There are wide geographical differences in the prevalence of knee osteoarthritis (kneeOA): lowest in Asians, followed by black Africans, and highest in white Europeans.1 Symptomatic kneeOA (pain on most days plus positive radiological findings) occurs in approximately 10% of those aged 55 years or more.2 The prevalence of kneeOA is correlated with age. People who are overweight develop kneeOA more often than those who are not overweight.3 4 Previous major knee injuries increase the risk of kneeOA.5 Elite runners and soccer players appear to be at increased risk for kneeOA in later life, while moderate regular running or soccer playing has low, if any, risk of leading to kneeOA.6

The aim of the present study was to critically review the epidemiological evidence for a possible causal relationship between kneeOA and an exposure to heavy lifting, kneeling, climbing stairs and exposure to heavy lifting combined with kneeling or squatting. The review is based on a scientific report7 made on behalf of the Scientific Committee of the Danish Society of Occupational and Environmental Medicine for the use of the Danish National Board of Industrial Injuries in their evaluation of whether kneeOA and hipOA caused by specified physical work demands should be included on their list of occupational diseases that may be compensated for through the Danish Worker Compensation Act. The relationship between hipOA and occupational work demands is presented in another paper.8

MATERIALS AND METHODS

Search strategy

The relevant studies were identified through searches in the following literature databases: Medline (1966 to April 2007), NIOSH-tic (1990 to April 2007), Embase and HSE-line (1990 to April 2007). The following keywords were used: knee and (osteoarthritis or osteoarthrosis) and (work or occupation). Further relevant literature was found by screening the reference lists of all relevant articles identified. A study was selected for a more detailed review if it fulfilled the following criteria: (1) one of the aims of the study was to investigate an association between kneeOA and the physical demands “heavy lifting, kneeling, heavy lifting combined with kneeling/squatting, or climbing stairs or working on ladders”; (2) the literature was published in English, German, Danish, Swedish or Norwegian languages; (3) it was a full text article; (4) the subjects studied had (a) radiological verified kneeOA using a case definition including the Kellgren and Lawrence criteria for kneeOA, (b) a diagnostic code from WHO International Classification of Diseases ICD-8 = 713.01 or ICD-10 = M17 for kneeOA, (c) had a total knee replacement (TKR) or (d) were on a waiting list for knee replacement; (5) the study had a controlled design. All type of exposures: job title, classification in low- and high-exposure groups, and further description of assessment of physical activity using questionnaires or interviews are included.

Assessment of the study quality

The quality of the studies, and the strengths and the weaknesses of the studies were evaluated by the author and the following aspects were included in the evaluation:

Design and material (0–3)

• Study design and inclusion/exclusion criteria described (1)

• Participation rate >70% for cases and controls (1)

• Sufficient description of baseline characteristics, size of study group, information on completers vs. withdrawals (1)

Exposure (0–3)

• Exposure assessment by video observation (3), or

• Exposure measured by interviews specified in different physical activities (2), or

• Exposure defined by job title (1)

Outcome (0–3)

1. Radiological examination (0–3). Sum of:

2. • Standardised for cases and controls (1)

   • Weight-bearing (1)

   • Blinded (1), or

3. TKR (0–3). Sum of:

4. • Assessed identical in studied population (1)

   • Criteria for TKR validated (1)

   • Radiological examination re-assessed (1), or

5. Diagnosis code (1)

Adjusted for (0–3)

• Age or sex (1), or

• Age and sex (2), or

• At least age, sex and one of earlier traumas, smoking, sports activities, Heberden’s nodes (3)

Data presentation and statistical analysis (0–3). Sum of:

• Frequencies of most important outcome were given (1)

• Appropriate analysis techniques were used (2)

Each of the five items was scored (0–3), and the quality of the study was assessed by the sum of the scores, the maximum being 15 points. The papers were subdivided into: poor quality (score 1–5); medium quality (score 6–10); high quality (score 11–15). The quality assessment is shown in table 1.

Assessment of causal relationship

The studies were considered to be heterogeneous with regard to the population studied, outcome measures for kneeOA and exposure measurement. It was not possible to pool the data statistically so we performed a “best evidence” synthesis instead.33 34 The studies were classified according to the type of study design. The prospective cohort study was judged as the preferred design, followed by the case–control study and then by the cross-sectional study. The studies were then ranked by their methodological quality score. The overall evaluation of the degree of evidence of a causal relationship between kneeOA and physical demanding work have been made using specific criteria established by The Scientific Committee of the Danish Society of Occupational and Environmental Medicine.7 The paper was discussed with the Scientific Committee, and the final decision on the level of evidence was assessed together with the Committee.

The following ranking of the levels of evidence was used:

• Strong evidence: A causal relationship is very likely (chance, bias and confounding could be ruled out with reasonable confidence) between an exposure to a specific risk factor and a specific outcome. A positive relationship has been observed between exposure to the risk factor and the outcome in several studies (two or more high quality cohort studies, several case–control and cross-sectional studies)

• Moderate evidence: Some convincing epidemiological evidence exists (chance, bias and confounding are not the likely explanation) for a causal relationship between an exposure to a specific risk factor and a specific outcome. A positive relationship has been observed between exposure to the risk factor and the outcome in several studies (one high-quality cohort study and two or more case–control studies or in three or more high-quality case control studies)

• Limited evidence: Some convincing epidemiological evidence exists in some studies (one or two case–control studies or in more cross-sectional studies) for a causal relationship between an exposure to a specific risk factor and a specific outcome. A positive relationship has been observed between exposure to the risk factor and the outcome, but it is not unlikely that this relationship could be explained by chance, bias or confounding

• Insufficient evidence of a causal association: The available studies are of insufficient quality, consistency or statistical power to permit a conclusion regarding the presence or absence of a causal association

• No evidence is provided when no studies could be found.

RESULTS

Identification and selection of literature

The electronic search retrieved 919 references dealing with kneeOA. Of these only 20 epidemiological studies with a controlled design met the inclusion criteria.5 10 12^–14 16 18^–20 22^–25 26 27 29^–32 35 After screening the reference lists of the included studies, another five studies were included.9 15 17 21 28 Figure 1 shows the results of the searches and the number of epidemiological studies that remained after applying the inclusion criteria on the search.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 KneeOA: materials, and number of references in the electronic search and literature references.

Description of the studies

Heavy lifting

The association between kneeOA and heavy lifting has been investigated in 17 studies.12^–18 20 22 23 25^–31 Study details are shown in table 2. Nine studies showed a significant association with an odds ratio (OR) range of 1.9–7.31 or relative risk (RR) range of 1.9–14.3. Based on the study design, the size of the population and the exposure measurement, six studies were considered as being of the highest quality and are discussed in detail below.12 16 25 26 27 30

In a case–control study by Manninen et al,30 226 women and 55 men, aged 55–75 years, who had undergone a TKR because of primary kneeOA were selected as cases and compared with 526 from the general population matched by age and sex. The exposure was obtained by interview and assessed up to 49 years of age. The exposure was calculated as the cumulative lifted weight (kg) in each occupation and divided in three categories: low (no regular lifting), medium (<100 000 kg) and high (⩾100 000 kg). After adjustment for BMI and knee injuries high level of lifting for men showed an OR 0.92 (not significant (NS)) and for women 1.11 (NS). For physical workload was reported an OR 1.93 (95% CI 1.18 to 3.16) (medium level) and OR 2.19 (95% CI 1.32 to 3.64) (high level) (men and women). The study was well-designed, but included only few men. The exposure assessment was retrospective (participants had to remember physical demands before the age of 49 years), which might involve a certain degree of non-differential misclassification, giving a dilution of true risk for those highly exposed and an overestimation or underestimation of the risk for those medium exposed. The subjects with kneeOA may also remember their physical demands as more knee-demanding than the controls (recall bias).

In a case–control study by Sandmark et al,27 325 men and 300 women who had undergone a TKR because of primary kneeOA were defined as cases. Male (n = 264) and female (n = 284) controls were randomly selected from the general population. Subjects with earlier trauma or surgery of the knee, rheumatoid arthritis, poliomyelitis, rickets or musculoskeletal malformations were excluded. The information on exposure (life-time) was collected 1–4 years after TKR by questionnaire; the questions included details about the duration of exposure, stairs climbed (number per day), lifting (frequency and weight) and kneeling (h/day). The exposure was accumulated for each physical activity and classified into low, moderate and high exposure. After adjustment for BMI, smoking, sports activities and age, high level of lifting had an OR 3.0 (95% CI 1.6 to 5.5) (men) and OR 1.7 (95% CI 1.0 to 2.9) (women). The study is well-described but it has the same limitations as all case–control studies as already described.

In a case–control study by Coggon et al,25 205 men and 313 women (aged 47–93 years) listed for TKR were compared with an equal number of controls. Radiographs for each case were reviewed and 78% of the cases had kneeOA grade 3–4 according to the criteria of Kellgren and Lawrence.37 Exposure was collected by interviews of life-time occupation and classified by different physical activities. After adjustment for BMI, previous knee injuries and Heberden’s nodes lifting >10 kg at least 10 years at least ten times per week showed an OR 1.9 (95% CI 1.0 to 3.3) (men) and 1.5 (95% CI 1.0 to 2.3) (women). For lifting >25 kg results were only significantly increased for women. For lifting >50 kg no significant increase was found. The strengths of the study are the high number of participants, and that information on work demands was specified as different physical activities. The information about the occupational activities was collected retrospectively, which might be susceptible to recall bias. One of the limitations of the study may be the low participation rate (55%). It is possible that subjects with poorer social background were less willing to participate, but adjustment for social class led to only small differences in the outcome. A case-definition “placed on a waiting list for TKR” may lead to a selection bias.

Lau et al26 undertook a case–control study in Hong Kong with the same design as Coggon et al.25 The information on occupational activities was collected by interviews. For lifting >10 kg more than ten times per week, ORs for kneeOA 5.8 (95% CI 3.1 to 10.8) (men) and 3.0 (95% CI 1.6 to 5.5) (women) were found. The study has the same strengths and limitations, but the design is not as well-described, as the study by Coggon et al.35 Relatively few women were exposed to heavy lifting and this may influence the results. Despite these limitations the study showed a relationship between heavy lifting and kneeOA, as in the study by Coggon et al.25

Anderson and Felson12 used data from the US First National Health and Nutrition Examination Survey 1971–1975 (HANES I), and included 5193 participants of which 315 had radiological kneeOA grade 2–4. The study design was cross-sectional. The radiographs were single, non-weight-bearing x-rays for both knees, which may lead to an underestimation of kneeOA. Exposure was obtained by using current occupation when the subjects joined the study. Subjects were graded in relation to heavy work (lifting or kneeling) by experienced occupational professionals. No significant increases were shown in men or women beyond the age of 55 years. At age 55–64 years the ORs for lifting were 1.88 (95% CI 0.88 to 4.0) for men and 3.13 (95% CI 1.04 to 9.4) for women. The strengths of the study are the high number of participants and being a well described study. Physical demands were coded by using job titles. Although more than 300 job titles were used, this may lead to a risk of misclassification. Non-differential misclassification may occur if low-exposure occupations were classified as high-exposure or visa versa, and possible risk in some groups may then be underestimated.

As a part of the Framingham Heart study, Felson et al16 carried out a longitudinal cohort study including 569 men and 807 women with a mean age of 73 years. KneeOA was defined as grade 2–4 changes (Kellgren and Lawrence) on weight-bearing radiographs. Exposure was defined by current occupation (the occupation that they had from examination 1 to examination 6). The results showed an OR 0.96 (NS) for men and OR 2.53 (NS) for women for lifting medium, heavy or very heavy demands compared with no lifting. The strength of the study was the longitudinal design, and the fact that the radiographs were taken weight-bearing. In general there may be a high risk of misclassification in this study and a dilution of the results, because it only included few occupations with really heavy physical work demands and physical demanding jobs were uncommon both among the men and the women.

Dose–response relationship for lifting

In the studies, there seemed to be a dose–response relationship, with higher risks for the high-exposure groups than for the medium-exposure groups when both were compared with the low-exposure groups. In two studies, medium-exposure compared with low exposure had a positive association with kneeOA with OR range of 2.5–4.5.17 27 In five studies high-exposure compared with low-exposure had an OR range of 1.4–14.3.12^–14 17 27 In six studies, the weight of the lifts was specified into lifts >10 kg, >25 kg or >50 kg and into frequency per week and duration. In only one of these studies was a dose–response relationship shown, with an increase in OR for men ranging from 1.7 for lifts >10 kg one to ten times per week, OR 5.8 for lifts >10 kg more than ten times per week, OR 3.5 for lifts >50 kg one to ten times per week, and OR 7.1 for lifts >50 kg more than ten times per week. In four studies a significant increased risk of getting kneeOA was shown when comparing workers in heavy jobs (at least 10 years or at least 25 years) with those unexposed.25 27^–29

Kneeling

The association between kneeOA and kneeling has been investigated in 12 studies. Study details are shown in table 3. Eight studies showed a significant association, with an OR range of 2.2–6.9.

Based on the study design, the size of the population and the exposure measurement, six studies were considered as being of the highest quality.12 16 25–27 30 The studies have already been described in relation to lifting.

The cross-sectional study by Anderson and Felson12 showed a relationship between kneeling and kneeOA for subjects, 55–64 years of age, with an OR of 2.45 (95% CI 1.21 to 4.97) for men and 3.49 (95% CI 1.22 to 10.5) for women. Three of four case–control studies showed a significantly increased risk of kneeOA in relation to kneeling or squatting.25^–27 The exposure was collected retrospectively and was defined as kneeling/squatting >1 h/day35 or >2 h/day. The cohort study by Felson et al16 showed no significant increase for knee-bending combined with sedentary work, but an increase for knee-bending combined with medium, heavy or heavy demands with an OR 2.22 (95% CI 1.38 to 3.58). The classification of exposure was done by using job titles, which increases the risk of misclassification. The occupations that normally would be related to kneeling work positions were placed in the group with combined kneeling/squatting and heavy workload. The cohort included only a few participants with heavy physical work tasks, especially among women.

Dose–response relationship

In two studies, there was a dose–response relationship for men between kneeling and kneeOA, with higher risks for the high-exposure group than for the medium-exposure group. The increase was shown for both “total number of squatting/knee-bending” (OR 2.9 vs 1.3) and for kneeling (min) (OR 2.1 vs 1.4)27 and for low-moderate, high, and very high exposure (compared with no exposure) with ORs 2.96, 4.2, and 4.92, respectively.38 In two studies, the importance of the duration of the exposure was investigated, and a significant increased risk of getting kneeOA was shown when comparing workers in heavy jobs (at least 10 years or at least 25 years) with those unexposed.25 29 Coggon et al25 found no dose–response relationship in men or women for kneeling >1 h/day 1–9.9 years, 10–19.9 years and >20 years with ORs 3.3, 1.3, and 1.7, respectively.

Occupations involving heavy lifting and/or kneeling/squatting

The association between kneeOA and occupations involving heavy lifting and/or kneeling have been investigated in 12 studies. Study details are shown in table 4.

Two studies showed an increased frequency of kneeOA among miners compared with a control group.9 35 The association between working in the construction industry (not further defined) and kneeOA has been investigated in five studies.14 17 21 27 32 All studies showed significantly increased risk for kneeOA. In one study on concrete-reinforcement workers and painters no significant difference in the risk for kneeOA was found.10 In two studies on floor-layers, one study showed no significant differences for tibio-femoral kneeOA but did for patello-femoral OA and osteophytes19, and the other showed an increased risk of kneeOA.24

Heavy lifting and kneeling/squatting

The association between kneeOA and kneeling combined with heavy lifting have been investigated in three case–control studies and one cohort study. Study details are shown in table 5. The four studies showed an increased risk for kneeOA with OR range of 2.2–5.4. The association between kneeOA and kneeling combined with heavy lifting showed a more pronounced difference for subjects employed in occupations that involved both heavy lifting and kneeling/squatting than for heavy lifting alone. Cooper et al20 found ORs of 1.4 for lifting and 5.4 for combined lifting and kneeling/squatting; Felson et al16 found ORs of 0.96 for lifting and 2.2 for combined lifting and kneeling/squatting. No studies have investigated the dose–response relationship for the combination of heavy lifting and kneeling, either in relation to the amount lifted, the frequency of lifting, the duration, or the aggregate of years with kneeling- and lifting-work activities.

Climbing stairs or ladders

Five case–control studies have investigated the association between kneeOA and climbing stairs (stairs or ladders − one study). Study details are shown in table 6.

Four of five studies showed significantly increased risk between stair-climbing and development of kneeOA for men and only one study showed an increased risk for women.

All the studies have the same design (case–control). This may be a problem because information on exposure in these studies are retrospective. Only one study on the association between kneeOA and climbing stairs used radiographic OA as the inclusion criterion; the association in this study was statistically significant, with OR 2.7 (95% CI 1.2 to 6.1).20 The measures of exposure differed in the studies from climbing stairs >30 min/day, to >30 times/day and >15 flights/day. No dose–response relationship has been investigated.

KneeOA and sex

Eleven studies have investigated the relationship between heavy lifting and kneeOA for women.12 14^–16 22 25^–29 31 All the studies revealed more significant results for men than for women. Six studies showed significant positive relationship with OR range of 1.7–3.1. Two studies29 31 only included women, and the association between heavy lifting and kneeOA in these studies was positive, showing OR range of 1.9–7.3 (statistically significant in only one of the studies).

Ten studies have investigated the relationship between kneeOA and kneeling/squatting for women. Three of these studies have shown significantly positive relationship for women with OR range of 1.8–4.2. The numbers of women in occupations which had heavy workloads or kneeling have been few in many of the studies, probably one of the reasons for the non-significant results. In general, women do not have work-tasks with the same degree of physically heavy workloads or kneeling in their occupations as men do, and they traditionally work in other trades. It is not easy, therefore, to recruit a sufficient quantity of women with high exposure into the studies.

Best-evidence synthesis

The studies was considered to be of high quality if the methodological quality score was >10.

Of 17 studies on the association between kneeOA and lifting, six studies reach the level of high quality.12 16 25^–27 30 One had a cohort design,16 four had a case–control design25^–27 30 and one a cross-sectional design.12 The reported ORs were 1.88, 2.53, 1.9, 7.1, 3.0 and 3.0. Hence there is moderate evidence for a positive relationship between kneeOA and heavy lifting.

There were 12 studies reporting the association between kneeOA and kneeling or squatting. Of these, six reached the level of high quality.12 16 25^–27 30 One had a cohort design,16 four had a case–control design,25^–27 30 and one a cross-sectional design.12 The reported ORs were 2.45, 1.07, 3.0, 1.4, 2.9 and 1.81. Furthermore, many studies on the relationship between trades with kneeling work demands showed a positive relationship. So, with the best-evidence synthesis, it can be concluded that there is moderate evidence for a positive association between kneeOA and kneeling.

For the four studies assessing kneeOA and the combination of heavy lifting and kneeling/squatting two studies reached the level of high quality. One had a cohort design16 and one a case–control design.25 No studies have investigated a dose–response relationship either in relation to the amount lifted (kg), the frequency of lifting (times per day), the duration of lifting-work (years) or to the aggregate of years with kneeling and lifting-work activities. Therefore the degree of evidence was considered as moderate.

Five case–control studies investigated the relationship between kneeOA and climbing stairs of which four case–control studies reach the level of high-quality studies. The reported ORs were 1.7, 2.3, 6.1 and 3.06. Although all the studies showed a positive association, no studies showed a dose–response relationship. All were case–control studies. This may be a problem because exposure information in these studies are retrospective. People in occupations that include stair-climbing may seek treatment earlier than other workers because they experience worse pain when climbing stairs. These studies in particular are therefore at high risk of selection bias and recall bias. The evidence of a causal association between kneeOA and climbing stairs is therefore considered to be limited and for ladders insufficient.

DISCUSSION

Study design and bias

This review may have some limitations. Although great effort was made to identify the relevant literature, some studies may have been missed because different keywords were used in the databases, because they were not indexed in databases or because they were indexed in databases other than the databases used for this review. Only studies published in English, German or Scandinavian languages were considered. This may also have excluded some relevant studies.

Publication bias, in which negative studies are less likely to be published than positive studies, may be a risk, but is presumably more likely when studies are not well designed.

The assessment of the quality of the studies was done by scoring a number of items on different aspects of quality and summing the scores. This method should be considered with some reservation. The final classification depends on the specific item evaluated, on the number of items, on the scores attached to each item and on the way the scores are summed. A quality assessment will always be subjective: when carried out by only one author there will be a risk of misclassification. However, this review and its conclusions have been assessed by a scientific committee, which may protect against serious misclassifications.

Some of the limitations in the included studies are related to use of different diagnostic criteria and a poor description of the exposure. Some studies only included a small number of participants among cases, and furthermore only a small number of cases and controls were been employed in occupations with heavy lifting.18 20 31 This could dilute the associations; the risk will probably then be underestimated.

Unfortunately, there was only one cohort study of the association between kneeOA and lifting, kneeling or squatting, and no studies on the association between kneeOA and climbing stairs could be found.16 The cohort design is known to be the most valid. Data from prospective cohort studies would have provided a more precise and more valid conclusion.

Measurement of outcome

In the seven studies on the association between kneeOA and heavy lifting, the criteria established by Kellgren and Lawrence were used.12 13 15 16 20 28 31 In six of the studies “grade 2–4” was used as diagnostic criterion. In the study by Seidler et al28 the case-definition included grade 1–4. This may result in a non-differential misclassification of the cases that may in reality not be cases, and this may dilute the true risk.

In some studies the radiographs were taken non-weight bearing, which may lead to an underestimation of kneeOA.12 24 It is presumed that the underestimation will be the same among subjects with heavy physical work demands (cases) as among subjects without (controls).

Vingård et al14 17 found a higher risk for hospitalisation or for receiving a disability pension because of kneeOA among workers in occupations with heavy physical work, such as farmers and construction workers.14 17 A possible explanation for this association is that people with highly physically demanding jobs will seek TKR or disability pension earlier than people in less demanding occupations, because they are more disabled because of their OA and not because of a higher prevalence of OA. These factors could be a problem in most of the case–control studies using TKR or waiting for TKR as the diagnostic criteria.25^–27 31

The differences between the classification criteria may explain some of the differences between the ORs reported in the studies, and the use of different cut-off points may lead to diagnostic misclassification. However, there is no reason to believe that the misclassifications will go in a particular direction, and there was no specific pattern of higher risk estimates when using TKR comparing radiological kneeOA or the diagnostic code ICD-8. Only for the study using disability pension as an outcome was the OR higher.17

The development of OA normally takes many years. Pain-disabled people who work in heavy occupations and who cannot meet the requirements for managing their physically heavy job tasks will tend to leave their trade; this results in the “healthy worker effect”.10 19 Many of the studies in this field have also included people who are no longer working in their former trade, thus avoiding the healthy worker effect.

Measurement of exposure

Some of the major problems in the reviewed studies are related to measurement of the occupational exposure. The heterogeneous nature of the exposure, the variation over time and the long duration from first exposure to the development of OA makes it difficult to obtain relevant exposure-measurements. Many studies classified the level of exposure by job title. Non-differential misclassification may occur if low-exposure occupations have been classified as high-exposure or visa versa. This may lead to a bias towards zero and possible risks in some groups may be underestimated. In case–control studies using retrospective data, the level of exposure is difficult for the subjects to remember precisely, especially many years after the event, and misclassification due to memory deficit (recall bias) can occur.25^–27 31 Subjects with knee pain may also recall their physical demands differently from subjects without knee pain and may have a tendency to overestimate their physical workload. Remembering past situations may be a problem for both cases and controls (recall bias). In a case–control study by Sahlström and Montgomery23 lifting/carrying was defined as light exposure, lifting objects from one level to another as medium exposure, and jumping as heavy exposure. This may misclassify some subjects with heavy lifting/carrying as light exposure, and a possible risk in the exposure groups may then be underestimated.

Only two studies used video-recordings, which are a more valid exposure measurement.18 24

Patho-physiological mechanisms

The pathogenesis of kneeOA in relation to workloads has not been clarified. In the standing position, the weight on either knee corresponds to 40% of body weight, during normal walking the pressure on the knees increases to 2–4 times body weight, and to six times during stair-climbing.39 40 During kneeling, approximately 70% of the body weight rests on a few square centimetres of the tibia and the patella, which may lead to damage to cartilage and bone. In a study on deep knee flexion, the estimated forces on the tibio-femoral joint were 4.7–5.6 times body weight in the vertical direction and 2.9–3.5 times body weight in the horizontal direction.41 In a study of 12 healthy subjects, the sagital moment in normal gait was measured as 15 Nm, while the moment for lifting with flexed knee was 50 Nm.42 The increase in the extensor force during deep knee flexion will increase the stress on the patella tendon and joint contact forces. At more than 150° of flexion, the extensor mechanism will also apply a posteriorly directed force on the tibia and this will increase the total force at the knee.

The dose–response relationship documented in the studies on heavy lifting and climbing stairs supports the hypothesis of a biological gradient. A biomechanical model supports the notions that the load can cause damage to the knee joint because of heavy lifting and climbing stairs.

If the mechanical effect is the primary cause in developing kneeOA, obesity may also increase the risk of kneeOA by increasing the load on the weight-bearing joints. Alternatively, because obesity and OA are both associated with a genetic predisposition, it has been proposed that the two conditions could be linked if the genes that cause obesity also predispose to OA, but this could not be verified in a large twin study.43 Another theory has been that obesity, by changing the hormone balance, may change the risk of OA. This is supported by the fact that obesity also may increase the risk of hand OA.

Another hypothesis is that subchondral micro-fractures may induce OA.44 Micro-fractures may occur when the joint is in extreme positions or when physical workload exceeds a critical level. Radin et al45 reported micro-fractures appearing in the subchondral bone due to repeated high forces across a joint. The overlying cartilage has to absorb more force, which will cause degeneration of the cartilage. These studies indicate that OA is caused by a mechanical effect and/or micro-fractures during repeated physical workload.

CONCLUSION

Nine of 17 studies with different study designs showed a significantly increased risk of kneeOA in subjects with heavy lifting, and eight of 12 studies showed a significantly increased risk in subjects with kneeling work demands. Studies in specific occupations (construction and mining) support the results. The best-evidence synthesis of a causal relationship is considered as moderate. For the combination of kneeling and heavy lifting, the association seems stronger than for heavy lifting alone, but there are only a few studies. No studies have investigated a dose–response relationship. The evidence for a causal relationship was therefore also considered as moderate. There are relatively few studies investigating the association between climbing stairs and development of kneeOA. Although the studies showed a positive association, all the studies had the same design, and no studies showed a dose–response relationship. The evidence of a causal relationship was considered to be limited (stair-climbing) and insufficient (ladders).

Future research should focus on well-designed prospective cohort studies with adequate follow-up time; high-quality studies on the risk of kneeOA among different workers in construction industry (e.g. pipe-fitters, tile-setters, bricklayers); studies on the dose–response relationship with quantification of the amount lifted, frequency of lifting, duration of kneeling, duration and amount of stair-climbing; new methods for outcome measurements (e.g. MRI); and on exposure measurements for epidemiological studies. Furthermore there is a need of future research clarifying the patho-physiological mechanism of the development of occupational kneeOA.