Blood Pressure and Bone Mineral Density in Premenopausal and Postmenopausal Women

Abstract

Introduction:

Previous studies regarding the associations between blood pressure (BP) and bone mineral density (BMD) have shown conflicting results. However, menopausal status and pharmacotherapy may modify this relationship. The objective of this study was to explore the association between systolic BP (SBP) and diastolic BP (DBP) and BMD in pre- and postmenopausal women, and to assess the extent to which this association is mediated by menopausal status and pharmacotherapy.

Methods:

A cross-sectional study was conducted using a sample of 4,058 pre- and postmenopausal women aged 40 years or older (N = 991 and 3,067, respectively), who participated in NHANES III. BMD measurement of the femur neck was used as the primary outcome measure. Regression models were used to examine the association between SBP or DBP and BMD.

Results:

The unadjusted models for systolic and diastolic BP were positively and significantly associated with femoral BMD in premenopausal women (p = 0.0039 and p = 0.0065, respectively) as well as in postmenopausal women (p < 0.0001 for both SBP and DBP). After adjusting for covariates in the multivariate models, the association between BP and BMD in postmenopausal women no longer prevailed. In premenopausal women, the association between SBP or DBP and BMD was modified by hormone therapy (p = 0.0278 and p = 0.0025, respectively). Once the stratum-specific adjusted models by hormone therapy use were examined, the association between SBP or DBP and BMD was no longer significant.

Conclusions:

The study results suggest that there is no link between BP and BMD in pre- and postmenopausal women.

Introduction

Since the late 1990s, epidemiological studies have emerged suggesting that elevated blood pressure (BP) is associated with low bone mineral density (BMD). Two cross-sectional studies conducted on men found a negative association between diastolic BP (DBP) and BMD in several measurement sites.^1,2 In studies conducted in women, systolic BP (SBP) was negatively associated with BMD.^3,4 A large prospective study, conducted in white elderly women, established a positive association between SBP and BMD loss among elderly women.³ The results showed a significant increase in the yearly change rates of femoral BMD from a baseline visit to a follow-up visit; these bone loss rates rose with increasing baseline SBP. These studies also suggested that calcium metabolism is involved in the potential mechanism between BP and BMD.^5
–7 However, as opposed to these studies, a more recent study found no association between hypertension and BMD in postmenopausal women.⁸

Assessing the potential relationship between elevated BP (SBP or DBP) and low BMD is complicated by the fact that both conditions share several risk factors. Menopausal hormone therapy is associated with a higher BMD and has been found to slow the increase in SBP over time compared with nonusers.^9,10 Similarly, thiazide diuretics, while reducing high blood pressure, were also reported to be associated with a decrease in the rate of BMD loss.^11

–14 Thus, therapy used to treat hypertension may be directly associated with BMD. The objective of this study was to explore the association between SBP and BMD and between DBP and BMD in a nationally representative sample of both pre- and postmenopausal women, and to assess the extent to which this association is mediated by antihypertensive medications and hormone therapy.

Materials and Methods

Data were obtained from the National Health and Nutrition Examination Survey (NHANES) III. The NHANES III was a cross-sectional survey, conducted from 1988 through 1994, that was designed to provide data on health status, risk factors, and nutritional profiles from a nationally representative sample of a noninstitutionalized, U.S. civilian population aged 2 months and older. NHANES III used a stratified, multistage probability sample design. The plan and the operation of NHANES III have been described in detail elsewhere.^15,16

The initial study sample consisted of 5,342 women, aged 40 years and older, who were interviewed in their homes and physically examined in a mobile examination center (MEC). From this crude sample, 1,284 women (24%) were excluded to ensure that most women in the sample had valid outcome measures, especially BMD measure, as the main outcome of interest. The following criteria were used to exclude women from the initial sample: invalid SBP or DBP measurement (n = 54); unacceptable scan of bone densitometry determined after review at the Mayo Clinic (n = 801)¹⁷; undetermined menopausal status (n = 422) or premenopausal women above 70 years old (n = 3); pregnancy (n = 5); breastfeeding (n = 2); or serum creatinine ≥1.5 mg/dL (n = 616), as an indication for renal insufficiency (in the absence of glomerular filtration rate [GFR] measure), which might affect calcium metabolism and thus might be a confounder.

Menopausal status was determined based on the occurrence of the last menstrual period reported during the examination interview. Women were asked to provide an interval period during which their last period occurred. Premenopausal women were defined as women who had their last menstrual period no longer than 3 months prior to the interview. Perimenopausal women had at least 4 months but less than 12 months of amenorrhea.¹⁸ Women with no menses for at least 12 consecutive months, with no obvious cause, such as pregnancy or lactation; or women with cessation of menses due to either removal of the uterus and at least one ovary or removal of both ovaries, with or without removal of the uterus (surgical menopause), were categorized as postmenopausal women.¹⁹ Because of their small percentage, perimenopausal women (2%) were grouped together with the premenopausal women for their physiological similarities.

All available (1 to 6) SBP and DBP measurements obtained for each participant at home and at an MEC were used to calculate average BP measurements. The majority of women (82%) had all 6 measurements to compute average SBP and DBP. Average SBP and DBP were categorized in accordance with the cutoffs established by the Sixth Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC VI).²⁰ Women had high SBP if they had an average SBP of 140 mm Hg or greater; women had high DBP if they had an average DBP of 90 mm Hg or greater.

BMD measurements were performed in five regions of the proximal femur at the hip area using dual-energy X-ray absorptiometry (DEXA). Female examinees who were younger than age 60 with a positive or uncertain pregnancy test, or examinees with both hips fractured, were excluded from testing.²¹ Considered as the gold standard, BMD measurements at the femur were categorized, according to the WHO diagnostic criteria comparing the T-score of women in the sample to the mean of young adults.²² The following general diagnostic categories were applied to all women to determine their BMD status: (1) normal bone mass—defined as a value for BMD within 1 standard deviation (SD) of young adult reference group mean; (2) osteopenia (low bone mass)—defined as a value for BMD more than 1 SD below the young adult mean but less than 2.5 SD below this value; and (3) osteoporosis—defined as a value for BMD 2.5 SD or more below the young adult mean.²² Women aged 20–29 years from NHANES III were selected as a reference group for consistency with the timing of peak bone mass occurrence in women.^23,24

Drug use was self-reported and recorded from the label of the prescribed medication container during the home interview. A pre-prepared list of antihypertensives (including diuretics, beta blockers, calcium channel blockers, angiotensin converting enzyme [ACE] inhibitors, and alpha agonists) and hormone therapy (HT) (including estrogens, conjugated; estrogens, esterified; estradiol valerate; estrogenic substances; estrogens, esterified and methyltestosterone; estropipate; and quinestrol) was compiled from the American Hospital Formulary Service (AHFS) to identify the corresponding drugs from a list of standardized generic drug names provided in the prescription medication documentation file.²⁵

Other measurements were also extracted from NHANES files to characterize the study population. The total nutrient intake file was used to obtain the quantitative information on dietary calcium, sodium, or caffeine intake collected in the 24-hour dietary recall. Dietary sodium intake was categorized as high (>2.4 g) or moderate (≤2.4 g) based on the USDA's dietary guidelines and the JNC IV report recommending no more than 2.4 g of sodium consumption (or 6 g sodium chloride) per day.^20,26 Information on physical activity was collected in the survey, and women were categorized as physically active if they engaged in any physical activity in the previous month (e.g., walking, running, and dancing). Smoking status was also self-reported, and women were categorized into current, former, and never-smokers.

Statistical analysis

The cohort selection and the data management were conducted using SAS software (Version 9.1).²⁷ The average SBP and DBP measurements were used as independent variables in the models, and femur BMD was used as the outcome measure. Descriptive statistics and multivariate estimates were calculated on the weighted data to account for the multistage sampling probability design.²⁸ All results are provided on the weighted sample unless otherwise mentioned.

Multivariate linear regression analyses were performed to examine the association between BP and BMD adjusting for potential confounders. Absolute BMD measurement (gm/cm²) was used as a continuous dependent variable modeling SBP and DBP separately. The following covariates were included in the model: age (continuous), race (white, other), BMI (continuous), antihypertensive drug use (yes, no), HT (current use, no-use), smoking status (nonsmoker/current or former smoker), physical activity (active/not active), dietary calcium intake (continuous), and dietary sodium intake (continuous). Age at last menstrual period was added to the adjusted models in postmenopausal women. The correlation between age and age at last menstrual period was examined and found to be low (unweighted data, Pearson correlation coefficient = 0.34). It should be noted that data for dietary calcium intake and dietary sodium intake were missing for about 2%–3% of the women in the sample. Values for age at last menstrual period were missing for about 10% of postmenopausal women in the sample.

Interaction terms between the study independent variables—SBP and DBP—and the other covariates were generated to test the potential effect modification of the covariates on the association between BP and BMD using the 0.05 significance level as the criterion for keeping the interaction in the model. Only significant interaction terms were reported. Stratum-specific models were used to explore the identified interactions further.

Results

The final study sample consisted of 4,058 women aged 40 years and older representing an estimate of 43.6 million women in the United States. Seventy percent (weighted) of these women (30.5 million women) were classified as postmenopausal, and the remaining women were categorized as premenopausal women (13.1 million women).

Postmenopausal cohort

All women in the postmenopausal cohort (N = 3,067) were 40 years of age or older [mean age = 63 years, standard error (SE): 0.5] (Table 1). The mean age at last menstrual period was 44 years (SE:0.3). The average time passed since their last menstrual period was 18 years (SE: 0.4). The majority of the postmenopausal women were overweight (33%) or obese (29%); 15% reported HT use. More than half of the women in the postmenopausal cohort were classified as having osteopenia in the femur neck (54%); and the prevalence of osteoporosis was 18% (Table 2).

Table 1.

Characteristics of Pre- and Postmenopausal Women Aged 40 Years and Older, NHANES III, 1988–1994

Characteristics	Premenopausal women (unweighted N = 991) Weighted %	Postmenopausal women (unweighted N = 3,067) Weighted %
Age groups (years)
40–49	82.0	13.8
50–59	15.9	26.3
60+	2.1	59.9
Race
White	85.0	88.9
Other	15.0	11.1
Body mass index (kg/m²)
Low–normal (<25.0)	48.1	38.5
Overweight–obese (≥25.0)	51.9	61.5
Physical activity
Active	86.2	77.5
Nonactive	13.8	22.5
Smoking
Never	56.6	51.8
Current smokers	18.6	20.0
Past smokers	24.8	28.2
Family history of osteoporosis
Yes	7.1	7.4
Hormone therapy
Users	6.2	14.9
Antihypertensive drug use
Current users	9.7	36.0
Dietary sodium intake
Moderate (≤2400 mg)	38.5	50.7
High (>2400 mg)	61.5	49.3
Dietary calcium intake
	674 mg (SE = 19.0)	649 mg (SE = 10.6)

Based on National Health and Nutrition Examination Survey (NHANES) III, 1988–94.

Table 2.

Weighted Distribution of Blood Pressure Classification by Femoral Bone Mineral Density Status in Pre- and Postmenopausal Women

	Femoral bone mineral density status
Blood pressure classification	Normal	Osteopenia	Osteoporosis
Premenopausal women (100%)	59.8%	39.4%	0.8%
Normal blood pressure	53.0%	36.8%	0.7%
High blood pressure	6.8%	2.6%	0.1%
Postmenopausal women (100%)	28.4%	54.0%	17.6%
Normal blood pressure	20.0%	34.7%	9.4%
High blood pressure	8.4%	19.2%	8.3%

Based on NHANES III, 1988–1994.

Antihypertensive medications were frequently reported in the study population (28%), particularly in the postmenopausal cohort. More than 50% of the postmenopausal women were classified as hypertensive based on an elevated BP and/or reported use of antihypertensive medications. About 36% of postmenopausal women used antihypertensive drugs (Table 1). Among women reporting antihypertensive medication use, diuretics were the most frequently used class of antihypertensives, with a similar percentage in pre- and postmenopausal women (45.9% and 53.8%, respectively, p = 0.1903) (Table 3). Regardless of menopausal status, approximately three quarters the of women who reported diuretic use used thiazide diuretics (75.3% in the premenopausal and 75.5% in the postmenopausal cohorts) (data not shown).

Table 3.

Weighted Distribution of Antihypertensive Therapeutic Classes Among Women Reported Antihypertensive Medication Use by Menopause Status

	Premenopausal women (N = 113)	Postmenopausal women (N = 1,185)	Total (N = 1,298)
Antihypertensive therapeutic classes	% ^a	% ^a	% ^a
Antihypertensives	10.4	89.6	100.0
Diuretics (nonthiazides)	11.3	14.1	14.1
Thiazide diuretics	34.6	40.0	40.0
β-blockers	34.3	30.4	30.4
ACE inhibitors	22.9	20.4	20.4
Calcium channel blockers	24.0	28.1	28.1
α-adrenergic agents	1.4	5.8	5.8
Other antihypertensives	0.7	6.8	6.8

Antihypertensive therapeutic classes could add up to more than 100%.

Based on NHANES III, 1988–1994.

In the linear regression models for SBP, SBP was negative and significantly associated with BMD (p < 0.0001) in the unadjusted model (Table 4). After adjusting for covariates, the association between SBP and BMD was no longer statistically significant (β = 0.00002, SE = 0.000138, p = 0.8573). There were no statistically significant interactions in either of the adjusted models. Examining the stratum-specific model by HT further showed that there was no significant association between SBP and BMD.

Table 4.

Weighted Regression Coefficients and Standard Errors of Systolic/Diastolic Blood Pressure from Linear Regression Models of Femoral Bone Mineral Density in Pre- and Postmenopausal Women

Model type	Coefficient estimate	Standard error	P-value
	Postmenopausal women
Systolic BP
Unadjusted model	−0.00090	0.000152	< 0.0001
Adjusted model^a	0.00002	0.000138	0.8573
Stratum-specific adjusted models^b
HT users (n = 317)	0.00061	0.000416	0.1506
HT nonusers (n = 2,429)	−0.00005	0.000140	0.7191
Diastolic BP
Unadjusted model	0.00232	0.000361	< 0.0001
Adjusted model^a	0.00003	0.000353	0.9327
Stratum-specific adjusted models^b
HT users (n = 317)	−0.00028	0.000745	0.7097
HT nonusers (n = 2429)	0.00009	0.000367	0.8173
	Premenopausal women
Systolic BP
Unadjusted model	0.00108	0.000356	0.0039
Stratum-specific adjusted models^c
HT users (n = 48)	−0.00127	0.001228	0.3096
HT nonusers (n = 920)	−0.00006	0.000461	0.8911
Diastolic BP
Unadjusted model	0.00189	0.000665	0.0065
Stratum-specific adjusted models^d
HT users (n = 48)	0.00378	0.002001	0.0705
HT nonusers (n = 920)	0.00013	0.000568	0.8208

Adjusted for age (continuous), race (white, other), body-mass index (continuous), antihypertensive drug use (yes, no), hormone therapy (current use, no-use), smoking status (nonsmoker, current or former smoker), physical activity (active, not active), dietary calcium intake (continuous), dietary sodium intake (continuous), and age at last menstrual period (continuous).

Adjusted for the same variables as listed above (a) excluding hormone therapy.

Adjusted for age (continuous), race (white, other), body mass index (continuous), antihypertensive drug use (yes, no), hormone therapy (current use, no-use), smoking status (nonsmoker, current or former smoker), physical activity (active, not active), dietary calcium intake (continuous), and dietary sodium intake (continuous).

Adjusted for the same variables as listed above (c) excluding hormone therapy.

BP, blood pressure; HT, hormone therapy.

Based on NHANES III, 1988–94.

Similar patterns were found for the DBP models. The coefficient estimate for DBP in the unadjusted model suggested a positive association between DBP and BMD (p < 0.0001) (Table 4). After adjusting for the same covariates as in the SBP model, there was no association between DBP and BMD (β = 000003, SE = 0.000353, p = 0.9327).

Premenopausal cohort

Women in the premenopausal cohort (N = 991) had a mean age of 45 years (SE ± 0.3) (Table 1). A small percentage of this cohort reported HT use (8%). Antihypertensive medications were reported in 10% of this population. The prevalence of osteopenia among these premenopausal women was 40%, though osteoporosis was uncommon (1%) (Table 2).

The linear regression models for SBP showed that SBP was significantly and positively associated with BMD (p = 0.0039) (Table 4). After adjusting for covariates, HT use was identified as an effect modifier, modifying the association between SBP and BMD (p-value for the interaction term: 0.0278). To explore this interaction term, two stratum-specific models were used. The first model was built for premenopausal women who used HT, and the second model was built for premenopausal women who did not used HT. There was no association between SBP and BMD among women who used HT and among women who did not used HT (p = 0.3096 and p = 0.8911, respectively).

In the linear regression models for DBP, the coefficient estimate for DBP in the unadjusted model was positive and significantly different from zero (β = 0.00189, SE = 0.000665, p = 0.0065) (Table 4). After adjusting for covariates, HT use was identified as an effect modifier, modifying the association between DBP and BMD (p-value for the interaction term: 0.0025). Once the stratum-specific adjusted models of HT users or HT nonusers were examined, the association between SBP and BMD was no longer significant (p = 0.0705 and 0.8208, respectively).

Discussion

Findings from this nationally representative study of pre- and postmenopausal women suggest that there is no association between BP and BMD, regardless of menopausal status or the type of BP measurement (i.e., SBP or DBP). While both systolic and diastolic BP were significantly associated with femoral BMD in the unadjusted models for both pre- and postmenopausal women, the association no longer prevailed after adjusting for covariates. HT modified the association between SBP or DBP and BMD in premenopausal women. However, after stratification by HT and adjusting for potential confounders, this association was no longer significant.

Results from this study are consistent with a previously published study by Mussolino and Gillum that found no association between hypertension and BMD in the same data source.⁸ The study was conducted in postmenopausal women aged 50 years and older, and examined the association between low BMD with an increased prevalence of hypertension.⁸ They used hypertension as the outcome and found that the age- and race-adjusted relative odds of hypertension were decreased in the first BMD quartile as compared to the fourth quartile (odds ratio [OR], 0.50; p < 0.01); however the association was no longer significant after adjusting for additional risk factors in multivariate models (OR, 0.92; p = 0.62).

Although this study was performed in the same database, some major and important differences are noteworthy. Unlike the previous Mussolino's study, this study examined the association between BP and BMD in a younger population of both pre- and postmenopausal women. In addition, BP was used as an independent variable adjusting for the antihypertensive group as a confounder, since it was shown that thiazide diuretics may be associated with a higher BMD.²⁹ Also, BMD was used as the outcome measure and not as an independent variable. Next, this study examined the association between SBP or DBP and BMD separately, rather than examining hypertension in general. This differentiation between SBP and DBP is important, since it was suggested that in women SBP, and not DBP, was associated with BMD.^3,4 Another difference is that this study explored the effect modification of HT and found it to be significant in premenopausal women. This finding led to further exploration of the stratum specific-models by HT. Lastly, the multivariate regression models were adjusted for potential confounders that were not previously examined, including dietary calcium and sodium intakes and age at last menstrual period (for postmenopausal women).

While consistent with the prior study conducted in the same database, findings from this study differed from a large prospective longitudinal Study of Osteoporotic Fractures (SOF).³ The SOF reported an inverse association between BP and BMD in elderly women. They observed a significant increase in the yearly change rates of femoral BMD from a baseline visit to a follow-up visit; these bone loss rates rose with increasing baseline SBP. A possible explanation for the difference in the study results may stem from the differences in the study outcome and the study design. Our study was a cross-sectional study and BMD was captured at one point in time. In contrast, the SOF was a longitudinal study, which enabled researchers to examine the association between BP and the change in BMD over time. Since the SOF did not report on the relationship between BP and BMD at any one visit as a cross-sectional study, it is not known if the study results would have been different from this study. Moreover, 23% of the women in the SOF did not complete the follow-up. The effect of their loss to follow-up would have led to an exclusion of women with higher BP and lower BMD, thereby making the association harder to detect.³ Other differences between the two studies are related to the inclusion criteria for the study population. The SOF study was limited to white women only, and the use of thiazides was an exclusion criterion. In contrast, this study included mostly white women but also women of other races (15%). Also, the use of antihypertensive medications was controlled for in the multivariate models.

The effect modification of HT in premenopausal women reported in this study was not suggested in previous studies, possibly because premenopausal women as a group were not studied before. Stratification by HT revealed that BP was no longer associated with BMD in women who used HT. It should be noted, however, that the sample size of premenopausal women who used HT was very small (n = 48) and that may have produced unstable estimates and insignificant results. A further exploration of this group suggested that the mean age of these women was 55 years, and about 20% of them reported a family history of osteoporosis. Nonetheless, these results should be interpreted cautiously as HT is not indicated for premenopausal women. A possible explanation for the use of HT in premenopausal women is the irregularity in estrogen levels once women approach menopause, which results in prescribing HT (or an oral contraceptive) to provide more stable levels of estrogen. It is also possible that in this group there was misclassification in their menopausal status.

Evidence for the association between BP and BMD from animal and human studies is compelling; however, the mechanism underlying this potential relationship remains unclear. The suggested mechanism for this relation involves central blood volume as the causal pathway to bone demineralization rather than BP per se.⁷ It was suggested that a genetic defect, such as salt sensitivity, could be involved in the kidneys' failure to excrete sodium and thus further affect calcium levels in patients with essential hypertension.^6,7 Sodium retention may have a role in the increase in the central blood volume (a manifestation of high BP), and eventually in enhanced urinary calcium excretion, followed by stimulation of the parathyroid, thereby releasing hormones that ultimately lead to bone demineralization. This suggested mechanism was based on accumulated findings of abnormalities in a calcium metabolic profile of hypertensives that included elevated parathyroid hormone, lower serum ionized calcium, and increased calcium excretion.^30

–34 Tsuda et al. also examined calcium metabolism and found that calcium excretion, measured by the calcium/sodium ratio in the urine, was consistent with a lower BMD and was significantly greater in hypertensive women than in normotensive controls.⁴ This hypothesized mechanism may explain why hypertensive individuals in the long term are more likely to develop bone demineralization, as demonstrated in some animal models of hypertension.⁷ Furthermore, this potential mechanism was the basis for using BMD as the outcome in this study rather than as an explanatory variable.

There are several limitations in the study design that need to be considered. The study is cross-sectional by design, thus operational definitions of BP and BMD were based on cross-sectional measurements. While these measurements allowed us to differentiate between SBP and DBP on a continuous scale, it also limited the inferences from the study regarding a temporal association between BP and BMD. Another potential limitation derives from some self-reported information collected in the survey. This information, particularly the reproductive responses, was subject to recall bias and thus could lead to misclassification bias in the classification of menopausal status. Especially, menopausal status was determined based on the reported last menstrual period, and it was found that those classified as premenopausal women also reported that they were taking HT. In addition, the definition of surgical menopause may have introduced a misclassification bias since women with a hysterectomy with only one ovary removed were categorized as postmenopausal. Yet, it was shown that even after a hysterectomy, the remaining ovary may continue to secrete hormones and may function for five years.³⁵ However, since no information was available on the timing of the surgery and thus on the transition point when these women became menopausal, classifying these women as premenopausal women might have introduced misclassification bias. No adjustment was made for the use of oral contraceptives, since only two women in the sample reported the use of contraceptives. Despite these limitations, the study was based on a large representative sample of the U.S. population, and data were systematically collected to assure their reliability and to minimize measurement error.

Conclusions

While no single study is sufficient for causal inference, this study suggests that there is no association between BP and BMD in pre- and postmenopausal women from a nationally representative sample.

Disclaimer

The views expressed in this article do not necessarily represent the views of the Food and Drug Administration.

Disclosure Statement

This study was conducted as part of the doctoral dissertation of the author S. Kaplan at the University of Maryland. The study was supported by a research award granted to S. Kaplan by the University of Maryland Women's Health Research Group. No competing financial interests exist. The authors have no conflicts of interest to report.

References

Metz

, Morris

, Roberts

, McClung

, McCarron

. Blood pressure and calcium intake are related to bone density in adult males. Br J Nutr, 1999; 81:383–388.

Jankowska

, Susanne

, Rogucka

, Medras

. The inverse relationship between bone status and blood pressure among Polish men. Ann Hum Biol, 2002; 29:63–73.

Cappuccio

, Meilahn

, Zmuda

, Cauley

. High blood pressure and bone-mineral loss in elderly white women: a prospective study. Study of Osteoporotic Fractures Research Group. Lancet, 1999; 354:971–975.

Tsuda

, Nishio

, Masuyama

. Bone mineral density in women with essential hypertension. Am J Hypertens, 2001; 14:704–707.

Cruz

. The renal tubular Na-Cl co-transporter (NCCT): a potential genetic link between blood pressure and bone density? Nephrol Dial Transplant, 2001; 16:691–694.

Cappuccio

, Kalaitzidis

, Duneclift

, Eastwood

. Unravelling the links between calcium excretion, salt intake, hypertension, kidney stones and bone metabolism. J Nephrol, 2000; 13:169–177.

MacGregor

, Cappuccio

. The kidney and essential hypertension: a link to osteoporosis? J Hypertens, 1993; 11:781–785.

Mussolino

, Gillum

. Bone mineral density and hypertension prevalence in postmenopausal women: results from the Third National Health and Nutrition Examination Survey. Ann Epidemiol, 2006; 16:395–399[Epub 2005 Oct 11].

Cauley

, Robbins

, Chen

et al. Effects of estrogen plus progestin on risk of fracture and bone mineral density: the Women's Health Initiative randomized trial. JAMA, 2003; 290:1729–1738.

10.

Scuteri

, Bos

, Brant

, Talbot

, Lakatta

, Fleg

. Hormone replacement therapy and longitudinal changes in blood pressure in postmenopausal women. Ann Intern Med, 2001; 135:229–238.

11.

Wasnich

, Davis

, Ross

, Vogel

. Effect of thiazide on rates of bone mineral loss: a longitudinal study. BMJ, 1990; 301:1303–1305.

12.

Morton

, Barrett-Connor

, Edelstein

. Thiazides and bone mineral density in elderly men and women. Am J Epidemiol, 1994; 139:1107–1115.

13.

LaCroix

, Ott

, Ichikawa

, Scholes

, Barlow

. Low-dose hydrochlorothiazide and preservation of bone mineral density in older adults: a randomized, double-blind, placebo-controlled trial. Ann Intern Med, 2000; 133:516–526.

14.

Reid

, Ames

, Orr-Walker

et al. Hydrochlorothiazide reduces loss of cortical bone in normal postmenopausal women: a randomized controlled trial. Am J Med, 2000; 109:362–370.

15.

Ezzati

, Massey

, Waksberg

, Chu

, Maurer

. Sample design: Third National Health and Nutrition Examination Survey. Vital Health Stat, 1992; 2:1–35.

16.

National Center For Health Statistics. Plan and operation of the Third National Health and Nutrition Examination Survey (NHANES III), 1988–94. Reference manuals and reports. [CD-ROM] Hyattsville, MD, 1996.

17.

Wahner

, Looker

, Dunn

, Walters

, Hauser

, Novak

. Quality control of bone densitometry in a national health survey (NHANES III) using three mobile examination centers. J Bone Miner Res, 1994; 9:951–960.

18.

Brambilla

, McKinlay

, Johannes

. Defining the perimenopause for application in epidemiologic investigations. Am J Epidemiol, 1994; 140:1091–1095.

19.

McKinlay

. The normal menopause transition: an overview. Maturitas, 1996; 23:137–145.

20.

Joint National Committee on the Detection, Evaluation and Treatment of High Blood Pressure. The sixth report of the Joint National Committee on prevention, detection, evaluation, and treatment of high blood pressure. Arch Intern Med, 1997; 157:2413–2446.

21.

National Center for Health Statistics. Bone density manual. NHANES III: reference manuals and reports. [CD-ROM] Hyattsville, MD, 1996.

22.

World Health Organization (WHO). Assessment of fracture risk and its application to screening for postmenopausal osteoporosis. Report of a WHO study group. Technical Report Series, 1994; 843:1–129.

23.

Looker

, Orwoll

, Johnston

et al. Prevalence of low femoral bone density in older U.S. adults from NHANES III. J Bone Miner Res, 1997; 12:1761–1768.

24.

Slemenda

, Longcope

, Peacock

, Hui

, Johnston

. Sex steroids, bone mass, and bone loss: a prospective study of pre-, peri-, and postmenopausal women. J Clin Invest, 1996; 97:14–21.

25.

AHFS drug information 1999. Bethesda, MD: American Society of Health System-Pharmacists, Inc., 1999.

26.

Dietary Guidelines Advisory Committee. Dietary guidelines for Americans. Washington, DC: Agricultural Research Service, U.S. Department of Agriculture, 2000. http://www.cnpp.usda.gov/publications/dietaryguidelines/2000/2000DGcommitteeReport.pdf. 2010 March 18.

27.

SAS Institute Inc. SAS software, Version 8.2. Cary, NC: SAS Institute, 1999–2001.

28.

National Center for Health Statistics. Analytic and reporting guidelines. NHANES III: reference manuals and reports. [CD-ROM] Hyattsville, MD, 1996.

29.

Cauley

, Cummings

, Seeley

et al. Effects of thiazide diuretic therapy on bone mass, fractures, and falls. The Study of Osteoporotic Fractures Research Group. Ann Intern Med, 1993; 118:666–673.

30.

McCarron

, Pingree

, Rubin

, Gaucher

, Molitch

, Krutzik

. Enhanced parathyroid function in essential hypertension: a homeostatic response to a urinary calcium leak. Hypertension, 1980; 2:162–168.

31.

Hvarfner

, Bergstrom

, Morlin

, Wide

, Ljunghall

. Relationships between calcium metabolic indices and blood pressure in patients with essential hypertension as compared with a healthy population. J Hypertens, 1987; 5:451–456.

32.

Young

, Morris

, McCarron

. Urinary calcium excretion in essential hypertension. J Lab Clin Med, 1992; 120:624–632.

33.

Grobbee

, Hackeng

, Birkenhager

, Hofman

. Raised plasma intact parathyroid hormone concentrations in young people with mildly raised blood pressure. BMJ, 1988; 296:814–816.

34.

Brickman

, Nyby

, von Hungen

, Eggena

, Tuck

. Calcitropic hormones, platelet calcium, and blood pressure in essential hypertension. Hypertension, 1990; 16:515–522.

35.

Chalmers

, Lindsay

, Usher

, Warner

, Evans

, Ferguson

. Hysterectomy and ovarian function: levels of follicle stimulating hormone and incidence of menopausal symptoms are not affected by hysterectomy in women under age 45 years. Climacteric, 2002; 5:366–373.