In this extensive 10 year analysis combining HF and meteorological data, a decline in HFi among people aged ≥ 65 years in Catalonia, Spain, was observed from 2010 to 2019. There was a distinct seasonal pattern, with higher rates in autumn and lower rates in summer, with more pronounced fluctuations in the older age groups. Solar radiation exhibited a robust negative correlation with HFi, higher than that of temperature.

While the elderly population has steadily increased in Catalonia over the last decade, the number of HFs has not risen at the same rate. These data confirm the tendency of a decrease in the standardized incidence in our [7, 19, 20] and other, mainly Western, countries [3, 21,22,23]. The decreasing rates in most countries in Europe, North America, and Oceania, observed since the last decade of the twentieth century, differs from the increasing incidences observed in most Asian countries [8]. The combination of several factors could explain this downward trend: sociodemographic changes such as historical cohort effects have been described in Spain, which would also influence the differences in incidence and trends between eastern and western communities [24]. Moreover, improvement in healthy habits such as increasing physical activity [25] and cessation of tobacco consumption [26] have been related to a decrease in HF risk in postmenopausal women. In that sense, over the last decades some interventions have been implemented in Catalonia to increase the proportion of adults complying with physical activity recommendations [27] and decreasing smoking habits [28]. Whether or not due to these health policies, the prevalence of healthy levels of physical activity has risen and the percentage of tobacco consumption has decreased, while obesity, a protective factor [29], has climbed [30]. Regarding the possible relationship between the use of anti-osteoporosis medications and the trend in HFi, the beginning of our time series coincides with the concerns raised due to the association of bisphosphonates with atypical fractures, which led to a dramatic decrease in the use of those medications in Spain [31]. However, to better understand the influence of anti-osteoporosis medications on the HFi trend, it would be necessary to analyze other data, such as adherence rates, the baseline risk of fractures in the treated population, and the comparative effectiveness of medications used in different periods of time. Secondary prevention programs (Fracture Liaison Services) in line with International Osteoporosis Foundation guidelines have had an increasing impact in our country in recent years. These units have been shown to improve the outcomes of patients with FF and to reduce fracture risk [31]. Finally, climate change, with increasing temperatures over time, could have influenced HFi to some extent. However, based on our results, the HFi would decrease approximately by 1% for every 1 °C increase in temperature. Given that air temperature is increasing 0.25 °C every decade in Catalonia [32], the expected decrease in HFi due to rising temperatures over a ten-year period would be 0.25%, which is insignificant compared to the observed one.

By sex, the standardized HFi rate decreased more in women than in men. This has been described in ours and other countries and could be related, in part, to a lower awareness in the diagnosis of osteoporosis and the prescription of preventive drugs in men [33]. There were also some differences between age groups; a significant downward trend in the age-specific incidence ratio was found in people ≥ 75 years, while it was more pronounced from 75 to 85 years than in people aged ≥ 85. The downward trend previously described in people ≤ 75 years up to 2014 in Catalonia [19] was lost.

HF rates were clearly affected by seasonal variations, with a higher prevalence in cold seasons and a more pronounced effect in older people and in men. The average daily solar radiation in our series was 16.2 MJ/m2, ranging from a minimum of 9.0 MJ/m2 in autumn to a maximum of 22.8 MJ/m2 in spring. According to our ARIMA seasonal regression analysis, for every MJ/m2 increase in average daily solar radiation, the monthly HFi rate × 100,000 decreased by 2.72 points in people over 84 years of age. Given that the average monthly HFi rate for people in this age group is 191.3, this represents a reduction of 1.42% for every MJ/m2. The same increase in solar radiation was only associated with a 0.92% decrease in HFi in people aged 75 to 84 years. Consequently, the lower levels of radiation in the colder months may have had a more pronounced impact on older people, due to a decrease in the skin's ability to produce vitamin D with age [34, 35]. Besides, colder temperatures can heighten the risk of falls by affecting neurosensory abilities [36, 37]. In our time series, autumn was the season with the lowest average insolation and with the highest global HFi in the overall population and among women. Conversely, men exhibited the highest HFi during winter, the coldest season. The seasonal pattern was more pronounced in men than in women, with women-to-men ratio much lower in winter and higher in spring. Differences between sexes in the seasonality of HF have been described, although only in some geographic areas. In a study focused on New York City, seasonality had a more marked effect in men than in women, with a greater susceptibility to HF in men during cold months. The greater propensity for men to go outdoors in worse weather conditions was mentioned as a possible explanation [38]. Otherwise, other cohorts from Canada and Taiwan found no differences between age groups or sexes [21, 39]. The autumn predominance of the HFi is the most frequently observed pattern in Spain [13, 14]. In a geographically close cohort from the Mediterranean region in Spain, in which the day was used as a time unit, the overall pattern of seasonality was similar to ours. However, the authors’ findings differed in that a close relationship between HFi and wind intensity was found, especially in younger people [13]. A possible explanation comes from the different time units used in the analyses. When the day is used as a unit of time, meteorological factors that increase the probability of falls, such as wind, snow, or ice, could be better evaluated. In contrast, when the month is used, those climatic variables that have a more long-term effect would be more apparent in the results.

Solar radiation, followed by temperature, were the two climate variables that exhibited a strong and negative association with HFi in the bivariate analysis. Moreover, in the seasonal ARIMA regression analysis, insolation, but not temperature, retained a statistically significant association with HFi. Two systematic reviews have examined the relationship between cyclical changes in fracture rates and climatic parameters, revealing a protective effect of higher temperatures [12, 17]. One of these reviews, specifically focused on HFs, also summarized prior studies investigating the influence of sunshine on HF risk. These studies, generally using monthly aggregates, consistently showed a negative association [12]. Those studies estimated insolation based on a timed measure of sun exposure (in minutes, hours, or days). However, the amount of solar radiation on earth not only depends on the time of solar exposure, but also on other factors, such as the hour of the day, the season of the year, the ozone layer, the surface reflection, the altitude, and the latitude [41]. Our analysis is based on the measurement of direct and diffuse incident solar radiation measured at a wavelength that corresponds to the spectrum spanning UV to infrared. As the cutaneous synthesis of vitamin D depends on the amount of UVB light, we believe that our estimate is more reliable than those based only on hours of sunshine. Moreover, the design of our analysis, based on monthly data, is better suited to examine the effects of insolation on fracture incidence. While the impact of sun exposure on vitamin D synthesis from pre-vitamins is rapid [40], its beneficial effects on bone resistance are gradual. A relationship between cumulative UV exposure, bone mineral density and the risk of falls and fractures has already been documented [41]. In contrast, another study conducted in inland Spain found a negative association between UVB light radiation and HFi in the short term. The authors hypothesized that increased sunshine could also exert its effect through enhanced visibility, thereby reducing the risk of falls [14]. A work performed in Boston, MA (USA), a city with the same latitude as Catalonia, showed that from November to February there was insufficient UBV radiation to synthesize previtamin D from 7-dehydrocholeterol. Beginning in March; however, the skin was capable of synthesizing vitamin D [42]. In a study of patients with HF in our geographical area, serum vitamin D levels correlated with the solar radiation received during the 2–3 months prior to the fracture, while the seasonal pattern of HF was inversely proportional to the seasonality of vitamin D levels [43]. Finally, a recent Japanese study exploring the external causes of death from 1979 to 2015 in relation to ambient temperature found that both cold and heat exposures were associated with increased falls-related deaths [44]. In view of all the above, we hypothesize that gradual increase in solar radiation with a cumulative effect on bone mineralization, along with good visibility and pleasant temperatures without extreme heat, would have led to a minimum HFi in June and the opposite trend in December.

In addition to sunshine, only relative humidity showed a protective effect in general, and in all subgroups (except in people ≤ 75 years of age), in the regression model, which was more pronounced in people over 85 years of age. There are few studies that have analyzed the relationship between relative humidity and the risk of HF. In one aforementioned study performed in Spain, the authors described a different pattern in people under 75 years of age, in which high relative humidity was associated with a higher HFi compared to older people, in whom the effect was almost null [13]. Another study from Israel involving subjects with a mean age of 78 years [45] did not find any associations. The rest of the parameters (atmospheric pressure and average wind speed) similarly failed to show any consistent association.

There are certain limitations in our study that deserve some mention. Firstly, the data provided comes from administrative sources that may be subject to notification and registration errors. Local privately owned hospitals do not have incentives to record discharge information in a systematic manner, which could lead to under-registration in some cases. Nevertheless, as Catalonia has universal and free access to public healthcare system, only a minimal percentage of HFs are admitted to private centers [46]. In addition, the relationship between climatic variables and fracture rates are temporal in nature and could be influenced by other uncontrolled factors, such as differences in mobility and clothing between seasons [39]. In fact, the ecological design of the study could lead to an ecological fallacy—the possibility of making incorrect conclusions about individual-level associations when only using aggregated data. Another limitation of our study is that we lack other important information related to the individual fracture risk: the mechanism of fracture or the place where the fracture occurred (indoors or outdoors). Finally, as we chose to analyze the time series using periods of one month, we believe that the effects of meteorological phenomena such as wind speed or icy days, which have immediate effects on fracture risk and are scarcely present in our environment, could be underestimated.

In summary, the data that we present from southern Europe show a decreasing trend in the rate of fractures in our population, especially in elderly women, until 2019. After the start of the Covid-19 pandemic, most studies from different geographical areas, including ours, reported a sharper decrease in the HFi and other FFs [47, 48]. We observed a marked seasonality in incidence rates, with a peak in autumn and a nadir in summer, especially in older people. Based on our data, insolation could be more responsible than temperature for the fluctuation of HFi in our country.