Extreme Cold, Not Winter Itself, Raises Early Death Risk in a Distinct Heart Failure Group

When winter arrives, cardiologists have long braced for a familiar surge: more heart failure admissions, more decompensations, and more deaths. Decades of observational research from France, the United States, and elsewhere have painted a consistent picture of a winter-dominant mortality peak in heart failure, generally attributed to respiratory infections and cold-triggered sympathetic nervous system activation. But a new analysis from Germany suggests that this seasonal dogma may not apply uniformly across the spectrum of heart failure phenotypes, and that the real culprit may not be the season at all, but the precise temperature a patient experiences in the days surrounding a hospital admission.

The study, published as a letter to the editors of Clinical Research in Cardiology, comes from a team at the University Medical Centre Mannheim led by Henning Johann Steffen and Tobias Schupp. The researchers focused on a specific and often overlooked group of patients: those with heart failure and mildly reduced ejection fraction, abbreviated HFmrEF, in which the heart’s left ventricular ejection fraction, a measure of pumping efficiency, sits between 41 and 49 percent. This phenotype was only formally recognized as a distinct category by the European Society of Cardiology in 2016, renamed from mid-range to mildly reduced in 2021, and, in a twist that gives the new findings added urgency, discontinued as a separate entity in the 2026 ESC guidelines, which merged it into an expanded heart failure with reduced ejection fraction spectrum covering all patients with an ejection fraction below 50 percent.

Almost all previous evidence on seasonality in heart failure derives from cohorts dominated by patients with reduced or preserved ejection fraction, and it relies on coarse seasonal categories rather than granular temperature measurements. Whether seasonality influences the profile or short-term prognosis of HFmrEF patients had never been specifically examined. The Mannheim team set out to fill that gap using their HARMER registry, a single-center database established in 2016 specifically to investigate the newly defined HFmrEF entity. The registry’s design allowed comprehensive collection of echocardiographic, laboratory, and comorbidity parameters in this population, which is a significant advantage when studying a phenotype that has historically been lumped together with its neighbors.

The researchers retrospectively analyzed 2,184 consecutive patients hospitalized with HFmrEF between January 2016 and December 2022 at their center, which sits in one of the warmest regions of Germany. Admissions were stratified by season of presentation: spring from March to May, summer from June to August, autumn from September to November, and winter from December to February. Crucially, the team went beyond simple seasonal buckets. They obtained daily meteorological data from the German Meteorological Service climate station in Mannheim and defined extreme temperature as the lowest and highest 5th percentile of the temperature at admission, as well as of the three-day mean temperature preceding admission. Day-to-day temperature fluctuation was defined analogously. Mortality analysis used multivariable Cox regression, adjusted for age, sex, acute decompensated heart failure, myocardial infarction, and ischemic etiology, with a sensitivity analysis additionally adjusting for calendar year.

The first striking finding was what the researchers did not see. Admissions were distributed almost perfectly evenly across the year: 556 in spring, 548 in summer, 518 in autumn, and 562 in winter. A calendar-month-level analysis confirmed the absence of clustering, with monthly admissions ranging from 149 in May to 213 in March, the modal month, and no discernible winter predominance. Baseline characteristics were closely balanced across seasons, with sex distribution, heart failure etiology, valvular disease burden, NYHA functional class, and the large majority of comorbidities showing no meaningful differences. Only diabetes mellitus, ranging from 32.5 percent in summer to 40.7 percent in winter, and current smoking, ranging from 15.8 percent in autumn to 22.3 percent in spring, differed statistically, and neither followed a consistent seasonal gradient. Discharge prescription of guideline-directed heart failure medications was likewise similar year-round.

At 30 days, 92 patients, or 4.2 percent of the cohort, had died. Season of admission showed no association whatsoever with 30-day mortality, in either univariable or multivariable analysis, with a log-rank p-value of 0.809. This null result stands in sharp contrast to the pronounced winter excess mortality described in French and American cohorts dominated by reduced ejection fraction or undifferentiated heart failure. The authors propose several explanations. First, HFmrEF is pathophysiologically distinctive: preserved contractile reserve relative to more severely reduced ejection fraction may buffer the hemodynamic and infection-related stressors implicated in winter decompensation of advanced phenotypes. Consistent with this, neither C-reactive protein nor N-terminal pro-B-type natriuretic peptide concentrations at admission showed any seasonal gradient, arguing against subclinical, season-driven inflammatory or congestive priming. Second, guideline-directed medical therapy was prescribed at similar rates throughout the year, removing a potential seasonal treatment gap.

But the granular temperature analysis revealed something the seasonal categories had hidden. Admission during extreme cold, defined as the lowest 5th percentile of ambient temperature at or below 0.8 degrees Celsius, was associated with significantly higher 30-day mortality: 8.6 percent among the 105 patients admitted during such cold snaps, compared with 4.0 percent among the remaining 1,972 patients. The association held in both univariable analysis, with a hazard ratio of 2.216, and multivariable analysis, with a hazard ratio of 2.012 and a p-value of 0.049. A similar association emerged for extreme cold in the three-day mean temperature preceding admission, at or below 1.33 degrees Celsius, with a multivariable hazard ratio of 2.058. In other words, patients admitted during the coldest days, or after several days of sustained cold, faced roughly double the risk of dying within a month, regardless of their age, sex, or underlying cardiac pathology.

Equally notable was what did not predict death. Extreme heat, defined as the highest 5th percentile at or above 23.3 degrees Celsius, showed no association with 30-day mortality, with a multivariable hazard ratio of 0.849. Day-to-day temperature fluctuation in the top 5th percentile was similarly uninformative, with a hazard ratio of 0.854. A sensitivity analysis additionally adjusting for calendar year left the extreme cold estimates materially unchanged, and calendar year itself was not independently associated with mortality. The pattern suggests that short-term extreme cold exposure, rather than coarse seasonal categorization, may be the more sensitive determinant of early risk in HFmrEF. This aligns with recent evidence that ambient temperature and atmospheric pressure predict heart failure hospitalization and decompensation severity across exposure windows of days rather than months.

The picture is not entirely uniform across the literature. A recent Austrian cohort found that ambient temperature predicted emergency decompensations in patients with reduced and preserved ejection fraction but not in those with mildly reduced ejection fraction, suggesting that cold sensitivity may depend on the specific outcome, exposure window, and population studied. Divergent findings from Japan, where heart failure mortality peaks in summer rather than winter, together with heterogeneous regional patterns in the United States, further indicate that seasonality in heart failure is climate-dependent and phenotype-dependent rather than a fixed biological law. The Mannheim findings add an important data point: in a warm-region German cohort, the calendar season was irrelevant, but the thermometer was not.

The authors are candid about the limitations. The retrospective, single-center design means residual confounding cannot be excluded. Respiratory infection surveillance data, a presumed mediator of winter seasonality, were unavailable. The temperature analysis relied on a single weather station and therefore does not capture individual-level exposure, humidity, or air pollution coexposures. Most importantly, the extreme temperature subgroups were small, with 105 and 107 patients in the lowest and highest 5th percentiles respectively, and the resulting wide confidence intervals warrant cautious interpretation and external replication. The registry also lacks a directly comparable reduced or preserved ejection fraction cohort within the same dataset, although the broader literature documents substantial parallels between HFmrEF and reduced ejection fraction heart failure, including a higher prevalence of ischemic etiology, which reached 57.6 percent in this cohort, and similar benefit from neurohormonal therapies, reflected here in high discharge prescription rates of beta-blockers at 74.9 percent and ACE inhibitors or ARBs at 71.3 percent.

The takeaway is both scientifically intriguing and practically actionable. Season of admission carried no prognostic weight in this HFmrEF population, whereas extreme cold doubled the short-term mortality risk, and extreme heat carried none. If confirmed prospectively in multicenter settings, this reframing could sharpen how clinicians and health systems target preventive resources: not at winter as a blanket category, but at the specific cold snaps that precede dangerous decompensations. It also raises a pressing question for the post-2026 guideline era, in which HFmrEF has been absorbed into an expanded reduced ejection fraction spectrum: whether the temperature-mortality association documented here extends to the full range of patients with an ejection fraction below 50 percent, or whether cold sensitivity remains a fingerprint of this particular phenotype.

Subject of Research: Seasonal and temperature-related variation in short-term mortality among patients hospitalized with heart failure and mildly reduced ejection fraction

Article Title: Seasonal variation and extreme cold exposure in heart failure with mildly reduced ejection fraction

Article References: Seasonal variation and extreme cold exposure in heart failure with mildly reduced ejection fraction. (n.d.). https://doi.org/10.1007/s00392-026-03038-7

Image Credits: AI Generated

DOI: 10.1007/s00392-026-03038-7

Keywords: heart failure, HFmrEF, ejection fraction, seasonal variation, extreme cold, 30-day mortality, cardiology, ambient temperature, ESC guidelines, Cox regression, hospitalization, climate