Reference | Study population | Age (years) | Participation rate (%) | Exposure measured by | Diagnostic criteria | Adjusted for/Matched by | Comparisons | Results (OR, 95% CI) | Design | Quality of the study (+, ++, +++)/Strength/Weakness |
Anderson and Felson, 198812 | 2428 men and 2765 women from a population study (HANESI) kneeOA:105 men210 women | 35–74 | 75 | Questionnaire graded in relation to heavy work by experienced occupational professionals | Kellgren and Lawrence grade 2–4 TFJSingle AP, non-weight-bearing of both knees | Race, BMI education | Heavy lifting45–54 yearsMenWomen55–64 yearsMenWomen | 1.05, 0.45 to 2.41.09, 0.31 to 3.51.88, 0.88 to 4.03.13, 1.04 to 9.4 | Cross-sectional | +++Strength: Well-described design and material. Each job title coded in relation to physical demandsWeakness: Non-weight bearing x-rays may underestimate kneeOAExposure: Measured by current occupation (risk of misclassification) Use of job titles (risk of misclassification) |
Kohatsu and Schurman, 199013 | Cases: 46 with TKRControls: 46 random sample from a large community sample | >55 | 35 | QuestionnaireOccupational history classified in light, moderate, and heavy work | Kellgren and Lawrence grade 3–4 OA | – | Moderate/very heavy vs light workAge 20–29 yearsAge 30–39 yearsAge 40–49 years | 2.3, 0.9 to 6.13.4, 0.9 to 10.83.0, 0.9 to 11.4 | Case–control | +Strength: NoneWeakness: Few participantsLow participation rate. Exclusion criteria not describedCase definition: TKR (risk of selection bias) Data presentation inadequate. No adjustment for confounders |
Felson et al, 199116 | Cases: 176 men, 279 women with kneeOAControls: 569 men, 807 women from the Framingham Heart Study | Mean: 73 | 96 | InterviewPhysical demands scored by trained coders | KneeOA, TFJ, grade 2–4Weight-bearing radiographs | AgeBMISmokingHistory of knee injuryEducation | Lift medium, heavy or very heavy demands vs no lift/no knee bendingMenWomen | 0.96, 0.49 to 1.872.53, 0.82 to 7.85 | Cohort | +++Strength: Longitudinal designCase-definition: weight-bearing radiographsWeakness: Exposure classified by job title (risk of misclassification) Few subjects with heavy physical work especially among women |
Vingård et al, 199114 | High exposure: 116 581 men (321 OA) 18 434 women (66 OA) Low exposure: 91 057 men (200 OA) 24 145 women (48 OA) | Born 1905–1945 | Register-based | OccupationScored by experienced occupational physicians | Hospitalised in 1981–83ICD-8: 713.01; ICD-10: M17 | Age, residence | High exposure vs low exposureMenBorn 1905–24Born 1925–45WomenBorn 1905–24Born 1925–45 | RR 1.2, 0.9 to 1.5RR 1.4, 1.1 to 1.9RR 1.4, 0.6 to 3.2RR 1.9, 1.3 to 2.9 | Cohort | ++Strength: High number of participantsWeakness: Exposure classification: job title, classified in light, medium, heavy (risk of misclassification) Case-definition = hospitalised with kneeOA (risk of selection bias) |
Bagge et al, 199115 | 136 men, 207 women from two subsamples of 79-year-olds in a population study from Göteborg | 79 | – | InterviewsClassified in: no, light, moderate, and heavyCategorised as an arbitrary score | Weight-bearing radiographs, Kellgren and Lawrence grade 2–4 | – | MenScore 0–1Score 2–3Score 4–9Score 10–16WomenScore 0–1Score 2–3Score 4–9Score 10–16 | 68%39%58%47%48%56%69%67% | Cohort | +Strength: NoneWeakness: Results not well-described. No analysis or adjustment for confoundersPrevalence of kneeOA high and independent of earlier workload, (risk of underestimation). Few men (n = 17) with high exposure |
Vingård et al, 199217 | Cases: 181 men, disability pension due to kneeOAControls: 298 from the general Swedish population | Born 1915–1934 | – | InterviewOccupational historyJob title classified by experienced persons | Primary diagnosis kneeOA made by a physician | Age | Medium exposure vs low exposureHigh exposure vs low exposure | RR 4.5, 2.6 to 7.6RR 14.3, 8.1 to 25.4 | Case–control | ++Strength: Well-described design and materialsWeakness: Diagnosis from physicians certificates (risk of misclassification); subjects in physically demanding occupations may have an increased risk of getting disability pension (risk of selection bias) High workload on knees may not be related to heavy lifting (risk of misclassification) |
Schouten et al, 199218 | 105 subjects from the general populationFollow-up 12 years | ⩾45 | Only 25% from baseline used in analyses | QuestionnaireOccupational history number of years employed, heavy lifting | Cartilage lossWeight bearing AP | Age, BMI, sex | Lifting heavy objects vs no liftingMediumHigh | 1.0, 0.3 to 3.020.65, 0.19 to 2.3 | Cohort | ++Strength: DesignWeakness: Few participants from first study included in the follow-up. More restrictive scoring of radiographs (risk of misclassification). Exposure: Low frequency of participants with heavy lifting. Participants with heavy lifting may have left the physical demanding occupations (risk of healthy worker effect) |
Cooper et al, 199420 | Cases: 109 (30 men, 79 women grade 3–4 OAControls: 218 age- and sex-matched (without knee pain) from a large general practice | 55–90Mean: 73 | – | QuestionnaireDetails of the main job entailed eight specific physical activities, including, heavy lifting | Symptomatic kneeOA Kellgren and Lawrence grade 3–4 TFJ±PFJWeight-bearing AP | BMI, Heberden’s nodes | Lifting weights >25 kg in an average working day | 1.4, 0.5 to 3.7 | Case–control | ++Strength: Well-described design and materials, weight-bearing radiographs, inclusion of PFJ OAInclusion of subjects with moderate/severe symptomatic OAWeakness: Exposure measured retrospectively by interviews (risk of recall bias) Small number of male participants and subjects with heavy lifting |
Elsner et al, 199637 | Cases: 115 men, 86 women with kneeOA from an orthopaedic clinicControls: 95 men, 87 women from general practice in the same area | <45: 43%>55: 57% | 61 among cases | Physical demands reported in a diary book | All degrees of radiological kneeOA | Age | Lift weights >20 kg yes/no MenWomen | 1.3, 0.73 to 2.351.5, 0.56 to 4.18 | Case–control | ++Strength: Well-described design and materialWeakness: Case definition: grade 1–4 OA (risk of misclassification) Exposure: Lifting weights (no frequency, no duration) (risk of misclassification) retrospective (risk of recall bias). Low participation rate among casesControls collected over 4 years; selection of the controls not defined (risk of selection bias) |
Sahlström and Montgomery, 199723 | Cases: 266 with kneeOAControls: 463 age- and sex-matched (sex not stated) | 47–96Mean: 72 | 71 | QuestionnaireLight, medium, heavyClassified by industrial hygienists | At least grade 1 Ahlbäck (3 mm) Weight-bearing | SittingOverweightKnee injury | Weight-bearing knee bending and lifting/carrying UnadjustedAdjusted | 1.9, 1.4 to 2.71.1, 0.7 to 1.8 | Case–control | ++Strength: Case-definition: weight bearing radiographsExposure: Assessed in two waysWeakness: Definition of exposure: medium = lifts of objects from one level to another; heavy = combined jump and lifting (risk of misclassification). Adjustment not defined |
Coggon et al, 200025 | Cases: 518 (205 men, 313 women) Controls: 518 (205 men, 313 women) from the same communityFrom three English districts in a 2-year period | 47–93Mean: 73 | 55 | InterviewOccupation held for >1 year from school ageSpecified physical activity | Cases waiting for surgery | BMIHeberden’s nodesPrevious knee injuriesMatched by sex and age | Lift at least 10 years ten times per week vs no lift>10 kgMenWomen>25 kgMenWomen>50 kgMenWomen | 1.9, 1.0 to 3.31.5, 1.0 to 2.31.7, 0.9 to 3.01.7, 1.0 to 2.81.7, 0.9 to 3.21.2, 0.6 to 2.4 | Case–control | +++Strength: Well described design, high number of participants, interviews with specification of different physical activitiesWeakness: Exposure collected retrospectively (risk of recall bias). Low participation rate especially among controls. Case definition: placed on a waiting list for TKR (risk of selection bias) |
Lau et al, 200026 | Cases: 166 men, 492 womenControls: 166 age- and sex-matched from general practice in the same regionOver a 3-month period | – | – | InterviewPhysical activity in job in which they had work for the longest period before symptoms | 28% had TKR, 15% waiting for TKR, 57%Grade 3–4 OA | Matched by sex and age | Lift 10 kg vs no lift1–10 times/weekMenWomen>10 times per weekMenWomen50 kg1–10 times per weekMenWomen>10 times per weekMenWomen | 1.7, 0.9 to 3.21.5, 1.0 to 2.25.8, 3.1 to 10.83.0, 2.2 to 4.13.5, 1.4 to 8.80.9, 0.5 to 1.77.1 3.1 to 16.22.9, 1.9 to 4.5 | Case–control | +++Strength: Well described design and materialsHigh number of participants. Exposure: Collected by interviews specified in different physical activities. Same design as Coggon et al15 Weakness: Exposure collected retrospectively (risk of recall bias) Case-definition: Subjects seeking hospitals for knee OA (risk of selection bias). Controls with knee complaints excluded |
Sandmark et al, 200027 | Cases: 325 men, 300 women with TKRControls: 264 men, 284 womenFrom 14 counties in Sweden between 1991 and 1995 | 55–70 | 80 | Interviews, questionnaireOccupational historySpecified physical activity including lifts at work in kg | TKR between 1991–1993 | BMI, smoking, sports activityAge-matched | Lifts at work vs unexposedMediumMenWomenHighMenWomen>10 years in heavy jobsMenWomen | 2.5, 1.5 to 4.41.2, 0.7 to 1.93.0, 1.6 to 5.51.7, 1.0 to 2.92.5, 1.7 to 3.62.5, 1.6 to 3.9 | Case–control | +++Strength: Well described design and material, high number of participantsWeakness: Case definition: TKR (risk of selection bias) Exposure assessment retrospective (risk of recall bias) |
Seidler et al, 200128 | Cases with kneeOA: 105 men90 womenControls without OA: 105 men90 women from an orthopaedic clinic108 from general practice | CasesMean: 53ControlsMean: 35 | 64 | QuestionnaireSpecified physical demands, weight lifted in kg/day and duration | TFJ grade 1–4PFJ excluded | Age, BMI, sports activities | MenLifting daily 20–50 kg1–10 years>10 years Lifting >50 kg/d1–10 years >10 yearsWomen | 0.5, 0.1 to 1.41.2, 0.4 to 3.01.0, 0.2 to 3.93.4, 0.7 to 17.2NS | Case–control | ++Strength: Design and materials described. Controls without radiological kneeOA and subjects with meniscal lesions excludedWeakness: Case-definition: OA grade 1–4. 66% of cases had grade 1 OA, only 10% grade 3–4 OA (risk of misclassification) Retrospective exposure assessment (risk of recall bias) |
Dawson et al, 200229 | Cases: 29 women placed on waiting list for TKR during past 12 monthsControls: 82 women age-matched from general practice | 50–70 | 45 | InterviewedOccupational risk factors in job | Waiting for TKR | AgeGeneral practitioner | Lifting >25 years vs lifting <25 yearsLifting >33 years vs lifting <25 years | 7.31, 2.01 to 26.73.58, 0.89 to 14.4 | Case–control | ++Strength: Design and materials described.Weakness: Participation rate low and high exclusion rateExposure assessment: Retrospective (risk of recall bias). Exposure defined as lifting two times per week (weight and frequency of lifting not defined) (risk of misclassification). Housewives included. Duration of exposure divided in three categories; lowest category = exposure <24 years (risk of misclassification) |
Manninen et al, 200230 | Cases: 55 men226 womenControls: 524 age- and sex-matched from same population | 55–75 | Cases: 78.5Controls: 65.6 | Telephone interviewPhysical demands before the age of 49 years | First surgery for kneeOA | BMIKnee injury | Liftingvs low exposureMedium levelMenWomenHigh levelMenWomen | 1.35, 0.57 to 4.160.90, 0.55 to 1.50.92, 0.5 to 2.391.11, 0.71 to 1.75 | Case–control | +++Strength: High number of female participants, interviews with specification of different physical activitiesWeakness: Exposure assessment: retrospective (risk of recall bias) Few men (n = 55) Cases defined as TKR (risk of selection bias) |
Yoshimura et al, 200431 | Cases: 101 women with kneeOA from six hospitals in three cities of JapanControls: 101 women, random sample from the local population | 45 | Cases: 84 Control: 59 | QuestionnaireSpecific physical demands | Radiological grade 3–4 kneeOA | Age-matchedAdjusted for potential risk factors (not further defined) | Lifting weights >25 kg vs no lifting on an average dayFirst jobMain job | 1.0, 0.50 to 2.001.91, 0.92 to 3.96 | Case–control | ++Strength: Design and materials well describedWeakness: Differences in participation rate among cases and controls. 50% of cases had a previous knee injury compared with 14% controls. Few exposed to heavy lifting. Cases defined as TKR (risk of selection bias). Exposure assessment: retrospective (risk of recall bias) |
AP, anterior-posterior; PFJ, patello-femoral joint.
+, Poor quality score 1–5; ++, medium quality score 6–10; +++, high quality score >10.