Studies of general injury profile in rowing

A search was carried out using a combination of the terms: rowing, injury, rowers. EMBASE, Pubmed, PEDro and AMED were searched. The final search was carried out in May 2014. No date or language limits were set. Two authors, (FW and AM) screened titles (initially) and abstracts and papers independently to decide inclusion. A third author (CG) was available where they may be disagreement. One hundred and sixty-four studies were accessed initially. Of these, the following were excluded: review papers/letters and editorials (38); studies of stress fractures only (26); studies that were not specific to rowing such as the use of rowing as rehabilitation in spinal cord injury (27); and studies of single injury in rowers (48). This left a small number of studies; of these, 11 were specifically studies of injury in rowing.2–12 Five studies used a retrospective questionnaire of injury rate,2–4 ,6 ,12 and three studies a retrospective review of athlete attendances at a sports injury clinic.5 ,8–10 Furthermore, the population group was frequently limited to elite rowers, most of whom were international standard at a specific training camp or international competition.2 ,4 ,7–12 These studies are summarised in table 1.

The varied methodologies of the studies make comparisons inappropriate. Of the 11 studies, only 3 collected data prospectively and of those, only 2 studies6 ,11 reported the injury incidence/1000 h at 1.5 and 3.67, respectively. Of the 10 studies2–12 that reported location of injury, nine2–12 cited the lumbar spine as the most commonly reported site in rowing, ranging from 2.4% (of all sports) to 51% of injuries.

While there is a need for additional, well-constructed, prospective injury surveillance in rowing, there is consensus that the most common injury is to the lumbar spine. Such data have more meaning when put in the context of reported incidence in the general population. Of the studies that reported 12-month incidence,1 ,2 ,4 ,8 ,12 the incidence of low back pain ranged from 32 to 51% of the cohort. The most recent systematic review of 12 month incidence of low back pain in the general population ranged from 6.3 to 15.4% (first ever episode) to 1.5 to 36% for ‘first ever or recurrent episode’.13 So, the incidence of low back pain in rowers is likely to be higher than that of the general population, particularly when other risk factors for back pain, such as obesity and psychosocial factors, are taken into account.

Studies specifically examining lumbar spine injury

Few studies14–19 have specifically examined back pain in rowing. A common issue with all studies is the variation (or absence) of a clear definition of low back pain. Indeed this is observed in many other studies of low back pain, which is not surprising when it is considered that it is a symptom rather than a disease.

Bahr et al15 compared the prevalence of low back pain retrospectively in endurance athletes in several different sports. Over half (55%) of the rowers reported low back pain in the previous 12 months, which compared with 63% in skiing, 49.8% in orienteering and 47.5% in the control group. Of note, the rowers reported the most missed training sessions because of injury and the greatest number of injuries requiring hospitalisation compared to the other groups.

A survey carried out on a large cohort of intercollegiate rowers is limited in some aspects of methodology but permits some useful conclusions. Former intercollegiate rowers (N=1632) were surveyed to investigate training methods and back pain before and during college rowing.16 Thirty two percent of respondents reported back pain that developed during college rowing. Factors significantly associated with the development of back pain included age at the time of the survey; history of rowing before age 16; use of a hatchet oar; training with free weights, weight machines and ergometers; and ergometer training sessions lasting longer than 30 min.

Back pain was most likely to develop in the winter months (39% of cases) compared to spring (33%), autumn (25%) and summer (4%). This reflects previous research11 and likely reflects the high volume of land training in winter. Many respondents associated the onset of back pain with a specific event; outdoor rowing was the most common at 72% of cases followed by weight lifting (50%) and ergometer training (29%). When training types were assessed, the only significant predictors of back pain for men were ergometer sessions that lasted longer than 30 min. Wilson et al11 also found that time spent ergometer training was the most significant predictor of onset of low back pain. The 20-year period of this survey16 is probably its greatest limitation and data is subject to recall bias. Also, the response rate of 35% of 4680 athletes originally surveyed introduces a further risk of bias. However, it did highlight an aspect of training (more than 30 min of ergometer training) that is a predictor for back pain.

The initial study was followed up to examine if rowers who developed back pain in college are more likely than the general population to have back pain later in life.17 Those who had developed back pain in college had more subsequent back pain later (78.9% vs 37.9%), although the lifetime prevalence of back pain in former college rowers was no different to the general population. A further survey on the same cohort noted that participants who had previous back pain before rowing were more likely to develop back pain during their college-rowing career than participants without pre-existing pain (57.1% vs 36.6%).18

Ng et al19 surveyed adolescent rowers and found a point prevalence of 64.5% (males) and 52.8% (females), compared to the mean point prevalence in the general population of 18.1%.13 In a self-report questionnaire, participants reported primary aggravating factors for low back pain as ergometer training, ‘long row’ sessions and sweep rowing (single oar rowing). This finding is in conflict with a previous study in adolescent rowers,4 which found no significant association between mean length of ergometer training sessions and low back pain. It is notable that studies in adolescent populations found a higher incidence of low back pain in males compared to females,4 ,19 with such gender differences not found in adult rowers.16

Although all these surveys were retrospective, they are the first to establish a profile of lumbar spine injury in rowers. The studies suggest that a number of factors such as volume of ergometer training and a history of low back pain may be associated with the onset of low back injury and raise questions regarding injury predictors. Studies in the general population20 and the sporting population21 have also identified pre-existing back pain as one of the best predictors of future injury. There is an absence of investigation of other predictors in rowers; notably the influence of occupation. Many studies have examined international rowers, most of whom do not have another occupation. The influence of work-associated loading of the lumbar spine, combined with rowing stresses, merits research as this may identify modifiable factors, particularly in recreational rowers.

Mechanisms that may underpin rowing-related back pain

Low back pain is clearly associated with rowing, but what are the mechanisms behind this association? Highly repetitive rowing actions require high levels of consistency, coherence, accuracy and continuity. Additionally, the rower has to draw on physical strength and endurance, and translate this through skill and coordination into performance. A number of factors are proposed to relate to causation. These include rowing technique, the associated training (such as weight-training) and the use of the rowing ergometer, as well as other issues such as changes in the design and shape of the rowing oar.11 ,22–24 To solve this conundrum relies on knowledge of how a rower’s body moves during the rowing stroke. There is a growing body of work exploring how understanding rowing kinematics can help prevent injury and enhance performance.25–31

Research has been laboratory based with a focus on rowing ergometer technique. This has the advantage of controlling for external influencing factors such as wind speed, water condition, etc, as well as allowing exploration of factors such as fatigue, but does not reflect ‘on-water’ technique. A number of studies25 ,26 ,28 ,32 highlighted the importance of rotation of the pelvis on the hips as well as the rotation of the spine (lumbopelvic motion), showing that to achieve the extreme positions of the rowing stroke necessitates good anterior rotation of the pelvis rather than extreme flexion of the lumbar spine (catch position) and posterior rotation of the pelvis as opposed to hyperextension of the lumbar spine (finish position). Rowing technique deteriorates during continuous rowing leading to increased lumbar flexion26 and frontal plane motion,33 which is attributed to fatigue; although this could be dependent on rowing ability and experience. There is also an indication that the ergometer exaggerates these changes in technique compared to ‘on-water’ rowing.29

Novice rowers use high levels of lumbar flexion with limited pelvic rotation, deteriorating further with higher work intensities.30 While similar changes are seen in elite rowers, these are of a much lower magnitude.31 What is particularly interesting in both groups of athletes is the drive to maintain power output at the cost of technique. This has implications with respect to injury, particularly in an athlete returning to competition following injury or a young athlete attempting to gain a place in a highly competitive team.

The relationship between poor technique and injury is as yet unclear, although a study in international junior rowers4 showed that rowing experience (and thus enhanced skill) might reduce risk of injury in this group. Statistical modelling was able to use kinematic data to distinguish between rowers with and without a history of back pain.34 Mathematical modelling of rowing has shown that lumbopelvic motion patterns do influence spinal loading, showing that poor lumbopelvic motion (an inability to rotate the pelvis and flex the lumbar spine in unity) is associated with higher loads and moments generation at L4/5 and L5/S1 (Buckeridge 2013—Imperial College London, PhD thesis). The challenges then remain as to how we can use this information to manage and prevent injury.