Life course longitudinal growth and risk of knee osteoarthritis at age 53 years: evidence from the 1946 British birth cohort study

Objectives To examine the relationship between height gain across childhood and adolescence with knee osteoarthritis in the MRC National Survey of Health and Development (NSHD). Methods Data are from 3035 male and female participants of the NSHD. Height was measured at ages 2, 4, 6, 7, 11 and 15 years, and self-reported at ages 20 and 26 years. Associations between (i) height at each age (ii) height gain during specific life periods (iii) Super-Imposition by Translation And Rotation (SITAR) growth curve variables of height size, tempo and velocity, and knee osteoarthritis at 53 years were tested. Results In sex-adjusted models, taller height at 4 and 6 years were modestly associated with decreased odds of knee osteoarthritis at age 53 (ORs per 1cm increase in height at age 6: 0.97 and age 4: 0.98 (95% CI: 0.95-1.00)). These associations were attenuated after adjustment for potential confounders. Similarly, taller adult achieved height measured at 26 and 53 years of age were associated with decreased odds of knee osteoarthritis (OR per 1cm increase in height: 0.98 (95% CI 0.96 to 1.00)). No associations were found between height gain during specific life periods or the SITAR growth curve variables and odds of knee osteoarthritis. Conclusions There was some limited evidence to suggest that taller height in childhood is associated with decreased odds of knee osteoarthritis at age 53 years in this cohort. This work enhances our understanding of osteoarthritis predisposition and the contribution of life course height to this.


Abstract 23
Objectives 24 To examine the relationship between height gain across childhood and adolescence with knee 25 osteoarthritis in the MRC National Survey of Health and Development (NSHD). 26 Methods 27 Data are from 3035 male and female participants of the NSHD. Height was measured at ages 28 2, 4, 6, 7, 11 and 15 years, and self-reported at ages 20 and 26 years. Associations between (i) 29 height at each age (ii) height gain during specific life periods (iii) Super-Imposition by 30 Translation And Rotation (SITAR) growth curve variables of height size, tempo and velocity, 31 and knee osteoarthritis at 53 years were tested. 32

33
In sex-adjusted models, taller height at 4 and 6 years were modestly associated with decreased 34 odds of knee osteoarthritis at age 53 (ORs per 1cm increase in height at age 6: 0.97 and age 4: 35 0.98 (95% CI: 0.95-1.00)). These associations were attenuated after adjustment for potential 36 confounders. Similarly, taller adult achieved height measured at 26 and 53 years of age were 37 associated with decreased odds of knee osteoarthritis (OR per 1cm increase in height: 0.98 38 (95% CI 0.96 to 1.00)). No associations were found between height gain during specific life 39 periods or the SITAR growth curve variables and odds of knee osteoarthritis. 40 Conclusions 41 There was some limited evidence to suggest that taller height in childhood is associated with Introduction 46 Joint health is reliant upon the preservation of the articular cartilage and, its degradation is one of the main hallmarks of the degenerative joint disease osteoarthritis. Osteoarthritis, 48 characterised by articular cartilage loss, subchondral bone thickening and osteophyte 49 formation, is a major health care burden throughout the world. It is estimated that worldwide 50 at least 10% of men and 18% of women aged over 60 years have symptomatic osteoarthritis. 51 Osteoarthritis causes much pain and disability, and yet its underlying molecular mechanisms 52 are not fully understood. Indeed, even the precipitating pathology remains a matter of debate 53 and we are still unable to identify those at most risk of developing the disease. 54 Our previous work in a spontaneous murine model of ageing-related osteoarthritis, the STR/Ort 55 mouse, revealed accelerated long bone growth, increased growth plate chondrocyte 56  preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 21, 2020. . https://doi.org/10.1101/2020.08.18.20177485 doi: medRxiv preprint Herein, we use one of these studies, the MRC NSHD, to examine the relationship between 71 childhood and adolescent height growth and knee osteoarthritis at 53 years. Our aims were to: 72 (1) test associations between height at different ages in early life and knee osteoarthritis in 73 adulthood; (2) assess how patterns of height growth during childhood and adolescence are 74 associated with knee osteoarthritis. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 21, 2020. . https://doi.org/10.1101/2020.08.18.20177485 doi: medRxiv preprint Height was measured by nurses using standardised protocols at ages 2, 4, 7, 11 and 15 years, Participation in leisure-time physical activity at age 53 was ascertained by asking participants 115 to report whether they had undertaken any sports, vigorous leisure activities, or exercises in 116 their spare time, not including getting to and from work, in the last 4 weeks prior to the 117 examination and if so on how many occasions they had done these activities. This was 118 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 21, 2020. . https://doi.org/10.1101/2020.08.18.20177485 doi: medRxiv preprint categorised into three groups: inactive (no participation); moderately active (participated 1 to 119 4 times); most active (participated ≥5 times). 120 Statistical analysis 121 To address the two main aims, we used logistic regression models to test associations between:

122
(1) height at each age (aim 1); (2) conditional changes in height during specific life periods 123 (early childhood: 2-4 years; late childhood: 4-7 years; childhood to adolescence: 7-15 years; 124 adolescence to young adulthood: 15-20 years) (aim 2) and; (3) each SITAR height variable 125 (aim 2) and odds ratios (ORs) of knee osteoarthritis. In initial models, we formally tested for 126 interactions between sex and each main independent variable and where no evidence of 127 interaction was found, models were fitted with men and women combined and adjustment for 128 sex. In each set of models we first adjusted for sex (where there was no evidence of interaction) 129 before then also adjusting for early and adult life factors (sporting ability + physical activity + 130 occupational class). In models to address aim 2, we generated conditional changes in height by 131 regressing each height measure on the earlier height measure for each sex and calculating the 132 residuals.
[13] The residuals were standardized (to have mean 0 and SD of 1) to ensure their 133 comparability and these were included as the main independent variables. To maximise 134 statistical power each set of models were run on the sample with valid data for the outcome, 135 the specified independent variable and covariates. Data were analysed using Stata statistical 136 software (version SE 14.2). 137 Sensitivity analyses 138 In additional models we adjusted for adult weight at 53 years, considered a potential mediator. 139 We also adjusted for weight at each age for aim 1, conditional weight gain (aim 2) and the 140 SITAR weight variables (aim 2) to assess the contribution of weight during growth. To assess 141 the potential impact of having to exclude those participants lost to follow-up and with missing 142 data, comparisons were drawn between those included and those excluded in the main analyses. 143 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in Life course height and knee osteoarthritis 158 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 21, 2020. . https://doi.org/10.1101/2020.08.18.20177485 doi: medRxiv preprint In sex-adjusted models, taller height at age 6 (OR of knee osteoarthritis 0.97 per 1cm increase 159 in height, 95% CI 0.95 to 1.00) was modestly associated with lower odds of knee osteoarthritis 160 at age 53 years (Model 1; Table 2). There was also weak evidence of an association at age 4 161 (OR of knee osteoarthritis 0.98 per 1cm increase in height, 95% CI 0.95 to 1.00) ( Table 2).   Height growth and knee osteoarthritis 174 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 21, 2020. . https://doi.org/10.1101/2020.08.18.20177485 doi: medRxiv preprint No associations were found between height gains during any of the four periods assessed and 175 odds of knee osteoarthritis at 53 years (Table 3). There was also no evidence of associations 176 between height size, tempo or velocity (SITAR variables) and knee osteoarthritis at 53 years 177 (Model 1;   Comparison of the characteristics of those individuals with complete data, vs those excluded 189 are described in Tables S1.1 & S1.2. We found that higher proportions of those included were 190 female (50.7% vs 43.7%; p<0.001; Tables S1.1 & S1.2). No significant differences were 191 observed in height between ages 2 -15 years but at age 20, those included reported shorter 192 heights than those excluded (171.0 cm vs 169.4 cm; Tables S1.1 & S1.2). 193 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 21, 2020.  (Table S3.3). 203 When models were rerun on the maximum available samples including all available 204 participants (Tables S43.1 -S4.3), there were no substantive differences in findings. 205 When we excluded those participants with potential secondary knee osteoarthritis from our 206 analyses, there were no substantive differences in associations between height (Table S5.1) or 207 conditional height gain (Table S5.2) and primary knee osteoarthritis at 53 years. A weak 208 association was observed between SITAR height size when adjusted for sex (OR of knee 209 osteoarthritis 0.97 per 1cm increase in height, 95% CI 0.95 to 1.00) (Model 1; Table S5.3). 210 This was however attenuated when adjusted for early and adult life factors (Model 2; Table   211 S5.3).

213
In this nationally representative British birth cohort study, associations between greater height 214 at ages 4 and 6 years and marginally lower odds knee osteoarthritis at age 53 were observed in 215 sex-adjusted models, but these were attenuated after adjustment for early and adult life factors. 216 No associations were observed between height changes during early childhood, late childhood, 217 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 21, 2020. . https://doi.org/10.1101/2020.08.18.20177485 doi: medRxiv preprint childhood to adolescence or adolescence to young adulthood or SITAR parameters and knee 218 osteoarthritis. 219 A major strength of our study is the availability of multiple prospectively ascertained perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 21, 2020. . https://doi.org/10.1101/2020.08.18.20177485 doi: medRxiv preprint the cohort are still relatively young, and osteoarthritis prevalence (7.31% in men; 13.06% in 243 women) is lower than that seen currently in primary care at this age. It would therefore be of 244 interest to further examine these potential associations in older individuals. 245 Primary osteoarthritis is described as naturally occurring or ageing-related osteoarthritis, while 246 secondary osteoarthritis is associated with other causes including trauma. Our previous 247 findings in the STR/Ort mouse examined primary murine osteoarthritis [1] and therefore to 248 examine the influence of secondary knee osteoarthritis on the patterns of height growth in the 249 NSHD, we ran a sensitivity analysis in which we excluded individuals who had reported 250 consulting a Doctor about a knee injury. However, whilst we found no substantive differences 251 in findings, this highlights the need to examine the risk of osteoarthritis in aged individuals 252 where primary knee osteoarthritis is more prevalent. 253 Our study extends a previous study examining this British birth cohort in which prolonged 254 exposure to high BMI through adulthood increased risk of development of knee osteoarthritis 255 at age 53.
[3] This is consistent with our sensitivity analyses in which adjustment for weight 256 strengthened the associations between SITAR height size and odds of knee osteoarthritis. Wills 257 et al., also found that BMI increases from childhood to adolescence (7-15 years) were 258 positively associated with knee osteoarthritis, however this was in women only.
[3] In our 259 analyses, we found no evidence of differences in association by sex. We did find that in our 260 cohort with complete data, women had a higher prevalence of knee osteoarthritis, similar to 261 that reported previously in the NSHD, and in primary care. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in were not observed in this study. 274 Together, in this relatively large population-based cohort study, there was limited evidence to 275 suggest that height in childhood is associated with odds of knee osteoarthritis at age 53 years. 276 However, there were no associations with height gain during specific periods of growth, or perpetuity.
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 21, 2020. . https://doi.org/10.1101/2020.08.18.20177485 doi: medRxiv preprint