Nulliparity and Osteoporotic Fracture Risk

Abstract

We thank Drs Robbins, Schott, and Meunier for validating our results that nulliparity increases the risk of hip fracture in older women (1) among their large EPIDOS cohort. The importance of replication among different study settings is a critical piece for scientific proof of causality of any study's initial findings which we endorse.We now have several additional years of follow-up of the Study of Osteoporotic Fractures cohort (SOF), which now makes the age of our cohort more closely comparable with what Robbins et al. presented with EPIDOS. We have done additional analyses to allow better comparison between the two cohorts. Among our 9704 SOF women, with now an average of 12.8 years of follow-up available for analysis, there were 926 fractures. The average age of the SOF cohort was 71.7 years at the baseline exam, and thus the average age of survivors after 13 years of follow-up is essentially identical to Drs Robbins, Schott, and Meunier's findings from EPIDOS. Furthermore, both cohorts are of primarily Northern European ancestry-the highest risk group for osteoporosis-although environmental and/or lifestyle differences may be likely between their cohort and ours.We first did age-and weight-adjusted Cox proportional hazards models to mirror the EPIDOS analysis, and nulliparous women in SOF had an increased risk of hip fracture (HR 1.20; 95% CI, 1.03-1.41). As the issue raised by Robbins et al. is the contribution of bone mineral density (BMD) to this risk of hip fracture among nulliparous women, we next limited our analyses to the 7928 SOF women who also had BMD measured at our second visit in 1989 -1990. Among this group, there were 754 hip fractures-the strong relationship of nulliparity to increase hip fracture remained significant after adjustment for BMD (HR 1.26; 95% CI, 1.06 -1.49). Further multivariate adjustment for covariates used in our original paper (1) with BMD was similar (HR 1.31; 95% CI, 1.10 -1.55).We would like to note that when we initially tested for a relationship between parity and hip fracture with BMD in SOF, this relationship was borderline significant after multivariate adjustment (mean duration of follow-up at that time was 4.1 years).(2) However, with longer follow-up and additional hip fracture cases, the effect of nulliparity on increasing hip fracture risk became evident. We would argue that the effect size did not change appreciably in the EPIDOS study with the addition of BMD (HR decreased from 1.38 to 1.31), and that it is very possible with additional follow-up that the EPIDOS study will also find this increased risk is significant after adjustment for BMD.We completely agree with Drs Robbins, Schott, and Meunier that the clinically important time for hip fractures is in older women of ages similar to the women in the SOF and EPIDOS cohorts. However, we respectfully disagree that clinicians should consider only BMD when assessing a women's risk of hip fracture-our results would suggest increasing parity operates through a different mechanism than is detected by BMD to incr...

Female reproductive history and the skeleton—a review

“…So, having many children will probably lead to not only a high BMD but also reduced fracture risk. This view is further supported when data infer that the association between parity and reduced hip fracture risk may operate through a different mechanism than is detected by BMD 66,67 . The BMD loss found to be associated with a pregnancy and lactation must be regarded as a finding without clinical relevance, as the deficit recovers after weaning.…”