Article Text

Download PDFPDF

Original article
Cumulative years in occupation and the risk of hip or knee osteoarthritis in men and women: a register-based follow-up study
  1. Susan Andersen,
  2. Lau Caspar Thygesen,
  3. Michael Davidsen,
  4. Karin Helweg-Larsen
  1. National Institute of Public Health, University of Southern Denmark, Copenhagen, Denmark
  1. Correspondence to Susan Andersen, National Institute of Public Health, University of Southern Denmark, Øster Farimagsgade 5A, 1353 DK Copenhagen, Denmark; sua{at}


Objectives Occupational workload has been associated with an increased risk of osteoarthritis (OA), but only little research has been conducted among female workers. The objective of this study was to analyse if men and women in farming, construction or healthcare work have increased risk of developing OA of the hip or knee.

Methods A follow-up study based on register data of the whole Danish working population in the period 1981 to 2006 followed up for hip or knee OA during 1996 to 2006. Cumulative years in occupation were calculated for assessment of dose–response relationship. Gender-specific analyses were carried out with Cox regression models using age as timescale and adjusting for calendar period, income, unemployment and previous knee injury, and done separately for hip and knee OA.

Results Male floor layers and bricklayers and male and female healthcare assistants had the highest risks of knee OA, and farmers had the highest risk of hip OA. Male farmers had increased risk of hip OA already after 1–5 years in occupation (HR, 1.63) and a dose–response-related risk of hip OA (HR up to 4.22). Generally, the risk of OA increased with cumulative years in the occupation in both men and women.

Conclusions Occupations with heavy physical workload present a strong risk for hip and knee OA in both men and women, and the risks increase with cumulative years in occupation and noticeable hip OA among male farmers.

  • Osteoarthritis
  • occupational groups
  • dose–response relationship
  • follow-up studies
  • gender
  • epidemiology
  • exposure assessment
  • workload

Statistics from

What this paper adds

  • There is evidence that occupational workload is associated with an increased risk of OA. This is known for male workers.

  • This study examines the association between several occupations with heavy workload and occurrence of hip and knee OA.

  • In particular, this study adds to the evidence of whether occupations with heavy workload are a causative factor for OA in women.

Policy implications

  • Preventive strategies may benefit from assessment of occupational workload to prevent physical wear.


In Western societies, osteoarthritis (OA) is a common chronic disease and is becoming increasingly important due to its prevalence in ageing populations. OA is a major contributor to functional impairment,1 2 and work-related musculoskeletal disorders are believed to be a major cause of disability in the labour force.3 Thus, OA is a public health problem, and identification and monitoring of high-risk groups is important. Several studies have found that age, female gender, injuries, overweight and vigorous physical activity in sport and work are associated with increased risk for OA of the hip or knee.4–11 While a number of studies have proven an increased risk in construction workers,10 12 13 floor layers,13 14 forestry workers10 and farmers,10 15–17 only little research on the risk in other occupations has been conducted and very few studies among female workers. Thus, there are limited data to describe the possible occupational risk factors for OA among women. However, one study found high physically demanding tasks to be risk factors for hip OA in women.18 Another study has shown that among women farming and exposure to physically demanding tasks at home were associated with knee OA10 and, further, working in the clothing industry, cleaning and agriculture have been identified for the risk of OA of the hip, knee and hand in women.19 Contrary to this, other studies have reported that several occupations with workload, for example, cleaning, farming and healthcare were not related to OA in women.12 19 20 Most studies that examined the relationship between occupation and OA are either case–control or cross-sectional studies and only few studies have been designed as prospective follow-up studies.13 16 17 In Denmark, using register-based data, there are unique opportunities to obtain nationwide individual data over a long period of time. This approach is used in the present study to examine occupations we expected to have sufficient workload to be a causative factor for OA of the hip and knee. As OA may be dose-related to time with workload and build up over years through cumulative exposures,21–24 we wanted to study the cumulative effect of number of years in the occupation on hip and knee OA. We used comprehensive register data covering employees and identified cases of hip or knee OA in the National Patient Register.

Materials and methods

All data are derived from individual-level registers at Statistics Denmark. For each of the years 1981–2006, information on labour market affiliation for each person living in Denmark and aged 16 years or older was obtained from Integrated Database for Labour Market Research (IDA),25 and using the unique Danish personal identification number,26 each person was followed over the period. Furthermore, each person was in 1996–2006 followed in the Danish Patient Register27 in order to define first surgically treated hip and knee OA and in the Danish Civil Registration System to see if she dies or emigrates.

Study population

Persons in our study are from one of the following five occupational groups comprising both skilled and unskilled workers:

  1. Floor layers/bricklayers/pavers (hereafter referred to as floor- and bricklayers): work with flooring and wall covering, bricklaying and paving.

  2. Construction workers: total construction work such as construction of buildings, insulation, carpentry, painting, plumbing and electrical work. This group also includes floor- and bricklayers.

  3. Farmers: agricultural and horticultural work.

  4. Healthcare assistants: nursing assistants, nursing aides and similar health caretakers in the healthcare and social system.

  5. Office workers: clerical staff, accounting staff and computer staff within banks, financing and insurance.

Persons in group 1–4 were considered to be exposed, while group 5 serves as control. We assume that controls were comparable with our study groups with regard to socioeconomic status and not exposed to jobs with high physical load.

Exposure definition

We used several sources to define the occupation groups in order to ensure good coverage.28 A person was in a specific group if he or she either (a) had appropriate work skills/position or (b) worked at an appropriate establishment25 and had an appropriate vocational oriented education.

IDA in itself contains information from several registers.25 From 1981 to 1991, work skills/position was coded according to a classification based on the work position and from 1992 according to skills required to perform work functions using the Danish version of the International Standard Classification of Occupations (ISCO 88). Both classifications are based on information from the Danish tax system using the income register.29

IDA provides the possibility to link information on establishment to individual-level information for each person each year.25 Information on the industry of each establishment is coded according to the Danish version of the European Industrial Classification of all Economic Activities (NACE). Only information on the establishment where the person was employed full time was used.

Information on highest fulfilled level of vocationally oriented education30 is coded with an eight-digit education code. The first two digits provide information on the level of education (eg, high school, vocational, bachelor), the following two digits indicate the type of education (eg, building construction), digits 5–6 provide a more detailed grouping (eg, carpenter) and the last two digits are the separate education titles available within each discipline.

The criteria (a) and (b) mentioned above may lead to two different results regarding occupational group membership. To avoid this, the following prioritisation was used: (1) floor layers, (2) construction workers, (3) farmers, (4) healthcare workers and (5) office workers. All information was constructed to be valid at 1 January each year.

Outcome variables

OA was defined based on the National Patient Register.27 This register contains data on all persons admitted to hospitals since 1977. The register includes discharge diagnoses of hospitalised patients indicating the main medical reason for diagnostic procedures or treatment. Discharge diagnoses are coded by physicians, and since 1994, they have been coded according to the 10th version of the International Classification of Diseases (ICD-10). Surgical procedures are registered by operation codes since 1996, coded according to the Danish version of the Nordic Classification of Surgical Procedures. A person was defined to have a hip OA if she (a) had a discharge diagnosis of specified OA in several joints in hip, knee or foot (ICD-10, M15–M17, M19) and (b) within the same calendar year had a hip replacement surgery performed (operation code KNFB) and a knee OA if she had (a) and (b) a knee replacement (KNGB) within the same year.

Background variables

Age was computed as the age at 1 January each year. Calendar year was categorised into three periods: 1981–1989, 1990–1999 and 2000–2006. Annual personal income derived from the income register29 was for each year categorised in quartiles based on the whole Danish population. Annual personal unemployment available from the register on Coherent Social Statistics was categorised in proportion of the year: ≤20%, 21%–80% and >80%. A person was considered to have had a previous knee injury if the person since 1977 had been hospitalised with an appropriate discharge diagnosis (yes/no).

Statistical analysis

Each employee was identified for each year from 1 January 1981 until 1 January 2006 and followed for first diagnosis of hip or knee OA from 1 January 1996 until 31 December 2006 or until death or emigration, whichever came first. All analyses were performed separately for men and women and for each exposure group. Initially, age-standardised incidence rates of OA were calculated by direct standardisation, using the age distribution for both sexes combined (with 5-year age groups) of the office workers as the standard population. HRs and 95% CIs for OA were estimated using Cox proportional hazards regression models with age as the underlying timescale. For a given exposure group, a person was included as exposed from the age at which she the first time was seen in the group. A person was considered to be in the reference group if she at no time during the follow-up was in any of the exposure groups but in at least 1 year was in the control group—the age at entry was age at first year as control. Age at exit was age at event/censoring31—thus for example, a floor- and bricklayer was considered to be floor- and bricklayer from entry to exit, even though she during the period had had another occupation. Cumulative years in the studied occupation was summed dynamically adding 1 year for each year a person was in the given occupation and categorised as 1–5, 6–10 and >10 years. The Cox regression models were adjusted for year of entry (three groups), annual personal income (four groups), annual personal unemployment (three groups) and previous knee injury (yes/no). Each of these variables was treated as being time-dependent. Analyses were performed using SAS (Proc TPHREG, SAS Institute Inc), V.9.1.


Median age at entry to follow-up was between 36 and 44 years (table 1), and mean follow-up time was between 8.1 and 9.2 years. For the four occupational groups studied, a higher proportion had no education beyond primary school compared to office workers (35.2%–57.4% vs 27.2%). Floor- and bricklayers were more likely than office workers to have a low income. The proportion with a history of knee injury varied between 3.5% (floor- and bricklayers) and 1.9% (farmers). The age-standardised incidence rates were generally higher for surgically treated OA of the hip than the knee (table 2).

Table 1

Characteristics* of the occupational groups

Table 2

Age-standardised incidence rate per 100 000 person-years of hip and knee OA according to occupation, stratified by gender

Occupation and knee or hip OA in men

Compared to office workers, male farmers had an increased risk of developing hip OA and male floor- and bricklayers had an increased risk of developing knee OA (table 3). The risk increased by number of years in the occupations (table 4). Male farmers had an increased risk of hip OA following 1–5 years of work as farmer (HR, 1.63) and threefold increased risk after 10 years. Among male healthcare assistants, an elevated risk of developing knee and hip OA was only seen in those who had worked >10 years (HR, 2.44 and 1.71, respectively).

Table 3

HR (95% CI) for surgically treated OA in the hip or knee according to occupational group and gender

Table 4

HR (95% CI) for surgically treated OA in the hip or knee according to cumulative years in occupation among men and women

Occupation and knee or hip OA in women

Female healthcare assistants who had worked >5 years had increased risk of developing knee and hip OA (table 4). The risk of developing knee OA was twofold in female construction workers who had worked for 6–10 and >10 years. Likewise, female farmers who had worked for 6–10 and >10 years had about a similar increased risk of hip OA (HR, 1.59 and 1.62, respectively). We did not find strong evidence for neither a risk of hip OA among female construction workers nor a risk of knee OA among female farmers.

Dose–response relation of number of years in the occupation

A dose–response relation with higher risks of OA the more years of construction work, farming and healthcare work compared to office work was observed, and the risk of OA increased even further when cumulative years in occupation was >20 years (figure 1). However, the dose–response effect seemed more pronounced for some of the occupations studied than others. In male farmers, there seemed to be a direct dose–response relation between cumulative years in farming and hip OA resulting in HR of 4.20 among those who had worked >20 years. Similarly, a dose–response-related risk of knee OA was found among male construction worker (HR up to 2.79) and among male and female healthcare assistants (HRs up to 3.50 and 2.88, respectively).

Figure 1

Cumulative years in occupation and risk of knee osteoarthritis (OA) or hip OA in (A) men and (B) women, according to occupation.


In this nationwide register-based follow-up study, we found that construction, farming and healthcare work compared to office work increases the risk of OA in both men and women. Further our results indicate that the risk of OA increases with cumulative years in the occupation. Farmers have the highest risk of hip OA, and construction workers and healthcare assistants have the highest risk of knee OA. In general, the risk estimates are higher for men than for women. An exception to this is among construction workers, where the risk estimates of knee OA are similar or even slightly higher for women.

To our knowledge, the majority of previous studies have dealt with either hip or knee OA and mainly among men. Few of these studies have been based on long-term follow-up. We analysed the risk of both hip and knee OA in different occupations and by years of occupation as well as by sex. Our results of an increased risk of hip OA among farmers and knee OA among floor- and bricklayers and construction workers are in accordance with previous Nordic follow-up studies. A Danish study17 investigated time trends of hospital admissions for hip OA in four consecutive cohorts of men during 1981–1999 with 3–6 years of follow-up and found that self-employed farmers had a consistently high risk of hip OA and employed farm workers had a relative lower risk, the risk increased during the calendar period. In our study, we did not distinguish between self-employed and employed workers. It is possible that self-employed workers are more concerned about their jobs, and hence, frequency and duration of physical load may be considerably higher at the expense of their physical capabilities. In a Swedish study16 with rural men and urban referents, the results showed that farmers had an increased risk of hip OA (HR, 3.0) and OA in any site (HR, 2.1). Our results support that farming contributes to a greater risk of hip OA than knee OA. In another Swedish study based on a cohort of male construction workers,13 floor- and bricklayers had an increased risk of surgically treated knee OA compared to white-collar workers (RR, 4.72 and 2.14, respectively). This study further reports that construction work presents a greater risk of knee OA than of hip OA, which is partially supported by our results. However, we found a similar risk of OA in the knee and hip among male floor- and bricklayers who had worked for >10 years.

Our findings suggest that occupations characterised by high physical strain after 5 or 10 years imply a significant risk of OA for both sexes. Apart from construction workers where a similar risk of knee OA is seen among men and women, men have a higher risk of OA than women. The gender differences may be a consequence of women not having the same amount, frequency and duration of occupational physical load. Few studies have focused on the risk of OA in professions with many female employees such as healthcare worker, and the studies have not included men. The study of Sandmark et al10 showed an increased risk of replacement surgery due to knee OA (OR, 2.2) among women who had at least 10 years of physically demanding task at home such as nursing and taking care of an older or handicapped person (jobs considered to entail high physical demands on the knees) compared to women who had never had jobs. These results are comparable to our findings of a high risk of both knee and hip OA among female healthcare assistants. Contrary to these results, another Swedish case–control study12 reported that compared to population controls, healthcare work >1 year among women (nursing assistants, nursing aides and nurses) showed no elevated risk of radiographically diagnosed knee OA. However, it was a small study based on the population in the southern part of Sweden and the authors used another study design and definition of OA than the present study.

In our large-scale study, the risk of OA generally increased with cumulative years in the studied occupations; however, the number of individuals decreased with cumulative years in occupation, especially pronounced for female construction workers and female farmers. Dose–response relation has previously been reported between combined hours per day by years with specific work task and OA.7 10 22 32–34 To our knowledge, only few studies12 21 have investigated dose–response relation by occupational title. In a Danish study, Jensen21 found a positive dose–response relationship between combined daily time spent in kneeling or squatting with years working as floor layers or carpenters and the development of radiographic knee OA. Holmberg et al12 found that men who had worked for 11–30 years in construction work had an increased risk of radiographic knee OA (OR, 3.7).

Our results were solely based on register data using only occupational codes to measure occupational strains. Register data may be biased by an imprecise definition of exposure. Our approach does not take into account the possible differences in exposure within a given occupation. Thus, farming might be considered rather heterogeneous regarding raising livestock or crops.35 However, we based our exposure classification upon strict criteria so that the groups reflect the actual workers with relevant job functions. Our analyses did not include survey data and hence we were not able to include analyses on potential confounders, such as obesity. If obesity acts as a confounder, it may have attenuated the association between occupations with workload and OA because obesity might be positively associated with OA and negatively associated with engagement in occupational workload. However, manual work tends to be associated with obesity, and then confounding would be in the opposite direction.

The strength of the present study design is that we included the whole Danish working population over a long period of time, and information on exposure was not biased by recall bias both because information on exposure was registered before OA and because exposure information was not self-reported. Furthermore, due to our register-based analyses, there was no non-response bias.

There exist difficulties in defining OA. Radiological screening, clinical investigation, hospitalisation due to joint complaints or joint replacements as well as self-reported joint symptoms are frequently used as definitions of OA. Radiological screening might be sensitive, but several studies have found poor correlation between radiographic findings and clinical symptoms.36–39 Joint replacement is influenced by several factors, for example, the patient's perceptions of the indications for and risks associated with the operation and the health professional decisions regarding indications for joint replacement.3 40 41 Hence, our strict definition of OA, joint replacement due to OA, may have been influenced by referral biases. The Danish healthcare system is publicly funded, which would minimise bias due to socioeconomic status. However, we cannot exclude the possibility that those in physically demanding jobs are more likely to be impeded by OA and therefore are treated at an earlier phase of symptoms than those who are not engaged in strenuous work. On the opposite, a Swedish study42 found farmers to have a relatively low inclination to seek healthcare for musculoskeletal symptoms, which would result in reduced risks of OA. Contrary to other professions, farmers may also have personal motivations to delay a change in occupation due to familial affiliation.20 Several previous studies have not adjusted for bias in regard to socioeconomic status and thus exposure group and reference group have not been comparable. In our study, the groups were comparable with regard to socioeconomic factors, and all analyses were adjusted for income and unemployment to minimise bias due to potential confounders related to social class.43

In conclusion, workers with 6–10 or >10 years cumulative work characterised by physical strain have strongly increased risk of OA. Among male farmers, the risk of hip OA was already increased after 1–5 years. The risk estimates were except for knee OA among construction workers higher for men than for women. In both men and women, there seemed to be a dose–response relation between cumulative years in occupation and development of OA, at least of hip OA among farmers and of knee OA among healthcare assistants. Long-term strenuous working conditions are an important public health issue. Future preventive strategies may benefit from assessment of occupational physical load to prevent physical wear.



  • Funding The study is supported by a research grant from the Danish Ministry of Interior and Health.

  • Competing interests None.

  • Patient consent Personal information is fully anonymised due to register based data on the whole Danish population.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.