Bleeding Injuries in Professional Football: Estimating the Risk for HIV Transmission

Methods

Participants and Data Sources

A sample of 11 of the 28 professional football teams of the National Football League (NFL) was observed during 155 regular season games from September through December 1992. The observing physicians, who recorded the bleeding injuries, were either the team's own physicians or selected physicians who worked closely with the certified athletic trainers. A standard data collection instrument was developed, and the physicians were trained to complete it. The physicians recorded the number and types of bleeding injuries (abrasions or lacerations), the number of players who sustained bleeding injuries during the games, and various environmental and athletic factors, such as the date of the game, the playing surface, the type of stadium, the game location, the outcome of the game, and the margin of victory or loss. The physicians were stationed on the sidelines and accompanied the teams into the locker and training rooms. They recorded every bleeding injury they observed, regardless of whether the injured player sought treatment, in addition to those treated and recorded by the team's trainers.

Statistical Analysis

We used the results of chi-square and Student t-tests, and odds ratios with 95% confidence intervals, to determine associations between dependent variables (bleeding injuries and players with bleeding injuries) and independent environmental and athletic variables [8]. We then used a forward and, separately, a backward model for stepwise regression analysis of all significant associations [9].

Estimation of Risk for HIV Transmission

Specific data on the risks for transmission of HIV during athletic competition are lacking. We developed a straightforward model for estimating the probability of HIV transmission during a single contact between two randomly chosen professional football players in one game (one “game contact”). We defined this probability to be the product of four factors (probability of HIV transmission = prevalence of persons infected with HIV x risk for percutaneous HIV transmission x risk for laceration in an opponent x risk for any bleeding injury per game for each player).

Because the prevalence of HIV infection among athletes is not known, we used the HIV seroprevalence rate for college men, cited as 1 per 200 [10], as a surrogate for HIV seroprevalence among professional football players. Although some professional football players undoubtedly participate in risk behaviors for HIV, whether such behaviors occur at a frequency similar to that among college men is unknown. However, we believe that the seroprevalence rate among college men is a reasonable approximation of the rate among professional football players because almost all players attended college and many are still near college age.

Secondly, we assumed that a laceration during athletic competition carried no more risk for HIV transmission than did a percutaneous injury occurring in a health care setting when the source of the injury was documented to be the blood of a patient infected with HIV; the latter carries an estimated risk of 1 transmission per 300 percutaneous injuries [11-13]. Although this type of exposure may not be ideally analogous, it is the best estimate available of the risk for HIV infection due to blood contact with nonintact skin.

Results

Roughly equal numbers of the 155 observed games were played early or late in the season, on grass or on an artificial playing surface, and at the home facility or at the opposing team's facility: Sixty-eight games (44%) were played in September or October, 81 games (53%) were played on artificial surfaces, and 85 games (55%) were held at the home facility of the team surveyed. Forty-one games (26%) were played in domed stadiums. The teams in this study won 82 (53%) of the games; the margin of victory or loss of the teams surveyed was more than 10 points in 79 games (51%). Eighty-six games (55%) involved a surveyed team whose season-end record during the 1992-1993 football season included more wins than losses (>0.500).

Bleeding Injuries

Of the 575 bleeding injuries observed, 503 (87.5%) were abrasions and 72 (12.5%) were lacerations. Games with at least 1 bleeding injury were more frequent among games played on artificial surfaces (77 of 81 games) than among those played on grass (65 of 74 games) (odds ratio, 2.67); in domed stadiums (39 of 41 games) than in open-air stadiums (103 of 114 games) (odds ratio, 2.08); and on teams with a season-end losing or even record (66 of 69 games) than on teams with a winning record (76 of 86 games) (odds ratio, 2.89; (Table 1). When these three factors were incorporated into a forward and, separately, a backward model of regression analysis, only the type of playing surface remained significant [t = 3.62; P = 0.0004]. Bleeding abrasions were more frequent in games played on artificial surfaces (t = 3.61; P = 0.001). The frequency of bleeding lacerations was not associated with any environmental or athletic factor studied.

Players with Bleeding Injuries

Because an individual player may have more than one bleeding injury, we determined the rate per game of players with bleeding injuries: 3.5 ±0.2 players with bleeding injuries per game. Seventy-five of the 81 games played on artificial surfaces (93%) had at least one player per team with a bleeding injury. However, the rate was lower for games played on grass (odds ratio, 4.3; 95% CI, 1.5 to 13.0; P =0.002); in 55 of the 74 games (74%), at least one player per team had a bleeding injury.

The mean number of players with bleeding injuries per team was higher for games played in domed stadiums (t = 2.51; P = 0.013). When these factors were entered into a stepwise regression equation, only the type of playing surface remained significant (t = 4.095; P = 0.0001) in both a forward and a backward model.

When analyzed separately, players with bleeding abrasions were observed more frequently in games played on artificial surfaces (t = 3.83; P = 0.0001) and in games played in domed stadiums (t = 2.04; P = 0.043). After these factors were entered in a regression analysis equation, only the type of playing surface remained significant (t = 3.781; P = 0.0002). This finding was observed in both a forward and backward model. At least one player had a laceration in 20 of 41 games (49%) played in domed stadiums but in only 32 of 114 games (28%) played in open-air stadiums (odds ratio, 2.4; 95% CI, 1.1 to 5.4; P = 0.02).

Using the frequencies we observed for players with a bleeding laceration and players with any type of bleeding injury per game, we calculated the risk to each player to be less than 1 infection per 85 million game contacts (1 infected player/200 players × 1 HIV transmission/300 exposures × 0.41 lacerated players per game/45 players per game × 3.46 bleeding players per game/45 players per game = 1 HIV transmission per 85 647 821 game contacts).

Discussion

Bleeding injuries, particularly abrasions, were observed more often in games played on artificial surfaces. Although preventing such injuries is a worthwhile goal, the merits of doing so need to be weighed against other alternatives. Because different types of artificial surfaces are used in stadiums where football is played, the relation between bleeding injuries and type of playing surface should be studied further. Risk factors for bleeding injuries should be identified and minimized, if possible, by interventions such as training players, changing rules, using protective gear, and modifying playing surfaces.

Our study is the first attempt to calculate the risk for HIV transmission during athletic competition in professional football. Studies that quantify risks for the transmission of blood-borne pathogens, especially HIV, are important in developing credible prevention messages. Revelations that prominent professional athletes have been infected with HIV have heightened the need for this information, especially as it applies to athletic competition.

The limitations of our study include the atypical nature of the study participants. Professional football players are clearly not representative of all athletes or even of all football players. Although it is possible that the 11 teams in our study do not represent all 28 professional football teams, this is unlikely. The 11 teams are located in all geographic regions of the United States; furthermore, the 155 games we observed were distributed almost equally among the 11 teams and were similar with respect to the environmental and athletic circumstances relevant to our study. Finally, the teams were evenly divided between teams with season-end winning and those with losing records.

The risks for a bleeding injury and for transmission of HIV were calculated without respect to the player's position or degree of participation in the game. Some of the players prepared to play may not play at all; others may play only for short periods. Thus, risks will probably vary considerably according to factors such as playing time and playing intensity that would be difficult to measure.

We recognize that the calculations used in our study tend to overestimate the risk for HIV transmission. The direct contact between the bleeding injuries of two players is almost impossible and is impractical to study in the context of a professional football game. Consequently, the real risk for transmission of HIV may be considerably less than that calculated in our study.

A professional football player can expect to participate in a maximum of fewer than 50 games per year (including practice, exhibition, regular-season, playoff, and other games). The average number of plays per game in 1992 was 145, and the average career of a professional football player lasts approximately 8 years (NFL. Personal communication). Thus, the risk of a particular professional football player for acquiring HIV infection during athletic competition is extremely remote.

When we extrapolated our data to include the entire population of professional football players, we found that the risk for infection of any player during an entire season may be 10 000 times higher (16 games per team × 45 players per team × 28 teams ÷ 2 teams per game = 10 080). Even so, the resulting calculation for the probability of a single HIV transmission during an NFL season would be an estimated 0.017, meaning that HIV might be transmitted less than once in 58.6 seasons. Although the number of contacts that a player may incur in a play is unknown, a player is unlikely to have much more than one contact in a single play and almost no players participate in more than 100 plays in a game. Given this information and the low frequency of bleeding injuries reported in this study, the calculation of the risk for transmission of HIV during an NFL season is likely to be an overestimate.

When compared with the risk for HIV transmission associated with sexual and drug-using behaviors, and the risk for death associated with other activities in which professional athletes engage, the risk for HIV transmission during athletic competition is infinitesimally small. For example, the risk for HIV transmission from woman to man during unprotected sexual contact ranges from 26 in 10 000 to 58 in 100 [14-17]. By contrast, air travel, a necessary component of a career in professional football, is associated with deaths of passengers or crew in 1 in 1.6 million flights [18].

Professional football players, like other persons, can do more to reduce the spread of HIV by abstaining from unsafe sex and injecting drug use than by modifying their activities during athletic competition. Thus, our findings support the HIV and AIDS policies promulgated by the World Health Organization, the NFL, and other agencies [5-7].