Seasonal variation of hospital bed usage is a well known phenomenon [1–4]. On one hand, seasonal increases in hospital admissions may be weather dependent [3, 5–13]. On the other hand, in certain areas tourism may cause seasonal variation in health service utilization by temporarily increasing a population at risk during holiday seasons [14]. Whereas climate-dependent variations involve the local population of an area, vacation-induced increases in hospital admission rates may largely derive from nonlocal residents.

Switzerland can be divided into tourist and nontourist regions. Most of the tourist areas are located in the mountains, which afford ski and snowboard activities in winter, and mountaineering and sports such as mountain biking during the summer. In general these mountain regions have a lower local population density than nontourist areas, although their population fluctuates greatly during holiday seasons. The provision of sufficient numbers of hospital beds in such areas is of key importance for resource allocation in health care, and making efficient and effective use of those available cannot rely only on local population size. Resorts providing seasonal leisure sport activities such as skiing or mountaineering may be expected to have high emergency admissions of nonlocals during peak seasons, whereas nontourist areas usually do not show seasonal effects. The estimation of emergency service treatment beds in tourist areas therefore presents a planning challenge that can be addressed only through careful analysis of admission patterns over time [15]. One method for such analysis is small area analysis based upon hospital service areas [16–19]. Using orthopedic discharge data and following the method described by Klauss et al[16], Switzerland can be divided into 85 orthopedic hospital service areas (HSAo). In short, each discharge is labeled with a residence code called medstat (Switzerland is divided into 612 medstat regions), these medstat regions are then aggregated into HSAo according to hospital usage patterns. HSAo contain at least 1 and a maximum of 27 hospitals, with a high hospital density seen in HSAo including large major cities such as Zürich or Geneva to only one hospital in most rural HSAo. These HSAo were analyzed for seasonal effects in order to evaluate seasonal variations in emergency admissions due to tourism in mountain sport resort areas. The aim of this evaluation is to provide healthcare planners with detailed information on seasonal variations in admission rates in different area types.

Federal discharge data for orthopedic procedures (according to CHOP [18]- and/or ICD10 codes[19]) from the Swiss hospital discharge master file from 2000–2002 were used (Swiss Federal Office of Statistics). Commercial GIS-compatible vector files for medstat regions (an aggregate of several postal code regions, built-up according to socio-economic and geographic coherence criteria) were obtained from MicroGIS (MicroGIS Ltd, Baar, Switzerland).

The inclusion criteria for the total orthopedic dataset were: Primary or additional procedure CHOP codes 77.00–84.90 and/or primary diagnosis ICD10 M00.0-M25.9, M40-M43.9, M45-M51.9, M53-M54.9, M60-M63.8, M65-M68.8, M70-M73.8, M75-M77.9, M75-M77.9, M79-M96.9, M99-M99.9.

The inclusion criteria for the analysis of seasonal emergency admissions were:

Swiss and foreign residents admitted during the years 2000–2002 and emergency admission type (needing treatment within 24 hours after admission).

HSAo population counts were calculated from the 2000 census (Swiss Federal Statistical Office, Section Geoinformation, Espace de l'Europe 10, 2010 Neuchâtel)

Swiss HSAo were constructed according to the method described by Klauss et al. [16] and Goodman and Green [17] using all admission types and only data of Swiss residents.

A basic data set per HSAo contains 5 indicators that permit the calculation of further indices (numbers or ratios). The 5 basic indicators per HSAo are number of local residents (l-res) admitted in HSAo hospital(s), number of nonlocal residents (nl-res) admitted in HSAo hospital, total number of admissions in HSAo hospital(s), length of stay, and age. On the HSAo level the following indicators per month/year can be calculated:

A monthly admission index for l-res, nl-res, and total hospital admissions for each HSAo was calculated [15] by dividing each month's total number of admissions into the average monthly number of admissions for a year (total number of admissions for a year divided by 12). An overall pattern of demand is calculated by averaging month indices over years. This index removes the effect of overall population growth and compensates for the statistics of small population sizes, especially in mountain area HSAo.

Winter sport HSAo were defined using guest house accommodation nights during the year 2003, provided by the Swiss Federal Statistical Office (UNT Division – Business, Espace de L'Europe 10;CH-2010 Neuchâtel). Switzerland offers ski slopes between 1500 and approximately 3500 meters altitude, and corresponding lodging opportunities are found from 1000 meters altitude upwards. Therefore, HSAo offering guest beds at or above 1000 meters, showing a seasonal accommodation rate (number of accommodations per 10000 HSAo residents) according to a winter-summer pattern -steep increase during the winter months starting in December, followed by a steep decrease in April and May, an increase during the summer months, and a decrease in October and November- were defined as winter sport areas (WSA). HSAo not offering guest beds above 1000 meters altitude, or HSAo with guest beds above 1000 meters only showing increases in accommodation rates during the summer (HSAo with summer pattern), were defined as regular areas (RegA). Additionally, every HSAo was evaluated for seasonal hospital admission rates (hosp_ar).

A Wilcoxon rank sum test was used to analyze differences of rates and indices between WSA and RegA HSAo because variables for several months were skewed. These analyses were performed for each month; the year of admission was an additional cofactor in the model. Statistical analyses where performed with SAS 9.1^® (SAS Institute Inc., Cary, NC, USA). The significance level was set at p < 0.05 throughout the study.

For geographic presentations ArcGis, ArcView8.2^®, ESRI, Redlands CA, USA was used.

Most of the ambulatory and stationary treated patients in winter resorts in Switzerland have head and extremity injuries that are mostly the result of ski/snowboard accidents (85%)[20]. An orthopedic dataset including trauma and non-trauma related diagnoses were used to evaluate to which extent these cases stress the overall hospital service utilization in such areas. We decided to include non-trauma (such as degenerative diagnosis) related diagnosis because the tourist population may as well contain (older) people with degenerative conditions. These tourists possibly will be at higher risk of having a winter condition related accident in winter sport than in flat land areas. An even better dataset would have included all hospitalized patients, but a dataset including all necessary variables was not available at the time of analysis.

Using the method of small area analysis, Switzerland can be divided into 85 orthopedic hospital service areas (HSAo), of which 24 can be defined as a winter sport area (WSA) based upon an accommodation pattern of the guest beds above 1000 meters altitude. Seventeen of the HSAo defined as WSA show a seasonal winter-summer pattern (SeH) in hospital admission rates, with single HSAo showing up to 7–9 times higher admission rates in the winter season than in the intermediate season. The remaining 7 WSA HSAo do not show a seasonal hospitalization pattern. Whereas WSA HSAo showing a seasonal admission pattern all contain large nationally and internationally recognized winter sport resorts, resorts in WSA HSAo without a seasonal hospitalization pattern are, with one exception, only nationally known. None of the HSAo with guest beds above 1000 meters that have a summer season accommodation pattern show a seasonal pattern in hospital admission rates.

HSAo defined as WSA differ in several ways: injuries of nonlocal residents can be treated locally, resulting in a seasonal pattern of hospital admissions; or, nonlocal residents can be treated outside their HSAo of injury because, for example, the necessary health service is not available there, thus leaving too few nonlocal residents treated locally to produce a seasonal pattern. In addition, more geographically isolated areas, such as found in Graubünden or Valais, have to treat their patients locally, whereas WSA HSAo adjacent to RegA HSAo may transfer emergency cases to those regular HSAo, where they dissolve in the mass. Also, larger, international ski resorts may produce more emergencies (when compared to smaller, more local resorts) that involve foreigners, who may be less easily transferred elsewhere.

As an alternative hypothesis, one may postulate that WSA HSAo that do not show a seasonal hospitalization pattern provide better injury prevention measures, resulting in fewer injuries during the winter than in SeH. Although this hypothesis cannot be verified with our data, it would be of great interest to investigate the link between admission rates and the number of people practicing leisure sports during a given period.

Nonlocal resident admissions in WSA, especially during the winter months, are significantly younger than admitted local residents and their average length of stay is shorter. This supports the hypothesis that winter leisure sport tourism causes the mean observed seasonal peak in hospital admission rates.

In general, emergency cases stress the health care system more in WSA than in RegA. With large admission fluctuations over the year, these cases require tremendous flexibility in resource planning in WSA (especially in SeH within WSA). As nonlocal residents are the main originators of the emergency admissions, in order to avoid an imbalance between demand and supply hospital bed and staff planning cannot be based on the local population size alone. The provision of enough beds in winter may mean a half empty hospital and suspension of staff during the intermediate season, which could have substantial financial and employment consequences for mountain regions with mostly lower than average population densities and job opportunities.

Small area analysis is an appropriate method to study seasonal effects in hospitalization rates. Concerning orthopedic emergency admissions, Switzerland can clearly be separated into areas with a seasonal and constant admission rate patterns. HSAo showing a seasonal admission pattern all lie within regions containing winter sport resorts. Because the seasonality in admission rates is caused only by nonlocal residents, it can be concluded that seasonality for orthopedic emergency admissions in Switzerland is largely due to leisure sports injuries deriving from tourism. Large variability in admission rates over the year in such areas demands great flexibility in resource planning. Further analysis of types of injuries (leisure sport related or not) and financial implications of seasonality should refine these conclusions.