A negative T‐wave in electrocardiogram at 50 years predicted lifetime mortality in a random population‐based cohort

Abstract Background Severe electrocardiographic (ECG) abnormalities in asymptomatic subjects correlate with cardiovascular risk. Hypothesis The role of minor ECG abnormalities is less well‐known. We evaluated the association between a negative T‐wave and mortality, as a possible marker for prognosis. Methods A prospective, population‐based cohort, examined at 50 years, and followed until death. Time to death (event rates) and predictive role of a negative T‐wave (Cox regression) were analyzed. Results Participants (n = 839) with a negative T‐wave (7.3%) had significantly higher blood pressure (BP) (mean systolic 157.9 mmHg vs 136.8 mmHg without negative T‐wave, P = <.0001). A negative T‐wave correlated with elevated risk (hazard ratio [HR] [95% CI] [confidence interval]) for all‐cause and cardiovascular (CV) death (1.59 (1.20‐2.11) P = .0012 vs 1.91 (1.34‐2.73) P = .0004). The association remained after excluding coexisting Q/QS patterns and ST‐junction/segment depression ECG abnormalities (1.66 [1.13‐2.44] P = .0098 for all‐cause vs 1.87 [1.13‐3.09] P = .015 for CV death). Death from other causes was not associated with a negative T‐wave. A major negative T‐wave carried higher risk than a minor (2.17 [1.25‐3.76] P = .0062 vs 1.78 [1.13‐2.79] P = .012) for CV death. Conclusion A negative T‐wave at 50 years, in asymptomatic individuals, carried an increased risk of all‐cause and CV death during lifetime follow‐up.


| INTRODUCTION
The electrocardiogram (ECG) is a powerful and easily available tool for detection of cardiac disorder in asymptomatic individuals, often preceding symptoms by considerable time. [1][2][3][4][5] Several studies have linked ECG abnormalities with CV events and prognosis in asymptomatic adults [6][7][8][9] predominantly for major ECG aberrations. [10][11][12] Minor ECG abnormalities are common from middle age, 13,14 and frequently seen in clinical practice in asymptomatic subjects. They are generally regarded as unspecific, not linked to specific myocardial disease. 13,[15][16][17][18] The prognostic value of minor ECG abnormalities is not fully known, but has a potential for preventive considerations. Here, we present a longitudinal population-based observational cohort study of 50-year old men, investigating the correlation between a negative T-wave in resting ECG and mortality during 48 years of follow-up.

| Study population
The "study of men born in 1913" is a longitudinal, prospective, population-based study, which recruited men born in 1913 from the city of Gothenburg (approximately 500 000 inhabitants at start of study) in western Sweden. From the population registry, a sample was drawn in 1963, consisting of all the men born in 1913 on a day of the month divisible by three (ie, the third, sixth, and ninth day of each month, and so forth). These criteria were fulfilled by 973 men born in 1913, of whom 855 (87.9%) participated in a health examination. Participants and nonparticipants have been previously described. 19,20 Informed consent was obtained at each examination, orally during the first examina- (1.6%) were lost to follow-up over time, primarily due to emigration.

| Outcome associated with a negative T-wave
The mortality associated with a negative T-wave in ECG was computed (

| Factors that may affect prediction of death by a negative T-wave
We tested the impact on mortality, all-cause and CV, of all clinical and demographic variables presented in our models, in the presence of a T A B L E 3 Mortality (all-cause, CV and other death) and follow-up times by T-wave negativity during the study. Unadjusted Cox proportional hazard models for prediction of time to death, CV death and other death by T-wave negativity and other concomitant ECG pathology in 50-yearold men Although cohorts studied at the time differ, ECG abnormalities were generally more prevalent, 2,5 than in recent observations. 27 The difference is likely to reflect improved CV health and healthcare at comparable ages over time, also reflected in the remarkably high overall prevalence in hypertension and smoking in our study (Table 1). Similar observations are not expected in the present day, due to improved detection, advances in pharmacologic treatment, and lower cutoff levels for therapy in treatment guidelines. 28,29 Simultaneous other ECG abnormalities were much more common when a negative T-wave was present, but limited to ST-segment/junction depression, most frequently, and also Q/QS pattern. The former would be expected since nondiagnostic ST/T changes are the most common ECG changes overall. 30 Also conditions causing ECG changes often concomitantly affect ST-junction/segment levels and T-wave axis. Thus, we found ST-segment/junction depression almost 14 times more common when a negative T-wave was present. Regarding Q/QS pattern, we cannot completely exclude prior silent MI in our cohort, although all confirmed MIs were excluded. Also, a minor negative T-wave was much more frequent than a major, making up more than 60% of cases, similar to previously reported. 13 4.2 | A negative T-wave as an independent lifetime predictor for risk for death and cardiovascular death The higher event rate for all-cause and CV mortality associated with a negative T-wave was consistent with the presence of more hypertension in this group, and of ECG abnormalities, in general, as markers of compromized CV health. [24][25][26] The association of a negative T-wave with CV death, but not non-CV death, strengthens its role as a lifetime marker for increased CV risk.
Consistent with the above, the HR for death in the presence of a negative T-wave was also significantly elevated both for all-cause and CV death, almost doubled for the latter. The increased risk persisted after exclusion of concomitant Q/QS pattern and concomitant STjunction/segment depression. This implies that a negative T-wave marks an elevated risk of death, particularly CV death, and the risk persists, also in the absence of other ECG abnormalities. This is an important finding since the significance of isolated negative T-waves in an asymptomatic individual has been unclear.
Furthermore, although both major and minor T-wave inversions carried significantly increased risk for all-cause and CV death, the risk was most expressed for a major negative T-wave. Although our focus was on analysis of isolated T-wave inversion, we propose a similar rationale as for general ECG findings: Major ECG abnormalities are more solidly linked to CV disease and prognosis in asymptomatic individuals, than minor. [1][2][3][4]11,14 The physiology behind greater aberration from normality is likely to reflect more extensive CV pathology.
Therefore, it is reasonable that a major negative T-wave carries a higher risk than a minor. To our knowledge, there is no prior documented observation of this greater prognostic risk for isolated major T-wave abnormalities.
Despite adjustment for numerous risk factors, a negative T-wave remained an independent lifetime predictor for risk for all-cause and CV death for lower cholesterol levels. For all-cause death, the increased risk included also lesser smoking and lower hematocrit. All three conditions per se are associated with lower overall CV risk. Our results indicate that a negative T-wave in absence of other evident CV risk factors may carry a distinctive message of enhanced risk, not to be ignored. When a negative T-wave is present in middle age, albeit an otherwise favorable risk factor profile, the enhanced risk appears comparatively even larger, and must always be regarded as a marker of CV risk.
However, this study has some limitations to consider. First, the cohort was collected in the 1960s, which makes the results of limited contemporary applicability. Moreover, only men in a limited geographic area were included and the findings cannot be generalized to women and other populations. Second, the size of the cohort was moderate. However, the long-term follow-up still gave statistical strength.

| CONCLUSION
A negative T-wave in ECG registered at 50 years of age in asymptomatic individuals from a randomized observational cohort carried an increased risk of all-cause and CV death during lifetime follow-up. The