eTable 7. The results of the studies included.
| Study | Clinical endpoint (outcome) | Prevalence or incidence, n (%) | Risk estimator | Remarks | ||||
| Effect estimator | Exposure group or cases | Comparison group or controls | Effect estimator | Effect size [95% CI] | Adjustment for confounders | |||
| Retrospective cohort study | ||||||||
| Kujala et al. (1994) | Hospital admission due to knee osteoarthritis | Outcome incidence | 5 (2.0) | 18 (1.3) | Odds ratio | Professional soccer: 1.55 [0.57; 4.21]*1 | Not adjusted Sex: Only men included Age: Similar average age of exposure and comparison groups |
For the sake of better comparability of results, the odds ratio was calculated, rather than the relative risk as is more usual for incidence. The relative risk was similarly high, at 1.54 [0.58; 4.11]. |
| Case–control studies | ||||||||
| Sandmark & Vingard (1999) | Knee joint prosthesis due to primary tibiofemoral knee osteoarthritis | Prevalence of exposure | Professional soccer: Men: 41 (12.6) Women: 0 (0.0) Amateur soccer: Men: 101 (31.1) Women: 5 (1.7) |
Professional soccer: Men: 19 (7.2) Women: 0 (0.0) Amateur soccer: Men: 77 (29.2) Women: 1 (0.4) |
Odds ratio | Professional AND amateur soccer: Men: 2.0 [1.4; 2.8] Professional soccer: Men: 1.86 [1.05; 3.29] *1 Amateur soccer: Men: 1.10 [0.77; 1.56]*1 Women: 4.80 [0.56; 41.31]*1 |
Professional AND amateur soccer: Age body mass index, physical strain at work/in the household/in leisure time: adjusted Professional soccer: Not adjusted Amateur soccer: Not adjusted Valid for all models: Sex: Only men included |
– |
| Thelin et al. (2006) | Radiographic diagnosis of tibiofemoral knee osteoarthritis or knee joint operation (osteotomy or joint replacement) | Prevalence of exposure | Men: 140 (41.4) Women: 5 (1.1) |
Men: 91 (31.2) Women: 6 (1.5) |
Odds ratio | Professional AND amateur soccer: Men: Model 1: 1.56 [1.12; 2.17] Model 2: 1.52 [1.04; 2.20] Model 3: 0.94 [0.61; 1.44] Women: Model 1: 0.6 [0.23; 2.50] |
Model 1: Not adjusted Model 2: Smoking, body mass index, heredity, work in construction: adjusted Model 3: Smoking, body mass index, heredity, work in construction, previous knee joint injury (fracture, ligament or tendon injury, meniscus injury): adjusted Valid for all models: Sex: Only men or only women included Age: Similar average age of cases and controls |
The study did not distinguish between professional and amateur soccer, so no odds ratios specifically for professional soccer are available. The analyses of models 1–3 for men were also adjusted for professional-level sport; however, this did change the results. These data were not included in the study. |
| Vrezas et al. (2010) | Radiographic diagnosis of knee osteoarthritis | Prevalence of exposure | No soccer: 178 (60.3) > 0–1 660 h: 29 (9.8) 1 660–4 000 h: 41 (13.9) 4 000–7 800 h: 32 (10.8) ≥7 800 h: 15 (5.1) |
No soccer: 208 (63.6) > 0–1 660 h: 35 (10.7) 1 660–4 000 h: 34 (10.4) 4 000–7 800 h: 19 (5.8) ≥ 7 800 h: 16 (4.9) |
Odds ratio | Professional AND amateur soccer: Model 1: Reference category: 1.0 > 0–1 660 h: 1.3 [0.7; 2.3] 1660–4 000 h: 1.9 [1.0; 3.1] 4 000–7 800 h: 2.2 [1.1; 4.4] ≥ 7 800 h: 1.2 [0.5; 2.8] Model 2: Reference category: 1.0 > 0–1 660 h: 1.1 [0.5; 2.1] 1 660–4 000 h: 2.0 [1.0; 3.8] 4 000–7 800 h: 2.2 (1.0; 5.0] ≥ 7 800 h: 1.4 [0.6; 3.6] Dichotomous analysis (no soccer vs. soccer at any time): 1.62 [1.04; 2.51] |
Model 1: Age, region: adjusted Model 2: Age, region, body mass index, cumulative lifting/carrying, kneeling/squatting, jogging/athletics: adjusted Dichotomous analysis: Age, region, body mass index, cumulative lifting/carrying, kneeling/squatting, jogging/athletics: adjusted Valid for all models: Sex: Only men included |
The study did not distinguish between professional and amateur soccer, so no odds ratios specifically for professional soccer are available. The odds ratio for dichotomous analysis was derived from a personal communication from one of the study authors (A.Seidler). |
| Cross-sectional studies | ||||||||
| Fernandes et al. (2018) | Diagnosis of knee osteoarthritis by a physician | Outcome prevalence | 341 (28.3) | 500 (12.2) | Relative risk | Professional soccer: Model 1: 3.53 [3.15; 3.96] Model 2: 3.73 [3.33; 4.17] Model 3: 2.69 [2.36; 3.07] Model 4: 2.18 [1.73; 2.77] |
Model 1: Not adjusted Model 2: Age, body mass index: adjusted Model 3: Age, body mass index, knee joint injury: adjusted Model 4: Age, body mass index, knee joint injury, Heberden’s nodes, knee joint malalignment at age of around 20 years, high-risk occupation for knee osteoarthritis, index finger/ring finger length ratio, comorbidities: adjusted Valid for all models: Sex: Only men included |
The adjustment for knee joint malalignment may be excessive, as professional soccer players are more likely to have a varus knee joint than the general population. The result in model 2 (3.61 [2.90; 4.49]) deviates somewhat from the figures in the study due to different calculation parameters in RevMan 5.3. |
| Knee joint replacement | 134 (11.1) | 157 (3.8) | Professional soccer: Model 1: 2.88 [2.31; 3.60] Model 2: 3.61 [2.90; 4.49] Model 3: 2.33 [1.84; 2.95] Model 4: 2.10 [1.42; 3.14] |
|||||
| Iosifidis et al. (2015) | Clinical knee osteoarthritis | Outcome prevalence | 11 (9.1) | 15 (8.2) | Odds ratio | Professional soccer: 1.11 [0.49; 2.50]*1 |
Not adjusted Sex: Only men included |
The study reported only odds ratios in relation to the whole population of all athletes investigated. In the case of calculation of an odds ratio for a prevalence of > 10%, the value could be overestimated in relation to the corresponding relative risk. |
| Radiographic diagnosis of knee osteoarthritis | 20 (22.0) | 21 (12.9) | Odds ratio | Professional soccer: 1.90 [0.97; 3.74]*1 |
||||
| Kujala et al. (1995) | Radiographic diagnosis of tibiofemoral and patellofemoral knee osteoarthritis | Outcome prevalence | 9 (29.0) | 1 (3.4) | Odds ratio | Professional soccer: Model 1: 12.3 [1.35; 112.06] Model 2: 5.21 [1.14; 23.8] |
Model 1: Age: adjusted Model 2: Type of sport, age, body mass index, occupational stress from heavy work work involving kneeling and squatting, knee joint injury: adjusted Valid for all models: Sex: Only men included |
The result in model 1 (12.3 [1.35; 112.06]) deviates somewhat from the figures in the study due to different calculation parameters in RevMan 5.3. |
| Radiographic diagnosis of tibiofemoral knee osteoarthritis | 8 (25.8) | 0 (0.0) | n.d. | – | – | |||
| Radiographic diagnosis of patellofemoral knee osteoarthritis | 5 (16.1) | 1 (3.4) | Odds ratio | Professional soccer: 5.38 [0.59; 49.20]*1 |
– | |||
| Roos et al. 1994) | Radiographic diagnosis of tibiofemoral knee osteoarthritis | Outcome prevalence | Professional soccer: All: 11 (15.5) With knee injury: 5 (33.3) Without knee injury: 6 (10.7) Amateur soccer: All: 3 (4.2) With knee injury: 4 (12.5) Without knee injury: 5 (2.7) |
All (matched to age of professional soccer players): 4 (2.8) With knee injury: 2 (15.4) Without knee injury: 7 (1.3) |
Odds ratio | Professional soccer: a) 11.47 [4.57; 28.79]*1 b) 6.32 [1.94; 20.66]*1 c) 5.73 [2.01; 16.29]*1 d) 9.46 [3.06; 29.24]*1 Amateur soccer: a) 2.7 [1.0; 6.8] b) 2.73 [1.07; 6.98]* 1 |
Professional soccer: a) Not adjusted b) Not adjusted, age: use of age-matched comparison probands c) Knee joint injury: adjusted d) Exclusion of participants with knee joint injury (non-adjusted) Amateur soccer: a) n.d. b) Not adjusted Valid for all models: Sex: Only men included |
Calculation of the odds ratio for c) with adjustment for knee joint injury was done by means of the Cochran–Mantel–Haenszel test. The odds ratio for amateur soccer players was reported in the study, but it was not made clear whether any adjustment was carried out. We therefore calculated this odds ratio—without adjustment—and found a slightly different value of 2.73 [1.07; 6.98]. |
| Tveit et al. (2012) | Diagnosis of knee osteoarthritis by a physician | Outcome prevalence | 67 (18.2) | 163 (13.0) | Odds ratio | Professional soccer: Model 1: 1.52 [1.11; 2.07] Model 2: 1.46 [1.04; 2.05] Model 3: 1.13 [0.75; 1.72] |
Model 1: Age: adjusted Model 2: Age, body mass index, occupational load: adjusted Model 3: Age, body mass index, occupational load, soft-tissue injury of knee joint: adjusted Valid for all models: Sex: Only men included |
The result in model 2 (1.15 [0.56; 2.36]) deviates somewhat from the figures in the study due to different calculation parameters in RevMan 5.3. |
| Knee joint replacement | 12 (3.3) | 30 (2.4) | Professional soccer:Model 1: 1.40 [0.71; 2.77] Model 2: 1.15 [0.56; 2.36] Model 3: 1.21 [0.38; 3.84] | |||||
*1 The review authors calculated this risk estimator based on the prevalence or incidence data given in the original study. No adjustment could be made for these calculations.
Bold type indicates statistical significance.
CI, Confidence interval; h, hours; n.d., no data