Snow looked like the cause of winter hip fractures until the temperature was added
In 2,292 hip fractures across six cities in Nagano, snowfall raised the odds of admission — until the models adjusted for how cold it was, at which point the association disappeared.
| Group | Value (value) |
|---|---|
| Snowfall 2 days before admission | 1.16 (1.01 to 1.33) |
| Snowfall 1 day before admission | 1.15 (1 to 1.33) |
Hip fractures cluster in winter, and the intuitive explanation is ice underfoot. A study published in Environmental Research tested that explanation directly, and found that most of what looks like a snow effect is a temperature effect wearing snow's clothes [s1].
The design
The authors collected daily hospital admissions for hip fracture and meteorological variables in six cities in Nagano Prefecture, Japan, from November to March between 2014 and 2019, in adults aged 65 years and older [s1]. They used a time-stratified case-crossover design, in which each patient serves as their own control on other days in the same month — a structure that removes confounding by anything about the person that does not change week to week [s1].
A total of 2,292 hip fracture cases were included [s1].
The paper's stated motivation is a gap in the earlier literature: previous studies have suggested snowfall is a risk factor for hip fracture, but most did not adjust for ambient temperature, which has itself been associated with hip fracture incidence [s1].
The result, in two stages
In models not adjusted for ambient temperature, snowfall — defined as more than 0 cm against 0 cm — two days before admission was associated with significantly higher hip fracture incidence (adjusted odds ratio 1.16; 95% CI 1.01–1.33) [s1]. Snowfall on the day before admission showed the same pattern (adjusted odds ratio 1.15; 95% CI 1.00–1.33) [s1].
Both intervals scrape the null, and one touches it exactly. These are the estimates a study that stopped there would have reported as positive.
After further adjustment for ambient temperature, the estimates attenuated toward the null, and no significant associations were observed [s1].
Meanwhile, the paper found a significant inverse association between ambient temperature and hip fracture incidence that was independent of snowfall [s1] — colder days, more fractures, regardless of whether snow fell.
The authors' conclusion: snowfall was not associated with hip fracture after accounting for ambient temperature, and lower ambient temperature rather than snowfall itself may be the primary driver of hip fracture incidence [s1].
Why this is more than a statistical footnote
Snow and cold travel together. Any analysis that includes one and omits the other will assign the combined effect to whichever variable it happened to measure. Nagano is a good place to separate them precisely because it has plenty of both, across six cities and six winters.
The practical implication points away from the obvious intervention. If the driver were slippery ground, gritting and footwear would be the levers. If the driver is temperature, the mechanism is more likely to run through the body than the pavement — cold affects muscle function, balance, reaction time and the amount of clothing worn — and the levers are different ones.
The study does not identify which of those mechanisms operates. It is a time-series association, not a physiological experiment, and it says nothing about any individual's risk on any given day.
A second climate says something similar
Evidence from a much milder climate points the same way. A serial cross-sectional study published on August 26 examined adults aged 50 and over treated operatively for hip fracture at a single Southern California academic health system between 2017 and 2022 [s2].
Among 499 patients — average age 77.2 years, 62.7% women — monthly hip fracture counts showed significant seasonality (p = 0.02), peaking in January and troughing in July [s2]. Southern California does not have icy pavements in January.
The analysis found a significant interaction between daylight duration and temperature on the odds of a fracture falling in a high-volume month (p for interaction = 0.02) [s2]. At cooler temperatures, shorter daylight was associated with higher odds (OR 2.6; 95% CI 1.48–4.58); at warmer temperatures, both shorter (OR 0.17; 95% CI 0.06–0.50) and longer (OR 0.52; 95% CI 0.29–0.92) daylight durations were associated with lower odds than medium daylight [s2].
The authors read this as evidence of seasonal influences beyond weather-related fall hazards [s2]. It is a small single-centre study, and daylight and temperature are themselves entangled, so the interaction should be treated as a hypothesis rather than a finding. But two studies in very different climates now point past the ice.
What to watch
Whether winter fall-prevention programmes, which are largely built around surfaces and footwear, are ever evaluated against interventions aimed at cold exposure itself. On present evidence nobody knows which would work better, and the question has barely been asked.
Sources
- Associations between snowfall and hip fracture among adults aged 65 years and older in six cities in Nagano Prefecture, Japan — Environmental Research, 2026-08-29
- Hip fracture seasonality in a mild climate: the joint association of daylight duration and temperature — Archives of Osteoporosis, 2026-08-26
Sources
- Associations between snowfall and hip fracture among adults aged 65 years and older in six cities in Nagano Prefecture, Japan — Environmental Research , August 29, 2026
- Hip fracture seasonality in a mild climate: the joint association of daylight duration and temperature — Archives of Osteoporosis , August 26, 2026
More on
Dairy prevented fractures in one large trial and nothing in the cohort data. Both hold.
Adding milk, yoghurt and cheese cut hip fractures 46% in Australian aged-care residents. In the general over-50 population, dietary calcium intake shows no association with fracture risk at all.
Senolytics: two randomised trials, and no confirmed clinical benefit yet
A phase 2 trial of dasatinib plus quercetin missed its bone endpoint. A senolytic injected into the eye produced a 5.6-letter vision gain whose confidence interval crosses zero.
A DXA body-fat Z-score flagged twice as many athletes with absent periods as BMI did
In 388 female athletes, a per cent fat Z-score below −1.0 more than doubled sensitivity for amenorrhoea against standard BMI thresholds. It gave up specificity to do it.
Three months of an osteoporosis drug matched twelve on bone density in a 50-woman trial
The LIDA trial tested whether romosozumab's bone-building window closes early enough to shorten treatment. It met non-inferiority on hip bone density — at a single centre, with 50 participants, over one year.