Associations of Tackling Characteristics, Player Position, and Head Contact Risk During Game Play in College Football
Sport-related concussion is a common injury among National Collegiate Athletic Association football athletes. Beginning with the 2016 season, Ivy League Conference coaches voted to ban player-on-player tackling from all in-season practices. Dartmouth College’s program has enforced a no-tackle approach in practices since 2010. To examine the association between tackling technique and head contact risk and to compare base rates of techniques used in the 2016 season between an Ivy League team with a long-standing no-tackle practice policy (Dartmouth) and the rest of the league. Cross-sectional study. Ivy League College Football Conference. Two hundred thirty-seven Ivy League defensive football players who participated in the 2016 season. Tackles (N = 3701) across 237 Ivy League defensive football players in the 2016 season were coded based on predetermined classifications, which were combined to create unique tackle combinations/techniques. Associations among tackling techniques, head-impact risk, and teams (Dartmouth vs other Ivy League teams) were evaluated using logistic regression, yielding odds ratios (ORs) for head contact. Low-risk tackle characteristics for head contact during a tackle were neutral neck position (OR = 0.1), back contact area (OR = 0.3), pursuing momentum (OR = 0.5), and quarterback sack momentum (OR = 0.3). Low-risk tackle techniques were high-back-neutral (OR = 0.1), low-back-neutral (OR = 0.2), and medium-back-neutral (OR = 0.1). High-risk tackle characteristics were flexion (OR = 14.2) and extension (OR = 3.8) neck positioning, front contact (OR = 2.2), blowup (OR = 2.5), and cut (OR = 3.0). High-risk tackle techniques included low-side-flexion (OR = 4.9), low-front-flexion (OR = 9.9), medium-side-flexion (OR = 15.5), and medium-front-flexion (OR = 11.4). Relative to Dartmouth, other teams demonstrated higher odds of using high-risk techniques (low-side-flexion OR = 3.5; low-front-flexion OR = 3.9; medium-side-flexion OR = 6.3, medium-front-flexion OR = 2.3) and reduced odds of using low-risk tackle combinations (high-side-neutral OR = 0.4; high-back-neutral OR = 0.6; medium-side-neutral OR = 0.8). Tackling techniques are associated with head contact risk, and by extension, player safety. Dartmouth College Football, which has a long-standing policy of practicing without player-on-player tackling, showed reduced use of high-risk tackling techniques.Context
Objectives
Design
Setting
Patients or Other Participants
Main Outcome Measure(s)
Results
Conclusions
Key Points
Tackling techniques with low (neutral neck position, back contact area, and pursuing and quarterback sack momentums) and high (flexion and extension neck positioning, front contact area, blowup and cut tackle types) risk of head injury were identified.
A team with a well-established no-tackle practice policy (Dartmouth) showed a lower prevalence of high-risk tackling techniques relative to other Ivy League teams.
Teaching specific tackle techniques rather than focusing on player-on-player tackling, in tandem with safe practice policies, may be a promising avenue for preventing football-related head injuries.
Approximately 1.6 million to 3.8 million sport-related concussions (SRCs) occur in the United States each year.1 In the National Collegiate Athletic Association (NCAA), SRC accounted for 6.2% of all injuries reported during the 2009–2010 and 2013–2014 academic years.2 Football accounts for the greatest number of concussions compared with other sports in the NCAA and within the Ivy League, with a notable increase in incidence over time.2,3 Level of play and physical contact increases the risk of concussion, with a rate of 1.26 concussions per 1000 athletic exposures in the Ivy League and Big Ten conferences between the 2013–2014 and 2017–2018 academic years.3,4 Beyond SRCs, mere head contact is ubiquitous among athletes. To illustrate, in epidemiologic head-impact telemetry data from 6 Division I NCAA football programs (N = 658 collegiate football players), 528 684 incidents were recorded across the 2015 and 2019 seasons.4 On average, players sustained over 400 recorded head impacts (Med = 415, IQR = 190–727) each season.4
Critically, repetitive head-impact exposure confers increased risk of SRC. One study of 50 concussed NCAA Division I football athletes found that 72% of these players had the highest or second highest cumulative head-impact exposure compared with team- and position-matched controls.5 A separate biometric study of 502 Division I college football players across the 2015 and 2017 seasons found that those who sustained SRCs experienced nearly 100 (mean = 93.7) more recorded head impacts (including more high-magnitude impacts) than physically matched controls.6 Thus, as expected, athletes who sustain the highest number of head impacts (relative to their peers) have the greatest risk of SRC, even when controlling for player-level traits (eg, height, weight, age, race, SRC history) thought to affect concussion tolerance (ie, predisposition or vulnerability to sustaining SRCs). Other risk factors for head and neck injury (and by extension SRC) among football players include level of player physical maturation and quality/skill level of coaching.7
Authors of epidemiologic studies of concussion in college football published in the last decade found rates were higher in game play than during practice, but the majority of concussions occur during practice.2,3 Indeed, according to large-scale data from 6 NCAA football teams spanning 5 seasons (2015–2019), nearly half of the incident SRCs across the study period occurred during preseason (despite preseason encompassing only about one-fifth of the football season).4 Relatedly, the same observational cohort study revealed a 2-fold increase in head-impact incidence in the preseason compared with the regular season (325 vs 162 head impacts per team per day), and found that most head-impact exposures (67%) and SRCs (72%) occurred during practices.4 In this context, preseason training and practices across the football season are important targets for intervention.
Given the robust evidence implicating repetitive exposure to head impacts as a predisposing factor for incident SRC, there have been efforts to increase player safety over time through refinement of tackling techniques and rule changes, which are aimed to reduce head-impact exposures, as well as improvements in equipment to reduce the effect of head impact.7–9 Some of the rule changes at the professional (2017) and collegiate (2016) levels included modification of the kickoff rule, which resulted in a significant reduction in concussions.10 Research has also informed suggestions made for helmet modifications based on force transference during tackling.9
Moreover, considering the compelling evidence of disproportionately high incidences of head impacts and SRCs sustained in preseason and football practices, prevention efforts have targeted these exposure periods. Dartmouth College Football discontinued player-on-player tackling in practice in 2010 in an effort to reduce the risk of head injuries occurring in practice. It also incorporated a mobile tackling dummy into practices in 2015 to reduce head contact exposure and resultant injuries.11 Although the NCAA issued new football practice guidelines limiting the number of scrimmages and contact practices in 2014, the Ivy League implemented a more stringent rule in 2016 that uniformly banned player-on-player tackling during practice. In this context, the Dartmouth College football program had ample experience (ie, a 4-season “head start”) implementing the no-contact practice guidelines relative to when the other Ivy league teams adopted this policy.
A recent cohort study investigated trends in SRC frequency across 14 seasons (2006–2019 and 2021) in Dartmouth’s team, and across 7 seasons (2013–2019) for other teams in the Ivy League football conference.12 Keeping in mind the practice policy changes detailed above, a robust inverse association was observed between season year and frequency of SRCs sustained in game for Dartmouth and other Ivy League teams, such that SRCs trended downward in later vs earlier seasons. Follow-up analysis revealed reduced in-game SRC frequency for Dartmouth in seasons with no-contact and tackle dummy practice policies compared with seasons without these guidelines/rules. Analysis of in-game SRC frequency among other Ivy League teams showed a similar trend of lower in-game SRCs sustained for seasons with the no-contact practice policy vs contact in practices. Finally, for Dartmouth, both practice policies accounted for 51% of the variance in SRCs sustained in games across the 14 seasons analyzed, though the tackle dummy intervention was a superior predictor in the model compared with the no-contact rule. In keeping with the epidemiologic literature,2,3 no consistent effects of the practice policies were observed for SRCs sustained in practice.12
Although this Ivy League study provided preliminary evidence of reduced recorded SRCs possibly in relation to practice policy changes, several limitations hindered the strength of such findings.12 Critically, the study focused on large-scale outcomes without consideration of more granular covariates such as tackling mechanics/techniques and player position. As such, the findings were limited by unmeasured player- and team-level variables that would have helped contextualize the observed effects (eg, by way of prevalence of and changes in tackling techniques used by players). Here, we sought to address these limitations by conducting a follow-up study of detailed player-level variables, particularly tackling characteristics and player position in relation to risk of head contact, and comparing these in Dartmouth vs other Ivy League teams.
The goals of this study were 2-fold. First, we sought to characterize defensive football player tackling techniques in relation to head contact risk in the Ivy League during the 2016 season. Second, considering Dartmouth’s early adoption of banning player-on-player tackling during practice (ie, a “no-tackle” practice policy), we examined differences in base rates of tackle characteristics and combinations between Dartmouth and other Ivy League teams. A priori, we refrained from making predictions regarding the association between tackle characteristics and head contact but broadly hypothesized that Dartmouth would show lower base rates of using high-risk tackling characteristics/combinations.
METHODS
Design
We used a comprehensive sample of video footage from all 2016 games played across the Ivy League Conference accessed as part of a joint agreement among Ivy League teams. Plays were parsed within DV Sport (DV Sport Inc) game and practice data video analysis software. The provided footage encompassed 2 distinct vantage points for each play: an elevated position of the side of the field from the press box and an elevated position behind the end zone. The quality of the video was comparable among teams and plays. Only tackles from Ivy League players were analyzed. All extracted retrospective data were deidentified.
Participants
Data presented in this project are at the level of defensive tackles. That is, we analyzed characteristics of tackles (N = 3701) from 237 defensive football players across the 8 Ivy League conference teams. Although games involved both conference and nonconference opponents, only tackles made by Ivy League players were coded and analyzed. Each game in which an Ivy League team participated during the 2016 fall football season was included (10 games in total, 7 of which were against another Ivy League team).
Procedures
Our method for identifying and labeling tackle combinations was rooted in an informed observational approach. A series of preset classifications were determined from an analysis and characterization of all in-game tackling patterns of Dartmouth football players during the 2016 season. Data included head and neck position at contact, how the tackler approached the ball carrier, and the way in which the tackler completed the tackle. Tackle characteristic categories were selected a priori by a subject-matter expert (E.F.T.) with over 30 years of coaching Division I college football. Using those criteria, each play was then classified by members of the football staff. Each tackle was then rereviewed by a special assistant to the subject-matter expert, and approximately 25% of the most difficult-to-classify cases were discussed with the expert for additional expert opinion. Cases in which the contact was not discernable enough to classify were excluded. In total, a minimum of 5 trained individuals reviewed tackles from each play. One of the staff members who reviewed tackles had years of experience in football coaching. Four additional team members were students trained by the subject-matter expert mentioned above.
Tackle Characteristics and Combinations
Expanded details for each tackle characteristic coded are presented in Table 1. Briefly, tackles were coded according to whether there was head contact (directly attributable to the tackle action rather than, for instance, a player hitting their head on the ground after the tackle), defensive player position of the tackler (defensive backfield [DB], defensive line [DL], or linebacker [LB]), neck position (neutral, flexion, or extension), contact area (side, front, or back of player), strike zone (high, low, or medium), tackle momentum (attack, blocked, multiple tacklers, follow-through, gathering, pursuing, quarterback sack), and tackle type (blowup, cut, follow-through, roll, swipe, or wrap). In addition, we computed tackle combinations based on an amalgam of neck position, contact area, and strike zone (27 possible permutations). The present data yielded 26 unique combinations. We retained the most frequent tackle combinations (defined as those with a base rate of greater than 2% relative to all tackles that occurred in the 2016 season) for analyses. Thirteen distinctive tackle combinations (which, collectively, represented 93% of all tackles from the 2016 season) were analyzed: high-side-neutral, high-front-neutral, high-back-neutral, low-side-neutral, low-side-flexion, low-front-neutral, low-front-flexion, low-back-neutral, medium-side-neutral, medium-side-flexion, medium-front-neutral, medium-front-flexion, and medium-back-neutral.
Analysis Plan
First, we examined the association of broad tackle characteristics (eg, neck position) with head contact via omnibus χ2 tests of independence. Next, individual elements of tackle characteristic categories (eg, flexion neck position) were dummy coded and entered into binary logistic regression models to predict the head contact tackle outcome (ie, 0 = no head contact; 1 = head contact). We conducted an identical set of analyses on the 13 unique tackle combinations mentioned above to determine their association with the odds of resultant head contact. We then compared the base rates of the 13 tackle combinations between Dartmouth College and the other Ivy League football teams aggregated (note: data were weighted according to total number of tackles within each team). That is, binary logistic regression models were constructed to determine whether respective tackle combinations predicted group membership (ie, football team, coded as 0 = Dartmouth, 1 = other teams). Finally, we conducted post hoc exploratory analyses to determine whether defensive player position was associated with tackle variables (ie, broad tackle characteristic categories, such as neck position, and any of the 13 computed combinations) using χ2 tests.
Importantly, tackle characteristic categories and computed tackle combinations are orthogonal, or mutually exclusive, from one another. To illustrate, tackles were coded such that they could not possess both neutral and flexion neck position, and each tackle could only have 1 unique combination (eg, high-side-neutral). For χ2 analyses, effect strength was determined by Cramér ϕ (ϕc; small = 0.20, moderate = 0.50, strong ≥ 0.80).13 Odds ratios (ORs) and 95% CIs were computed to evaluate effect strength for logistic regression models (small = 2.0, moderate = 3.0, strong ≥ 4.0).13 We also followed the Ferguson recommended effect threshold for what may be considered a “practically” significant effect (ϕc ≥ 0.20; OR ≥ 2.0 or ≤ 0.5).13 To control for α inflation due to multiple comparisons, all P values were adjusted using the Benjamini-Hochberg false discovery rate method. Significance threshold αFDR was set to .05 (2-tailed).
RESULTS
Descriptive information for tackles analyzed (eg, player position, tackle characteristics, rate of head contact), stratified by team, is presented in Table 2. Notably, the base rate of head contact outcome (ie, the tackler hitting their head) was 28.1% across all tackles (N = 3701). Table 3 shows supplementary analyses of tackle characteristics and combinations as a function of defensive position. Defensive positions were broadly equally represented across tackles. Base rates of head contact were likewise generally commensurate across positions. The only “clinically meaningful” effects were observed for broad tackle momentum (ϕc = 0.34) and tackle combination (ϕc = 0.22). Greatest differences according to position type were observed for attack momentum tackle (DB = 52.2%, DL = 8.0%, LB = 35.0%) and quarterback sack (DB = 0.8%, DL = 11.4%, LB = 3.4%) characteristics and low-front-flexion tackle combination (DB = 12.9%, DL = 0.6%, LB = 3.4%). Statistical assumptions of binary logistic regression were evaluated and upheld. Thus, we proceeded with our planned models.
Tackle Characteristics and Head Contact
Table 4 displays base rates (%) of specific tackle characteristics, as well as respective rates and ORs of head contact, across all tackles aggregated (N = 3701). Broad and specific neck position characteristics were strongly associated with head contact. Notably, neutral neck position was associated with 14.7 times (95% CI = 12.3, 17.5) lesser odds of head contact, whereas flexion neck position was associated with 14.2 times (95% CI = 11.8, 17.2) higher odds of head contact. A moderate to strong effect for extension neck position was also observed, such that odds of head contact were 3.8 times (95% CI = 2.6, 5.5) higher for tackles with this characteristic. Smaller relations with head contact were observed for broad and specific contact area and strike zone characteristics. Front contact and back contact areas were associated with 2.2 times (95% CI = 1.9, 2.5) greater and 3.2 times (95% CI = 2.5, 4.0) lesser odds of head contact, respectively. Clinically meaningful relations were observed only for pursuing and quarterback sack tackle momentums, which were associated with 2.2 times (95% CI = 1.7, 2.8) and 3.4 times (95% CI = 2.0, 5.6) lower odds of head contact, respectively. Blowup and cut tackle types were related to 2.5 times (95% CI = 1.7, 3.6) and 3.0 times (95% CI = 2.3, 4.0) higher odds of head contact, respectively, whereas roll and swipe were both associated with 2.1 times lower odds of head contact (95% CI = 1.6, 2.6 and 1.3, 3.2, respectively).
Tackle Combinations and Head Contact
Table 5 displays base rates (%) of tackle combinations, as well as respective rates and ORs of head contact, across all tackles aggregated (N = 3443). The omnibus test indicated a considerable association between tackle combinations and head contact (ϕc = 0.55). At the individual level, particularly robust (strong) ORs were observed for several combinations. Tackle permutations associated with markedly higher odds of head contact included low-side-flexion (OR = 4.9; 95% CI = 3.3, 7.2), low-front-flexion (OR = 9.9; 95% CI = 7.2, 13.4), medium-side-flexion (OR = 15.5; 95% CI = 9.6, 25.0), and medium-front-flexion (OR = 11.4; 95% CI = 8.0, 16.2). Conversely, high-back-neutral, low-back-neutral, and medium-back-neutral were associated with 7.8 (95% CI = 2.9, 21.3), 5.0 (95% CI = 10.2, 2.4), and 7.7 (95% CI = 4.8, 12.5) times lower odds of head contact, respectively.
Tackle Combinations Reflected in Dartmouth vs Other Ivy League Teams
Table 6 displays base rates (%) of tackle combinations, stratified and compared by team. Broadly, tackle combination was weakly associated with football team (Dartmouth vs other, ϕc = 0.16). Nevertheless, several tackle combinations appeared to differentiate Dartmouth from the other Ivy League teams in this sample. Largest effects were observed for low-side-flexion, low-front-flexion, and medium-side-flexion, such that odds of these tackle combinations were 3.5 (95% CI = 1.4, 8.6), 3.9 (95% CI = 2.0, 7.6), and 6.3 (95% CI = 2.0, 20.0) times higher for other Ivy League teams vs Dartmouth. Worth mention, analyses identified these 3 combinations as being associated with considerably higher odds of head contact. To add, although effects were smaller in magnitude, other Ivy League teams showed lower odds of using several “safer” tackle combinations compared with Dartmouth: high-side-neutral (OR = 0.4; 95% CI = 0.3, 0.7), high-back-neutral (OR = 0.6; 95% CI = 0.3, 0.9), and medium-side-neutral (OR = 0.8; 95% CI = 0.6, 0.98).
DISCUSSION
The overarching aim of our study was to expand upon a recent pilot study reporting preliminary evidence of lower documented SRCs among Ivy League teams for seasons in which practice policies (such as no-tackle) were implemented.12 Specifically, here, we captured finer player-level variables within teams (as opposed to gross metrics such as SRC frequency across teams). We first sought to examine characteristics of tackling and their associations with head contact. Examination of individual tackling characteristics demonstrated lower odds of head contact associated with neutral neck position, making contact with the back, tackling momentum of pursuing and quarterback sack, and proceeding through the tackle using a roll or swipe technique. Similarly, tackling combinations with the neutral neck position and contact in the back regardless of strike zone were less likely to result in head contact. Conversely, tackling characteristics that led to higher odds of head contact included a flexion neck position, making contact in the front, and blow up or cut tackle types. Tackling combinations with high odds of head contact all involved the flexion neck position with low or medium strike zone in the front or side contact areas. Notably, DBs were more likely to use a low-front-flexion tackle combination than other positions, a technique with the third-highest risk of head contact (OR = 9.85). Further, although head contact did not vary significantly by position, tackling momentum reflected variation in position characteristics: attacking was more common for DBs and quarterback sacks for DLs.
Consistent with our observing higher odds of head contact related to flexion neck positioning, multiple studies indicate neck flexion before contact in rugby and football is associated with a broad range of adverse outcomes including concussion, cervical spine injury, quadriplegia, and mortality.7,9,14,15 In a similar vein, Suzuki et al investigated the relation between tackling characteristics and concussion in rugby and found that the head-down position before contact increased the risk of concussion by a factor of 4.67.15 The flexion neck positioning increases the risk of concussion not only in the defensive player, but also in the player being tackled, due to more momentum being transferred in a head-down tackling position.9,16 In recognition of the dangers associated with this neck positioning, rules prohibiting deliberate spearing and use of the top of the helmet as the initial point of contact were first instituted in 1976 in the NCAA and the National Federation of State High School Associations.9 Further, helmet rules (eg, initiating contact with the helmet) were instituted in the National Football League in 2018 and were reemphasized during the 2021 season.17 Finally, the dangers of this neck position are clearly reflected in the 2022 National Athletic Trainers’ Association position statement, which includes 14 recommendations for equipping American football athletic training/coaching personnel with tactics for reducing such head-first contact.18 Some examples include developing compulsory up-to-date education on the harmful effects of head-first contact to the head and neck for players, intensive evidence-based instruction that teaches players “progressive techniques for avoiding head-first contact behavior” before the exposure of tackle football (eg, in preseason), and eliminating or modifying drills in football practices that go against the ethos of “proper and safe tackling and blocking behaviors or techniques.”18
Regarding other aspects of tackling, tackling techniques taught by the 2012 USA Football’s Heads Up program emphasize primary contact being made to the midsection while maintaining the neutral neck position. This tackling technique is based on those used in collegiate rugby, leading to reduced incidence of concussion when performed correctly.14,15 Our results demonstrate that striking a player in the midsection did not increase their odds of head contact if the tackler maintained a neutral neck position, and the odds of head contact decreased further when contact was made with the ball carrier’s back. Nevertheless, the broad neutral neck position characteristic was related to a 93.2% reduction in odds of head contact in our data, one of the most robust findings in this study. Heads Up also teaches that the player should grab the jersey and continue through the tackle (consistent with a wrap or roll tackle type), which, along with a wrap tackle type, was found to reduce the likelihood of head contact. Similarly, in rugby, making contact with the ball carrier’s back and using a roll tackle type were found to reduce the likelihood of concussion compared with tackles in which the head was positioned in the front or to the side of the ball carrier or tackles that did not involve pulling, gripping, or wrapping the ball carrier with their arms.15 Therefore, it follows that there is an increased risk of head contact associated with cut-block or blow-up tackles, because these do not involve wrapping or gripping the ball carrier.
The second purpose of our study was to compare Dartmouth tackling with that of other Ivy League teams. This goal was motivated by the findings of a recent pilot study that observed strong effects of practice policies in attenuating SRCs sustained across 12 seasons (2009–2019 and 2021) for Dartmouth, particularly those in game.12 We sought to compare identified high-risk and low-risk tackle characteristics/combinations between Dartmouth and other Ivy League football conference teams to further clarify whether a team that has been engaged in specific preventative measures for SRC longer than other teams demonstrates any systematic differences in tackling techniques in game. That is, Dartmouth has been implementing a no-tackle practice policy since 2010, whereas at the time of this study’s data collection (2016), other teams were just beginning to incorporate this rule change into practices.11 The results indicated that Dartmouth used the 3 tackling combinations associated with the highest odds of head contact (low-side-flexion, low-front-flexion, and medium-side-flexion) significantly less than other Ivy League teams. Further, Dartmouth showed evidence of using several “safe” tackle combinations (high-side-neutral, high-back-neutral, and medium-side-neutral) more than other Ivy League teams.
One possible explanation for these findings is that Dartmouth implemented no-tackle practices 6 years before the other Ivy League teams. That is, Dartmouth College Football focused on teaching proper tackling techniques without tackling other players. Therefore, the coaches and players had more time to adjust to and experience using this method of practicing more than other teams. In this sense, the concept of not tackling in practice may translate best to gameplay and player safety when it is part of a culture of safety, rather than simply thought of as a rule or policy initiative. Another explanation is that Dartmouth introduced a mobile tackling dummy into its practice approach in 2015, which recent literature suggests predicted reduced SRC frequency above and beyond the no-tackle policy.12 Thus, the team’s unique access to this technologic innovation may have contributed to enhanced skills in the context of the no-tackle practice approach. Conceptually, the key focus in practice is on tackling form rather than bringing other ball-carrying players to the ground, which is espoused in formal league games. Epidemiologic findings underscore the importance of teaching and practicing safely (with respect to reducing odds of head injury–related sequelae) because the majority of concussions occur during practices (despite actual gameplay having highest rates).2,3 Ideally, experiential learning that occurs across practices will transfer to gameplay and contribute to reduced odds of head injuries.
Limitations
The current study was limited by the lack of concussion diagnostic data to contextualize our findings, because head contact does not inherently lead to head or neck injury. Although lack of corroborating evidence regarding concussion incidence in relation to the tackles we analyzed limits the breadth of inferences we can draw from our data, this was out of the scope of the current project, which focused on head-impact exposure (in relation to defensive tackle characteristics and player position). Conversely, a strength of the study is the inclusion of head contact, rather than only concussions. Head contacts may be easier to visually observe in a game than bona fide concussions, which often go undiagnosed and underidentified.19 However, the objectivity of our methods of documenting player head contacts was weakened by lack of a concurrent head-impact sensor system to confirm observer ratings of head contact. Although sensors are useful in head-impact research because they provide a variety of force parameters (eg, peak linear acceleration, angular velocity, angular acceleration magnitude) to grade impact severity, some limitations have been noted. One study found instances of a mouthguard sensor missing up to approximately 30% of helmet contact events classified by video, while also detecting too many impacts at times (ie, far exceeding those cross-verified in video).20 Next, our study did not use a previously validated methodology for establishing the predetermined tackling characteristics and combinations. Observer training for counting head impacts may also differ across programs. Indeed, tackling instructions, associated nomenclature, and key performance metrics (and their measurement) more likely than not vary among coaches and football programs. Future authors should investigate tackling techniques by examining incidence of both head contact and concussion, because this may provide a more comprehensive picture.
Additionally, the scope of the study was limited to tackles performed by defensive players on Ivy League teams during the 2016 football season. In this context, our data were limited to a cross-sectional (vs longitudinal) sample of an Ivy League football season. As such, we advise cautious interpretation of the relation between the no-contact practice regulation change and tackling techniques used in games. The conclusions that can be drawn from our findings would be strengthened (in either direction) if expanded tackle data from seasons before and after the uniform ban on player-on-player tackling during practice in the Ivy League in 2016 were available. We unfortunately did not have such data to study changes in tackle techniques in relation to the no-tackle practice policy. Future investigations of tackling techniques and their association with head contact should span across seasons and include offensive and defensive players.
Next, we were unable to quantitatively index the degree of interrater reliability/agreement (eg, intraclass correlation and Cohen κ statistics) because the requisite data were not available to us. Nevertheless, 5 or more trained athletics personnel reviewed tackles from each play, and therein, interrater discrepancies were carefully processed and rectified. In a separate vein, we were unable to quantitatively analyze various idiosyncratic team-level variables (eg, quality of teams, coaching staff, athletic programs, and training facilities; player size, talent level, position, and experience; overall team ability; team cultural differences; and other in-house practice/game policies) that may differ from team to team and perhaps influence tackling techniques players use. Such nuanced team- and player-level variables may play a role in head injury/contact–related outcomes. As such, future studies should consider empirically incorporating these elements. Next, we did not consider, nor did we have access to, additional biometric assays such as in-helmet accelerometers to analyze additional metrics of head impacts resulting from various tackle types. Future researchers may wish to evaluate measured biomechanical and actual concussion-diagnostic outcomes of the high-risk tackle combinations identified from our analysis. In a separate vein, we also did not code information on play type (eg, runs, passes). We recommend that future authors examine the relation between play type and base rates of high- and low-risk tackles.
Finally, we were unable to formally assess whether tackling characteristics and combinations were associated with team success/win-loss record and odds of winning games. This was out of the scope of the aims of the current project but a promising topic for future investigators to consider. However, 2 recent studies analyzed tackling techniques in relation to their effectiveness (ie, success of the tackle). The first of these examined 1000 defensive tackle attempts that occurred in the NCAA Southeastern Conference 2021 season and found that head-down tackles below the offensive player’s waist yielded lower success rates than head-up tackles and at or above the waist.21 These findings were mirrored in a study published a year later that assessed 1000 defensive tackle attempts that occurred across 6 English Rugby Premiership matches in the 2022 season.22 Interestingly, the tackle characteristics linked with poorer success rates in these 2 recent studies were linked with higher odds of head contact in our data (ie, head-down = flexion neck position and below waist = low strike zone in the present study’s classification system). Next, our data revealed lower odds of head contact for the characteristics linked with higher tackle success rates (ie, head-up = neutral neck position and at or above waist = medium or high strike zone). Thus, it remains possible that promoting safe (ie, low risk for head injury) tackling may have indirect (downstream) effects on tackle success, and, by extension, team success. Anecdotally, Dartmouth averaged 6.6th place in Ivy League standings in the 10 seasons (ie, 2000–2009) before it banned tackling from practices. In contrast, its average team standing within the conference increased to 3.3 across the 13 seasons (ie, 2010–2019, 2021–2023) after setting the no-tackle practice policy in place (with adjunctive implementation of mobile tackling dummies in 2015). Moreover, across these later 13 seasons, it won or shared 4 division titles, juxtaposed with no titles from the earlier 10 seasons. We stress that additional data are needed to clarify whether practice policies are associated with changes in tackle techniques as well as defensive (and team) success before anecdotes like the one we presented above can be given weight.
CONCLUSIONS
Our study echoes prior literature thereby suggesting that the use of a proper tackling technique may reduce the chances of head contact, and, by extension, subsequent head and neck injury.2,3,11,12,17 Reducing the chance of head impact in organized football through refinement of tackling techniques and rule changes is important for player safety. Although our results suggest a relation between the elimination of player-on-player tackles in college football practices and a reduced chance of head contact in games, additional longitudinal studies are necessary to clarify whether this is attributable to changes in tackle technique. Future studies will also be important for determining whether our results generalize to other levels of play, such as high school and youth sports. Indeed, the majority of football injuries presenting to emergency departments occur in children under 14 years, with the primary diagnosis being concussion.23 Notably, this likely reflects coaching experience, ongoing physical development of the player, and fewer resources (eg, athletic trainers) available to assess and manage injuries in youth sports. However, it also highlights the need for more teaching of proper tackling techniques at younger ages and implementation of rules that reduce injury risk. Implementing a no-tackle practice approach for developing players may control more injury-risk variables. Learning safe tackling techniques would increase player safety in practice but may also lead to players maintaining better form during gameplay.
Contributor Notes