Seasonal variations in incidence of femoral fractures in the state of Rio Grande do Sul, southern Brazil

Introduction: Femoral fractures are a major cause of morbidity and mortality, mainly among older people. Objective: To examine the effect of seasonality on hospitalizations due to femur fracture among people residing in the Rio Grande do Sul state, southern Brazil, from 2008 to 2019. Methods: Ecological study based on secondary data from the SUS Hospital Information System (SIH/SUS). A total of 74,374 reports of hospital admissions was considered. The generalized additive model (GAM) approach was employed to assess the seasonality of the time series, with stratification by sex and age groups and considering the monthly average number of events of femoral fractures per day as a dependent variable. Results: A considerably higher incidence of femoral fractures in women aged 70 years or more was described. Among people aged less than 50 years, there is not an apparent seasonal effect. Men aged 70 years or older and women aged 50 years or older have a higher frequency of hospitalizations due to femur fractures in the colder months. Conclusion: Among older people, more femoral fractures occurred during the winter compared to summer. This supports findings from other studies, although reasons for this seasonal variation are uncertain. The knowledge of these seasonal variations can help to plan the health care in the public health system.


Introduction
Among the fractures that most affect the elderly population, femur fracture is the most prevalent.
Regardless of the anatomical location, femur fractures are considered severe and a significant health problem. 1,2 This is because this injury demands an extended period for patient recovery and, in some cases, evolves with complications and sequelae that can be much more serious in the elderly population due to the physiological characteristics of this age group and associated diseases. Among the elderly, hospitalization time is longer, often requiring admission to intensive care units due to complications. The rehabilitation period is usually prolonged, and the sequelae may reach partial or total loss of independence in activities of daily living, in addition to high economic costs for health services and considerable social demands for the maintenance of life and health of those affected. [3][4][5][6] Studies carried out in Northern Hemisphere countries have raised discussions regarding the determinants of fracture incidence in the elderly, in which the role of climate conditions, such as freezing and slippery winters, has been shown to be very relevant. These studies all agree that a deeper understanding of the relationship between climate and fracture incidence in different climatic conditions can lead to a better understanding of fracture etiology and seasonality of their incidence, allowing gains in preventive strategies, particularly among the elderly. 7-11 In a systematic review by Burget et al., 8 it is discussed that knowledge about seasonal patterns in relation to geriatric fracture incidence still has many gaps and that divergent findings have been

Design and ethical considerations
This is an ecological study conducted from January  16 where the mean of Y t is given by the expression a + s 1 (trend t ) + s 2 (month t ).
In this formulation, a is an intercept, "trend" is the time  17 For the fit of the model to the data to be considered adequate, the residuals (differences between the observed values of the series and those predicted by the model) should not be autocorrelated, that is, there should be no correlation between successive values of the residuals over the period studied. We used plots of the autocorrelation function (ACF) and the partial autocorrelation function (PACF) to verify this assumption.
We use the "mgcv" package of the R program version 3.6.0 for fitting the GAM. 18 The analysis was stratified by increase in fracture rates in the 20-29 age group, perhaps due to more exposure to risky work activities or sports practices. The rates decrease in the following age groups and increase rapidly after the age of 50.
The graphs in Figure 1 describe the monthly average number of events of femoral fractures per day, considering all age groups and both sexes. Panels (b) to (g) in Figure 1 show that the number of hospitalizations due to femoral fractures is higher among men than among women up to the age of 59. After 70 years of age, the number of hospitalizations is much higher among women than among men.          The results from the GAM analysis of the data from RS provided convincing evidence of seasonality among the elderly population, which is in agreement with most previous studies where seasonality with a winter peak was found. 7-11 For example, in a systematic review including 24 studies, Shi et al. 24 showed that the incidence of