITT), LDL cholesterol, HDL cholesterol; in females: physical activity, hypertension, PAI-1.

QT13.4 (10.2)n.s.39.8 (13.9)0.00416.5 (9.1)0.07222.5 (13.2)0.090
QTc-Bazett5.9 (8.1)n.s.−3.7 (11.3)n.s.13.9 (7.2)0.054−15.2 (10.7)n.s.
QTc-Fridericia8.5 (7.1)n.s.11.2 (10.0)n.s.14.8 (6.4)0.021−2.5 (9.5)n.s.
QTc-Framingham8.3 (7.4)n.s.9.6 (10.0)n.s.14.7 (6.6)0.026−4.2 (9.4)n.s.
QTc-Hodges9.7 (7.5)n.s.15.6 (10.1)n.s.14.7 (6.7)0.0282.8 (9.6)n.s.
QT60-Rautaharju8.3 (7.3)n.s.7.0 (10.1)n.s.14.5 (6.6)0.028−5.0 (9.5)n.s.
Figure 1.

Graph depicting the mean QTc-Fridericia intervals plotted as a function of intima media thickness (IMT) of the internal carotid arteries (ICA). In male subjects with an ICA IMT up to roughly 0.95 mm, the mean QTc-Fridericia did not differ very much resulting in a rather flat plot. However, the significant association between QTc-Fridericia and ICA IMT was related to a steep increase of both the ICA IMT >1 mm and the mean QTc intervals.

Figure 2.

This plot shows the relationship between the mean values of the uncorrected QT intervals and intima media thickness (IMT) of the common carotid arteries (CCA) in females. A highly significant association exists between these two variables. The graph depicts the correlations for unadjusted and adjusted data separately (see text for further details).

Although the associations between QTc and ICA IMT are statistically significant in males, the strength of these associations is rather weak, as can be appreciated by the effect size (η2) of 1.9–2.6% in the crude, and 0.5–0.8% in the adjusted analysis. The effect size η2 can be interpreted as the proportion (here in %) of the variance of the dependent variable, which can be explained by CCA IMT or ICA IMT after adjusting for the other explanatory variables in the linear model. In females, a strong association was noted between the uncorrected QT interval and CCA IMT (P = 0.004), with an increase in the effect size from 2.2% in the crude to 2.7% in the adjusted analysis. No significant relationship existed at all between QT/QTc and ICA IMT in females.

Further analysis was performed dichotomizing CCA IMT at 0.75 mm, which has been determined as the median of CCA IMT for both sexes. With this categorization above and below 0.75 mm statistically significant differences were found in males between the adjusted mean QT/QTc values (P < 0.05 and P ≤ 0.03 for QT and QTc respectively), with exception of QTc-Bazett (Table 5). The mean differences in QT/QTc levels varied between 2.9 and 4.0 ms in males. However, in females only the difference in the mean uncorrected QT interval remained statistically significant after adjustment (9.0 ms; P = 0.004) when CCA IMT was dichotomized at 0.75 mm.

Table 5.  The mean values of QT/QTc intervals and mean differences of QT/QTc intervals for intima media thickness (IMT) of common carotid arteries (CCA) dichotomized at 0.75 mm and for internal carotid arteries (ICA) dichotomized at 1 mm
CCA ≤0.75 mm (n = 341)CCA >0.75 mm (n = 316)Diff. (SE)P-valueCCA ≤0.75 mm (n = 173)CCA >0.75 mm (n = 128)Diff. (SE)P-value
  1. Data (expressed in ms) are adjusted for age, body mass index and further risk factors, i.e. in males: physical activity, hypertension, insulin resistance (K-ITT), LDL cholesterol, HDL cholesterol; in females: physical activity, hypertension, PAI-1. Diff. = mean differences of QT/QTc intervals. Numbers in brackets are SD and SE.

QT402.92 (26.76)406.93 (27.19)4.01 (2.03)0.049405.62 (26.36)414.61 (27.50)8.98 (3.13)0.004
QTc-Bazett401.19 (22.09)404.10 (20.28)2.91 (1.61)0.071419.55 (21.93)419.54 (21.83)−0.01 (2.55)n.s.
QTc-Fridericia401.49 (19.19)404.77 (18.62)3.28 (1.43)0.022414.60 (18.53)417.68 (20.22)3.08 (2.25)n.s.
QTc-Framingham399.91 (19.81)403.32 (18.90)3.41 (1.47)0.020414.28 (18.34)417.09 (20.41)2.81 (2.24)n.s.
QTc-Hodges403.49 (20.06)406.84 (19.31)3.35 (1.49)0.025414.43 (18.52)418.32 (20.85)3.90 (2.28)0.088
QT60-Rautaharju402.33 (19.96)405.54 (18.63)3.21 (1.46)0.028416.70 (18.75)418.87 (20.36)2.17 (2.27)n.s.
 ICA ≤1 mm (n = 645)ICA >1 mm (n = 66)  ICA ≤1 mm (n = 281)ICA >1 mm (n = 20)  
QT403.73 (26.62)414.17 (29.07)10.44 (3.47)0.003408.73 (27.36)420.11 (21.35)11.39 (6.25)0.070
QTc-Bazett401.94 (21.38)407.84 (20.47)5.90 (2.75)0.032419.67 (22.14)417.98 (17.81)−1.69 (5.07)n.s.
QTc-Fridericia402.27 (18.85)409.66 (19.38)7.40 (2.44)0.003415.75 (19.70)418.37 (12.12)2.62 (2.95)n.s.
QTc-Framingham400.78 (19.41)407.79 (19.23)7.01 (2.51)0.005415.42 (19.61)416.58 (13.89)1.16 (4.46)n.s.
QTc-Hodges404.22 (19.59)412.46 (20.42)8.24 (2.54)0.001415.79 (20.01)420.63 (11.65)4.84 (2.86)n.s.
QT60-Rautaharju403.09 (19.37)410.33 (19.05)7.24 (2.50)0.004417.52 (19.86)419.36 (12.39)1.84 (4.51)n.s.

According to the definition of a plaque formation ICA IMT was dichotomized at 1 mm. In male subjects, the adjusted differences of mean QT/QTc values were statistically significant [P = 0.003 and P ≤ 0.005 for QT and QTc (except Bazett) respectively]. The mean differences in QT/QTc levels ranged between 5.9 and 10.4 ms (Table 5). In contrast to that the mean differences of QT/QTc intervals did not reach statistical significance in female subjects when ICA IMT was dichotomized at 1 mm.


The purpose of the present study was to relate a computerized measurement of the QT interval to carotid artery IMT measured at the time of risk assessment. This approach was selected to examine the clinical utility of the QT interval as surrogate marker for silent atherosclerosis in a screening programme. There is evidence that the thickening of the arterial wall observable with B-mode ultrasonography represents one initial step of preclinical atherosclerotic disease. Early detection of subclinical disease may increase our ability to predict the subsequent cardiovascular risk and lead to optimal disease prevention and treatment strategies [28].

Festa et al. were the first to report a positive association of HR-corrected QT interval with subclinical carotid atherosclerosis in nondiabetic subjects without clinically overt coronary artery disease [8]. However, in that study no significant interaction of gender existed on the association of CCA IMT with HR-corrected QT interval. Further, QT interval was related only to CCA IMT but not to ICA IMT.

The findings in the present study are not in equivocal accordance with the observation of a linear association between QTc and carotid IMT. Our results rather indicate that the relationship between the QT/QTc intervals and carotid artery IMT is complex and different for both genders. The significant but low correlation coefficients might be explained by the fact that the recent study is an evaluation of the relationships between two surrogate measures for preclinical vascular disease, namely the QT interval as marker of coronary atherosclerosis and the IMT as marker of cerebral atherosclerosis. From a clinical perspective it is important to realize that the levels of carotid IMT in the present study are indicative for subclinical atherosclerotis and do not reflect the presence of an arterial stenosis in the CCA or ICA.

Subjects with prevalent type 2 diabetes were excluded from the recent analysis. Manifest diabetes has been associated with both increased carotid IMT and prolonged QT interval duration [2, 6, 12]. As expected, adjustment for cardiovascular risk factors, such as hypertension, obesity and hyperlipidaemia, reduced the magnitude of associations between QT and carotid IMT considerably (Tables 2 and 4). However, as our interest was to assess whether QT/QTc interval is independently associated with carotid IMT, we wanted to adjust for significant factors promoting atherosclerosis (Table 3). There is growing evidence that risk factors for increased IMT might be different in the CCA and ICA.

Several aspects of the present study need to be addressed specifically. The relationship between QT/QTc and IMT was remarkably different for different arterial segments. Due to the site-specific nature of carotid atherosclerosis, IMT of ICA and CCA were analysed separately. IMT of ICA may be a better marker of focal atherosclerosis than IMT of CCA, because plaques have a tendency to form near areas of nonlaminar flow [29]. Similarly, it is conceivable that the level of coronary atherosclerosis may also be different in view of the very specific type of coronary flow. Prospective studies have found ICA IMT to be a better predictor of cardiac events than CCA IMT [16]. Gender-related differences in carotid artery IMT are known to occur physiologically in adults. Delay in arterial thickening in women could be related to a protective effect of oestrogen. Experimental and epidemiological data suggest that oestrogen may have beneficial effects on endothelial function and atherosclerosis, raising the possibility of sex differences in arterial remodelling [30].

The results of the present study may be provocative with respect to the necessity of HR correction of QT interval, which is deeply rooted in clinical practice. Amongst the many physiological and pathological factors that contribute to the QT interval, the HR and the autonomic tone play a major role. Thus, many formulae have been suggested previously to correct the QT interval for the HR, but the problem in terms of a truly universal correction formula has not been satisfactorily resolved. Due to its simplicity Bazett's formula is the most widely used method, although it leads to overcorrection of QTc intervals at high and to undercorrection at low HRs. In order to achieve a more accurate estimate of HR-corrected QT interval we included four additional correction formulae based on different mathematical concepts. However, one has to be aware that there is no ‘physiologically’ correct relationship between QT and RR intervals that is common to all healthy subjects and perhaps also to patients with various cardiac diseases. In our study, correction of QT interval for HR turned out to be useful to detect associations with carotid artery IMT in males, but not in females. Consequently, the uncorrected QT interval has to be considered as important as the corrected QT interval in this type of analysis. Of note, Bazett formula performed worst, not only with respect to rate correction of QT interval, but also in the linear model analysis for QTc on IMT in males.


Due to the epidemiological approach on which the analysis of the data is based, it is not possible to state whether an association of carotid IMT and QT interval exists in a single individual. Another limitation is the smaller sample size of females compared with that of males. Thus, gender differences as shown might also be explained by differences in statistical power to detect associations. Finally, no precise or invasive evaluation of real atherosclerosis has been performed that may help to determine which surrogate measure, QT interval or carotid IMT, is the most useful noninvasive test.


The QT interval on standard surface ECG – reflecting the total duration of ventricular myocardial depolarization and repolarization – tends to be associated with carotid artery IMT and may be considered an indicator of subclinical atherosclerosis. In males, only the corrected QT intervals (except for QTc-Bazett) showed significant associations with ICA IMT. In females, the opposite was the case and the uncorrected QT interval was related to CCA IMT. These findings suggest, that differences in carotid IMT and ventricular repolarization amongst genders might be related to hormonal but also nonhormonal effects. Due to its wide availability and ease of evaluation QT interval measurement on surface ECG appears to provide a promising method for identifying high-risk individuals, who could potentially benefit from intensive cardiovascular risk factor management in populations at large.

Conflict of interest statement

No conflict of interest was declared.