The objective of this study was to determine the effect of oral contra
ceptive pills on bone turnover. The design consisted of a cross-sectio
nal analysis of a prospective cohort. There were 52 women taking oral
contraceptives and 156 nonuser controls from a large cohort of 1039 he
althy women, aged 31-89 years (OFELY study). Most users were taking co
mbined oral contraceptives containing 30 mu g ethinyl estradiol and th
e mean duration of pill use was 6.7 +/- 6.4 years, Users and nonusers
were matched for age [mean age (years): 39.3 +/- 3.5 vs. 40.5 +/- 4.3,
range 35-49 years for both]. Main outcome measures included three mar
kers of bone formation (serum osteocalcin, bone-specific alkaline phos
phatase, and C-terminal propeptide of type I collagen) and two markers
of bone resorption that are pyridinoline crosslinked peptides (Crossl
aps(TM) and NTX), Users and nonusers did not differ for weight, height
, alcohol and tobacco use, dietary calcium intake, parity, exercise ac
tivity, body fat and lean composition, and calcium chemistry tests. In
pill users all bone formation and resorption markers were decreased c
ompared with controls: osteocalcin, 7.7 +/- 2.7 vs. 10.1 +/- 3.1 ng/mL
(-24%, p < 0.001); bone-specific alkaline phosphatase, 7.5 +/- 2.3 vs
. 8.8 +/- 2.7 ng/mL (-15%, p < 0.003); C-terminal propeptide of type I
collagen, 77.2 +/- 93.1 vs. 93.1 +/- 31.9 ng/mL (-17%, p = 0.001); Cr
osslaps(TM): 175 +/- 91 vs. 211 +/- 105 mu g/mmol Cr (-17%, p = 0.03);
and NTX, 16.2 +/- 5.8 vs. 22.5 +/- 9.4 nmol of bone collagen equivale
nt/mmol Cr (-28%, p < 0.001). There was no significant difference in w
hole body BMC and BMD, lumbar spine, total hip, and distal radius BMD
between oral contraceptive users and controls. Oral contraception is a
ssociated with a moderate, but significant, decrease in bone turnover
that may have a beneficial influence on bone mass only after prolonged
use. However, given the large interindividual variability of bone mes
s, such an effect could not be established by this cross-sectional stu
dy and a longitudinal design is required.