Are daylight saving time transitions associated with changes in myocardial infarction incidence? Results from the German MONICA/KORA Myocardial Infarction Registry

More than 1.5 billion people in over 70 countries worldwide are subject of transitions to daylight saving time (DST) since some decades. In Germany, DST was introduced in 1980. It begins on the last Sunday in March. The end was on the last Sunday in September from 1980–1995, and is one month later since 1996.

Very little is known about the effects of the disruption of circadian rhythms caused by DST shifts and the study results are overall inconsistent [1–5]. Studies on DST time shift, however, provide a unique possibility to investigate these effects in a “natural experiment” without confounding from individual characteristics since everyone is exposed at a given time point.

Some published studies have addressed the impact of transitions to DST on acute myocardial infarction (AMI) incidence [6–9]. Data from the Swedish AMI registry showed a significantly higher AMI incidence for the first 3 workdays and the whole week after the spring transition. In contrast, after the autumn transition only the Monday was affected significantly and showed a reduced AMI incidence [6]. However, this study only investigated age and sex as factors that might influence the association between DST time shifts and AMI incidence. A further study from the Swedish authors examined subgroups of AMI cases and found a higher risk of AMI for the spring transition in individuals taking cardiac medication or having low blood lipids, and a lower risk of AMI for the autumn transition in persons with hyperlipidemia, and persons taking statins or calcium-channel blockers, but these differences were not statistically significant [7]. Moreover, this study sample was restricted to hospitalized AMI cases and used the time of hospital admission instead of onset of AMI symptoms, which may have influenced the results. Potential meteorological confounders such as air temperature, relative humidity, and barometric pressure were not analyzed.

A recent study from Croatia on 2,412 hospitalized AMI survivors confirmed the significant increase of AMI incidence for the first 4 workdays after spring transition with a particular excess on Monday [8]. In contrast to the Swedish results, the authors reported a significant increase after the autumn transition for the first four workdays with a peak on Tuesday and Thursday. However, this study did not include fatal AMI cases and consider meteorological variables as potential confounders.

A smaller study performed by Jiddou et al. [9] on 935 hospitalized U.S. AMI survivors finally found a significantly increased AMI incidence for the first day (Sunday) after spring transition but no significant effects in terms of the autumn shift. Limitations of this study refer to its small sample size, an exclusion of fatal AMI cases, the use of the time of hospital admission as AMI onset, and the lacking consideration of meteorological confounders.

In addition, it remains unclear to what extent findings can be generalized across countries with different geographical location, since latitude and longitude influence light–dark cycles and circadian rhythms of humans. Thus, the objective of this study was to examine the association of DST transitions with AMI incidence using data of coronary deaths and non-fatal AMIs recorded in the MONICA/KORA Myocardial Infarction Registry, located in Southern Germany.

Our study is the first one which investigated the association of DST shifts and AMI incidence in Germany and considered meteorological variables as potential confounders. In contrast to previous studies, we found no significant changes of AMI incidence in the entire sample for the first 3 days and 1 week after the spring transition to DST and the autumn transition from DST [6–9]. Analyses for each weekday separately yielded increased risks for Mondays and Tuesdays after the spring transition and reduced risks for Mondays after the autumn transition which are consistent with the Swedish results [6]. Although risk ratios in our study were greater compared with the estimates reported by Janszky et al. [6], they are lacking statistical significance due to smaller sample size.

Investigators of previous studies commonly explained the adverse associations of transitions to and from DST with AMI incidence by the disruption of the human circadian system, which may be associated with sleep deprivation particularly in the spring transition. Sleep deprivation itself negatively affects the cardiovascular system by increase of the sympathetic activity and proinflammatory biomarkers [15, 16]. The ability to adapt to DST may also depend on the chronotype. Chronotype is an attribute reflecting at what time of the day a persons’ physical functions, including sleeping, are active, change or reach a certain level. People can be divided into early (morning) and late (evening) chronotypes. Some studies reported that late chronotypes have more problems in adjusting to the spring transition to DST [17, 18], whereas the fall transition was more disturbing for early chronotypes [18]. In contrast, a recently published study found no differential adjustment of the chronotypes to DST transitions [19]. Chronotypes have been shown to depend not only on genetic factors and age [20, 21]. Studies indicated that the distribution of chronotypes among a population varies according to geographical location [22]. Latitude and longitude are key factors that influence how light–dark cycle interacts with the circadian rhythm of humans. It seems possible, that the different geographic locations of the studies which investigated the association of AMI incidence and DST shifts, namely Sweden, USA, and Croatia, may contribute to the difference of the results compared with our data from South Germany. To give an example, Sweden lies between latitudes 55° and 70°N whereas Augsburg is located at the 48°N latitude. On spring DST transition in 2010, sunrise was 30 min earlier in Stockholm than in Augsburg. Thus, individuals with late chronotypes in Sweden may experience a greater reduction in sleep duration than in Augsburg, leading to an excess of AMI cases.

Although no overall significant effects were found in our study, subgroup analyses revealed a significantly higher AMI incidence 3 days after the spring transition for persons with ACE inhibitor medication prior to infarction. Notably, the 1.489-fold increased risk was the highest estimate among all subgroup analyses. Janszky et al. [7] also found an increased risk for persons with prior ACE inhibitors (incidence ratio 1.141) but this result failed to be statistically significant. However, in contrast to Janszky et al. [7], in our study beta-blockers and calcium channel blockers tended to be associated with a lower risk for the spring transition. Thus, our findings do not fully support the hypothesis that prior cardiac medication may reflect more-progressed coronary atherosclerosis which increases the persons’ vulnerability to external AMI triggers such as time shifts.

In contrast to Janszky et al. [6, 7] but in line with Culic [8] we found men to be more prone to the spring shift. This may be explained by the fact that men are more likely to be later chronotypes than women also in later adulthood [21, 23]. The reasons for the conflicting study results are unclear. One may speculate that differences in study designs have contributed to these findings. In addition, as stated above, geographical location may be an overall factor that affects the association of DST transitions and AMI incidence.

In terms of the autumn DST transition, we found that patients with previous infarction had a higher risk to have a re-infarction in the week following autumn shift from DST compared with patients without prior AMI. This finding may also be related with the more-progressed coronary atherosclerosis and the possibility that minor disruptions of the circadian system, even if they may be associated with a longer sleep duration rather than a sleep deprivation, serve as a trigger of an AMI event.

The strengths of this study include the population-based sample of consecutive cases with coronary death and survivors with validated AMI, inclusion of patients in a defined area and according to defined criteria, and the standardized collection of risk factors, treatments and mortality data. Contrary to most prior investigations which were restricted to hospitalized patients with AMI [7–9], we have also considered coronary deaths and used the time of symptom onset instead of time of hospital admission as indicator of AMI onset [7, 9]. A major strength is the adjustment of the regression models for air temperature, relative humidity, and barometric pressure, which were reported to trigger AMI events [13]. Moreover, in the subgroup of AMI survivors, we were able to consider a number of characteristics which might influence the effect of DST on AMI incidence.

Our study shares one major limitation with previous studies on the same subject, namely the lack of data regarding sleep duration and quality, and chronotype. Besides weather data, no other potential triggers such as physical exertion, emotional states, or distress could be considered. Most of the subgroup analyses could only be performed for persons who were hospitalized and lived long enough to be interviewed. Thus, the results of the subgroup analyses can only be applied to AMI survivors. The statistical power of those analyses was limited and we cannot exclude that the significant findings may also be the result of type I error associated with multiple testing. In addition, the excess model for single weekdays has limited statistical power. Finally, the study does not include patients older than 74 years.

In conclusion, our study showed no significant overall change of AMI risk associated with shifts to and from DST among the population in Southern Germany. However, the results indicate that specific subgroups such as men and persons with a history of AMI or prior treatment with ACE inhibitors may have a higher risk for AMI during DST shift. Further studies which include data on chronotype and sleep duration are needed in order to confirm these results.