1. Top of page
  2. Abstract


To determine the associations between cumulative occupational physical load (COPL) and 3 definitions of knee osteoarthritis (OA).


Cross-sectional analyses were performed from 2 population-based cohorts (n = 327). Eligible symptomatic participants were those with pain, aching, or discomfort in or around the knee on most days of a month at any time in the past and any pain in the past 12 months. Asymptomatic participants responded “no” to both knee pain questions. Self-reported COPL was calculated over each participant's lifetime and then categorized into quarters (QCOPL). Radiographic OA (ROA) and symptomatic OA (SOA) were defined by Kellgren/Lawrence grade ≥2, with SOA also including pain. Magnetic resonance imaging (MRI) OA was defined using criteria by Hunter et al. Logistic regression, adjusted with population weights, examined the associations between QCOPL and each of ROA, SOA, and MRI-OA after controlling for covariates and two-way interactions.


Participants had a mean ± SD age of 58.5 ± 11.0 years and a mean ± SD body mass index of 26.3 ± 4.7 kg/m2. Of those, 109 (33.3%) had ROA, 102 (31.2%) had SOA, and 131 (40.1%) had MRI-OA. Compared with QCOPL-1, increased odds of ROA were found for QCOPL-4 (odds ratio [OR] 3.15, 95% confidence interval [95% CI] 1.02–9.70) and QCOPL-3 (OR 4.19, 95% CI 1.55–11.34). Statistically significant relationships were found in SOA (QCOPL-4: OR 8.16, 95% CI 1.89–35.27; QCOPL-3: OR 5.73, 95% CI 1.36–24.12) and MRI-OA (QCOPL-4: OR 9.54, 95% CI 2.65–34.27; QCOPL-3: OR 9.04, 95% CI 2.65–30.88; QCOPL-2: OR 7.18, 95% CI 2.17–23.70).


Occupational activity is associated with knee OA, with dose-response relationships observed in SOA and MRI-OA.


  1. Top of page
  2. Abstract

Osteoarthritis (OA) is the most common form of arthritis and joint disease in the world ([1]). As the leading cause of chronic pain and disability worldwide, it is a major public health problem. The knee is the most commonly affected joint. An in-depth understanding of risk factors leading to the development of knee OA can inform researchers and clinicians about important clues for disease prevention and early treatment strategies.

The pathophysiology of OA is undoubtedly multifactorial; it is thought to evolve from the complex interaction of multiple risk factors, both modifiable (e.g., weight, injury, abnormal loading) and unmodifiable (e.g., age, sex, genetics) ([2, 3]). As one source of joint loading, performing occupational tasks such as heavy lifting, kneeling, and squatting has been linked to knee OA ([4, 5]). This knowledge has been reflected in labor policy in Germany and Denmark, where OA is considered an occupational disease ([4, 6]). Yet, gaps remain in our understanding about the relationship of occupation as a risk factor for knee OA.

First, studies have often referred to occupation as paid employment outside the home ([1, 5]). More broadly, it can be defined as what people do to occupy themselves with some regularity or consistency by contributing to the productivity of their communities through engagement in the workforce or home ([7]). The narrower definition often excludes homemakers, students, and retirees in an analysis of occupational joint loading, thereby creating a disparity in our knowledge about occupational risk in women. In addition, previous work has often focused on current job ([8]), principle job ([9]), or longest held job ([10]) instead of a more comprehensive calculation of occupational exposure. Accurately and inclusively quantifying cumulative occupational physical load (COPL) to include all historical occupations as well as the role of homemakers may supply new perspectives on this OA risk factor. This study utilized a new, broader definition of occupation and includes exposures from all previous occupations to study the role of occupation in joint loading to a fuller extent.

Second, most studies involving occupation and knee OA have focused on moderate to severe radiographic disease. Examining individuals across the spectrum of disease severity may also provide further insight. To detect and quantify signs that may be indicative of early OA, magnetic resonance imaging (MRI) has become increasingly useful in research settings. Pathologic joint changes, including cartilage defects, bone marrow edema (BME), cysts, synovitis, or meniscal lesions visualized on MRI, may aid in studying the association between these changes and joint pain in the absence of radiographic findings ([11]). A new structural MRI-OA definition developed by a group of leading OA experts allows the measurement of earlier disease ([12]). Using this new disease definition, the relationship of occupational activities to early OA can be examined in a way not previously explored.

Therefore, the objective of the current study was to examine the relationship of COPL to knee OA using 3 distinct definitions: radiographic OA (ROA), symptomatic OA (SOA), and MRI-OA.

Box 1. Significance & Innovations

  • Using 3 definitions of osteoarthritis (OA), we found an independent relationship between occupational activities and OA in a population-based cohort.
  • A new definition of OA using magnetic resonance imaging was applied to identify people with early disease. As such, this research broadens the understanding of the role of occupational activities and the onset of early OA.
  • Our findings suggest that future attention should be given to developing and evaluating OA prevention strategies in the work place and home.


  1. Top of page
  2. Abstract


Two cohorts of participants were recruited from the greater Vancouver area, Canada. The first cohort, the Model for the Diagnosis of Early Knee Osteoarthritis (MoDEKO), is a population-based study that aimed to develop a clinical prediction model to identify early knee OA using a combination of clinical tests, imaging techniques, and biomarkers ([13]). The second cohort, the Asymptomatic Cohort for Early Knee Osteoarthritis (ACE-KOA), consisted of a population-based sample of people without knee pain. The current study used baseline data collected in MoDEKO between 2002 and 2004 and in ACE-KOA between 2008 and 2009.

Individuals eligible for MoDEKO were between ages 40 and 79 years; had experienced pain, aching, or discomfort in or around the knee on most days of the month at any time in the past; and had experienced pain, aching, or discomfort in or around the knee in the past 12 months. Individuals were ineligible if they had inflammatory arthritis/fibromyalgia, knee arthroplasty, knee injury or surgery within the past 6 months, or knee pain referred from the hips or back (determined by clinical examination). Those who were unable to undergo MRI were also excluded. In participants with bilateral knee pain, the more symptomatic knee was used as the study knee. To ensure appropriate distribution, recruitment was organized into 8 strata by sex and age (40–49 years, 50–59 years, 60–69 years, 70–79 years), with each stratum capped at 34 participants. The enrollment target for MoDEKO was determined by sample size calculations completed by researchers prior to recruitment.

The multistage recruitment strategy for MoDEKO was described in detail previously ([13]). Briefly, a random list of households was generated from the telephone directory listings and invitation letters were mailed to these households (n = 8,523). Of the 5,231 English-speaking people who were reached by telephone, 3,269 (62.5%) were screened and 265 (8.1%) met the eligibility criteria. Another 10 individuals were excluded during the physical assessment, leaving 255 participants in the MoDEKO cohort (Figure 1).


Figure 1. Study recruitment. One participant was excluded due to incomplete data. MoDEKO = Model for the Diagnosis of Early Knee Osteoarthritis; ACE-KOA = Asymptomatic Cohort for Early Knee Osteoarthritis; MRI = magnetic resonance imaging.

Download figure to PowerPoint

The ACE-KOA cohort was recruited from the same population as MoDEKO using a similar multistage protocol (Figure 1). Eligible participants were between ages 40 and 79 years and had responded “no” to both knee pain inclusion criteria used in MoDEKO. The exclusion criteria and stratum sampling by age and sex were identical to those of MoDEKO, except that each sex stratum was capped at 12 and each age stratum was capped at 22. The enrollment target for ACE-KOA was determined by sample size calculations completed by researchers prior to recruitment. Invitation letters were mailed to a randomly generated list of households (n = 4,300), followed by a standardized telephone screening. Households were excluded if contact could not be established (n = 1,745 [40.5%]) or if the residents could not speak English (n = 200 [4.6%]). Of the 2,355 English-speaking people who were reached by telephone, 1,091 (46.3%) were screened and 104 (9.5%) met the eligibility criteria. During the physical assessment, a further 31 individuals were excluded, leaving 73 participants in ACE-KOA. MoDEKO and ACE-KOA were combined into a single cohort (n = 328); 1 participant was excluded due to incomplete occupational data (n = 327). All participants gave written informed consent. The study was conducted in compliance with the Declaration of Helsinki and ethics approval was obtained from the University of British Columbia Clinical Research Ethics Board.

Outcome variables

We employed 3 definitions of OA. A participant was classified as having ROA if their knee radiograph received a Kellgren/Lawrence (K/L) grade ≥2. SOA was defined as having a K/L grade ≥2 in addition to knee pain. Participants were classified as having MRI-OA based on the definition by Hunter et al ([12]), excluding the bone attrition criterion (Table 1).

Table 1. Hunter et al definition of OA with MRI ([12])*
  1. OA = osteoarthritis; MRI = magnetic resonance imaging.

Tibiofemoral OA on MRI: the presence of both group A features or 1 group A feature and ≥2 group B features
Group A
Definite osteophyte formation
Full-thickness cartilage loss
Group B
Subchondral bone marrow lesion or cyst not associated with meniscal or ligamentous attachments
Meniscal subluxation, maceration, or degenerative tear
Partial-thickness cartilage loss
Bone attrition
Patellofemoral OA on MRI: all of the following involving the patella and/or anterior femur
Definite osteophyte formation
Partial- or full-thickness cartilage loss

Knee radiographs were performed using a fixed flexion technique with the SynaFlexor positioning frame in weight bearing and a skyline view in the supine position ([14]). Two blinded independent readers scored the radiographs on the K/L scale of 0–4 ([15]).

MRI was performed on a GE 1.5T magnet using a transmitter–receiver extremity knee coil. The protocol involved 4 MRI sequences: 1) fat-suppressed T1-weighted 3-dimensional spoiled gradient-recalled acquisition in the steady state sequence with images obtained in the sagittal plane with reformat images in the axial and coronal planes (repetition time [TR] 52 msec, time to echo [TE] 10 msec, flip angle 60°, field of view [FOV] 12 cm, matrix 256 × 128, section thickness 1–1.5 mm, with 1 signal averaged); 2) fat-suppressed T2-weighted fast spin-echo (FSE) sequence with images obtained in the coronal plane (TR 3,000 msec, TE 54 msec, echo train length [ETL] 8, FOV 14 cm, matrix 256 × 128, section thickness 4 mm, with an intersection gap of 1 mm with 2 signals averaged); 3) T1-weighted FSE sequence with images obtained in the oblique sagittal plane (TR 450 msec, TE minimum full, ETL 2, bandwidth 20 Hz/pixel, FOV 16 cm, matrix 384 × 224, section thickness 4 mm, with an intersection gap of 1 mm with 2 signals averaged); and 4) T2-weighted FSE sequence with images obtained in the oblique sagittal plane (TR 4,025 msec, TE 102 msec, ETL 17, bandwidth 20 Hz/pixel, FOV 16 cm, matrix 320 × 288, section thickness 3 mm, with an intersection gap of 0 mm with 4 signals averaged).

Six regions of the knee were scored by a single blinded reader: medial and lateral tibial plateaus, medial and lateral femoral condyles, patella, and trochlear groove. Cartilage defects, the presence of osteophytes, BME, cysts, and meniscal damage were each graded (Table 2).

Table 2. Magnetic resonance imaging scoring criteria ([35])
Normal cartilage0
Abnormal signal without cartilage contour defect1
Cartilage defect ≤50%2
Cartilage defect >50% but <100%3
Cartilage defect 100% with subjacent bone signal abnormality4
Small beak-like osteophyte1
Intermediate-size osteophyte2
Proliferative or mushroom-like osteophyte3
Bone marrow edema 
Subchondral cysts 
≤2 small cysts1
>2 small cysts or 1 large cyst2
≥2 large cysts3
Intrasubstance increased signal intensity1
Increased signal intensity extending to articular surface (i.e., tear)2

Exposure variables

All participants received a standardized knee assessment by a rheumatologist ([16]). They also completed a standardized questionnaire about their sociodemographic details, medical history, and OA risk factors. The latter included occupational history, defined as all occupations held for at least 12 months after age 18 years, including occupations such as homemaker, student, or retiree. Participants were asked to list all previous occupations, the number of years in that position, and their starting age. For each occupation, they rated the activity level on a 5-point scale (where 1 = sedentary and 5 = very heavy). Also, they reported whether the job involved knee bending or kneeling, where 0 = never, 1 = occasionally, and 2 = frequently. These scales have been adapted from previous research by Felson et al ([1]), where occupations were scored on associated strength demand from the Dictionary of Occupational Titles.

The occupational physical load (OPL) of each job held by a participant was calculated using the following equation:

  • display math

where Y = number of years in each job, A = activity level, and K = knee bending or kneeling score. To avoid a score of 0 in the OPL calculation, the knee bending score was recoded to the following: 1 = never, 2 = occasionally, and 3 = frequently.

The COPL for each participant was the sum of the OPL for every occupation reported. If a participant reported multiple concurrent jobs, the OPL for each occupation was adjusted proportionally with respect to the overlapping time period. For example, if a participant reported 3 concurrent occupations, then one-third of the OPL for each was included.

To examine the validity of COPL for measuring occupational knee joint loading, we assessed the agreement between COPL and scores calculated based on data from the Occupational Information Network (O*NET) database prior to the main analysis ([17]). O*NET is the primary source of information of more than 900 occupations in the US, including detailed descriptions of required tasks, knowledge levels, skills, abilities, and qualifications. Each occupational title has a score for “performing general physical activities” (PA) and “time spent kneeling, crouching, stooping, or crawling” (TSK). The O*NET Occupational Score (OOS) was calculated by multiplying the number of years in each job (Y) by these O*NET scores for each occupation that participants reported:

  • display math

The Cumulative OOS (COOS) for each participant was calculated by summing the OOS for every occupation reported by participants. As in the COPL calculation, concurrent multiple jobs were taken into account.

O*NET does not provide data on homemakers as an occupation and so we used the occupation of patrolman as a proxy based on previous research that analyzed the composition of homemakers' duties ([18]). A subset of 60 participants randomly selected from the MoDEKO database was used for assessing the agreement between the 2 measures.

Statistical analysis

We used Pearson's correlation coefficient to assess the agreement between COPL and COOS. A Bland-Altman plot was constructed to demonstrate this relationship.

Participant data were described by the frequency (percentage) or the mean ± SD, as appropriate, for the full sample and for each of the OA definitions. The continuous exposure variable, COPL, was grouped into quarters (QCOPL) to allow for possible nonlinearity in the statistical models. This was done unweighted to avoid unequal statistical power across the different comparisons. Box plots were constructed showing the median COPL and interquartile range per quarter. For the remainder of the analyses, a population-based stratum-sampling weight was applied to the data. This weight was originally calculated for each cohort separately. As mentioned previously, the MoDEKO sample was stratified by age group and sex. Potential participants were contacted at random and asked about age, sex, and knee pain. Once the predetermined number of individuals who fit the inclusion criteria was reached in a given age–sex group, recruitment stopped within that group. The first time this happened, the number of participants recruited at this point with knee pain and the number of people who reported no knee pain were used to provide the estimated population distribution of age and sex for those with and without knee pain. Participants were recruited until sufficient numbers were enrolled in each cell. The ACE-KOA sample was collected analogously at a later time. A sample weight was developed for the combined sample with both symptomatic and asymptomatic persons as the ratio of the estimated population proportion in a given age–sex–symptomatic status cell over the sample proportion in that cell. The weight was scaled to sum to the total sample size.

An overall weighted mean ± SD COPL score was calculated for each quarter for all of the OA subgroups. Bivariable and multivariable logistic regression was used to assess the association between QCOPL and the presence of each of ROA, SOA, or MRI-OA after adjusting for female sex, age, and body mass index (BMI). Bivariable analyses of age and BMI were also tested for nonlinearity. Two-way interactions between age, BMI, and female sex with QCOPL were examined. The level of statistical significance was set at P values of 0.05 or less. All analyses were performed using SAS, version 9.2.


  1. Top of page
  2. Abstract

The COPL agreement analysis involved 60 randomly selected participants from MoDEKO (29 [48.3%] women, mean ± SD age 59.0 ± 11.3 years, 11 [18.3%] with K/L grade ≥2). The mean ± SD COPL score was 191.8 ± 141.7 and the mean ± SD COOS was 27,970.2 ± 25,883.7. Pearson's correlation coefficient was 0.69 (95% confidence interval [95% CI] 0.52–0.80), indicating good agreement between the 2 measures. The Bland-Altman plot (standardized variables) was distributed around 0 and free of systemic patterns, indicating no consistent bias of one measure over the other (Figure 2). A box plot of COPL by QCOPL level confirmed QCOPL calculations by showing no overlap between quarters (Figure 3).


Figure 2. Bland-Altman plot with standardized variables. COPL = cumulative occupational physical load; COOS = Cumulative Occupational Information Network Occupational Score.

Download figure to PowerPoint


Figure 3. Box plot of cumulative occupational physical load (COPL) by level of COPL categorized into quarters (QCOPL; unweighted). ° = outlier.

Download figure to PowerPoint

The unweighted demographic data and characteristics of all participants are shown in Table 3. They had a mean ± SD age of 58.5 ± 11.0 years and a mean ± SD BMI of 26.3 ± 4.7 kg/m2. There were slightly more women than men. The weighted COPL scores varied from 12.0 to 729.0, with a mean ± SD weighted COPL score of 151.8 ± 124.0. QCOPL-1 contained COPL scores <82, QCOPL-2 from 82–156, QCOPL-3 from 157–281, and QCOPL-4 >281.

Table 3. Participant characteristics*
 Full sample (n = 327)ROA (n = 109)No ROA (n = 218)SOA (n = 102)No SOA (n = 225)MRI-OA (n = 131)No MRI-OA (n = 196)
  1. ROA = radiographic osteoarthritis; SOA = symptomatic osteoarthritis; MRI = magnetic resonance imaging; BMI = body mass index; WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index; COPL = cumulative occupational physical load; QCOPL = COPL categorized into quarters.

  2. a

    Aggregate score of pain, stiffness, and function; range 0–100, where a higher number = higher disability.

Age, mean ± SD years58.5 ± 11.063.6 ± 9.655.9 ± 10.763.5 ± 9.956.2 ± 10.762.0 ± 10.156.1 ± 10.9
Women, no. (%)167 (51.0)62 (56.9)105 (48.2)57 (55.9)110 (48.9)62 (47.3)105 (53.6)
Ethnicity, no. (%)       
White257 (78.6)86 (78.9)171 (78.4)80 (78.4)177 (78.7)103 (78.6)154 (78.6)
Asian44 (13.5)13 (11.9)31 (14.2)12 (11.8)32 (14.2)17 (13.0)27 (13.8)
African American2 (0.61)2 (1.8)02 (2.0)01 (0.76)1 (0.51)
First Nation3 (0.92)3 (2.8)03 (2.9)03 (2.3)0
Hispanic2 (0.61)02 (0.92)02 (0.9)1 (0.76)1 (0.51)
Other19 (5.8)5 (4.6)14 (6.4)5 (4.9)14 (6.2)6 (4.6)13 (6.63)
Marital status, no. (%)       
Married198 (60.6)67 (61.5)131 (60.1)64 (62.8)134 (59.6)78 (59.5)120 (61.2)
Divorced42 (12.8)15 (13.8)27 (12.4)14 (13.7)28 (12.4)19 (14.5)23 (11.7)
Separated13 (4.0)2 (1.8)11 (5.1)1 (1.0)12 (5.3)4 (3.1)9 (4.6)
Widowed30 (9.2)16 (14.7)14 (6.4)15 (14.7)15 (6.7)16 (12.2)14 (7.1)
Single44 (13.5)9 (8.3)35 (16.1)8 (7.8)36 (16.0)14 (10.7)30 (15.3)
Postsecondary education, no. (%)       
High school73 (22.3)25 (22.9)48 (22.0)24 (23.5)49 (21.8)32 (24.4)41 (20.9)
Trades19 (5.8)8 (7.3)11 (5.0)8 (7.8)11 (4.9)10 (7.6)9 (4.6)
College58 (17.7)14 (12.8)44 (20.2)12 (11.8)46 (20.4)14 (10.6)44 (22.4)
University128 (39.1)40 (36.7)88 (40.4)36 (35.3)92 (40.9)45 (34.4)83 (42.3)
Other18 (5.5)6 (5.5)12 (5.5)6 (5.9)12 (5.3)11 (8.4)7 (3.6)
Missing data31 (9.4)16 (14.7)15 (6.9)16 (15.7)15 (6.7)19 (14.5)12 (6.1)
Knee injuries, no. (%)145 (44.3)61 (56.0)84 (38.5)44 (43.1)85 (37.8)73 (55.7)72 (36.7)
BMI, mean ± SD kg/m226.3 ± 4.727.1 ± 5.325.5 ± 5.627.3 ± 5.325.8 ± 4.327.3 ± 5.225.6 ± 4.1
Family history of OA, no. (%)       
No159 (48.6)54 (49.5)105 (48.1)48 (47.1)111 (49.3)64 (48.8)95 (48.4)
Yes152 (46.5)55 (50.5)97 (44.5)54 (52.9)98 (43.6)63 (48.1)89 (45.4)
Unsure16 (4.9)016 (7.3)016 (7.1)4 (3.1)12 (6.1)
Smoking ever       
No, no. (%)155 (47.4)61 (56.0)94 (43.1)58 (56.9)97 (43.1)67 (51.2)88 (44.9)
Yes, no. (%)172 (52.6)48 (44.0)124 (56.9)44 (43.1)128 (56.9)64 (48.9)108 (55.1)
Mean ± SD years16 ± 4.923.2 ± 15.218.1 ± 12.523.4 ± 15.318.2 ± 12.522.9 ± 14.217.5 ± 12.6
WOMAC aggregate score, mean ± SDa15.5 ± 17.323.3 ± 18.011.6 ± 15.524.8 ± 17.611.3 ± 15.521.8 ± 18.111.3 ± 15.4
COPL, mean ± SD194.2 ± 137.8242.1 ± 136.5170.2 ± 132.4249.0 ± 137.4169.3 ± 131.0245.6 ± 140.6159.8 ± 125.0
QCOPL, no. (%)       
181 (24.8)10 (9.2)71 (32.6)8 (7.8)73 (32.4)13 (9.9)68 (34.7)
282 (25.1)24 (22.0)58 (26.6)22 (21.6)60 (26.7)27 (20.6)55 (28.1)
382 (25.1)39 (35.8)43 (19.7)36 (35.3)46 (20.4)43 (32.8)39 (19.9)
482 (25.1)36 (33.0)46 (21.1)36 (35.3)46 (20.4)48 (36.6)34 (17.3)

The weighted bivariable and multivariable analyses of the relationship between QCOPL and each of ROA, SOA, and MRI-OA are shown in Table 4. Of the 327 participants, 109 were classified as having ROA, 102 were classified as having SOA, and 131 were classified as having MRI-OA. Increased odds of ROA were found in QCOPL-4 (adjusted odds ratio [OR] 3.15, 95% CI 1.02–9.70) and QCOPL-3 (adjusted OR 4.19, 95% CI 1.55–11.34) compared with QCOPL-1. Age and female sex were also significant (adjusted OR 1.07, 95% CI 1.04–1.11 and adjusted OR 2.15, 95% CI 1.05–4.39, respectively). In bivariable analysis, age was classified into categorical variables after it was found to be significant for nonlinearity. None of the interaction terms reached statistical significance and were removed from the final model.

Table 4. Multivariable logistic regression models for ROA, SOA, and MRI-OA*
Independent variableCrude OR (95% CI)PAdjusted OR (95% CI)aP
  1. A population-based stratum-sampling weight was applied. ROA = radiographic osteoarthritis; SOA = symptomatic osteoarthritis; MRI = magnetic resonance imaging; OR = odds ratio; 95% CI = 95% confidence interval; QCOPL = cumulative occupational physical load categorized into quarters; BMI = body mass index.

  2. a

    Adjusted for age, sex, and BMI.

  3. b

    Statistically significant at P < 0.05.

ROA vs. no ROA    
QCOPL-2 vs. 12.43 (0.93–6.35)0.072.35 (0.86–6.36)0.09
QCOPL-3 vs. 15.47 (2.17–13.81)b< 0.01b4.19 (1.55–11.34)b< 0.01b
QCOPL-4 vs. 13.80 (1.34–10.79)b0.01b3.15 (1.02–9.70)b0.05b
Age, years  1.07 (1.04–1.11)b< 0.01b
50–59 vs. 40–498.81 (2.47–31.45)b< 0.01b  
60–69 vs. 40–4912.1 (3.26–44.98)b< 0.01b  
≥70 vs. 40–4915.78 (4.06–61.39)b< 0.01b  
Female sex1.79 (0.93–3.44)0.082.15 (1.05–4.39)b0.04b
BMI1.03 (0.96–1.10)0.471.01 (0.94–1.09)0.79
SOA vs. no SOA    
QCOPL-2 vs. 13.16 (0.74–13.52)0.123.06 (0.70–13.39)0.14
QCOPL-3 vs. 17.12 (1.78–24.44)b0.01b5.73 (1.36–24.12)b0.02b
QCOPL-4 vs. 110.56 (2.58–43.21)b< 0.01b8.16 (1.89–35.27)b0.01b
Age1.07 (1.03–1.12)b< 0.01b1.05 (1.01–1.09)b0.03b
Female sex1.24 (0.56–2.75)0.591.56 (0.67–3.63)0.30
BMI1.08 (1.01–117)b0.04b1.07 (0.98–1.16)0.13
MRI-OA vs. no MRI-OA    
QCOPL-2 vs. 17.49 (2.29–24.49)b< 0.01b7.18 (2.17–23.70)b< 0.01b
QCOPL-3 vs. 110.88 (3.30–35.88)b< 0.01b9.04 (2.65–30.88)b< 0.01b
QCOPL-4 vs. 112.01 (3.46–41.71)b< 0.01b9.54 (2.65–34.27)b< 0.01b
Age1.05 (1.02–1.08)b< 0.01b1.03 (1.00–1.06)0.07
Female sex0.99 (0.53–1.77)0.921.17 (0.61–2.23)0.63
BMI1.06 (1.00–1.13)0.051.06 (0.99–1.13)0.13

For SOA, a statistically significant monotonic relationship was found among those in QCOPL-4 (adjusted OR 8.16, 95% CI 1.89–35.27) and QCOPL-3 (adjusted OR 5.73, 95% CI 1.36–24.12) compared to QCOPL-1, whereas QCOPL-2 did not reach statistical significance (adjusted OR 3.06, 95% CI 0.70–13.39). Age also remained significant (adjusted OR 1.05, 95% CI 1.01–1.09). A statistically significant monotonic relationship was also found in MRI-OA (QCOPL-4: adjusted OR 9.54, 95% CI 2.65–34.27; QCOPL-3: adjusted OR 9.04, 95% CI 2.65–30.88; QCOPL-2: adjusted OR 7.18, 95% CI 2.17–23.70).


  1. Top of page
  2. Abstract

In this population-based study, our results revealed a statistically significant relationship between the level of COPL and the presence of OA. For MRI-OA and SOA, the risk of OA was increased between 5 and almost 10 times for those with increased levels of occupational loading compared to the reference category. Results for the ROA analysis were similar, although a dose-response relationship was not observed. This study clearly supports the role of clinicians to educate their patients about the risks of occupational exposure in both the work place and home for the development of OA.

Over the last few decades, there have been many observational studies examining the relationship between occupational exposure and knee OA. Yet, likely due to their costly nature, there have been only a few prospective cohort studies and their results have been inconsistent ([1, 19, 20]). Potential reasons for these inconsistencies have been suggested and include variability in exposure measurement, different knee OA outcome definitions, diverse subject eligibility criteria, the inclusion of covariates, and small sample sizes in higher-exposure categories ([21, 22]). When these earlier longitudinal studies are combined with the more recent high-quality case–control and cross- sectional studies in systematic reviews ([21, 23]), there is evidence for the likely role of regular heavy lifting and kneeling in the onset of knee OA, especially in men. Yet, both reviews suggest that future work, especially prospective studies, is still needed. The current study not only confirms the evidence on the role of occupation in the presence of knee OA, it also provides justification to further understand the dose-response relationship between exposure to cumulative joint loading and OA.

COPL was quantified with a new self-report measure that was found to have strong agreement with an ergonomist expert-based ratings scale. It was unique in that it accounted for the timeframe (i.e., number of years) that a person performed each occupation in the past and allowed for an investigation of a dose-response relationship, when many previous studies have used a dichotomous exposure variable ([9, 10, 24, 25]). In previous studies, a wide variety of methods have been used to define and measure occupational exposure as one domain of physical activity, although many have not been validated. Most validated methods focus on measuring energy expenditure and metabolic equivalents, rather than the mechanical joint loading force during physical activity ([22]). This is because these tools are designed to measure the benefits of physical activity, rather than to examine the potential detrimental effects of certain body positions, movements, and activities on the health of muscle, bone, and joints ([26]). Although the measure used in this study is based on self-report of lifetime occupational activities and therefore is subject to recall bias, it has shown good agreement with another method of measuring COPL, and utilizing a questionnaire is both cost and time effective for use in a population-based cohort with the goal of investigating the development of OA. Given the long induction time and asymptomatic latency period of OA, gathering a full, detailed history of lifetime occupational exposure using a questionnaire is ideal and has shown an acceptable level of accuracy, since direct observation of COPL is not feasible ([27]). With future validation, this measure of COPL has the potential to be used for further work in this area.

The current study provides novel results suggesting occupational exposure as a risk factor for OA defined by MRI criteria. The MRI-OA outcome produced slightly higher levels of risk than the other 2 outcomes. This may be because MRI can detect earlier disease than SOA, resulting in a cleaner separation of individuals into 2 more distinctive subgroups, with those in the no MRI-OA subgroup having substantially fewer pathologic joint changes than the no SOA subgroup, where individuals may have had MRI changes invisible on radiographs. In addition, MRI-OA was a purely structural definition and did not include pain, therefore eliminating the healthy worker effect, a phenomenon that occurs when those with early development of disease modify their occupational exposure by changing careers to a less physical job or retiring early, leaving those without morbidity to work longer in more vigorous occupations ([28]). Finally, the MRI-OA definition included both tibiofemoral (TF) and patellofemoral (PF) disease, whereas the radiographic definition using K/L grade only examined the TF joint. The biomechanics of the PF joint are different than the TF joint ([29]). For example, during squatting activities, shear forces on the patellar cartilage surfaces have been found to be higher than on the TF joint ([30, 31]). COPL quantified combined heavy lifting and kneeling, which may put additional strain on the PF joint. Two recent studies found a link between MR-imaged PF pathology and occupation-related squatting, kneeling, and heavy lifting ([8, 32]). Future work using MRI could possibly examine the impact of occupational exposure on each compartment of the knee. Although the costs associated with MRI are a major barrier to widespread clinical use, there is an opportunity for MRI-OA to be applied in research to identify risk factors for early OA such as occupational joint loading.

In the current study, the odds of ROA were slightly lower than SOA and not monotonic. One plausible explanation for the difference in the ROA subgroup is that those with knee pain might have attributed their pain to previous occupational exposure and consequently overreported COPL ([33]). Of note, there were only a few participants without pain in the ROA subgroup, but they all had lower COPL scores. The lack of a monotonic relationship across QCOPL in the ROA model may have been due to the healthy worker survivor effect described above, where those without morbidity work longer and harder, resulting in a higher COPL score ([28]).

This study had several limitations. First, the cross-sectional design, where exposure and outcome are obtained at a single time point, revealed an association; however, this relationship should not be used to infer cause and effect. Second, as a secondary data analysis, inherent shortcomings are present, since this specific investigation was not considered during the original study design. For example, data on the number of hours worked per week or part time versus full time versus seasonal work were not collected, which could have further refined the COPL calculation. Also, although participant history of previous knee injury was a potential confounder, it was not included in the analysis due to the quality of data. Third, for the sample size that was available, there was insufficient power to explore the relationship between COPL and OA in men and women separately. Finally, as discussed, the use of a self-report questionnaire to measure historical occupational exposure risks recall bias. However, since direct observation of COPL is not feasible for a study evaluating lifetime occupational activities, it is considered a reasonably sufficient alternative ([27]).

One strength of this study is the comprehensive recruitment strategy used to obtain a population-based cohort of individuals representing the full spectrum of OA. Consequently, the results have external validity extending to the entire greater Vancouver region. This study also involved detailed measurement of lifetime occupational exposure to lifting and kneeling in a manner that allowed the investigation of a dose-response relationship, which has potential clinical implications. For example, in the future, an intervention aimed at OA prevention may be based on guidelines limiting joint loading to an evidence-based threshold. Furthermore, this study included loading performed as a homemaker, student, and retiree in its quantification of occupational exposure, revealing loading activities that may have been underestimated in previous work.

These results highlight the importance of future work place prevention programs both at organizational and individual levels. A recent systematic review failed to discover any randomized controlled trials investigating prevention of work-related knee injuries or OA ([23]). They proposed interventions such as the new work methods, use of better tools and equipment, implementation of administrative controls, or better training and education of workers to minimize future harm needed to be created and evaluated. There is a role for clinicians to become engaged in primary prevention through the education of vulnerable workers on avoidance of high load positions, especially those who may already have other OA risk factors. Clinicians may partner with ergonomists to define and promote better principles of work design with the use of adaptive devices, tools, and equipment that may eliminate some high loading activities ([34]).

This study provides evidence that occupational exposure to the physically demanding tasks of heavy lifting and kneeling or knee bending is an independent risk factor for the presence of ROA, SOA, and MRI-OA of the knee. This evidence suggests that further longitudinal studies are justified in this area of joint disease and occupational health across the full continuum of OA and that future emphasis should be placed on the development of viable strategies for OA prevention in the work place and home. Additionally, future research should clarify the role of cumulative occupational loading as a risk factor for incident and progressive knee OA, especially in individuals with early disease.


  1. Top of page
  2. Abstract

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Li had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Ezzat, Cibere, Koehoorn, Li.

Acquisition of data. Cibere.

Analysis and interpretation of data. Ezzat, Cibere, Koehoorn, Li.


  1. Top of page
  2. Abstract

Centocor Research & Development had no role in the study design, data collection, data analysis, or writing of the manuscript, as well as approval of the content of the submitted manuscript. Publication of this article was not contingent on the approval of Centocor.


  1. Top of page
  2. Abstract
  • 1
    Felson DT, Hannan MT, Naimark A, Berkeley J, Gordon G, Wilson PW, et al.Occupational physical demands, knee bending, and knee osteoarthritis: results from the Framingham Study.J Rheumatol1991;18:158792.
  • 2
    Garstang SV, Stitik TP.Osteoarthritis: epidemiology, risk factors, and pathophysiology.Am J Phys Med Rehabil2006;85 Suppl:S211.
  • 3
    Zhang Y, Jordan JM.Epidemiology of osteoarthritis.Clin Geriatr Med2010;26:35569.
  • 4
    Jensen LK.Knee osteoarthritis: influence of work involving heavy lifting, kneeling, climbing stairs or ladders, or kneeling/squatting combined with heavy lifting.Occup Environ Med2008;65:7289.
  • 5
    Maetzel A, Makela M, Hawker G, Bombardier C.Osteoarthritis of the hip and knee and mechanical occupational exposure: a systematic overview of the evidence.J Rheumatol1997;24:1599607.
  • 6
    Seidler A, Bolm-Audorff U, Abolmaali N, Elsner G.The role of cumulative physical work load in symptomatic knee osteoarthritis: a case control study in Germany.J Occup Med Toxicol2008;3:14.
  • 7
    Townsend EA, Polatajko HJ.Enabling occupation II: advancing occupational therapy vision for health, well-being, and justice through occupation.Ottawa:Canadian Association of Occupational Therapists;2007.
  • 8
    Teichtahl AJ, Wluka AE, Wang Y, Urquhart DM, Hanna FS, Berry PA, et al.Occupational activity is associated with knee cartilage morphology in females.Maturitas2010;66:726.
  • 9
    Yoshimura N, Kinoshita H, Hori N, Nishioka T, Ryujin M, Mantani Y, et al.Risk factors for knee osteoarthritis in Japanese men: a case-control study.Mod Rheumatol2006;16:249.
  • 10
    Bernard TE, Wilder FV, Aluoch M, Leaverton PE.Job-related osteoarthritis of the knee, foot, hand, and cervical spine.J Occup Environ Med2010;52:338.
  • 11
    Sowers MF, Hayes C, Jamadar D, Capul D, Lachance L, Jannausch M, et al.Magnetic resonance-detected subchondral bone marrow and cartilage defect characteristics associated with pain and x-ray-defined knee osteoarthritis.Osteoarthritis Cartilage2003;11:38793.
  • 12
    Hunter DJ, Arden N, Conaghan PG, Eckstein F, Gold G, Grainger A, et al.Definition of osteoarthritis on MRI: results of a Delphi exercise.Osteoarthritis Cartilage2011;19:9639.
  • 13
    Cibere J, Zhang H, Thorne A, Wong H, Singer J, Kopec JA, et al.Association of clinical findings with pre–radiographic and radiographic knee osteoarthritis in a population-based study.Arthritis Care Res2010;62:16918.
  • 14
    Kothari M, Guermazi A, von Ingersleben G, Miaux Y, Sieffert M, Block JE, et al.Fixed-flexion radiography of the knee provides reproducible joint space width measurements in osteoarthritis.Eur Radiol2004;14:156873.
  • 15
    Kellgren JH, Lawrence JS.Radiological assessment of osteo-arthrosis.Ann Rheum Dis1957;16:494502.
  • 16
    Cibere J, Bellamy N, Thorne A, Esdaile JM, McGorm KJ, Chalmers A, et al.Reliability of the knee examination in osteoarthritis: effect of standardization.Arthritis Rheum2004;50:45868.
  • 17
    O*NET online. URL:
  • 18
    Arvey RD, Begalla ME.Analyzing the homemaker job using the position analysis questionnaire (PAQ).J Appl Psychol1975;60:5137.
  • 19
    Vingard E, Alfredsson L, Goldie I, Hogstedt C.Occupation and osteoarthrosis of the hip and knee: a register-based cohort study.Int J Epidemiol1991;20:102531.
  • 20
    Schouten JS, van den Ouweland FA, Valkenburg HA.A 12 year follow up study in the general population on prognostic factors of cartilage loss in osteoarthritis of the knee.Ann Rheum Dis1992;51:9327.
  • 21
    McWilliams DF, Leeb BF, Muthuri SG, Doherty M, Zhang W.Occupational risk factors for osteoarthritis of the knee: a meta-analysis.Osteoarthritis Cartilage2011;19:82939.
  • 22
    Ratzlaff CR.Lifetime physical activity and osteoarthritis.Vancouver:University of British Columbia;2011.
  • 23
    Fransen M, Agaliotis M, Bridgett L, Mackey MG.Hip and knee pain: role of occupational factors.Best Pract Res Clin Rheumatol2011;25:81101.
  • 24
    Allen KD, Chen JC, Callahan LF, Golightly YM, Helmick CG, Renner JB, et al.Associations of occupational tasks with knee and hip osteoarthritis: the Johnston County Osteoarthritis Project.J Rheumatol2010;37:84250.
  • 25
    Anderson JJ, Felson DT.Factors associated with osteoarthritis of the knee in the first national Health and Nutrition Examination Survey (HANES I): evidence for an association with overweight, race, and physical demands of work.Am J Epidemiol1988;128:17989.
  • 26
    Dolan SH, Williams DP, Ainsworth BE, Shaw JM.Development and reproducibility of the bone loading history questionnaire.Med Sci Sports Exerc2006;38:112131.
  • 27
    Falkner KL, Trevisan M, McCann SE.Reliability of recall of physical activity in the distant past.Am J Epidemiol1999;150:195205.
  • 28
    Shah D.Healthy worker effect phenomenon.Indian J Occup Envion Med2009;13:779.
  • 29
    Hinman RS, Crossley KM.Patellofemoral joint osteoarthritis: an important subgroup of knee osteoarthritis.Rheumatology (Oxford)2007;46:105762.
  • 30
    Dahlkvist NJ, Mayo P, Seedhom BB.Forces during squatting and rising from a deep squat.Eng Med1982;11:6976.
  • 31
    Besier TF, Gold GE, Beaupre GS, Delp SL.A modeling framework to estimate patellofemoral joint cartilage stress in vivo.Med Sci Sports Exerc2005;37:192430.
  • 32
    Amin S, Goggins J, Niu J, Guermazi A, Grigoryan M, Hunter DJ, et al.Occupation-related squatting, kneeling, and heavy lifting and the knee joint: a magnetic resonance imaging-based study in men.J Rheumatol2008;35:16459.
  • 33
    Rogers LQ, Macera CA, Hootman JM, Ainsworth BE, Blairi SN.The association between joint stress from physical activity and self-reported osteoarthritis: an analysis of the Cooper Clinic data.Osteoarthritis Cartilage2002;10:61722.
  • 34
    Palmer KT.Occupational activities and osteoarthritis of the knee.Br Med Bull2012;102:14770.
  • 35
    Disler DG, McCauley TR, Kelman CG, Fuchs MD, Ratner LM, Wirth CR, et al.Fat-suppressed three-dimensional spoiled gradient-echo MR imaging of hyaline cartilage defects in the knee: comparison with standard MR imaging and arthroscopy.AJR Am J Roentgenol1996;167:12732.