Daylight saving time debate: A critique unveils flaws in mortality study

By Priyanjana Pramanik, MSc.Mar 6 2024Reviewed by Susha Cheriyedath, M.Sc.

In a critique of a previously published article on daylight saving time (DST) and European mortality in 1998-2012, researchers discussed their concerns regarding the study's key results and interpretations. The critique published in Nature Communications highlights significant flaws in the statistical analysis and conclusions of the study.

In the original article's abstract, researchers reported a decrease in mortality during the two weeks following the Spring change with a reverse effect after the Fall change, using the multiple negative binomial regression model. However, Figure 1 of the article showed the opposite effect – a slight increase from an already decreasing trend in mortality in the Spring and a slight decrease from an already increasing trend in the Fall in the weeks after the DST transition. These data are consistent with previous research on this subject.

In the critique, the authors presented the following reasons for this discrepancy.

Matters Arising: Daylight saving time and mortality—proceed with caution. Image Credit: Iryna Imago / Shutterstock

Concurvity in the model

Concurvity in non-linear models is analogous to collinearity in linear models, indicating that certain variables are interdependent but that this lack of independence has not been accounted for in the model.

In this case, the interdependent parameters are weeks after the DST transition, month, season, temperature, humidity, and photoperiod. Concurvity is not explicitly tested for in the analysis to identify the need for potential adjustments to the model.

Further, the model used does not test whether mortality-related parameters affect mortality rates in association with the transitions to DST. For example, the authors of the critique suggest checking whether results are seen only for a particular year and whether sex or age composition is different before and after a transition.

Time series and seasonal considerations

The dataset used in the paper included the entire time series, while the critique suggests restricting it to the weeks of data immediately before and after transitions. This restricted dataset would require fewer covariates and also reduce bias due to seasonal variations in humidity, month, temperature, and photoperiod, which can have longer-term trends.

At the same time, the analysis presents comparisons in mortality rates based on one week before the DST transition, which does not take into account longer-term seasonal trends in mortality, showing a rise in Fall and a decrease in Spring. The authors of the critique suggest explicitly incorporating seasonal trends using a design such as an Interrupted Time Series with a thoughtfully defined pre- and post-interval period.

Country-level differences

The dataset contains data from 16 countries, but the paper does not describe how the pooled regression model accounted for possible country-level differences. Interactions between the countries and other variables of interest are also not explicitly tested, which could lead to incorrect estimator variances and biased estimators.

The critique also notes that statistical significance in the study results could be due to the large sample size, with the relative impact of DST transitions in the model being relatively low. The analysis also does not list causes of death and thus makes no distinction between deaths from cancer (which cannot be attributable to DST) and traffic accidents (which can). There is, therefore, limited evidence to support the claim in the title of the original paper that DST affects mortality.

Conclusions

The original article's overall findings are inconsistent with those of many previous studies. The DST transition is linked to an increase in strokes and heart attacks. Furthermore, misalignment in circadian rhythms and sleep disturbances can have long-term health impacts due to increased inflammation, stress, and altered metabolism. Neuropsychological function is also affected by DST, which may be associated with increases in traffic accidents and suicides.

The critique's authors believe that policymakers must consider these adverse effects and compare them to the short-term benefits of DST in terms of increased exposure to light during evening commutes from work. A large body of evidence suggests that permanently staying on standard time is the best option, as advocated by many health practitioners and researchers worldwide.