If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
The HUNT Research Centre, Department of Public Health and Nursing, Faculty of Medicine and Health Sciences, Norwegian University of Science and Technology (NTNU), Levanger, NorwayDivision of Orthopaedic Surgery, Oslo University Hospital, Oslo, Norway
Faculty of Medicine, University of Oslo, NorwayResearch and Communication Unit for Musculoskeletal Health, Division of Clinical Neuroscience, Oslo University Hospital, Oslo, Norway
The Norwegian Arthroplasty Register, Department of Orthopedic Surgery, Haukeland University Hospital, Bergen, NorwayDepartment of Clinical Medicine, Institute of Medicine and Dentistry, University of Bergen, Bergen, Norway
Faculty of Medicine, University of Oslo, NorwayResearch and Communication Unit for Musculoskeletal Health, Division of Clinical Neuroscience, Oslo University Hospital, Oslo, Norway
Faculty of Medicine, University of Oslo, NorwayResearch and Communication Unit for Musculoskeletal Health, Division of Clinical Neuroscience, Oslo University Hospital, Oslo, Norway
The HUNT Research Centre, Department of Public Health and Nursing, Faculty of Medicine and Health Sciences, Norwegian University of Science and Technology (NTNU), Levanger, Norway
To investigate whether parity, age at menarche, menopausal status, age at menopause, use of oral contraceptives (OC) or use of hormone replacement therapy (HRT) were associated with total knee replacement (TKR) or total hip replacement (THR) due to primary osteoarthritis.
Method
In a prospective cohort study of 30,289 women from the second and third surveys of the Nord-Trøndelag Health Study, data were linked to the Norwegian Arthroplasty Register (NAR) in order to identify TKR or THR due to primary osteoarthritis. Cox proportional hazards models were used to estimate the hazard ratios (HRs).
Results
We observed 430 TKRs and 675 THRs during a mean follow-up time of 8.3 years. Increasing age at menarche was inversely associated with the risk of TKR (P-trend < 0.001). Past users and users of systemic HRT were at higher risk of TKR compared to never users (HR 1.42 (95% confidence interval (CI) 1.06–1.90) and HR 1.40 (95% CI 1.03–1.90), respectively). No association was found between parity, age at menarche, menopausal status, age at menopause, oral contraceptive use or HRT use and THR.
Conclusion
We found that increasing age at menarche reduced the risk of TKR. Past users and users of systemic HRT were at higher risk of TKR compared to never users. Parity did not increase the risk of THR or TKR.
Osteoarthritis is probably the result of a complex interplay between genetic, cellular and biomechanical factors. A better understanding of the risk factors and, thereby, groups at risk, would make it possible to target effective public health preventions
. The findings from epidemiologic studies on reproductive history (parity, age at menarche, menopausal status and age at menopause) and hormonal factors (oral contraceptives (OC) and hormone replacement therapy (HRT)) in relation to osteoarthritis have been conflicting. Increasing parity has been reported as a risk factor for radiographic osteoarthritis in the knee
. However, some studies have not found any association between parity and radiographic joint space narrowing, osteophytes or changes in either cartilage volume or cartilage defects
The effect of estrogen plus progestin on knee symptoms and related disability in postmenopausal women: the Heart and Estrogen/Progestin Replacement Study, a randomized, double-blind, placebo-controlled trial.
The aim of this study was to investigate the association between reproductive history and use of hormonal therapies and the risk of TKR or THR due to osteoarthritis in a prospective cohort study.
all inhabitants of Nord-Trøndelag county ≥ 20 years of age were invited to participate in three surveys: HUNT1 (1984–1986), HUNT2 (1995–1997) and HUNT3 (2006–2008)
. Our study only included baseline data from HUNT2 or HUNT3 as these surveys included questionnaire and interview data on reproductive history and covariates. We included women aged ≥30 years at baseline, and our study population consisted of 11,746 participants from HUNT2, 20,459 participants of both HUNT2 and HUNT3 and 4652 participants from HUNT3 alone. For those who participated in both HUNT2 and HUNT3, we used baseline measurements from HUNT3 in order to include as much information as possible on reproductive history and eventual use of HRT. In this study we defined reproductive history as parity, age at menarche, years of menstruation and age at menopause. Hormonal therapies included use of OC and use of HRT. Height and weight were measured by trained personnel. Body mass index (BMI) was calculated as weight in kilograms divided by squared height in metres. Bilateral oophorectomy in premenopausal women induces premature menopause
, and women who undergo a hysterectomy with ovarian preservation may almost double their risk of premature menopause compared to women with intact uteri
. We therefore chose to exclude both of these groups at baseline (n = 3710). After also excluding 1183 participants with joint replacement before recruitment, 91 with missing date of operation, 436 with missing BMI and 1148 with missing information on smoking, the analyses included 30,289 women (Fig. 1).
For follow-up, we identified cases with a TKR or THR due to primary osteoarthritis, according to the operating surgeon, using information from the Norwegian Arthroplasty Register (NAR). This linkage was conducted using the 11-digit personal identification number that is unique to each Norwegian citizen. NAR contains a record of over 95% of all TKRs and THRs in Norway
. If a person had more than one arthroplasty, only the first procedure was considered as the event.
Cox proportional hazards models were used to estimate the hazard ratios (HRs) of TKR and THR according to parity (nulliparous, 1, 2, 3, 4+ births), age at menarche (≤11, 12, 13, 14, 15+ years), menopausal status (pre/peri- and postmenopausal), age at menopause (≤48, 49–51, 52+ years), years of menstruation (age at menopause minus age at menarche), oral contraceptive use (never or ever, and duration of use) and HRT use (never, past, current; local or systemic and duration of use). Age was used as the time scale in the analyses. Model 1 adjusted for BMI (continuous) and smoking (never, former or current). The fully adjusted model 2 also adjusted for physical activity (none, medium, hard) and other reproductive variables as appropriate for the individual exposures. Each exposure was analysed for its interrelationship with other potential hormone-related confounders in a direct acyclic graph (DAG), resulting in a slightly different set of confounders for each exposure (Table A, Appendix). In these DAG analyses, diabetes was only found to be a potential confounder to parity and age at menarche, and thus only adjusted for in these two analyses.
Information on education level was only available for 8745 participants from HUNT2, and an additional sensitivity analysis adjusting for education was performed on this group. Education level was evaluated to be a confounder to parity, oral contraceptive use and HRT (Tables B and C, Appendix), and was defined as the highest level of completed education (primary/vocational, secondary or post-secondary).
The analyses examining age at menopause were limited to postmenopausal women who had never used HRT. The tests for linear trends were based on the categorical variables scored as the mean of each category. All statistical analyses were two-sided with a significance level of P < 0.05. The analyses were performed using Stata 14.0/SE (StataCorp LP, College Station, TX, USA). Tests based on Cox regression methods showed no evidence that proportional hazard assumptions were violated.
Ethics
The participants signed written informed consent for participation in HUNT, NAR and linkage of data to national health registries. This study was approved by the Norwegian Regional Committee for Ethics in Medical Research (2013/151/REK Sør-Øst C).
Results
For the 30,289 women included in the study population, the mean age at baseline was 55.7 and mean follow-up time was 8.3 years (SD 4.5). In total, 430 participants had a primary TKR, and 675 had a THR, due to primary osteoarthritis.
Women who reported age at menarche of ≤11 years were older at baseline than those who reported menarche at ≥15 years (Table I). Never users of OC were older than ever users, and past or current users of HRT were older than never users. BMI slightly decreased with increasing age at menarche. A lower portion of the women with higher age at menarche smoked. There was a higher prevalence of diabetes in women who were never oral contraceptive users. Hard physical activity was more prevalent in premenopausal women and oral contraceptive users. Women that received a TKR or THR during follow-up were older, and there were a higher percentage of past or current HRT users than among those who did not get a joint replacement (Table II).
Table IStudy population characteristics at baseline (OC: Oral contraceptives; HRT: Hormone replacement therapy)
Table IIStudy population characteristics at baseline in all women, and those who received a total knee replacement (TKR) or total hip replacement (THR)
Increasing age at menarche was inversely associated with the risk of TKR (P-trend < 0.001) (Table III). Compared to women with early menarche, those with menarche at 14 years and ≥15 years had a significantly lower risk of TKR (HR 0.64, 95% confidence interval (CI) 0.43–0.95; and HR 0.52, 95% CI 0.34–0.80; respectively). The number of years of menstruation between menarche and menopause was not associated with TKR. Past users of HRT were at higher risk of TKR compared to never users (HR 1.42, 95% CI 1.06–1.90), but only those who used systemic HRT compared to local treatment (HR 1.40, 95% CI 1.03–1.90). Ever users of OC had a higher risk of TKR (HR 1.38, 95% CI 1.03–1.84), but this association was only borderline significant in the fully adjusted model (HR 1.36, 95% CI 1.00–1.86).
Table IIIReproductive history and use of hormonal medication, and risk of total knee replacement (TKR) (OC: Oral contraceptives; HRT: Hormone replacement therapy)
Adjusted for age, BMI, smoking and physical activity in all analyses. Additional adjustment for diabetes, parity, menarche, menopausal status, oral contraceptives and hormone replacement therapy as appropriate in each DAG analysis.
HR
95% CI
HR
95% CI
Parity
Nulliparous
1168
12,610
17
1
Ref
1
Ref
1
2966
26,173
27
0.79
(0.43–1.45)
0.57
(0.28–1.19)
2
10,649
86,907
127
1.20
(0.72–2.00)
0.91
(0.50–1.67)
3
8818
71,050
134
1.28
(0.77–2.12)
0.88
(0.48–1.61)
≥4
5322
44,731
111
1.04
(0.62–1.73)
0.7
(0.38–1.30)
P linear trend
0.55
0.55
Age at menarche, years
≤11
2683
21,502
48
1
Ref
1
Ref
12
5502
45,203
83
0.81
(0.56–1.15)
0.83
(0.56–1.23)
13
7554
62,328
105
0.77
(0.55–1.09)
0.70
(0.48–1.03)
14
7227
59,988
104
0.68
(0.48–0.96)
0.64
(0.43–0.95)
≥15
6129
50,698
78
0.58
(0.40–0.84)
0.52
(0.34–0.80)
P linear trend
0.002
0.001
Menopausal status
Pre/peri
10,336
91,247
46
1
Ref
1
Ref
Post
14,922
112,668
341
0.95
(0.63–1.44)
1.16
(0.72–1.87)
Age at menopause, years
≤48
4815
37,504
111
1
Ref
1
Ref
49–51
5090
38,187
105
0.92
(0.70–1.20)
0.88
(0.64–1.21)
≥52
5017
36,976
125
0.99
(0.77–1.29)
0.98
(0.73–1.32)
Years of menstruation
14,386
250,254
430
1.01
(0.99–1.04)
1.02
(0.99–1.04)
OC use
Never
6202
56,109
130
1
Ref
1
Ref
Ever
11,924
96,117
103
1.37
(1.03–1.84)
1.36
(1.00–1.86)
Years of OC use
11,488
90,646
94
0.99
(0.93–1.06)
1.01
(0.95–1.09)
HRT use
Never
22,203
175,094
243
1
Ref
1
Ref
Past
2536
18,035
69
1.45
(1.10–1.90)
1.42
(1.06–1.90)
Current
2003
16,964
57
1.36
(1.02–1.82)
1.25
(0.90–1.73)
HRT use by site
Never
22,203
175,094
243
1
Ref
1
Ref
Local
2197
16,539
62
1.33
(1.00–1.76)
1.23
(0.90–1.68)
Systemic
2342
18,460
64
1.49
(1.13–1.98)
1.40
(1.03–1.90)
Years of HRT use
3370
22,306
99
1.02
(0.99–1.05)
1.03
(1.00–1.06)
∗ Adjusted for age, BMI and smoking.
† Adjusted for age, BMI, smoking and physical activity in all analyses. Additional adjustment for diabetes, parity, menarche, menopausal status, oral contraceptives and hormone replacement therapy as appropriate in each DAG analysis.
No association was found between parity, age at menarche, postmenopausal status or oral contraceptive use and THR (Table IV). Current HRT users had increased risk of THR after adjustment for age, BMI and smoking, but this association was no longer significant in the fully adjusted model. There was, however, an increased risk of THR associated with years of HRT use (HR 1.04, 95% CI 1.01–1.07). The vast majority of past/current HRT users were postmenopausal women (n = 4046), compared to pre/perimenopausal women (n = 329) (data not shown).
Table IVReproductive history and use of hormonal medication, and risk of total hip replacement (THR) (OC: Oral contraceptives; HRT: Hormone replacement therapy).
Adjusted for age, BMI, smoking and physical activity in all analyses. Additional adjustment for diabetes, parity, menarche, menopausal status, oral contraceptives and hormone replacement therapy as appropriate in each DAG analysis.
Adjusted for age, BMI, smoking and physical activity in all analyses. Additional adjustment for diabetes, parity, menarche, menopausal status, oral contraceptives and hormone replacement therapy as appropriate in each DAG analysis.
95% CI
Parity
Nulliparous
1168
12,610
25
1
Ref
1
Ref
1
2966
26,173
55
1.17
(0.73–1.88)
1.12
(0.58–2.18)
2
10,649
86,907
209
1.43
(0.94–2.17)
1.56
(0.86–2.81)
3
8818
71,050
186
1.30
(0.85–1.97)
1.42
(0.79–2.57)
≥4
5322
44,731
184
1.29
(0.85–1.97)
1.34
(0.74–2.44)
P linear trend
0.57
0.61
Age at menarche, years
≤11
2683
21,502
49
1
Ref
1
Ref
12
5502
45,203
118
1.04
(0.75–1.45)
1.15
(0.78–1.71)
13
7554
62,328
155
0.96
(0.70–1.33)
1.00
(0.68–1.47)
14
7227
59,988
177
0.93
(0.67–1.28)
1.04
(0.71–1.53)
≥15
6129
50,698
163
0.92
(0.67–1.28)
1.07
(0.73–1.58)
P linear trend
0.352
0.968
Menopausal status
Pre/peri
10,336
91,247
65
1
Ref
1
Ref
Post
14,922
112,668
515
0.99
(0.70–1.41)
0.97
(0.67–1.40)
Age at menopause, years
≤48
4815
37,504
156
1
Ref
1
Ref
49–51
5090
38,187
185
1.13
(0.92–1.40)
1.14
(0.89–1.45)
≥52
5017
36,976
174
1.04
(0.83–1.29)
1.03
(0.80–1.32)
Years of menstruation
14,386
236,732
667
1.01
(0.99–1.03)
1
(0.98–1.03)
OC use
Never
6202
56,109
193
1
Ref
1
Ref
Ever
11,924
96,117
133
1.11
(0.87–1.42)
1.03
(0.79–1.35)
Years of OC use
11,488
90,646
120
0.94
(0.87–1.01)
0.96
(0.89–1.04)
HRT use
Never
22,203
175,094
391
1
Ref
1
Ref
Past
2536
18,035
88
1.12
(0.88–1.41)
1.03
(0.80–1.33)
Current
2003
16,964
95
1.32
(1.05–1.66)
1.19
(0.92–1.53)
HRT use by site
Never
22,203
175,094
391
1
Ref
1
Ref
Local
2197
16,539
100
1.26
(1.01–1.58)
1.16
(0.90–1.48)
Systemic
2342
18,460
83
1.16
(0.91–1.48)
1.05
(0.80–1.36)
Years of HRT use
3370
22,306
116
1.04
(1.01–1.07)
1.04
(1.01–1.07)
∗ Adjusted for age, BMI and smoking.
† Adjusted for age, BMI, smoking and physical activity in all analyses. Additional adjustment for diabetes, parity, menarche, menopausal status, oral contraceptives and hormone replacement therapy as appropriate in each DAG analysis.
In a sensitivity analysis of 8745 participants from HUNT2 on parity, oral contraceptive use, and HRT use, adjusted for education level, we found a reduced risk of TKR in women reporting 1 birth (HR 0.15, 95% CI 0.09–0.78) or ≥ 4 births (HR 0.18, 95% CI 0.22–0.97) compared to nulliparous women, but there was no significant trend across the categories (P = 0.37) (Table B, Appendix). Years of HRT use slightly increased the risk of THR, but past or current use of HRT was not associated with THR (Table C, Appendix).
Discussion
This prospective cohort study of over 30,000 women found that older age at menarche was associated with decreased risk of TKR. We also found an association between past and systemic HRT use and increased risk of TKR. Parity did not increase the risk of TKR or THR.
The observation that increasing age at menarche was inversely related to the risk of TKR has also been reported in a large prospective study of 1.3 million middle-aged women by Liu et al.
. The mechanisms underlying these associations are unclear, but there could be several possible explanatory factors. A recent cross-sectional study found an association between early age at menarche and chronic widespread musculoskeletal complaints later in life
. One may therefore speculate that an increased level of pain from knee osteoarthritis in this group could lead to a higher incidence of TKR. Early onset of menarche has also been linked to other conditions of ageing such as elevated blood pressure and glucose intolerance, independent of body composition
. A cross-sectional study by Kalichman et al. demonstrated a negative association between age at menarche and radiological hand osteoarthritis. They proposed that one possible explanation could be that early menarche was associated with an increased rate of the general ageing process
Systemic use of HRT increased the risk of TKR, and although we did not find any association between current use of HRT and joint replacement, our finding of increased risk of TKR in women with past use of HRT is in agreement with the results by Liu et al.
. They reported that past or current use of postmenopausal hormone therapy was associated with a significant increase in the incidence of THR and TKR. However, clinical and epidemiological studies have shown conflicting results, and a systematic review found no clear association between HRT and osteoarthritis
. Heterogeneity between the hormones used and outcome measurements also made statistical data pooling impossible. They concluded that the relationship was, perhaps, too complex, or that other factors play a role in the increased incidence of osteoarthritis in women aged >50.
Our study did not observe any association between parity and joint replacement. Previous studies on the association between parity and knee osteoarthritis have shown conflicting results
. However, the absolute numbers of joint replacements in the nulliparous group in our study were low (n = 25 and n = 17 for THR and TKR, respectively), which calls the power of this analysis into question. We cannot exclude the possibility that this may have weakened any association. Since both parity and joint replacement are associated with education level, we did a sub-analysis with additional adjustment for education in 8745 participants with data on education level; we revealed a reduced risk of TKR in women with 1 birth or ≥4 births, but there was no significant trend across the categories (P = 0.37). This could indicate a complex relationship between parity and TKR/THR that we were unable to clarify further in our study.
The healthcare system in Norway is publicly funded and free of charge for patients. Although socioeconomic status would not affect access to surgery, it could lead to a difference in those seeking surgery. In 2009, Statistics Norway reported that amongst women with musculoskeletal diseases, those with a higher level of education (university/college level) were more likely to contact specialist health services than those with lower levels of education (high school or lower)
Ever use of OC did not significantly increase the risk of TKR or THR in the fully adjusted model, although the point estimate of the P value was borderline significant for TKR, P = 0.053 (HR 1.36, 95% CI 1.00–1.86). Menopausal status and age at menopause were not associated with THR or TKR.
Strengths and limitations
Major strengths of this study were the large sample size, prospective population-based design, objective measurements and nearly complete registration of TKR and THR.
Our study used objective measurements of height and weight by trained personnel, and thus avoided potential information bias. The study by Liu et al.
used self-reported BMI. Self-reported BMI may be biased, and a recent study showed limited agreement with actual height and weight in overweight and obese individuals with clinical osteoarthritis
Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results from the Women's Health Initiative randomized controlled trial.
. The proportion of women using HRT could therefore have been lower in the HUNT3 study. In our data-set we found that 19% of participants in HUNT2 were past or current HRT users, compared to 16.3% in HUNT3. Therefore, HRT prescription did not differ substantially between the two surveys, and should not have greatly affected our results.
The design of this study is prospective since the baseline information was recorded prior to an eventual joint replacement. However, we cannot exclude the possibility that recall bias might have influenced some of the covariates, especially age at menarche. Table I shows a mean age difference of almost 10 years between the women that reported age at menarche ≤11 years, and those reporting age at menarche ≥15. As well, a Danish study from 2009 showed significantly earlier breast development among girls born more recently during a 15-year period
. This could indicate that the age at menarche may have decreased over time in our study population, thus creating a cohort effect. Adjusting for age may then be insufficient for correcting an eventual systematic information bias and a cohort effect bias.
At baseline, the mean age of our study population was 55.7, and 62.1% of the women in our study were postmenopausal. However, since our lower cut-off for age at inclusion was 30 years, the information on reproductive history and use of HRT or OC could have changed for some participants after baseline. This is especially relevant when it comes to parity, oral contraceptive use and HRT, and could have led to non-differential misclassification and thus weakened any associations. To increase the information on lifetime reproductive history and eventual use of HRT, we chose to use baseline measurements from HUNT3 for those that participated in both HUNT2 and HUNT3, even if this reduced follow-up time after baseline; the 9468 participants with baseline measurements from HUNT2 had a mean follow-up time of 13.0 years compared to 6.1 years for the 20,821 participants with baseline measurement from HUNT3. Lower incidences of TKR (1.2 %) and THR (1.8%) in the HUNT3 group, compared to TKR (1.8%) and THR (3.2%) in the HUNT2 group, might contribute to lower precision and underestimation of any associations.
A previous study from the HUNT2 material reported that women who had undergone unilateral oophorectomy entered menopause around 1 year earlier than women with two intact ovaries
(Separation between uni- vs bilateral oophorectomy was only available from HUNT2, as the HUNT3 questionnaire only asked about bilateral oophorectomy). We chose not to exclude participants that had had only one ovary surgically removed (n = 776), and additional adjustment for unilateral oophorectomy when analysing age at menarche did not change the results (data not shown).
In HUNT2 we had information on type of HRT medication in 2601 participants. Of these participants, 1456 (56%) used a combination of oestrogen and progesterone, and 1145 used oestrogen without progesterone. HUNT3 did not have information about the precise type of HRT used by each individual. A previous publication on HRT from HUNT3 reported that data from the Norwegian Prescription Database showed that during the time frame and region of the HUNT3 study, 83.5% of HRT users were prescribed a combination of oestradiol and/or oestriol and progesterone, 9.0% either oestradiol or oestriol without progesterone and 7.5% used the synthetic oestrogen tibolone
Although there was a 10-year period between HUNT2 and HUNT3, they both used the same source population: All inhabitants ≥20 years of age in the county of Nord-Trøndelag in Norway. But there could be several reasons why HUNT2 and HUNT3 did not have all the same participants:
-
The participation rate in HUNT3 was lower than in HUNT2 (58.7% and 75.5%, respectively). Some of the responders in HUNT2 could therefore have been non-responders in HUNT3.
-
We would expect some of the older participants in HUNT2 to have died before HUNT3. And people that were too young to participate in HUNT2 could be part of the study population in HUNT3.
The population in Nord-Trøndelag is relatively homogeneous, with less than 3% non-Caucasian, and is relatively stable, with few people moving in or out of the county
. So despite the limitations that arise from using the participants from two consecutive waves of the HUNT health survey, we would argue that the two surveys represent one source population.
The osteoarthrosis diagnoses from the NAR have not been validated
Total hip replacement in young adults with hip dysplasia: age at diagnosis, previous treatment, quality of life, and validation of diagnoses reported to the Norwegian Arthroplasty Register between 1987 and 2007.
Registration in the Danish hip arthroplasty registry: completeness of total hip arthroplasties and positive predictive value of registered diagnosis and postoperative complications.
. However, the operating surgeon reports whether each joint replacement is due to primary/idiopathic osteoarthritis, or due to other specified causes. We only included joint replacement due to primary/idiopathic osteoarthritis.
We used joint replacement as an indicator of severe osteoarthritis. Joint replacement is the most definitive treatment for osteoarthritis in the hip or knee, and has the advantage of being a strong indicator of severe clinical disease compared to other definitions of osteoarthritis
. The decision to do a total arthroplasty does, however, rely on several factors: the severity of pain, radiographic findings, comorbidities and the patient's motivation for undergoing surgery. Subjects who wish to maintain an active lifestyle may be more motivated to have surgery than less active persons
, even if they have less severe osteoarthritis. This potential healthy patient bias could lead to an underestimation of the effect of reproductive and hormonal therapies on osteoarthritis.
We found that increasing age at menarche reduced the risk of TKR. Past users and users of systemic HRT were at higher risk of TKR compared to never users. Parity did not increase the risk of TKR or THR.
Contributors
AIH participated in the study concept and design, obtained funding, performed the analysis, interpreted the data and drafted the manuscript. LN, MBJ, AL, GBF, OF, KS and JAZ were involved in the conception and design of the study. OF was also involved in the collection of THR and TKR data. AMF contributed with statistical expertise. All the authors revised the manuscript for important intellectual content and approved the final version of the manuscript.
Conflict of interest
None.
Funding
This study was supported by research grants from the Liaison Committee between the Central Norway Regional Health Authority (RHA) and the Norwegian University of Science and Technology (NTNU) (2014/23166), the Dr. Egil Kjeldaas Foundation (Lege Egil Kjeldaas legat), the Norwegian Orthopaedic Association and Levanger Hospital, Nord-Trøndelag Hospital Trust. The funding sources had no involvement in the study design, the collection, analysis or interpretation of data, writing the paper or the decision to submit the paper for publication.
Acknowledgements
The Nord-Trøndelag Health Study is a collaborative effort of the Faculty of Medicine at the Norwegian University of Science and Technology, the Norwegian Institute of Public Health and the Nord-Trøndelag County Council. The Norwegian Arthroplasty Register is owned by the Norwegian Orthopedic Association and administered by the Orthopedic Department at Haukeland University Hospital, Bergen, Norway.
Appendix.
Table ACovariates adjusted for in Model 1 and Model 2 (BMI: Body Mass Index; HRT: Hormone replacement therapy)
Exposure variable
Covariates adjusted for in Model 1
Additional covariates adjusted for in Model 2
Parity
Age, BMI, smoking
Diabetes, physical activity, age at menarche, menopausal status, HRT
Age at menarche
Age, BMI, smoking
Diabetes, physical activity, parity, menopausal status
Years of menstruation
Age, BMI, smoking
Diabetes, physical activity, parity
Menopausal status
Age, BMI, smoking
Physical activity, parity, age at menarche, HRT
Age at menopause
Age, BMI, smoking
Physical activity, parity, age at menarche, HRT
Use of oral contraceptives
Age, BMI, smoking
Physical activity, parity, age at menarche, menopausal status
Table BParity, oral contraceptives (OC), hormone replacement therapy (HRT) and risk of total knee replacement (TKR); sensitivity analysis with additional adjustment for education
Adjusted for age, BMI, smoking, physical activity and education level in all analyses. Additional adjustment for diabetes, parity, menarche, menopausal status, oral contraceptives and hormone replacement therapy as appropriate in each DAG analysis.
95% CI
Parity
Nulliparous
797
9743
14
1
Ref
1
Ref
1
961
13,486
10
0.52
(0.23–1.18)
0.15
(0.09–0.78)
2
2602
36,701
40
0.83
(0.45–1.53)
0.20
(0.27–1.12)
3
2131
29,165
40
0.87
(0.47–1.60)
0.20
(0.26–1.11)
≥4
1913
22,874
46
0.77
(0.42–1.41)
0.18
(0.22–0.97)
P linear trend
0.97
0.37
OC use
Never
2119
30,002
59
1
Ref
1
Ref
Ever
2176
35,524
21
0.92
(0.52–1.62)
0.95
(0.50–1.78)
Years of OC use
1883
30,955
16
1.04
(0.95–1.13)
1.05
(0.95–1.16)
HRT use
Never
4897
67,037
81
1
Ref
1
Ref
Past
414
5172
10
1.05
(0.54–2.03)
1.06
(0.50–2.23)
Current
704
8760
22
1.38
(0.86–2.24)
1.36
(0.79–2.36)
HRT use by site
Never
4897
67,037
81
1
Ref
1
Ref
Local
544
6269
20
1.50
(0.91–2.48)
1.56
(0.88–2.76)
Systemic
574
7662
12
0.99
(0.54–1.83)
0.93
(0.46–1.89)
Years of HRT use
319
3740
16
0.96
(0.81–1.14)
0.96
(0.80–1.14)
∗ Adjusted for age, BMI, smoking and education level.
† Adjusted for age, BMI, smoking, physical activity and education level in all analyses. Additional adjustment for diabetes, parity, menarche, menopausal status, oral contraceptives and hormone replacement therapy as appropriate in each DAG analysis.
Table CParity, oral contraceptives (OC), hormone replacement therapy (HRT) and risk of total hip replacement (THR); sensitivity analysis with additional adjustment for education
Adjusted for age, BMI, smoking, physical activity and education level in all analyses. Additional adjustment for diabetes, parity, menarche, menopausal status, oral contraceptives and hormone replacement therapy as appropriate in each DAG analysis.
95% CI
Parity
Nulliparous
797
9743
21
1
Ref
1
Ref
1
961
13,486
21
0.83
(0.45–1.53)
1.02
(0.44–2.33)
2
2602
36,701
77
1.16
(0.72–1.89)
1.53
(0.77–3.02)
3
2131
29,165
61
0.97
(0.59–1.59)
1.17
(0.58–2.36)
≥4
1913
22,874
81
1.03
(0.64–1.67)
1.1
(0.55–2.23)
P linear trend
0.87
0.88
OC use
Never
2119
30,002
91
1
Ref
1
Ref
Ever
2176
35,524
33
0.97
(0.62–1.52)
1.01
(0.63–1.62)
Years of OC use
1883
30,955
25
0.96
(0.87–1.05)
0.95
(0.86–1.04)
HRT use
Never
4897
67,037
144
1
Ref
1
Ref
Past
414
5172
14
0.77
(0.45–1.34)
0.85
(0.47–1.51)
Current
704
8760
35
1.04
(0.72–1.52)
1.03
(0.68–1.57)
HRT use by site
Never
4897
67,037
144
1
Ref
1
Ref
Local
544
6269
26
0.98
(0.64–1.49)
1.07
(0.67–1.71)
Systemic
574
7662
23
0.92
(0.59–1.43)
0.86
(0.53–1.41)
Years of HRT use
319
3740
17
1.13
(1.03–1.25)
1.18
(1.05–1.33)
∗ Adjusted for age, BMI, smoking and education level.
† Adjusted for age, BMI, smoking, physical activity and education level in all analyses. Additional adjustment for diabetes, parity, menarche, menopausal status, oral contraceptives and hormone replacement therapy as appropriate in each DAG analysis.
The effect of estrogen plus progestin on knee symptoms and related disability in postmenopausal women: the Heart and Estrogen/Progestin Replacement Study, a randomized, double-blind, placebo-controlled trial.
Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results from the Women's Health Initiative randomized controlled trial.
Total hip replacement in young adults with hip dysplasia: age at diagnosis, previous treatment, quality of life, and validation of diagnoses reported to the Norwegian Arthroplasty Register between 1987 and 2007.
Registration in the Danish hip arthroplasty registry: completeness of total hip arthroplasties and positive predictive value of registered diagnosis and postoperative complications.
The authors have found an error in Table 4 of the above published article where the data for Model 2 of Parity were listed incorrectly. The correct numbers are listed in Table 4 below under Parity and Model 2.
31060-9&id=gr1.jpg){kind=link}