What is the effect of the weather on trauma workload? A systematic review of the literature

Injury, Int. J. Care Injured 46 (2015) 945–953

Contents lists available at ScienceDirect

Injury journal homepage: www.elsevier.com/locate/injury

Review

What is the effect of the weather on trauma workload? A systematic review of the literature A.M. Ali *, K. Willett Kadoorie Centre for Critical Care Research; Nuffield Department of Orthopaedics, Rheumatology and Musculoskeletal Sciences, University of Oxford, United Kingdom

A R T I C L E I N F O

A B S T R A C T

Article history: Accepted 6 March 2015

Background: Hospital admission rates for a number of conditions have been linked to variations in the weather. It is well established that trauma workload displays significant seasonal variation. A reliable predictive model might enable targeting of high-risk groups for intervention and planning of hospital staff levels. To our knowledge there have been no systematic reviews of the literature on the relationship between weather and trauma workload, and predictive models have thus far been informed by the results of single studies. Methods: We conducted a systematic review of bibliographic databases and reference lists up to June 2014 to identify primary research papers assessing the effect of specified weather conditions including temperature, rainfall, snow, fog, hail, humidity and wind speed on trauma workload, defined as admission to hospital, fracture or a Road Traffic Accident (RTA) resulting in a seriously injured casualty or fatality. Results: 11,083 papers were found through electronic and reference search. 83 full papers were assessed for eligibility. 28 met inclusion criteria and were included in the final review; 6 of these related to the effect of the weather on trauma admissions, one to ambulance call out for trauma, 13 to fracture rate and 8 to RTAs. Increased temperature is positively correlated with trauma admissions. The rate of distal radius fractures is more sensitive to adverse weather than the rate of hip fractures. Paediatric trauma, both in respect of trauma admissions and fracture rate, is more sensitive to the weather than adult trauma. Adverse weather influences both RTA frequency and severity, but the nature of the relationship is dependent upon the timecourse of the weather event and the population studied. Important methodological differences between studies limit the value of the existing literature in building consensus for a generalisable predictive model. Conclusions: Weather conditions may have a substantial effect on trauma workload independent of the effects of seasonal variation; the population studied and timecourse of weather events appear critical in determining this relationship. Methodological differences between studies limit the validity of conclusions drawn from analysis of the literature, and we identify a number of areas that future research might address. ß 2015 Elsevier Ltd. All rights reserved.

Keywords: Weather Trauma Fracture Road traffic accidents

Contents Introduction . . . . . . . . . . Methods . . . . . . . . . . . . . Search strategy . . Eligibility criteria

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

* Corresponding author. Tel.: +44 07732 309444. E-mail address: [email protected] (A.M. Ali). http://dx.doi.org/10.1016/j.injury.2015.03.016 0020–1383/ß 2015 Elsevier Ltd. All rights reserved.

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

946 946 946 946

946

A.M. Ali, K. Willett / Injury, Int. J. Care Injured 46 (2015) 945–953

Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Papers using admission or ambulance call for trauma Papers using fracture as the outcome measure (13) . . Papers relating to RTAs (8). . . . . . . . . . . . . . . . . . . . . . Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conflict of interest statement . . . . . . . . . . . . . . . . . . . . . . . . . Funding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

.......... as outcome .......... .......... .......... .......... .......... .......... ..........

Introduction The precision of short and medium range weather forecasting has improved significantly in recent years with advances in observational methods, data assimilation and modelling techniques [1,2]. Hospital admission rates for a number of conditions, most notably in respiratory and cardiovascular disease [3–5], have been linked to variations in the weather, and it is well established that trauma workload displays significant seasonal variation [6,7]. Weather forecasts are used to good effect in the commercial sector to predict consumer demand; a reliable predictive model in healthcare might enable targeting of high risk groups for intervention and effective planning of hospital staff levels [8– 10]. Moreover, the recent trend toward networks for major trauma and centralisation of care means that relatively small changes in the incidence of trauma resulting from adverse weather conditions may lead to shifts in workload at major trauma centres sufficiently large to justify adding weather forecasts to resource planning [11]. The existing literature suggests a complex relationship between weather conditions and rates of injury. Unclear definitions of ‘trauma’ and of what constitutes an ‘adverse’ weather condition (with respect to duration or severity) limit the ability to interpret the significance of observed effects [12]. It is generally acknowledged that warmer weather is associated with a higher volume of trauma [13,14] and that trauma in children is more closely associated to the weather than adult trauma [12,15]. However, to our knowledge there have been no systematic reviews of the literature on this subject and there remains confusion as to the strength of the evidence for planning trauma services based upon weather forecasts. Predictive models have thus far been informed by the results of single studies without full consideration of the relationship between weather and trauma. Most models have been derived from studies with different designs, definitions of trauma, localities and population groups. Our objective was to assess the strength of the evidence supporting an association between specific weather conditions and trauma service workload. Published studies that recorded trauma admissions, fractures or RTAs resulting in a seriously injured casualty or fatality as the outcome measure were sought, as each of these place a considerable demand upon trauma services. RTAs were singled out as a cause of trauma as they represent the most common cause of major trauma and a significant economic burden to healthcare systems around the world [16,17]. Methods The methodology of this study is reported in accordance with the Preferred Reporting Items for Systematic Reviews and MetaAnalyses (PRISMA) Statement for systematic reviews [18]. Studies were identified by searching electronic databases and scanning reference lists of selected papers. There was no restriction in the population group considered by age or geographical location. The intervention considered was a specified weather condition(s) including temperature, rainfall, snow, fog, hail, humidity and wind

........... measure (7) . ........... ........... ........... ........... ........... ........... ...........

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

946 947 951 952 952 953 953 953 953

speed, with the comparator being an absence or reduced severity of this weather condition. The outcome measure was trauma that resulted in increased workload to trauma services, defined as admission to hospital, fracture or RTA resulting in a seriously injured casualty or fatality. Search strategy The Medline (1946–January 2014) and Embase (1974–January 2014) databases were searched using the following terms: [trauma*.mp OR fracture*.mp OR exp Accidents, Traffic/or road traffic accident*.mp OR exp Accidents, Traffic/or road traffic collision*.mp] AND [exp Cold Temperature/or exp Climate/or exp Weather/or weather*.mp or exp Seasons/]. The Cochrane Library was searched using the following terms: [trauma* OR fracture* OR road traffic accident* OR road traffic collision*] AND [weather* OR climate* OR season* OR temperature*]. The last search was run in January 2014 and we conducted a limited update literature search from January 2014 to June 2014 using Medline (Pubmed) to find related articles. We only included full papers or systematic reviews and papers in the English language. Studies related to catastrophic weather events (e.g. hurricanes, tornadoes) or to periods of time identified by the authors as displaying highly uncharacteristic weather for that region or those related to the effect of the weather on people undertaking specific activities (e.g. injuries associated with skiing, hot-air ballooning, windsurfing) were excluded due to lack of generalisability of the findings. Titles and abstracts of papers were reviewed to produce a list of studies for full-paper review. Eligibility criteria Papers were required to meet the following eligibility criteria for inclusion:  Unambiguous definition of weather condition(s) being considered (e.g. ‘rain’, ‘snow’ and not e.g. ‘clear’ or ‘adverse’ without further clarification).  Use of a defined, objective information source for weather information (e.g. not obtained retrospectively from patient).  Weather information provided at least daily (e.g. not a monthly average).  For papers that did not specify the type of trauma, an outcome measure that, at the minimum, was admission to hospital or severe injury prompting an ambulance call (not e.g. attendance at Emergency Department (ED) without subgroup analysis of severe injuries).  For papers related to RTAs, an outcome measure that indicates the severity of the crash and includes subgroup analysis of serious or fatal crashes. Results A total of 11,083 papers were identified by applying the search criteria. After removal of duplicates there were 9532 papers

A.M. Ali, K. Willett / Injury, Int. J. Care Injured 46 (2015) 945–953

remaining. 83 of these were considered suitable for full-text review and 28 papers were included in the final review. A flow diagram of the search strategy and results is shown in Fig. 1 and final list of papers in Table 1. Papers using admission or ambulance call for trauma as outcome measure (7) Six papers used admission to hospital for trauma as the outcome measure of interest, and in all cases the hospital concerned was a level I trauma centre [12–15,19,20]. In three papers admission alone was a satisfactory criteria for patient inclusion [13,15,20] and in three papers inclusion required admission for trauma together with either trauma team activation (2), admission to an intensive care unit (ICU) (2), admission for >48 h (2), admission for >72 h (1), need for interhospital transfer (1) or death on arrival or in the ED or during admission (3) [12,14,19]. Four of the six studies were single centre studies (three in United States of America (USA) [13,14,20] and one in the United Kingdom (UK) [15]). Friede et al. used data from three Level I trauma centres in the USA [19]. Parsons et al. used data from 21 core hospital centres across the UK that included 59,617 patients, making this the largest study to date by far on the association between weather conditions and trauma admissions [12]. A single study by Kim et al. used ambulance call out for trauma as the criteria for inclusion, with an average of 395 calls/day

over a two year period across a population that represented 47.1% of that of South Korea [21]. The effect of a variety of weather conditions was assessed. Papers primarily focused on daily temperature but other conditions included sunshine hours, wind speed, humidity and the presence of rain, snow, sleet, fog or hail. All papers assessing the effect of temperature showed a positive and highly significant correlation between increasing temperature and the rate of trauma admissions, with maximum daily temperature appearing to be the most important predictive variable. Paediatric admissions were much more sensitive to temperature in both papers that analysed children as a separate subgroup [12,15]. Parsons et al. found that a 5 8C rise in maximum daily temperature and each additional hour of sunshine caused an increase in adult trauma admissions of 1.8% and 0.95%, respectively, and in paediatric trauma admissions of 10% and 3%, respectively [12]: very similar figures for paediatric rates, 11% and 4%, respectively, were reported by Atherton et al. [15]. Kim et al. found that a 1 8C rise in mean temperature was associated with a 0.41% increase in emergency ambulance delivery for severe trauma but only a 0.01% increase for minor trauma [21]. The effect of increasing precipitation is less clear, with some papers finding a positive correlation with trauma [12,14] and others a negative correlation [13,15,19]. In a large study in the USA with 9408 patients, Bhattacharyya et al. found that precipitation had a strong negative correlation (R = 0.857, p < 0.00001) with

Search terms applied = 11083 papers Duplicates removed 9532 papers assessed by title and abstract Inclusion criteria: 1. Papers assessing the effect of a weather condition(s) on the volume or severity of physical trauma (including RTAs) 2. Full papers or systematic reviews 3. Papers in English language Exclusion criteria: 1. Papers related to catastrophic weather events (e.g. hurricanes/ tornado-associated injuries) = 48 papers 2. Papers assessing effect of weather on specific types of activity (e.g. injuries associated with skiing, hot-air ballooning, windsurfing) = 39 papers

83 full papers assessed for eligibility Papers excluded on the basis of: • • • • • • • •

947

Inadequate definition of weather condition = 19 papers Unclear where weather data were obtained from = 5 papers Weather information obtained from patients so possible response bias = 2 papers Weather conditions averaged across month or year = 11 papers Attendance at ED was outcome measure and not severity/admission = 1 paper RTA frequency only and not related to injury/crash severity = 9 papers Minor/serious/fatal RTAs grouped together with no subgroup analysis = 7 papers Trauma and non-traumatic conditions grouped with no subgroup analysis = 1 paper

28 papers included in final review: Six papers related to admissions for trauma and one to ambulance call out for trauma 13 papers related to fracture rate Eight papers related to RTA crash severity/fatal crashes Fig. 1. Search strategy to obtain list of papers for inclusion in final review.

948

Table 1 Final list of papers included in review. Papers using admission or ambulance call for trauma as the outcome measure (7). Paper

Population studied

Number of patients

Weather condition assessed

Outcome measure

Key results

Atherton et al. (2005) [15]

All ages; adults; paediatric; adults with proximal femoral fractures

2914 total; 2279 adults, 635 children; 693 fractured neck of femur

Daily maximum and minimum temperatures, rainfall, sunshine, presence of snow, sleet, fog, hail or snow on ground

Daily emergency admissions to inpatient beds for trauma; Leicester Royal Infirmary, UK; 1998

Bhattacharya (2001) [13]

All age groups assessed together

9408

Daily average wind speed, maximum temperature, departure from normal temperature, dew point temperature, precipitation, pressure, snowfall, relative humidity

Daily admissions for trauma; Level I trauma centre, US; 1992–1998

For total and paediatric admissions, higher maximum and minimum temperatures, more hours of sunshine and fewer mm of rainfall positively correlated with rate; maximum daily temperature is the single most important factor Adult admissions and proximal femoral fracture admissions not significantly affected by any of the weather conditions Strong positive correlation between maximum daily temperature and trauma admissions

All age groups assessed together

21,427

Daily maximum and minimum temperatures, precipitation, snowfall

Kim et al. (2012) [21]

All age groups assessed together

288,350 (395/day)

Daily mean temperature

O’Connor et al. (2012) [20]

All age groups assessed together

7308

Daily average wind speed and peak 5 s wind gust speed

Parsons et al. (2014) [12]

Adult and paediatric

59,617 total

Daily maximum and minimum temperatures, rainfall, sunshine, maximum gust speed, wind speed, presence or absence of snow/sleet/ fog

Rising et al. (2006) [14]

All age groups assessed together

8269

Maximum temperature in preceding 24 h, total precipitation in preceding 3 h, relative humidity, wind speed and presence of and snow or freezing rain or ice in preceding 6 h

Hospital discharges when admission was for trauma (trauma team activation, death on arrival or in ED or during admission, ICU admission, transfer from another hospital for trauma evaluation or length of stay >48 h); 3 Level I trauma centres, US; 2003–2007 Daily ambulance calls for all traumatic injuries, classified as severe if verbal or pain responses or unresponsiveness, systolic blood pressure less than 90 mm Hg, respiration rate less than 10 breaths per minute or more than 30 breaths per minute; 7 metropolitan areas (covering 47.1% of Korean population), South Korea; 2006–2007 Daily admissions for blunt trauma for falls, crashes and other accidents authors felt likely to have been affected by the wind, as well as need for surgery, ICU admission, need for mechanical ventilation, ventilator days, length of stay; Level I trauma centre, US; 2004–2009 Daily trauma admissions, admissions in which the patient subsequently died as a result of the injury, the patient received interhospital transfer, the patient required critical care or the inpatient stay was greater than 3 days; 21 centres across UK; 1996–2006

All admissions for trauma seen and evaluated by general surgery trauma service and admitted to hospital on the trauma service for >48 h or those who died in ED after evaluation; Level I trauma centre, US; 1996–2002

Maximum humidity associated with decrease in trauma but effect not significant when days with rainfall excluded Positive correlation between temperature and trauma admissions; maximum daily temperature is most important predictor Precipitation associated with fewer admissions and snowfall with more admissions although pattern for both less consistent than effect of temperature Rise in mean temperature associated with 0.41% increase in severe trauma but only 0.01% increase in minor trauma For all trauma: non-linear relationship between temperature rise and trauma, with a decrease in ambulance calls with increasing temperature in cold (5 8C) and hot weather (20 8C) No significant relationship between average or peak wind speed and number of trauma admissions

Increase in maximum daily temperature and number of sunshine hours associated with increased trauma admissions. Effect much more pronounced in paediatric group Adverse weather conditions more important in increasing adult trauma admissions (vs paediatric admissions). Adult trauma increases with a fall in minimum daily temperature, the presence of snow or sleet and more rainfall. Increased wind speed associated with fewer paediatric admissions Maximum daily temperature and precipitation in preceding 3 h associated with increase in trauma admissions (0.5 in. of precipitation associated with nearly 30% increase in trauma) No significant effect of relative humidity, wintery conditions or wind speed

A.M. Ali, K. Willett / Injury, Int. J. Care Injured 46 (2015) 945–953

Friede et al. (2009) [19]

Precipitation associated with fewer admissions, snow weakly associated with fewer admissions

Papers using fracture as the outcome measure (13) Chesser et al. Age >65 years 818 (2002) [22]

Maximum and minimum daily temperatures

Hip fracture excluding high velocity trauma or pathological fractures; District General Hospital, UK; 5 year period

Classification of pavement conditions as normal, slippery or very slippery pavement (taking into account temperature, precipitation, humidity and radiation) Daily maximum and minimum temperatures, depth of snow or ice, precipitation, snowfall, windspeed

Distal radius fractures; Oulu hospital, Finland; 2008

Age >16 years

285

Giladi et al. (2014) [29]

Age 65–99 years

21,507

Hove (1995) [30]

Age >20 years

600

Snow or no snow on ground, mean temperature

Distal radius fractures; Bergen, Norway; 1988

Jacobsen et al. (1995) [23]

Women age >45 years

1147

Daily presence or absence of snow, freezing rain/freezing drizzle/glaze, high wind or rain/drizzle

Neck of femur fracture (excluding those resulting from severe trauma or pathological fractures); 2 hospitals in Rochester, Minnesota; 1952–1989

Jacobsen et al. (1999) [31]

Age >35 years

1809

Daily presence or absence of snow, freezing rain/freezing drizzle/glaze, high wind or rain/drizzle

Distal radius fractures; 2 hospitals in Rochester, Minnesota; 1952–1989

Jantzen et al. (2014) [33]

All age groups considered

4892

Daily road surface temperature

Distal radius, humerus, hip and ankle fractures; Bispebjerg Hospital, Denmark; 2009–2011

Distal radius fractures; Medicare beneficiaries across entire US; 2007

Days with average temperature less than or equal to 32 8F, snow/ice on the ground at the start, of the day and freezing rain had an increased risk of fracture Increase risk of fracture on slippery days Mean number of fractures in women 3.6 times higher on days where there was snow on the ground compared to days with no snow Among women aged 45–74 years, risk of hip fracture increased significantly on days with snow or freezing rain. Rain or high wind had no significant effect. Among women >75 years, none of the weather conditions had a significant effect Among women aged 35–64 years, strong association of snow/blowing snow and freezing rain with fracture rate Among women aged >65 years, snow/blowing snow and freezing rain significantly associated with increased fracture rate but magnitude of effect less than for younger women Rate of distal radius, humeral and ankle fractures increased significantly with decreasing road surface temperature and the presence of ice alert. Rate of hip fractures showed no significant association with temperature

Levy et al. (1998) [24]

Adults age >50 years

18,455

Daily maximum temperature and amount of freezing precipitation

Hip fracture; 28 acute care hospitals in Montreal, Canada; 1982–1992

Parker and Martin (1994) [25]

Age >60 years

429

Daily presence of ground frost, air frost, minimum daily temperature

Hip fracture; District General Hospital, UK

Decreasing temperature was associated with a significant decrease in daily number of fractures for patients < 15 years, whereas patients > 30 years experienced a significant increase Lower temperatures, snow, and freezing rain associated with increased rate of hip fracture, with freezing rain having greatest risk. Warm, rainy days protective for hip fracture in both sexes Association between inclement weather and hip fracture is stronger among younger persons in both men and women. In women, strongest association is 50–64 years, no consistent association in >80s Association between day of fall and the presence of ground frost. No significant association for air frost or minimum daily temperature

A.M. Ali, K. Willett / Injury, Int. J. Care Injured 46 (2015) 945–953

Flinkkila et al. (2011) [28]

No significant effect of temperature on hip fracture rate, no increased rate when temperature below 0 8C No effect of patient characteristics on rate of fracture at different temperatures (age, sex, type of fracture, preinjury mobility, residence, functional and cognitive scores) Number of fractures was 2.5 times greater on slippery winter days compared to non-winter days and 1.4 times greater on non-slippery winter days compared to nonwinter days

949

950

Table 1 (Continued ) Paper

Population studied

Number of patients

Weather condition assessed

Outcome measure

Key results

Sinikumpu et al. (2013) [32]

Children <16 years

148

Daily maximum temperature, precipitation, maximum wind speed

Risk of fracture was 50% higher on dry days compared to rainy days Temperature and wind speed had no statistically significant effect on fracture rate

Tenias et al. (2009) [26]

Age >45 years

2121

Turner et al. (2011) [27]

Age >75 years

Hip fracture; all New South Wales, Australia; 1998–2004

Weston-Simons et al. (2011) [34]

All age groups assessed together

Not stated; catchment population of 4 million people 12,228

Maximum and minimum temperature, relative humidity, precipitation, incidence of snow, hail, storm, fog, dew, frost, duration of gales Daily temperature

Forearm shaft fractures involving both bones (excluding single-bone wrist and elbow fractures, fracture dislocations and pathological fractures); paediatric trauma centre, Oulu, Finland; 1997–2009 Hip fractures; 2 hospitals in Valencia, Spain; 1996–2005

Lower temperature significantly associated with higher fall-related hip fracture hospitalisations in >75 yearolds Referrals increased significantly on snow days. During snow days there were significant increases in the number of distal radius and ankle fractures referred but not fractured necks of femur

1.4 million fatal crashes [36]

Different fracture types (distal radius, ankle, neck of femur); Brighton Hospital, UK; April 2009–March 2010

Daily precipitation, snowfall, snow depth

Fatal and non-fatal crash data; US; 1975–2000

Negative and significant relationship between monthly precipitation and monthly fatal crashes. However, in the daily level analysis, a strong positive relationship is estimated, as in prior studies. The source of the contrasting results appears to be a substantial negative lagged effect of precipitation across days within a state– month i.e. fewer crashes on days following rainy days

Eisenberg (2005) [36]

All road user groups considered together

1.4 million fatal crashes

Daily snowfall and rain

Fatal and non-fatal crashes; US; 1975–2000

Jung et al. (2010) [37]

All road users considered together

255 crashes in rainy weather

Weather condition in 15 min before time of each crash

Crashes graded by severity; section of highway, southeastern Wisconsin, US; 2004–2006

The risk imposed by precipitation increases dramatically as the time since last precipitation increases Snow days had fewer fatal crashes than dry days but more nonfatal-injury crashes and property damage only crashes. The first snowy day of the year was substantially more dangerous than other snow days in terms of fatalities particularly for elderly drivers The most severe crashes were more likely to occur as rainfall intensity for 15 min was getting stronger

Crash fatalities (30 day survival); 12 counties across Florida, US; 2000–2006

Moderately strong wind speed is likely to decrease the fatal and incapacitating injury crashes Independent predictors for driver death: foggy and cloudy weather

Kim et al. (2012) [38]

Morgan and Mannering (2011) [39]

All road users considered together

All road users considered together

3,468,326 crashes

3157 minor injuries; 261 severe injuries

Weather condition at time of crash

Weather conditions specified at time of crash

Crash data with severity; Indiana, US; 2007– 2008

Significant protective factor was rainy vs clear weather conditions Average severe-injury probabilities of most driver groups increased by more than 100% under adverse surface conditions relative to dry-surface conditions. The exception is male drivers under 45 years old, which actually had their severe injury probabilities decrease by 17% on wet surfaces and 42% on snow/ice surfaces relative to dry-surface severe-injury probabilities

A.M. Ali, K. Willett / Injury, Int. J. Care Injured 46 (2015) 945–953

Papers relating to Road Traffic Accidents (8) Eisenberg (2004) All road user [35] groups considered together

Daily snow vs no snow

Windier days strongly associated with more fractures especially in patients under 75. The remaining meteorological variables were not associated in any significant fashion with the fracture incidence

Crash data with severity analysis; section of freeway in Colorado, USA; 2007–2011 Weather conditions at time of crash All road users considered together Yu et al. (2014) [42]

670 crashes

Crashes leading to seriously injured patient; China; 2000–2001 Weather conditions at time of crash 3365 severe crashes All road users considered together Qin et al. (2004) [41]

Rainfall data for each crash All road users considered together Pei et al. (2012) [40]

347 crashes; 59 crashes with seriously injured or killed casualty

Crash injuries with severity analysis; Hong Kong; July–September 2009

Lower temperature during snow seasons increases the likelihood of severe crashes

Rainfall positively related to likelihood of crash occurring, negatively related to likelihood of crash resulting in a casualty who was seriously injured or killed Most extremely severe RTAs occurred in fine weather days and in the daytime. Compared with other RTAs, extremely severe RTAs were more likely to happen under following conditions: on cloudy, snowing, misty and blustering days Low temperature increases the probability of severe crashes

A.M. Ali, K. Willett / Injury, Int. J. Care Injured 46 (2015) 945–953

951

trauma admissions and that snowfall had a very weak negative correlation (R = 0.0616, p < 0.00001); precipitation of one inch or more was associated with a 10% drop in admissions although notably no significant change in motor vehicle collisions, and snowfall of two inches or more was associated with a 12.8% drop in admissions [13]. However, Rising et al. found that an additional 0.5 in. of precipitation in the preceding 3 h was associated with a 30% increase in trauma [14]. Parsons et al. found that adverse weather conditions were important for adults in a manner not seen in children, with adult trauma admissions increasing by 2.2% for each additional 10 mm of rainfall, by 7.9% in the presence of snow or sleet and by 3.2% for each drop of 5 8C in the minimum daily temperature [12]. In a study of 7308 patients from the Midwest region of the USA that looked specifically at the effect of wind speed on trauma, O’Connor et al. reported no significant relationship between either average or peak wind speed and number of trauma admissions per day [20]. In addition, there was no significant correlation between average or peak wind speed and Injury Severity Score, Glasgow Coma Score or Length of Stay [20]. However, Parsons et al. showed that a 20 mph increase in mean wind speed was associated with a 13% decrease in the incidence of paediatric trauma [12]. Papers using fracture as the outcome measure (13) Thirteen papers used fracture as the outcome measure of interest, either hip fracture [22–27], distal radius fracture [28–32] or a combination of fracture types including, hip, distal radius, ankle and humeral fractures [33,34]. The majority of papers show that temperature has no significant effect on hip fracture rate, although one study from Australia showed that lower temperature was significantly associated with a higher rate of fall-related hip fracture hospitalisations in over 75 year-olds [27]. The effects of snow are unclear and appear to be dependent upon location. In a study from Minnesota, USA, Jacobsen et al. found that among women aged 45–74 years, the risk of hip fracture was increased on days with snow (RR = 1.41) or freezing rain (RR = 1.82), although among women aged 75 years and older, ice and snow were not strongly related to fracture occurrence [23]. In a study from Montreal, Canada, Levy et al. found that, for women, the strongest association between inclement weather and fracture rate was in the 50–64 years group, but in women over the age of 80 there was no consistent association; for men, inclement weather was associated with increased rates of hip fracture at older ages [24]. However, studies from the UK and Spain showed no significant effect of snow on hip fracture rate [26,34]. In a multicentre Spanish study including 2121 patients, Tenias et al. found that windier days were strongly associated with more fractures especially in patients under the age of 75 [26]. For distal radius fractures, all papers found an increase in fracture rate on snowy days and on days with slippery road conditions. The largest study in the literature by far is from Giladi et al. who analysed 21,507 distal radius fractures amongst 65–99 year old Medicare beneficiaries from across the USA [29]. The group found an increased risk of fracture in days with an average temperature less than or equal to 32 8F (IRR = 1.36; p < 0.001), snow/ice on the ground at the start of the day (IRR = 1.45; p < 0.001), and freezing rain (IRR, 1.24; p = 0.025). Jacobsen et al. found that, amongst middle-aged women in Rochester, USA, the influence of weather on the risk of forearm fractures and hip fractures was very similar [31]. In women aged 35–64 years, freezing rain, freezing drizzle or glaze ice was associated with a 63% increase in the risk of distal forearm fractures and days with snow or blowing snow with a 44% increase; the corresponding values for increased risk of hip fracture

952

A.M. Ali, K. Willett / Injury, Int. J. Care Injured 46 (2015) 945–953

amongst women aged 45–74 years under these conditions was with 60% and 22%, respectively [23,31]. The association between distal radius fracture risk and weather was also found to be stronger in younger women. In a study from Finland assessing distal radius fractures in children, fracture risk was found to be 50% higher on dry days compared to rainy days. In a Danish study of 4892 patients, the daily number of distal radius, humeral and ankle fractures increased significantly with decreasing road surface temperature and the presence of ice alert, although there was no significant association between temperature and hip fracture incidence [33]. Whereas decreasing temperature was associated with a significant decrease in the daily number of fractures for patients <15 years, those patients >30 years experienced a significant increase. In a UK study assessing ankle fractures there was a significant increase in the fracture rate on snowy days [34]. Papers relating to RTAs (8) Eight papers assessed the effect of the weather on serious or fatal RTAs [35–42]. In a large study from the USA covering 1.4 million crashes from 1975 to 2000, snow days had fewer fatal crashes than dry days (IRR = 0.93) but more nonfatal-injury crashes (IRR = 1.23) and property damage-only crashes (IRR = 1.45) [36]. Moreover, the first snowy day of the year was found to be substantially more dangerous than other snow days in terms of fatalities (IRR = 1.14), particularly for elderly drivers (IRR = 1.34). Morgan and Mannering showed that, in male drivers under 45 years, the severe injury probabilities decrease by 17% on wet surfaces and 42% on snow/ice surfaces relative to dry surface severe-injury probabilities [39]. Yu et al. found that severe crashes are less likely to occur in snowy conditions but more likely to occur at colder temperatures [42]. Pei et al. found that rainfall is also positively related to the likelihood of a crash occurring but negatively related to likelihood of a seriously injured or killed casualty [40]. In a study of over three million RTAs in the USA state of Florida, Kim et al. found that foggy and cloudy conditions significantly increased the risk of a fatal crash compared to a non-fatal crash whereas rainy conditions had the reverse effect [38]. Jung et al. showed that the most severe crashes were more likely to occur as rainfall intensity for 15 min was getting stronger [37]. Discussion This paper is a systematic review that seeks to assess the strength of the evidence supporting an association between weather conditions and serious trauma. The existing literature strongly supports the view that increased temperature is positively correlated with trauma admissions, the rate of distal radius fractures is more sensitive to adverse weather conditions than the rate of hip fractures, and that paediatric trauma, both in respect of trauma admissions and fracture rate, is more sensitive to the weather than adult trauma. Adverse weather conditions influence RTA frequency and severity, but the nature of the relationship appears critically dependent upon the timecourse of the weather event and the population being studied. Adverse weather in itself, particularly snow, may produce a tendency towards lower crash severity, but the onset of snowfall or a rainy downpour appears to be a particularly hazardous time. Important methodological differences between papers make further generalisations difficult, and our analysis has revealed a number of areas that future work might address. The majority of papers used a daily average of a specified weather condition and correlated this with the total number of admissions, fractures or RTAs occurring on the same day. However,

those that took an event-based approach and examined weather conditions just prior to each event found important temporal effects, with an episode of adverse weather prompting the greatest increase in accident frequency shortly after that episode had begun [14,35,37]. Another feature of the weather that most papers failed to recognise in their analysis is the lag effect; snowfall or rain on a particular day may influence injury rate not only on that day but on subsequent days, either directly through the snow or rain remaining or through behavioural effects [35,40]. Quantification of the type of adverse weather differed, with some papers defining the condition as present or absent, others using threshold values (e.g. greater than one inch) and others using linear regression modelling. Few papers provided adequate multivariate analysis that corrected for other weather conditions when presenting data. It is difficult to interpret, for example, the extent to which snowfall alters fracture risk above and beyond the effect of pre-existing ground frost or cold temperature. Papers that assessed the effect of both weather conditions and seasonality on rates of trauma generally supported the view that seasonal variation cannot be entirely explained by changes in the weather [23,24]. Indeed, in a study in the USA (Northern Hemisphere), Bhattarcharyya et al. showed that the months of July and August remain significant predictors of trauma volume even when the effect of maximum daily temperature is controlled for by logistic regression [13]. Thus, analysis of the weather as an independent variable is important. Studies included in this review related to serious trauma only as many cases of ‘trauma’ attending the ED or prompting an ambulance call would not be resource-demanding. Indeed, in a large Japanese study looking at ambulance transport for all cases of trauma in 226,339 patients, only 22.5% of calls resulted in admission to hospital [43]. A large number of studies relating to the effect of the weather on trauma were therefore excluded from our analysis as they did not provide subgroup analysis of serious injuries. The most rigorous papers provided a clear definition of severe trauma as admission with one or more of ICU admission, hospital stay >48 h or death [12,14,19]. To build an effective economic and political case for planning trauma services based upon weather forecasts, or expending resources on accident prevention, the impact of adverse weather conditions on patient outcomes is needed. In some of the papers related to RTAs this analysis was present: Eisenberg et al., for example, estimate that across the US around 30 fatal crashes and 600 non-fatal-injury crashes might be avoided on the first 3 snow days of the year if drivers were more prepared [36]. In those papers related to trauma admissions, such analysis was missing. Another difficulty in extrapolating the results of local studies is that the response of a particular population to the weather is likely to be highly context-specific. For example, it would not be valid to compare those studies assessing the effect of temperature or snowfall in European countries that have temperate climates and relatively mild winters to their effect in parts of North America where the population is more accustomed to harsh winters. This is illustrated by the effect of snowfall on hip fracture rate. Whereas studies from the UK and Spain found no significant effect of snow on hip fracture rate [26,34], studies from Minnesota, US, and Montreal, Canada, found that snowy conditions increase the risk of hip fracture [23,24]. This may be explained by behavioural differences that mean that in colder climates it is the younger people who are more likely to venture out in the snow. Indeed, all these studies suggest that adverse weather conditions have less impact on the elderly presumably because they limit outside activity regardless. The impact of the weather on trauma incidence may be from direct effects such as slipping on ice or collisions in fog, or behavioural changes such as driving more cautiously on snowy roads and being more active outdoor in warmer weather. This in

A.M. Ali, K. Willett / Injury, Int. J. Care Injured 46 (2015) 945–953

turn may be influenced by the familiarity of different populations to weather conditions (e.g. car tyres, footwear) and the preparedness of local infrastructure such as roads and hospitals for dealing with such conditions. With such a broad range of context-specific considerations, the quantification of the effect of adverse weather on trauma workload will be dependent on local data. There is surprisingly little comparative data in the literature on the effect of a specified weather condition on trauma in different localities to help draw out some of the key differences that may influence this relationship. Even in the multi-centre studies, data from different regions were aggregated rather than compared. The effects of climate change in altering weather conditions and the centralisation of trauma care means understanding the impact of the weather on trauma is more important than ever before [44]. Whilst the current literature provides useful direction, prospective, event-based, comparative studies are needed to help discern if and how trauma services should be planned according to weather forecasts. Moreover, there are no studies reported from the developing world, a missed opportunity to identify and help prevent trauma given the exponential increase in road traffic use in many developing countries. Conflict of interest statement No conflicts of interest required. Funding No funding required. Acknowledgements The authors would like to thank Owen Coxall of the Bodleian Healthcare Cairns Library for his technical support. References [1] http://www.metoffice.gov.uk/about-us/who/accuracy/forecasts [accessed 28.02.15]. [2] Simmons AJ, Hollingsworth A. Some aspects of the improvement in skill of numerical weather prediction. Q J Roy Meteorol Soc 2002;128:647–77. [3] Shiue I, Muthers S, Bearman N. The role of cold stress in predicting extra cardiovascular and respiratory admissions. Int J Cardiol 2014;172:e109–10. [4] Rocklo¨v J, Forsberg B, Ebi K, Bellander T. Susceptibility to mortality related to temperature and heat and cold wave duration in the population of Stockholm County, Sweden. Glob Health Action 2014;7:22737. [5] Chau PH, Wong M, Woo J. Ischemic heart disease hospitalization among older people in a subtropical city – Hong Kong: does winter have a greater impact than summer? Int J Environ Res Public Health 2014;11:3845–58. [6] Met Office Health Forecasting Unit. Forecasting the nation’s health: an evaluation by the forecasting unit, Met Office. London: Met Office; 2001. [7] White C. Weather reports to be used to forecast NHS workload. BMJ 2001; 323:251. [8] Stulec I. On weather sensitivity in retail industry. Int J Retail Manage Res 2013;3:1–10. [9] Harper PR, Minty J, Sahu S, Baffour B. MetSim: a simulation decision support tool using meteorological information for short-term planning of hospital services. In: Presented at: SIMULTECH 2012-2nd International Conference on Simulation and Modeling Methodologies, Technologies and Applications; July 2012. Available at: http://orca.cf.ac.uk/38794/ [accessed 28.02.15]. [10] Marno P, Chalder M, Laing-Morton T, Levy M, Sachon P, Halpin D. Can a health forecasting service offer COPD patients a novel way to manage their condition? J Health Serv Res Policy 2010;15:150–5. [11] Metcalfe D, Bouamra O, Parsons NR, Aletrari MO, Lecky FE, Costa ML. Effect of regional trauma centralization on volume, injury severity and outcomes of injured patients admitted to trauma centres. Br J Surg 2014;101:959–64. [12] Parsons N, Odumenya M, Edwards A, Lecky F, Pattison G. Modelling the effects of the weather on admissions to UK trauma units: a cross-sectional study. Emerg Med J 2011;28:851–5. [13] Bhattacharyya T, Millham FH. Relationship between weather and seasonal factors and trauma admission volume at a level I trauma center. J Trauma 2001;51:118–22.

953

[14] Rising WR, O‘Daniel JA, Roberts CS. Correlating weather and trauma admissions at a level I trauma center. J Trauma 2006;60:1096–100. [15] Atherton WG, Harper WM, Abrams KR. A year’s trauma admissions and the effect of the weather. Injury 2005;36:40–6. [16] National Audit Office, England. Major trauma care in England; 2010, Available at: http://www.nao.org.uk/wp-content/uploads/2010/02/0910213.pdf [accessed 28.02.15]. [17] World Health Organisation. Youth and road safety; 2007, Available at: http:// www.who.int/mediacentre/news/releases/2007/pr17/en/ [accessed 28.02.15]. [18] Moher D, Liberati A, Tetzlaff J, Altman DG, PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ 2009;339:b2535. [19] Friede KA, Osborne MC, Erickson DJ, Roesler JS, Azam A, Croston JK, et al. Predicting trauma admissions: the effect of weather, weekday, and other variables. Minn Med 2009;92:47–9. [20] O‘Connor ZJ, Helmer SD, Yates CL, Ward JG, Khandelwal A, Haan JM. An evaluation of the effect of wind speed on trauma admissions and injury severity at a midwest level I traumacenter. Am Surg 2012;78:614–6. [21] Kim Y, Kim H, Shin SD, Hong YC. Different influence of outdoor temperature on traumatic and nontraumatic injuries. J Trauma Acute Care Surg 2012;73: 944–9. [22] Chesser TJ, Howlett I, Ward AJ, Pounsford JC. The influence of outside temperature and season on the incidence of hip fractures in patients over the age of 65. Age Ageing 2002;31:343–8. [23] Jacobsen SJ, Sargent DJ, Atkinson EJ, O’Fallon WM, Melton 3rd LJ. Populationbased study of the contribution of weather to hip fracture seasonality. Am J Epidemiol 1995;141:79–83. [24] Levy AR, Bensimon DR, Mayo NE, Leighton HG. Inclement weather and the risk of hip fracture. Epidemiology 1998;9:172–7. [25] Parker MJ, Martin S. Falls, hip fractures and the weather. Eur J Epidemiol 1994;10:441–2. ˜ iguez C, Ballester F. Short-term [26] Tenı´as JM, Estarlich M, Fuentes-Leonarte V, In relationship between meteorological variables and hip fractures: an analysis carried out in a health area of the Autonomous Region of Valencia, Spain (1996–2005). Bone 2009;45:794–8. [27] Turner RM, Hayen A, Dunsmuir WT, Finch CF. Air temperature and the incidence of fall-related hip fracture hospitalisations in older people. Osteoporos Int 2011;22:1183–9. [28] Flinkkila¨ T, Sirnio¨ K, Hippi M, Hartonen S, Ruuhela R, Ohtonen P, et al. Epidemiology and seasonal variation of distal radius fractures in Oulu, Finland. Osteoporos Int 2011;22:2307–12. [29] Giladi AM, Shauver MJ, Ho A, Zhong L, Kim HM, Chung KC. Variation in the incidence of distal radius fractures in the U.S. elderly as related to slippery weather conditions. Plast Reconstr Surg 2014;133:321–32. [30] Hove LM, Fjeldsgaard K, Reitan R, Skjeie R, So¨rensen FK. Fractures of the distal radius in a Norwegian city. Scand J Plast Reconstr Surg Hand Surg 1995;29: 263–7. [31] Jacobsen SJ, Sargent DJ, Atkinson EJ, O’Fallon WM, Melton 3rd LJ. Contribution of weather to the seasonality of distal forearm fractures: a population-based study in Rochester, Minnesota. Osteoporos Int 1999;9:254–9. [32] Sinikumpu JJ, Pokka T, Sirnio¨ K, Ruuhela R, Serlo W. Population-based research on the relationship between summer weather and paediatric forearm shaft fractures. Injury 2013;44:1569–73. [33] Jantzen C, Jørgensen HL, Thomsen MT, Riis T, Sommer B, Duus BR, et al. Daily number of fractures is associated with road temperature in an urban area. Dan Med J 2014;61:A4794. [34] Weston-Simons J, Jack CM, Doctor C, Brogan K, Reed D, Ricketts D. The impact of snow on orthopaedic trauma referrals. Injury 2012;43:1033–6. [35] Eisenberg D. The mixed effects of precipitation on traffic crashes. Accid Anal Prev 2004;36:637–47. [36] Eisenberg D, Warner KE. Effects of snowfalls on motor vehicle collisions, injuries, and fatalities. Am J Public Health 2005;95:120–4. [37] Jung S, Qin X, Noyce DA. Rainfall effect on single-vehicle crash severities using polychotomous response models. Accid Anal Prev 2010;42:213–24. [38] Kim T, Rivara FP, Mozingo DW, Lottenberg L, Harris ZB, Casella G, et al. A regionalised strategy for improving motor vehicle related highway driver deaths using a weighted averages method. Inj Prev 2012;18:16–21. [39] Morgan A, Mannering FL. The effects of road-surface conditions, age, and gender on driver-injury severities. Accid Anal Prev 2011;43:1852–63. [40] Pei X, Wong SC, Sze NN. The roles of exposure and speed in road safety analysis. Accid Anal Prev 2012;48:464–71. [41] Qin HL, Zhao XC, Zhou JH, Qiu J, Yang ZL, Jiang ZQ, et al. Effect of environment on extremely severe road traffic crashes: retrospective epidemic analysis during 2000–2001. Chin J Traumatol 2004;7:323–9. [42] Yu R, Abdel-Aty M. Analyzing crash injury severity for a mountainous freeway incorporating real-time traffic and weather data. Saf Sci 2014;63:50–6. [43] Abe T, Tokuda Y, Ohde S, Ishimatsu S, Nakamura T, Birrer RB. The influence of meteorological factors on the occurrence of trauma and motor vehicle collisions in Tokyo. Emerg Med J 2008;25:769–72. [44] Parry ML, Canziani OF, Palutikof JP, van der Linden PJ, Hanson CE, editors. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge, United Kingdom and New York, NY, USA: Cambridge University Press; 2007.