Knee osteoarthritis in the former elite football players and the ordinary population: a comparative cross-sectional study

ABSTRACT A cross-sectional case–control study compared subjective knee function, quality of life and radiographic knee osteoarthritis (OA) between 45 former elite football players and an age-matched general male population. Participants completed the Knee OA Outcome Score (KOOS), a quality-of-life assessment (EQ-5D-3 L) and standing knee radiographs. Among the players, 24 (53%) sustained at least one moderate or severe knee injury, while 21 (47%) did not recall any injury. Players with previous knee injuries reported significantly lower knee-specific and general quality-of-life scores (KOOS 69; EQ-5D-3 L 0.69 (0.2)) compared to the non-injured players (KOOS 92; EQ-5D-3 L 0.81 (0.2)) or the control population (KOOS 90; EQ-5D-3 L 0.83 (0.2)). The injured knees had higher radiographic OA Kellgren-Lawrence (KL) scale grades 1.7 (1.3) than the knees of the non-injured players 0.8 (1.0) or the control knees 0.8 (1.0) Former elite football players who had previously sustained a moderate or severe knee injury reported inferior knee function and lower quality of life. Injured knees had higher levels of radiographic OA. Non-injured players reported similar knee and general function and their knees had similar grades of OA to those in the control group. The defining moment for long-term knee preservation in football should be injury prevention protocols.


Introduction
Knee injury and continuous sporting activity with injured knees have been identified as major risk factors for the development of early osteoarthritis (OA) (Buckwalter and Lane 1997;Garrick and Requa 2003;Buckwalter and Martin 2004).There is general consensus that previous knee injury or surgery, advanced age or lower bone density, female gender and obesity are risk factors for knee OA (Blagojevic et al. 2010;Gouttebarge et al. 2018;Freiberg et al. 2020;Parekh et al. 2021).Football is one of the sports with the highest biomechanical stress on the knee joint and has undergone significant changes in recent years due to increasing physical demands (Drawer and Fuller 2002;Ekstrand et al. 2011;Arliani et al. 2014).Knee OA is reported to be the most common chronic injury suffered by professional football players after the end of their careers (Turner et al. 2000;Junge and Dvorak 2004;Wong and Hong 2005) and the reported prevalence of knee OA in retired football players was between 14% and 80% (Elleuch et al. 2008;Klünder et al. 2009;Krajnc et al. 2010).Turner et al., who studied the quality of life of 284 football players, reported that players with OA in all joints had a worse quality of life than players without OA (Turner et al. 2000).Volpi et al. reported that former football players experience a higher incidence of hip and knee arthroplasty (Volpi et al. 2019).Although Lohkamp et al. confirmed a significantly higher incidence of knee OA in their recent systematic review, they could not prove a higher rate of joint replacements in the football players (Lohkamp et al. 2017).Albeit retired football players presented with a higher prevalence of OA, they reported less pain compared to military personnel (Paxinos et al. 2016).If the higher prevalence of OA in football players is preventable, it is important to take primary preventive measures.
The aim of the presented cross-sectional study was to compare subjective knee function, general quality of life and radiological grade of knee OA between a group of former elite football players and their age-matched male control group.The hypothesis was that players who experienced a severe knee injury during their career report lower knee function and quality of life and have higher levels of radiographic knee OA than the non-injured players or the control group.

Materials and methods
The protocol for this cross-sectional case-control study was approved by the National Medical Ethics Committee (No. 88/ 01/08).The study was conducted between January 2018 and December 2019 at the University Medical Centre Maribor, Slovenia.Fifty-four contacts of retired elite male Slovenian football players were provided by the club NK Maribor.The professional career of the players had ended at least one year before the invitation to study.KOOS pain was defined as the primary study outcome.An apriori total sample size calculation based on the KOOS pain minimally clinically important difference of 15 points within three groups was 41 (estimated effect size 0.65; estimated power 0.95).Forty-five players (mean age 50 (10) years; BMI 26.2 (2.8) kg/m2) responded positively to the invitation and gave their informed consent.The players were age-matched to a control group consisting of 45 randomly selected male visitors to the trauma emergency department who required a knee radiograph for a recent acute knee injury (mean age 51 (9) years; BMI 25.9 (2.4) kg/m2).Participants in the control group were selected according to the following criteria: no participation in high-risk sports since the age of 14, no gross deformity or history of injury to the lower limbs or knee instability, tolerable participation in low-impact recreational sports.

Clinical and radiographic evaluation
The players and the control group participants provided information on demographics (age, BMI), general and knee-specific medical history (major injuries, musculoskeletal surgeries, systemic diseases), intensity of daily activities and sports background (professional and overall career length, playing position, football leg dominance).Each participant completed a Knee Osteoarthritis Outcome Score (KOOS) translated into Slovenian language (Knee injury and Osteoarthritis Outcome Score (KOOS),) and a general quality of life assessment European Quality of Life in Five Dimensions 3-level (EQ-5D-3L) (EQ-5D-3L -EQ-5D 2022).A country-specific Time-Trade-Off (TTO) value set was used to determine the index numbers corresponding to the patients' 5-digit health status score (Prevolnik Rupel et al. 2020).Participants in the control group had completed the questionnaires according to knee function prior to the last acute injury.Standard bilateral standing AP radiographs were obtained and graded for knee OA using the Kellgren-Lawrence criteria (KL) (KELLGREN and LAWRENCE 1957).

Data management and statistical analysis
Descriptive statistics were used to present the participants' demographic data, medical history and sport participation.Categorical variables were presented as frequencies with associated percentages; continuous variables were presented as means with standard deviation (SD).The first statistical analysis (shown in Table 1) was based on data from one individual (total N = 90).The group of players was first divided into two subgroups: non-injured players (N = 21) and injured players (N = 24).Injured players (moderateto-severe injury) were regarded as those that required at least 4 weeks of absence from training.The two subgroups and the control group (N = 45) were analyzed for statistical differences by one-way ANOVA, followed by a Tukey posttest.The second statistical analysis (shown in Table 2) was based on one knee joint (total N = 180).The knee joints of the football players (N = 90) were further divided into three subgroups: non-injured knees (N = 45), knees injured during football carrier (N = 36), knees injured during and after carrier (N = 9).The three subgroups of players' knees and the knees of the control group were analyzed for statistical Note: MSK -musculoskeletal; KOOS -Knee Osteoarthritis Outcome Score; EQ-5D-3 L (TTO) -EuroQol-5D 3-level time-trade-off; KL -radiographic knee OA Kellgren-Lawrence grades.* -statistically significant differences between former football players and control group.
differences in the number of knee injuries, knee surgeries and the level of radiographic knee OA by one-way ANOVA followed by a Tukey post-test.Statistical analysis was performed using the statistical software IBM SPSS ® Statistics 23 (IBM Corp, Armonk, NY, USA).A level of significance in all tests was set at p < 0.05.

Results
During their careers, 24 (53%) players sustained at least one moderate to severe knee injury.A total of 95 major injuries were reported; the most injuries suffered by a single player was 5.All of these players were assigned to the injured football players (IFP) group.Twenty-one (47%) players could not recall a moderate-to-severe injury and were assigned to the noninjured football players (NFP) group.There was no statistically significant difference in professional career length between the two groups, age or BMI, but players from the IFP group had a shorter football career overall (p = 0.044).After their careers ended, 6 (25%) players from the IFP group and 3 (14%) players from the NFP group suffered a knee injury.In the IFP group, 14 (58%) players cited knee problems as the reason for their retirement, accounting for 68% of all reasons for retirement due to injury (p < 0.001).Twenty (83%) players from the IPF group required knee surgery during their career (p = 0.001).The average KL grading in NFP was 1.0 for the worse knee and 0.7 for the better knee (p = 0,088); in the IFP group, 2.6 for the worse knee and 1.8 for the better knee (p < 0.001).The players from the IFP group had a statistically significantly worse KOOS score and KL score (p < 0.001).EQ-5D-3 L values of NFP were like those of the control group and higher than those of the IFP group (p = 0.010).The IFP group had a 3.1 times higher prevalence (46.5% vs. 15.4%) of moderate and severe OA (grade III and IV) than players in the NFP group (p = 0,088).Six patients from the matched control group (13%) had suffered a minor knee injury and 2 patients had already undergone minor knee surgery.Radiographic OA was present in 11 (24.4%)knees (p = 0,027).Their mean KL grading was 1.0 for the worse knee and 0.8 for the better knee (Table 1).The IFP group had a statistically significant impact on current general quality of life, other MSK health issues, and compared to both NFP and control groups (p = 0.049).
Regarding knee injuries, we found that out of 90 knees, 45 were never injured, 36 were injured during the career and 9 were injured during and after the career.A knee injured during and after career required an average of 2 surgeries and had an average KL score of 3.4, followed by knees injured during career that required 0.64 surgeries (and had an average KL score of 1.7.The knees of the non-injured footballers and the knees of the control group had a similar KL score (0.82 vs 0.76) (Table 2).

Discussion
The most important findings of this cross-sectional study are an increased prevalence of knee OA, reduced knee function and general quality of life, and increased pain and analgesics use in football players who had sustained a severe knee injury during their career.The non-injured football players had similar knee function and quality of life as well as the same radiological OA scores as the corresponding non-sporting controls.
Severe knee injury appears to lead to more knee surgeries and higher radiographic OA.Previous knee injuries lead to more post-career surgical procedures, even in the absence of new injuries.The reported prevalence of OA in Slovenian professional football players was 64%, which is consistent with reports by other authors (Larsen et al. 1999;Elleuch et al. 2008;Arliani et al. 2014;Prien et al. 2020).Injured knees require more surgical interventions and have a subjectively worse quality of life.Our data support the association between knee injuries and knee OA.The prevalence of OA in previously injured Slovenian football players is 84.4% (almost 100% after ACL tear) compared to only 15% in uninjured players.Muckle (Muckle 1983) and Chantraine (Chantraine 1985) reported knee OA in all football players 10 years after surgery (meniscectomy), but only in one-third of the non-operated knees.Larsen reported a 92% prevalence of knee OA after meniscectomy and 100% after ACL reconstruction (Larsen et al. 1999).Deacon reported radiographic OA of the knee in 66% of players, which increased to 78% in players after an intra-articular lesion (Deacon et al. 1997).Sandmark reported a twofold higher likelihood of knee OA in uninjured football players compared with the general population (Sandmark and Vingård 1999) and Elleuch demonstrated the presence of knee OA in 80% of former football players (Elleuch et al. 2008).However, a recent meta-analysis by Spahn et al. showed only a slightly increased risk of developing OA in the absence of knee injury (Spahn et al. 2015).In contrast, we found no higher risk of knee OA in the group of uninjured players (15%) compared to the control group (22%).Authors analyzing the consequences of injury in the general population reported a lower likelihood of knee OA after intraarticular injury compared to injured professional football players (Bartz and Laudicina 2005;Roos 2005).Meredith reported a twofold higher likelihood of knee OA after total meniscectomy compared to partial meniscectomy (Meredith et al. 2005) while Prien reported, in her MRI study, that the integrity of the meniscus is key to the prevention of OA in the knee (Prien et al. 2020).Two recent meta-analyses showed a clear association between playing football and the development of knee OA, even after adjusting for severe knee injuries, although they agree that injuries play a major role in the development of OA (Driban et al. 2017;Freiberg et al. 2021).We found that injured players (IFP) have significantly more problems with injured knees in everyday life compared to non-injured players (NFP).
Various reasons besides injury status may influence prevalence of knee OA in former football players.They are connected to universal differences such as physical load differences in different playing positions and playing levels, as well as individual anatomical variation (varus knee axis) (Parekh et al. 2021).

Limitations of the study
When interpreting the results of our study, we have to take into account that the patients had different surgical procedures.Some had open knee surgery and arthroscopy or both, depending on the timing of the surgery.The possibility of misdiagnosis and inadequate treatment in older players must be considered.As MRI was not available more than 25 years ago, diagnosis was more difficult, and many pathologies could have been overlooked.The medical history data collected from the subjects could not be verified in their medical records, so some subjective bias might be present.Another limitation is the control group, which probably does not represent the general population.It consists of active males who are more likely to have a knee injury requiring a visit to the emergency department.We also did not consider their working environment, which could influence development of knee OA.It has been shown that the minimum clinically important difference in KOOS score is 8 to 10 points (76).In all five sections of the KOOS questionnaire, the difference between injured and non-injured players and the control group was significant (15 to 27 points).

Conclusions
Football-related injuries are a major risk factor for early OA of the knee.Former elite football players who had previously sustained at least one moderate or severe knee injury reported inferior knee function and lower quality of life.Their injured knees evidently demonstrated higher degree of radiographic OA.Quality of life also seems to be affected by previous injuries, subsequent surgeries, and decreased knee function.In contrast, the non-injured players reported similar knee and general function and their knees had similar degrees of OA to those in the control group.Knee injuries, rather than football participation, have a major impact on knee function.

Table 1 .
Demographics, medical history, and subjective evaluation of the non-injured, injured retired football players, and their age/sex matched control.

Table 2 .
Injury history and radiographic evidence of OA in non-injured, injured, and control knee joints.