Current public health recommendations strongly recommend regular physical activity to improve cardiovascular health and reduce the risk of chronic diseases.1 However, with increasing physical activity, the risk of musculoskeletal injury also increases. Owing to the very large and increasing number of injuries induced by physical activity among adolescents and young adults, such injuries are currently considered a marked public-health burden.2 3

Because of its anatomical location, the knee joint is subjected to tremendous forces during exercise and physical activity, thus, it is not surprising that this joint is the most common site for a sport-related injury, usually accounting for 15–30% of all such injuries.4 5 In addition, the knee joint is a common site for severe injuries, such as ruptures of the anterior or posterior cruciate ligament (ACL or PCL). These injuries frequently need surgical treatment and long-term rehabilitation and may result in functional impairment and permanent disability.2 4 6 7 They are also very costly.8 In the USA, which has a rate of about 250 000 cruciate ligament injuries per year, a conservative estimate of surgical and rehabilitative costs of one cruciate ligament injury is about US$17 000.6 9 Moreover, the long-term costs of a cruciate ligament injury may be significantly greater, as the risk of radiographically diagnosed knee osteoarthritis later in life is increased by up to 105 times.10

Despite these results, the frequency of and risk factors for knee injuries in adolescents are not well established. Therefore, the aim of this prospective cohort study was to investigate the incidence of a major knee ligament injury in a general adolescent population and to compare sports with socioeconomic, health and other lifestyle background variables to obtain insight into how strong a risk factor is sports participation for these injuries.

METHODS

Approval for the study was obtained from the institutional review board of the National Research and Development Center for Welfare and Health (6267/54/2002) for the use of the National Hospital Discharge Register and, from Statistics Finland (TK-53-1526-04) for the use of the Official Cause-of-Death Statistics.

Baseline cohort

Combining survey data with data from national injury registers we obtained longitudinal information on cruciate ligament injuries of the knee and associated risk factors. The Adolescent Health and Lifestyle Survey is a nationwide monitoring system of adolescent health and health-related lifestyle in Finland, conducted as a postal survey on alternate years since 1977.11 12 Two reminders are sent to non-respondents after 3 and 7 weeks. The materials with respect to sampling, research methods, questions and time of inquiry have been maintained as similarly as possible for each year.

Our sample of adolescents aged 14, 16 and 18 years was drawn from the National Population Register Center through the selection of all Finns born on certain days in June, July or August. The mean ages of the respondents were 14.6, 16.6 and 18.6 years. Baseline data for the purpose of the present study was collected from 1987 to 1997. The baseline population consisted of 59 403 people, of whom 46 531 responded to the Adolescent Health and Lifestyle Survey, the response rate being 78% (table 1).

Cruciate ligament injury data

The follow-up started after the survey ended on 30 April of each data-collection year. The end-points in the study were the date of the first cruciate ligament injury hospitalisation (obtained from the National Hospital Discharge Register), the date of death (obtained from the Official Cause-of-Death Statistics), or the end of the study on 31 December 2001. The respondents were followed up for an average of 9.3 years, the total follow-up time being 435 840 person-years.

The hospitalisation data were obtained from the statutory, computer-based National Hospital Discharge Register of Finland, which contains data on the diagnosis, length of stay, location and cause of injury, possible surgery, age and place of residence of all patients in this country. The information is systematically collected from all hospital categories (public, private, military and other). The main outcome variable was patients hospitalised with the main or secondary diagnosis of ACL or PCL injury.

The diagnoses in the National Hospital Discharge Register were coded using the 9th (1987–1995) and 10th (1996–2001) revisions of the International Classification of Diseases (ICD). Cruciate ligament hospitalisation was defined by the ICD-9 codes 8442A and 8442B and by the ICD-10 code S83.5. In the analyses, ACL and PCL injuries were combined because ICD-10 uses code S83.5 for both of these injuries. During the years 1987–1995, 8% of the injuries involved the PCL. A recent validation study showed that the coverage (92%) and accuracy (89%) of diagnosis of cruciate ligament injury of the knee in the Finnish National Hospital Discharge register are very good for epidemiological purposes.13

People who had been hospitalised with cruciate ligament injury before entering to the sample (ie before the response date), were excluded from analysis (n = 73), whereas people who died during the follow-up (n = 257) were followed up until the time of death. As noted above, the data concerning deaths were obtained from the Finnish Official Cause-of-Death Statistics, which is also a statutory, computer-based register covering the entire population.14

Physical activity variables as risk factors for the injury

Frequency of participation in sports clubs and other physical activity in adolescence was analysed with following alternatives: never, ⩽3 times/week and ⩾4–5 times/week. Other physical activity was measured by combining three variables describing physical activity; (1) organised by school or workplace, (2b) organised by associations other than sports clubs and (3) participated in alone or with friends or family members.

Other background variables as risk factors for the injury

Altogether, 11 additional categorical variables from the Adolescent Health and Lifestyle Survey were used in the analysis. These were obtained as adolescents’ self-reports, with the exception of age, sex and urbanisation level of residence, the latter derived from the sample information (Population Register Center). Respondent’s socioeconomic background was measured by the education of the father or other parent/guardian, family composition and urbanisation level of residence. Adolescents’ health was assessed on the basis of their self-reports on perceived health status, chronic disease or disability restricting daily activities and by counting a summary index of eight stress symptoms (stomach aches, tension, irritability, sleep difficulties, headache, trembling of hands, feeling tired or weak, feeling dizzy) perceived weekly. Body mass index (BMI) was calculated by dividing weight (kg) by the height in meters squared, (m) and the cut-off points for overweight were set according to Cole and colleagues.15

The timing of puberty was assessed by questions about the respondent’s age at the time of the first ejaculation (boys) and first menstruation (girls) and classified into three categories: early (⩽12 years in boys and ⩽11 years in girls), average (13 or 14 years in boys and 12 or 13 years in girls) and late (⩾15 years in boys and ⩾14 years in girls). Adolescents’ health-compromising behaviours were described by daily use of tobacco and the drinking style (abstinence, occasional drinking, recurrent drinking, recurring drunkenness).

Statistical methods

The statistical analyses were carried out in two stages, separately for boys and girls. There were two exclusion criteria: non-respondents to the questionnaire and respondents who had not answered the questions investigated in this study.

In the analysis, we first conducted Cox regression with 95% CI to analyse the association between the background variables and cruciate ligament injury hospitalisation, forcing age at baseline into the model. Thereafter, a multivariate Cox’s regression model was conducted including adjustment for occupation of father or other parent/guardian, family composition, urbanisation level of residence, perceived health status, chronic disease, number of stress symptoms weekly, overweight, smoking, drinking style and timing of puberty. Independent samples t test was used when comparing the age at time of hospitalisation between sexes.

Finally, we compared respondents and non-respondents across the follow-up years to determine if any specific differences might exist between the groups relevant to this study. The non-respondents to the baseline surveys had 1.1 times (95% CI 0.8 to 1.4) the risk of ACL/PCL hospitalisation than the respondents, thus indicating random or nonbiased selection of the study respondents.

RESULTS

Altogether 265 (0.6%) people (194 boys and 71 girls) from the total cohort of 46 472 people were hospitalised for a cruciate ligament injury of the knee during the follow-up, giving an injury incidence of 60.9 (95% CI 53.6 to 68.2) per 100 000 person-years. The injury incidence was 96.6 (95% CI 83.0 to 110.2) in male subjects and 30.0 (95% CI 23.2 to 37.3) in female subjects. The mean age at the time of the injury was 22.6 years for male subjects and 22.2 years for female subjects (p = 0.5).

Of the measured socioeconomic, health and lifestyle background variables, participation in organised sports showed the strongest association with hospitalisation for cruciate ligament injury (tables 2 and 3). Participation in organised sports ⩽3 times/week involved a slightly but statistically significantly increased risk for cruciate ligament injury (adjusted hazard ratio (HR) 2.0 (95% CI 1.1 to 8.6) for female subjects and 2.0 (95% CI 1.4 to 2.9) for male subjects). Higher-activity participation was associated with higher risk for the injury: in female subjects, the adjusted HR for a cruciate ligament injury was 8.5 (95% CI 4.3 to 16.4) and, in male subjects, the corresponding figure was 4.0 (95% CI 2.7 to 6.1) (table 3).

DISCUSSION

This population-based prospective cohort study showed that the general risk for a cruciate ligament injury of the knee is relatively low among adolescents and young adults, but participation to organised sports increases the risk significantly. The risk is especially high in active young women.

The current study had a number of strengths. First, it involved a large, prospective, nationwide sample of adolescents over a remarkably long follow-up period (a total of 435 840 person-years). Second, medical treatment is equally available to everyone in Finland, ensuring the overall completeness of the used hospitalisation database. In addition, the coverage and accuracy of the National Hospital Discharge Register have been shown to be very good.13 16 17

There were also some limitations to the study. First, although the overall response rates of the adolescents were very good, they declined somewhat over the years. However, the analysis of non-respondents showed that non-respondents had a similar risk for cruciate ligament injury to that of the respondents, allowing us to conclude that non-response did not have a significant effect on the results. Second, it is possible that a few individuals with a cruciate ligament injury were treated on an outpatient basis and thus not registered in the national hospital discharge register. However, during the study years, outpatient care was rare among patients with a cruciate ligament injury.

The findings of the current population-based study are supported by previous investigations conducted among athletic populations. Such studies have shown that female athletes have up to a 4–7-fold increased risk for a cruciate ligament injury compared with their male counterparts playing at similar levels in the same sports.18^–20 The highest risks for knee injuries have been found in pivoting sports such as downhill skiing, basketball, football, team handball, floorball, soccer and ice-hockey.2–4 20

The reasons for the increased risk of knee ligament injury in female athletes are multifactorial, the most common explanations being anatomical, hormonal and training related.21 22 In our very recent prospective study among top-level female athletes, up to 70% of the cruciate ligament injuries were caused without contact with other players, thus indicating that intrinsic, person-related factors are of importance.23 In that study, the incidence of cruciate ligament injury of the knee was 5348 per 100 000 person-years (5.3% per year). In the current study, the cruciate ligament injury incidence in our general female population was clearly lower, being only 30 (95% CI 23 to 37) per 100 000 person-years.

A major problem after a cruciate ligament injury is that, regardless of treatment, athletes with the injury retire from the active participation at a higher rate than athletes without this injury.24 25 This is usually due to residual knee instability, reduced range of motion, stiffness and pain, alone or in combination. In addition, unwillingness to take risks for further injury or distrust in the body’s capacity contribute to the decision to retire.26

Our findings emphasise that strong and immediate action should be taken to prevent sport-related knee injuries. In fact, it should be regarded as one of the most important factors in training and coaching.27 28 Caraffa et al conducted a prospective controlled, non-randomised study to assess the effect of proprioceptive training (including wobble-board exercises) in preventing ACL injuries among soccer players.29 They showed a significantly lower incidence of ACL injury in the training group compared with controls (0.15 vs 1.15 injuries per team/season, respectively). However, although balance-board training has been shown to be effective in preventing ankle injuries, there is not yet enough evidence that this type of training alone can reduce the risk for knee injuries.28 30 31

Multidimensional training trials have, in general, indicated that sport-related injuries could be prevented by using these prophylactic training regimens.28 For example, Hewett et al found that a specific three-phase plyometric neuromuscular training programme for female athletes significantly reduced the incidence of knee injuries in the intervention group.6 It is likely that the preventive effect of this and other training programmes was the sum effect of several individual methods. However, because of the complexity of designs of these studies, it is currently impossible to clarify the effectiveness of individual components of the intervention programme used. In addition, the effects of detraining on injury risk are largely unknown, although it is likely that a year-round and regular neuromuscular training is needed to maintain the injury-preventive effect.

CONCLUSION

Our prospective, population-based epidemiologic study showed that among adolescents engaged in organised sports the risk for cruciate ligament injury of the knee is clearly raised, and is especially high in athletic young females. Preventive measures should be adopted to decrease the short-term and long-term burden of this severe injury.