METHODS

Data were obtained from the National Health Interview Survey (NHIS), a face-to-face household survey conducted yearly by the National Center for Health Statistics. Data are collected from a nationally representative sample of the civilian, non-institutionalized population residing in the United States.¹⁷ The NHIS was redesigned in 1997 and began using computer assisted personal interviewing. The Injury Section of the redesigned NHIS contains questions about both injury and poisoning. Data are collected from an adult member of the family (18 years or older) regarding all medically attended injury and poisoning episodes occurring in the previous three months for each person in the family. A parent or guardian responds for children under age 18. The 1997–99 NHIS collected detailed information about the four most recent injury episodes for each family member.

This study examines all medically attended SR injury episodes reported in a verbatim response to an Injury Section question. The injury episodes were defined as any traumatic event during the past three months that caused an injury or injuries from an external cause (for example, colliding, tripping, or falling) where (1) the injured person was participating in a SR activity and (2) the injury required treatment by a health care professional, either in person or by telephone. We included only episodes involving organized and unorganized sports (for example, basketball, football) or recreational activities (for example, biking, jogging, exercising); we excluded any leisure activities (for example, playing in yard, climbing a tree, gardening, rough housing). The injury diagnosis is the physical harm caused by the traumatic event, such as a sprain or fracture. An injury episode can have multiple injury diagnoses for the same person. Analysis by external cause of injury was based on the first listed mechanism of injury as determined by the International Classification of Diseases, Ninth revision, Clinical Modification (ICD-9-CM) external cause of injury codes (E codes).¹⁸

All episodes that included at least one health condition classified as an injury using the ICD-9-CM diagnosis codes 800–959 and 990–999 were included. Because children aged 4 years and younger do not routinely participate in SR activities in the same way that older children and adults do, more in-depth analysis of the data excluded this young age group.

An unweighted total of 1762 SR injury episodes were recorded for the three month recall period from 1997 through 1999. During the NHIS interview, participants were asked if they sustained injuries during the past three months that warranted medical attention (that is, advice including phone contact with a health care provider, or treatment). If the response was “yes”, the respondent was asked a series of questions from the Injury Section of the NHIS survey. To categorize their activity at time of injury, respondents were asked a question on what they were doing when the injury episode occurred. Potential answers included: sports, leisure, unpaid work, or driving. They were asked to describe how the injury event happened and the answer was recorded as a verbatim response. In 1664 of these narratives, a sport or recreational activity was specified. Since ICD codes do not generally capture activity,¹⁹ we developed and applied a classification scheme for SR activities based on the International Classification of External Causes of Injury (ICECI).¹⁹ Two researchers independently classified each injury episode, and a third researcher reviewed these classifications and helped resolve any inconsistencies. For 98 injury episodes where narratives lacked details, activity was classified as “unspecified sport”.

For comparative purposes, SR categories reported in this study were combined to match those used in a recent study of SR emergency department visits using data from the National Hospital Ambulatory Medical Care Survey (NHAMCS).⁷

Annualized national estimates were computed by multiplying three month injury episode weights by four. Person weights used to calculate annualized weights accounted for the complex sample design and sampling ratios, were adjusted for non-response, and were post-stratified to the US population based on the US Bureau of Census totals for sex, age, and race/ethnicity for the year of the study.^20,21 Since NHIS interviews were equally distributed across all four quarters of the year, no adjustment was made for seasonality.

The Barell injury diagnosis matrix²² of ICD-9-CM diagnosis codes was used to categorize episode related injury diagnoses by body part affected and type of diagnosis. The recommended framework for presenting injury data based on ICD-9-CM E codes was used to categorize injury episodes by mechanism of injury.²³

Age adjusted rates were calculated using the year 2000 standard population.²⁴ PROC RATIO of SUDAAN version 8.0.0 was used to obtain the estimated annualized rate of injury episodes per 1000 persons and the associated standard error.²⁵ Annualized estimates and rates per 1000 population were considered unstable if the coefficient of variation was 30% or greater or if they were computed by using fewer than 20 unweighted cases for the three month recall period. Unstable estimates and rates are noted in the tables with a symbol.

RESULTS

From 1997 through 1999, Americans sustained an average annual estimate of 7.0 million SR injury episodes or 25.9 injury episodes per 1000 persons (95% confidence interval 24.4 to 27.4 per 1000 persons) (table 2). Sixty four percent of these SR injury episodes occurred among persons ages 5–24 years. Injury episode rates were highest among the 5–14 year old (59.3 per 1000 persons) and the 15–24 year old age groups (56.4 per 1000 persons). Males sustained 68% of all SR injury episodes (36.3 per 1000 persons), more than twice the rate for females (16.0 per 1000 population). SR injury episode rates were consistently lower for females than for males for all age groups except for those aged 4 years or younger where the rates were similar for both sexes. The crude rate for females was highest in the 5–14 age group, while the crude rate for males was highest for the 15–24 age group. The age adjusted rate of injury events among whites (28.8 per 1000 population) was 1.5 times higher than that of black people (19.0 per 1000 population). Most SR injury episodes occurred at a sports facility (30.7%), at school (19.7%), or around the home (16.5%). Three percent (0.8 per 1000 persons) of SR injury episodes required a hospital stay. Twenty percent of persons attending school lost one or more days resulting from their injury episode and 27.6% of those working lost one or more days for their injury event, and 3% (0.8 per 1000 population) required help with personal care for six months or longer.

Further analysis focused on persons aged 5 years and older. For persons aged 5 years and older, an estimated 6.8 million or 27.2 episodes per 1000 persons occurred annually (table 3). Basketball was the most frequently mentioned SR activity (14.4%), accounting for about four injury episodes per 1000 persons. Pedal cycling, recreational sports, exercising, and football were the next most common activities accounting for 36.6% of injury episodes. Males were most commonly injured while engaging in basketball, football, or pedal cycling activities, which accounted for 40.1% of all SR injury episodes. Among females, about 32% of all SR injury episodes occurred during exercising, gymnastics/cheerleading, or basketball.

The type of SR activity performed when the injury episode occurred also varied by age group (table 4). Pedal cycling was the most common SR activity associated with injury episodes among children aged 5–14 years. Basketball was the most frequently played sport among persons aged 15–24 years who sustained SR injuries. Persons 25 years of age and older were most frequently injured while participating in a recreational sport.

A majority of injury episodes from SR activities were to the upper (31.2%) and lower (38.9%) extremities (table 5). For those that occurred to the lower extremities, 33.6% were to the lower leg or ankle. The hands, wrists, or fingers were affected in 51.5% of the injuries to the upper extremities and another 26.4% of injuries were to the shoulder or upper arm. An estimated 1.1 million SR episode related injuries involve the head or neck region, of which 17.2% were internal head injuries.

Strains and sprains were the most frequent injury diagnosis (31.5%) followed by fractures (22%). Among those with head or neck injuries, the most frequent injury diagnosis was an open wound (56.1%). Among persons with extremity injuries, strains/sprains, and fractures accounted for 69.4% of injuries. The most common mechanisms of injury were struck by/against, fall, and overexertion accounting for 75% of the SR injury episodes.

For SR injuries for persons aged 5–24 years, annualized national estimates of injury episodes reported in NHIS were 41.7% higher than national estimates of injury related emergency department visits based on data from NHAMCS (table 6). The percentage difference varied by type of activity ranging from 81.9 for recreational sport injuries to almost no difference for pedal cycling and water sports.

DISCUSSION

This study is among the first to examine all medically attended injuries from SR activities across all age groups in the US. The rate of 25.9 injury episodes per 1000 persons associated with SR activities was greater than the rate for transportation related injury episodes in 1997.¹⁷ Our results highlight the number of SR injury episodes where the injured person received medical attention beyond treatment in a hospital emergency department and demonstrate the importance of broad surveillance and prevention activities. For persons aged 5–24 years, an estimated 10.7% of all injury related emergency department visits were for SR activities based on NHAMCS data⁷ compared with 22.1% of all SR injury episodes in our study. Our study indicates that about 42% of persons aged 5–24 years receiving medical attention for an SR injury did not go to an emergency department. Also, the annualized estimate of 7 million injury episodes in our study is 1.6 times higher than the 4.3 million injuries treated in US hospital emergency departments based on the National Electronic Injury Surveillance System All Injury Program.¹ One fifth of schoolchildren lost one or more days of school because of their injury or injuries, and more than one quarter of working adults lost one or more work days due to the injury event. This extent of lost activity emphasizes the burden that SR injuries place on our society and the health care system.

Our study findings suggest that prevention programs aimed at reducing SR injuries through targeting high risk activities, places of occurrence, risk behaviors,²⁶ and use of protective devices need to go beyond focusing on children and also consider physically active adults. In this study, two thirds of injuries occurred among 5–24 year olds. However, one quarter of the injuries occurred among 25–44 year olds. The common activities involved with injury episodes in the latter age group include recreational sports (for example, racket sports, golf, bowling) and exercising (for example, jogging, walking, aerobics). The participants greater than 25 years are not often targeted in injury prevention efforts. For instance, no state and only 29 localities currently have bicycle helmet laws covering adults²⁷ despite the Healthy People 2010 objective supporting such laws.²⁸

Half of the injuries reported in this study occurred at a school or sports facility. Policies and programs to promote the use of effective interventions such as protective gear (for example, appropriate gear in team sports, racquetball eye protection) and enforcement of rules of play designed for safety (for example, prohibiting slide tackles from behind in soccer or spearing and clipping in football) need to be implemented and evaluated to determine if these policies can bring about a decrease in injuries at these common sporting sites.²⁸

Bicycling and exercising (which includes walking and jogging) were common activities associated with 19% of the injuries reported. These activities require safe outdoor environments. Specific strategies have been developed for use by policy makers, educators, advocates, transportation experts, health and injury professionals, and others with an interest in promoting safe bicycling.²⁹ These strategies should be actively promoted, implemented, and evaluated.

Our national estimates were consistently higher than those for emergency department visits using NHAMCS emergency department data for all sports except pedal cycling and water sports. Higher estimates using NHIS data may be expected because some injured persons seek treatment and medical advice at health care facilities other than hospital emergency departments. However, the differences may partially be the result of using different methodologies to obtain the data; the NHAMCS data are abstracted from medical records from a national probability sample of hospital emergency department visits; the NHIS data are from a national household survey.

Our study reports a SR injury rate of 59.3 injury episodes per 1000 children among 5–14 year olds. A recently published study of NHIS data about leisure and SR activities among 6–17 year olds found an estimated 91.2 injury episodes per 1000 children.⁸ The previous study used a broader definition that included leisure activities and a different age range; however, their findings further illustrate the burden on the health care system from injuries sustained from activities associated with a healthy lifestyle.

The ICD codes have traditionally been used to classify injury incidents treated in health care settings.¹⁸ However, these codes do not routinely capture the activity of a person at the time of injury. Injury professionals, under the auspices of the World Health Organization, have developed an improved tool for characterizing injury information—the ICECI classification scheme.¹⁹ The ICECI helps researchers and prevention practitioners better describe the types of injuries and the circumstances surrounding the injury incident. We were able to classify all but 98 (5.6%) of the 1762 recorded narratives as a particular SR activity by using the ICECI classification scheme. The remaining 98 lacked specificity in the narrative description of the incident.

The Barell injury diagnosis matrix is a useful tool for presenting uniform data by nature of injury and body part affected.²² This matrix was easily adapted to categorize ICD-9-CM diagnosis codes used to classify type of diagnosis and body part affected for NHIS cases to provide national estimates.

This study has several limitations. We are unable to compare risks among different activities because the NHIS data lack exposure information. To determine the actual risk of a SR injury event, information about each participant’s exposure to the particular SR activity is needed.³⁰ For instance, in our study, four injury episodes per 1000 population resulted from playing basketball. However, information on how often basketball was played either in practice, in competition, or for recreation for injured persons was not obtained during the household interview. Basketball may be a higher risk activity, have more participants, or have participants who play more often or for longer periods than do other activities.

Information obtained retrospectively from proxy reports may be subject to recall bias.³¹ Another limitation is the possible effect of seasonality associated with a three month recall period. Since household interviews in NHIS are conducted year round, multiplying sample weights by four to compute annualized estimates seems reasonable for most SR activities, but may overestimate injury episodes of certain seasonal sports, such as snow skiing.

IMPLICATIONS FOR PREVENTION

The incidence of SR injuries is substantial, particularly among persons ages 5–24 years. Data from our study suggest that many people are commonly treated for SR injuries in health care facilities other than hospital emergency departments, including clinics and doctors’ offices. These settings provide the opportunity to educate participants about injury prevention in SR activities.^32,33 As health professionals continue to promote physical activity and Americans remain active as adults, further injury prevention efforts need to be promoted and evaluated.^28,29,34 Policies and programs to enforce the use of appropriate safety gear at schools and sports facilities, active community effort focusing on safe outdoor recreation (for example, promotion of bike paths or lanes, sidewalks, and cross walks), and laws that promote safety in SR activities for persons of all ages (for example, requiring all bicycle riders to wear helmets) are just examples of possible prevention strategies. Additionally, development of a standard instrument to capture exposure information along with SR injuries in population based surveys, such as NHIS, are needed. These data can be used to compute injury (or injury episode) rates per exposure time to compare risks of injury among activities, to help set priorities and to develop further recommendations for SR injury prevention research and programs.