Objectives: To examine the effect of reproductive history and use of hormonal therapies on the risk of hip and knee joint replacement for osteoarthritis.
Methods: A prospective study of 1.3 million women aged on average 56 years at recruitment and followed-up through linkage to routinely collected hospital admission records was conducted. The adjusted relative risk (RR) of hip and knee replacement for osteoarthritis was examined in relation to parity, age at menarche, menopausal status, age at menopause and use of hormonal therapies.
Results: Over a mean of 6.1 person-years of follow-up, 12 124 women had a hip replacement and 9977 a knee replacement. The risk of joint replacement increased with increasing parity and the effect was greater for the knee than the hip: increase in RR of 2% (95% CI 1 to 4%) per birth for hip replacement and 8% (95% CI 6 to 10%) for knee replacement. An early age at menarche slightly increased the risk of hip and knee replacement (relative risk for menarche ⩽11 years versus 12 years, 1.09 (95% CI 1.03 to 1.16) and 1.15 (95% CI 1.08 to 1.22), respectively). Menopausal status and age at menopause were not clearly associated with risk. Current use of postmenopausal hormone therapy was associated with a significant increase in the incidence of hip and knee replacement (RR 1.38 (95% CI 1.30 to 1.46) and RR 1.58 (95% CI 1.48 to 1.69), respectively) while previous use of oral contraceptives was not (RR 1.02 (95% CI 0.98 to 1.06) and RR 1.00 (95% CI 0.96 to 1.04) for hip and knee, respectively).
Conclusions: Hormonal and reproductive factors affect the risk of hip and knee replacement, more so for the knee than the hip. The reasons for this are unclear.
Statistics from Altmetric.com
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.
Findings from some epidemiological studies suggest that female sex hormones play a role in osteoarthritis, however the nature of their influence is uncertain. Overall, women have a higher prevalence of osteoarthritis, in particular of the knee, than men.1 The incidence of disease has also been reported to increase at around the menopause.2 Few studies have examined the role of reproductive factors such as the age at menarche, age at menopause, childbearing history or use of oral contraceptives on disease risk.3 A number of studies have examined the relationship between postmenopausal hormone use and the risk of osteoarthritis or joint replacement however there is no consensus on what the effects are.4 Hence, this report examines the relationship between reproductive history and use of hormonal therapies and the risk of primary hip and knee joint replacement for osteoarthritis in a large cohort of women in the UK.
Study design and data collection
The Million Women Study is a prospective study of 1.3 million middle-aged women recruited in 1996–2001 through National Health Service (NHS) breast screening centres in England and Scotland.5 At study entry participants provided information regarding their reproductive history including their age at menarche, menopausal status, age at menopause, parity and use of hormonal therapies. Information on other lifestyle variables as well as a brief medical history was also collected. Participants were resurveyed approximately 3 years after recruitment to update, among other things, information on use of hormonal therapies. Study questionnaires are available to view at http://www.millionwomenstudy.org.uk.
For follow-up, study participants were linked to death and cancer registry data and to national hospital admission databases as described previously.6 7 Briefly, linkage to the hospital records was conducted using the participant NHS number, a unique personal identifier on all NHS health records as well as the date of birth and other identifying details. The linked hospital data (the Hospital Episode Statistics8 for England and the Morbidity Records9 for Scotland) contain a record of every NHS-funded inpatient hospital admission from April 1997 in England and January 1981 in Scotland. For each hospital record up to 14 diagnoses and up to 12 procedures are coded according to the International Classification of Diseases version 1010 (ICD-10) and the Office of Population Censuses Classification of Surgical Operations and Procedures version 411 (OPCS-4), respectively.
Women were considered to have had a hip replacement for osteoarthritis if they had a hospital record with an OPCS-4 code for a primary total hip replacement (W371, W381, W391) and the corresponding main diagnosis field indicated osteoarthritis of the hip (ICD-10 M16). Women were considered to have had a knee replacement for osteoarthritis if they had a hospital record with an OPCS-4 code for a primary total knee replacement (W401, W411, W421) and the corresponding main diagnosis field indicated osteoarthritis of the knee (ICD-10 M17). The study has been approved by the Eastern Multi-Centre Research Ethics Committee and all study participants provided written consent to be included. Access and linkage to hospital records was approved by the Information Centre for Health and Social Care in England and the Information and Statistics Division in Scotland.
Women were excluded if they had an OPCS-4 code for any hip or knee replacement before recruitment (OPCS-4 codes W37–W39 and W40–W42) regardless of the indication, or if they had any cancer except non-melanoma skin cancer (ICD-10 C44) before recruitment as this could affect the likelihood of subsequent joint replacement surgery for osteoarthritis. In analyses, hip and knee replacements were considered as competing risks. Person-years were calculated from the date of entry to the study to the date of first hospital admission for the joint replacement, date of death or the last date for which the hospital records were complete, whichever came first. For women recruited in England the hospital records were complete up to 31 March 2005 and for women in Scotland up to 31 December 2003. For the small proportion of women (5%) recruited in England before 1 April 1997 person-years were calculated from this date as hospital records were not available in England before this time.
Cox proportional hazards models were used to estimate the relative risk of hip and knee replacement according to parity (in categories: nulliparous, 1, 2, 3, 4+; and as a continuous variable), age at menarche (⩽11, 12, 13, 14, 15+ years), menopausal status (pre/peri versus post), age at menopause (⩽48, 49–51, 52+ years) for postmenopausal women, oral contraceptive use (never, ever and duration of use) and hormone replacement therapy (HRT) use (never, past, current and according to duration of use and type in current users only). The underlying time variable used was attained age and all analyses were routinely stratified for the recruitment region (10 geographic regions) and adjusted for socioeconomic status (in 5 categories of deprivation6), body mass index (10 equal categories of 2.5 kg/m2 increments from <20 to 40+ kg/m2), height (<160, 160–164, 165+ cm), smoking (never, past, current), alcohol intake (never, 1, 2, 3+ units/day) and each of the reproductive variables examined. We examined the effect of additional adjustment for other medical illnesses and the frequency of strenuous physical activity reported at recruitment. As body mass index (BMI) is associated with some reproductive factors we also examined the relationship between each reproductive variable and risk of hip and knee replacement in subgroups of BMI (<25 kg/m2, 25–29.9 kg/m2, 30+kg/m2).
The analyses examining menopausal status were restricted to women who had never used HRT and were aged <55 years and follow-up was censored at age 55. The analyses examining age at menopause were restricted to postmenopausal women with a natural menopause who had never used HRT and were older than 52 years at recruitment. Analyses examining HRT use were restricted to postmenopausal women and adjustment was made for hysterectomy and bilateral oophorectomy in addition to the other factors. As the use of HRT may have changed over time, in these analyses women were categorised according to HRT use reported at recruitment and contributed person-years up to the time they returned a resurvey questionnaire when HRT use was updated according to what they had reported on the resurvey. For the 35% of women who did not return a resurvey, follow-up was censored at 48 months from last contact. Unless otherwise specified, in analyses the χ2 likelihood ratio test was used to test for heterogeneity and tests for linear trends were based on the categorical variables scored as the mean in each category. Due to the large numbers of cases and controls, tests for heterogeneity and trend were considered significant when p values were less than 0.01. The Stata 9.2 statistical program was used for all analyses (Stata, College Station, Texas, USA).
After excluding 3934 women with a joint replacement before recruitment and 43 399 with pre-existing cancer, these analyses include prospective data on hospital admissions for hip and knee replacement in 1 306 081 women. Over a mean follow-up of 6.1 person-years per participant, 12 124 women had a primary total hip replacement for osteoarthritis and 9977 had a primary total knee replacement for osteoarthritis, corresponding to incidence rates of 1.5 per 1000 person-years for hip replacement and 1.2 per 1000 person-years for knee replacement.
Table 1 shows the characteristics of the study population according to incident hip or knee replacement and table 2 shows some characteristics according to each of the reproductive and hormonal variables. Women with a joint replacement differed from the total population on a number of aspects, and this was particularly so for those with a knee replacement. Compared to the entire cohort, women with a hip replacement were on average older at recruitment and more overweight, and were less likely to have used oral contraceptives. Compared to the entire cohort, women with a knee replacement were on average older at recruitment and substantially more overweight. They were also more likely to be in a lower socioeconomic group, less likely to smoke, drank less alcohol, were less likely to have used oral contraceptives and were more likely to have had a hysterectomy.
Regarding the baseline characteristics according to the reproductive and hormonal variables (table 2) age, BMI and socioeconomic group at recruitment differed somewhat. Postmenopausal women were older than pre/perimenopausal women and never users of oral contraceptives were older than ever users. BMI increased with increasing parity and with younger age at menarche. Socioeconomic group varied somewhat by parity, menopausal status, age at menopause and use of HRT.
Table 3 shows the relative risk of hip and knee replacement for osteoarthritis according to various categories of reproductive factors adjusted for age, recruitment region, BMI, socioeconomic status, height, smoking, alcohol use as well as parity, age at menarche, oral contraceptive use and HRT use as appropriate. Compared to nulliparous women the relative risk of hip replacement and of knee replacement increased with increasing parity although the increases in relative risks were greater for knee replacement than hip replacement. Examining parity as a continuous variable, for hip replacement the increase in risk was 2% (95% CI 1 to 4%) per birth while for knee replacement it was 8% (95% CI 6 to 10%).
Women reporting a younger age at menarche (age ⩽11 years) had a greater risk of hip and knee replacement compared to women reporting menarche at age 12 (relative risk (RR) 1.09 (95% CI 1.03 to 1.16) and RR 1.15 (95% CI 1.08 to 1.22) for hip and knee replacement, respectively). The test for linear trend was significant for an increase in the risk of knee replacement (p<0.001) with a younger age of menarche however this appeared to be largely due to the increased risk in women reporting menarche at 11 years or younger. Menopausal status and the age of menopause were not significantly associated with hip or knee replacement (p = 0.5 and p = 0.06 for linear trend, respectively).
Table 4 shows the relative risk of hip and knee replacement for osteoarthritis according to use of hormonal therapies. All ever users of oral contraceptives were past users at the time of the study and compared to never users, ever use of oral contraceptives was not significantly related to the risk of either hip or knee replacement (RR 1.02 (95% CI 0.98 to 1.06) and RR 1.00 (95% CI 0.96 to 1.04), respectively). Current use compared to never use of HRT, was associated with an increased risk of joint replacement (RR 1.38 (95% CI 1.30 to 1.46) for hip replacement and RR 1.58 (95% CI 1.48 to 1.69) for knee replacement). Past users of HRT also had a higher risk than never users although relative risks were somewhat lower than those in current users. Among current users of HRT the relationship between risk and duration of use was inconsistent. There was a significant difference in the risk of hip replacement according to the type of HRT used (comparing oestrogen only, oestrogen plus progestagen and tibolone, heterogeneity p = 0.002) with higher risks in women using oestrogen plus progestagen formulations compared to the other types. However we did not find significant differences for knee replacement by type of HRT (heterogeneity p = 0.1).
Additional adjustment for high blood pressure, heart disease, stroke, venous thromboembolic disease, asthma, thyroid disease, diabetes and frequency of strenuous physical activity reported at recruitment did not change our results substantially (relative risks differed from the original estimates by less than 5%). Similarly, when we examined the relationship between each of the reproductive and hormonal variables in subgroups of BMI we found no evidence of heterogeneity.
In this large prospective study of middle-aged women we found that some reproductive and hormonal factors were associated with hip and knee replacement surgery and most effects were greater for the knees than the hips. Increasing parity results in an incremental increase in the risk of incident hip and knee replacement, with the increase in relative risk per birth being greater for knee than for hip replacement. A young age at menarche (⩽11 years) was also found to increase the risk of hip and knee replacement, but menopausal status and the age at menopause were not related. With respect to use of hormonal therapies, we found that use of postmenopausal hormone therapy was associated with an increase in the incidence of hip and knee replacement but that previous use of oral contraceptives was not associated with risk.
Previous studies examining the relationship between reproductive history or hormonal therapies and osteoarthritis or joint replacement have been inconclusive. With respect to parity, of eight studies12–19 of osteoarthritis or joint replacement that we identified, only one cross-sectional study19 reported a significant increased risk of hand osteoarthritis with increasing number of children. One out of six observational studies12–15 20 21 examining oral contraceptive use reported an increased risk of hip replacement which was of borderline significance.21 Neither of two previous studies17 22 reported a significant association between the age of menarche or the age of menopause and the risk of osteoarthritis. Regarding HRT, studies using radiographic changes to define osteoarthritis have in general suggested a protective effect,23 whereas findings are inconsistent in studies using hip and knee replacement as an outcome. One randomised controlled trial found no significant effect of HRT use on the incidence of hip and knee replacement24 and similarly a prospective study of hip replacement found no significant association with HRT use.12 However one out of four case–control studies of hip replacement13 21 and knee replacement14 20 reported a significantly increased risk of knee replacement with HRT use.20
Oestrogen receptors are found on bone and cartilage cells.4 25 Oestrogen is known to prevent bone loss and some studies show associations between high bone density and radiographic changes of osteoarthritis.26 27 Hence it is plausible that greater exposure to oestrogens, while preventing bone loss, may promote osteoarthritic changes that lead to disability and joint replacement. With respect to its effects on cartilage, some studies suggest that use of oestrogen therapy is associated with greater cartilage thickness28 while others suggest that they may result in damage to cartilage.29 Furthermore, it has been suggested that oestrogen may have different effects on the initiation and progression of osteoarthritis.23 Given these heterogenous effects it is difficult to make inferences on the biological mechanisms which underlie the associations that we found, however the effects of oestradiol on excessive bone formation and chondrocyte synthesis of collagen could be explored further in animal models and in vitro studies.
High BMI is known to increase the risk of osteoarthritis and joint replacement, and this is particularly the case for the knee joint,30 but it is unlikely that a woman’s current BMI would explain the associations found here as we adjusted all analyses for current BMI in very fine categories, and our findings were consistently observed within subgroups of current BMI. However as each pregnancy results in a period of increased body weight, it is possible that the relationship between parity and the risk of joint replacement in middle age is a long-term result of the increase in weight bearing experienced during pregnancy. Similarly, a younger age at menarche may also be a marker of other factors such as body habitus when young which could also influence the development of osteoarthritis and joint replacement.12
The association between use of HRT and the incidence of joint replacement may be due to non-biological factors. Women who take HRT may have more access to health services and as such, may be more likely to have a joint replacement for existing osteoarthritis. In this cohort, women who had a knee replacement were in a higher socioeconomic group and were more likely to have other operations such as hysterectomy (table 1).
This is one of the largest studies of the relation between reproductive factors and joint replacement surgery in women and hence we were able to examine the effects of reproductive factors in greater detail than was possible in many previous studies. Outcomes were ascertained through linkage to NHS hospital records and therefore close to complete ascertainment of incident hip and knee replacement is likely. While the hospital admission databases contain only NHS-funded admissions, we have previously shown that in this population, who were recruited through NHS breast screening centres, ascertainment of incident joint replacement through linkage to NHS hospital records compares well with self report, suggesting that few events are missed.31 Furthermore incidence rates of joint replacement in the study population are comparable with those reported in other UK populations32 33 suggesting that our ascertainment of outcomes was comprehensive.
Despite being self reported, the reproductive variables, parity, age at menarche, age at menopause and use of hormonal therapies are known to be reasonably reliable.34 35 Also, as this was a prospective study the exposures of interest were ascertained before joint replacement events so any misclassification of exposures would bias our results towards the null. As the women included in these analyses were middle aged at recruitment their parity, menarche, and use of oral contraceptives would not change during the follow-up period and information on use of HRT was updated during follow-up, wherever possible. Joint replacement is elective surgery and as such there are other influences that determine whether women with osteoarthritis will go on to have a joint replacement.33 36 While we could not account for all potential confounders such as some types of physical activity or treatments that may influence the progression of osteoarthritis, we were able to adjust for many important factors including age, region of recruitment, socioeconomic status and BMI as well as medical illnesses and frequency of strenuous physical activity reported at recruitment.
Given the large burden of osteoarthritis and the associated burden of joint replacement surgery in women worldwide,37 it is important to understand the role of potentially modifiable factors for these conditions. We found that parity, age at menarche and HRT use are all associated with the risk of hip and knee replacement and that the knee joint is affected more by these factors than the hip joint. However the underlying reasons for these findings remain unclear.
We thank all the women who participated in the Million Women Study, collaborators from the NHS Breast Screening Centres, members of the study co-ordinating centre, and the study steering committee. We also thank the Information Centre for Health and Social Care and ISD Scotland for the linkage to the hospital records.
Funding: This research was funded by Cancer Research UK, the NHS Breast Screening Programme and the Medical Research Council. The study sponsors were not involved in the design, analysis, interpretation or writing of this report.
Competing interests: None.
Ethics approval: The study has been approved by the Eastern Multi-Centre Research Ethics Committee and all study participants provided written consent to be included.