Sodium and Bone Health: Impact of Moderately High and Low Salt Intakes on Calcium Metabolism in Postmenopausal Women


  • The authors state that they have no conflicts of interest.

  • Published online on April 14, 2008


High salt intake is a well-recognized risk factor for osteoporosis because it induces calciuria, but the effects of salt on calcium metabolism and the potential impact on bone health in postmenopausal women have not been fully characterized. This study investigated adaptive mechanisms in response to changes in salt and calcium intake in postmenopausal women. Eleven women completed a randomized cross-over trial consisting of four successive 5-wk periods of controlled dietary intervention, each separated by a minimum 4-wk washout. Moderately low and high calcium (518 versus 1284 mg) and salt (3.9 versus 11.2 g) diets, reflecting lower and upper intakes in postmenopausal women consuming a Western-style diet, were provided. Stable isotope labeling techniques were used to measure calcium absorption and excretion, compartmental modeling was undertaken to estimate bone calcium balance, and biomarkers of bone formation and resorption were measured in blood and urine. Moderately high salt intake (11.2 g/d) elicited a significant increase in urinary calcium excretion (p = 0.0008) and significantly affected bone calcium balance with the high calcium diet (p = 0.024). Efficiency of calcium absorption was higher after a period of moderately low calcium intake (p < 0.05) but was unaffected by salt intake. Salt was responsible for a significant change in bone calcium balance, from positive to negative, when consumed as part of a high calcium diet, but with a low calcium intake, the bone calcium balance was negative on both high and low salt diets.


Diets high in sodium alter calcium metabolism by increasing urinary calcium excretion (calciuria), and a high salt intake is therefore assumed to be a risk factor for osteoporosis. However, the evidence is mainly based on acute salt-loading studies, which have little relevance to long-term diets, and the results of bone turnover studies are inconsistent.(1) There is therefore a need to clarify the effects of salt and calcium on calcium metabolism at dietary levels that are commonly consumed. The published literature on sodium and calcium metabolism indicates that the average loss of calcium is 1 mmol Ca (40 mg) per 100 mmol (2290 mg) of sodium, and without any adaptive compensatory mechanisms, a daily loss of 40 mg calcium would deplete 10% of the skeleton within a decade.(2) Interindividual variation in sodium-induced calciuria in cross-sectional and some salt-loading studies are explained by differences in intake of other dietary factors known to affect urinary calcium excretion, such as protein(3) and potassium.(4,5) Differences in sodium-induced calciuria in short-term controlled dietary interventions (7–14 days) suggest that genotype-related salt-sensitivity may also play a role.(6)

To maintain homeostasis, urinary calcium losses may be compensated for by an increase in the efficiency of calcium absorption and/or an increase in the rate of bone resorption. The conclusion from a 2-yr longitudinal study in postmenopausal women was that high sodium intakes were associated with increased bone loss at hip but not spine(7) and from the regression analysis, it was estimated that reducing the daily urinary sodium excretion from 3450 (8.8 g salt) to 1725 mg (4.4 g salt) would have an effect on BMD equivalent to an increase in daily calcium intake of 891 mg, indicating a potentially powerful effect of sodium on bone loss at the hip. However, the overall evidence from cross-sectional and prospective studies on salt reduction and BMD is inconsistent, and there seems to be no published data on the relationship between sodium intake and fracture risk.(1) With regard to calcium balance, it has not yet been conclusively showed that upregulation of calcium absorption compensates (partially or fully) for sodium-induced calciuria. The null hypothesis for our study was that the level of dietary intake (moderately high or low) of sodium or calcium would not affect the efficiency of calcium absorption or bone calcium metabolism. State-of-the-art stable isotope labeling techniques were used to determine calcium absorption and excretion in a controlled cross-over metabolic study in postmenopausal women.



Sixteen female subjects, at least 5 yr postmenopausal and <75 yr of age, were recruited. Study exclusion criteria included current use of hormone replacement therapy (HRT) or <2 yr since the use of HRT, osteoporosis or other diagnosed bone disease, rheumatoid arthritis, osteoarthritis, chronic illness, renal, gastrointestinal or hormonal disorders, hypertension, regular use of diuretics or antacids, smoking, and unwillingness to stop taking vitamin/mineral supplements for the study duration. Blood pressure was measured at the prestudy stage to ensure subjects were not hypertensive. Following British Heart Foundation recommendations, systolic blood pressure of 140 mm Hg and diastolic blood pressure of 90 mm Hg were the thresholds above which subjects were excluded. Eleven of the volunteers completed all four dietary interventions.

Ethical considerations

Ethical approval for the study was obtained from the Norwich Research Ethics Committee. Each participant gave written informed consent.

Study design

The study was a randomized, repeat cross-over trial of the effect of moderately high or low salt intake (11.2 g salt [4.4 g/d sodium] and 3.9 g salt [1.6 g/d sodium], respectively) on calcium and bone metabolism in postmenopausal women, consuming either a moderately high or low or high calcium diet (1284 and 518 mg/d, respectively); there were four successive 40-d dietary intervention periods, each separated by a minimum 4-wk washout period when subjects consumed their habitual diet. An identical basal diet low in calcium and sodium was consumed during all four interventions, to which appropriate amounts of calcium (as calcium carbonate supplements) and sodium (in preprepared meals and as supplements) were added to increase the intake. The four diets were −Ca−Na (low calcium, low sodium), +Ca−Na (high calcium, low sodium), −Ca+Na (low calcium, high sodium), and +Ca+Na (high calcium, high sodium).

A dual-tracer stable isotope technique (oral 42Ca and IV 43Ca) was used to measure calcium absorption after 30 days on each of the four dietary interventions. Kinetic measurements were made to undertake compartmental modeling to estimate bone resorption and balance. On the morning of day 30, after an overnight fast of at least 10 h, subjects attended the Human Nutrition Unit (HNU) and were given a standard breakfast (containing 183 mg calcium) of toast, butter, jam, and 100 ml of semiskimmed milk, extrinsically labeled with 20 mg 42Ca (as CaCl2). The glass was rinsed several times with deionized water, and the washings were consumed to make sure the full dose of oral isotope was ingested. Approximately 30 min later a solution of 43Ca (4.5 mg 43Ca as CaCl2) was infused over 20 min through a cannula inserted into the antecubital vein of the other arm. During this period, volunteers were only allowed to consume deionized water (MilliQ; Millipore, Watford, Herts, UK). After removal of the cannula, volunteers were given a standard light meal, and further 10-ml fasting blood samples were collected on days 31, 34, and 38, and a final blood sample (40 ml) was collected on day 40. Calcium absorption was measured from isotope ratios in fasting serum samples obtained on days 31, 34, 38, and 40. Subjects collected 24-h urine samples on 2 consecutive days before each intervention, on day 16 (midintervention) and days 30–40 after the isotope administration. Urine samples before each intervention were analyzed for calcium, sodium, phosphorus, and potassium. Urines collected on days 16 and 30–40 were analyzed for calcium and sodium and those collected on days 39 and 40 were analyzed for potassium and phosphorus. Individual fecal samples were collected into autoclavable bags from day 30 through to day 40 and analyzed for total calcium and isotopic enrichment.

Venous blood on days 1 and 40 was analyzed for markers of bone formation (bone alkaline phosphatase [B-ALP] and osteocalcin [OC]) and calciotropic hormones {PTH, 1,25-dihydroxyvitamin D [1,25(OH)2D], and 25-hydroxyvitamin D [25(OH)D]}. Volunteers collected 24-h urine samples in preweighed acid-washed 2.5-liter bottles 2 days before each intervention and during the 10 days after isotope administration. Aliquots of the urine collected before each intervention and during the last 2 days of intervention were pooled and analyzed for markers of bone resorption (N-terminal telopeptide [NTX], pyridinoline [Pyr], deoxypyridinoline [Dpyr]) and sodium, calcium, potassium, and phosphorus.

Blood pressure was determined on three different occasions before acceptance onto the study and monitored on a weekly basis during each of the four dietary interventions. All measurements were carried out by trained investigators using a fully automated blood pressure monitor (Omron 711; Omron Matsusaka).

Experimental diets

The diet was comprised of a 7-day rotating menu, calculated to be low in sodium (1.5 g/d) and calcium (500 mg/d). Volunteers selected 7 menus from a total of 12 available menus. All meals (breakfast, lunch, dinner, and snacks) during each intervention period were prepared in the Human Nutrition Unit (HNU) and delivered to subjects' homes three times a week. A low calcium intake was achieved by preparing all bakery products with white flour without the UK legislative addition of calcium, and the milk intake was restricted to 90 g/d. To make the diets palatable and hence ensure compliance up to eight portions of fruits and vegetables were provided each day. A low calcium water supply was obtained by providing volunteers with a water filter, a supply of replacement filter cartridges and a kettle. During the high sodium intervention, additional sodium (0.7 g/d) was provided from high salt bread. Salt was also added to some of the cooked meals during batch cooking for the high salt diet. A total of 13 batches of the menu selections on offer were prepared. Of these, four were analyzed by AAS to confirm the sodium and calcium content of the low and high salt diets. The intake of other nutrients was calculated with “Diet Cruncher” (Way Down South Software, Dunedin, New Zealand) using the McCance and Widdowson's food composition tables.(8) No additional food was allowed and subjects completed a food intake diary each day to monitor dietary and fluid intake.


Calcium (125 mg calcium as CaCO3) and sodium supplements (333 mg sodium as NaCl) were provided by Penn Pharmaceutical Services (Tredegar, Gwent, UK). Vitamin D supplements (10 μg cholecalciferol) were provided by Solgar Vitamins UK (Newcastle, Staffordshire, UK). For the high calcium and/or sodium diets, subjects were requested to take two calcium and/or sodium capsules with each main meal (i.e., three times a day, providing a total of 750 mg calcium/d or 2.0 g sodium/d from supplements) for the full 40-day intervention period. Vitamin D supplements were taken once every day, at breakfast, starting 4 wk before each intervention period and supplementation was continued until the end of each 40-day intervention period. A daily guide for supplement intake was included in the food intake diary and compliance was monitored by written self-reported intake.

Preparation of isotope labels

The 43Ca tracer for intravenous infusion (52.1% enrichment) was prepared by the Ipswich Hospital Pharmacy (Ipswich, UK) as follows: an accurately weighed amount of 43Ca-enriched CaCO3 was dissolved in concentrated HCl (molar ratio of 3:1), the solution adjusted to pH 6, and made up to volume with sterile water to give a final concentration of 0.75 mg/ml. Individual subsamples of the solution were transferred to glass ampoules, sealed, and heat sterilized, and the sterility of the solution was verified. For oral administration, 42Ca-enriched CaCO3 (90.8% enrichment) was converted to CaCl2 by adding 6 M HCl in a 2:1 molar ratio. The pH was similar to that of milk (pH 6.7). For each subject, an aliquot of the extrinsic label containing 20 mg CaCO3 was added to 200 g semiskimmed milk and allowed to equilibrate overnight at 4°C.

Dietary analysis

All meals and snacks that formed part of the 7-day rotating menu cycle were homogenized (Ultra-Turrax T-50 homogenizer; IKA) and freeze-dried. Aliquots were ashed in a muffle furnace for 48 h at 450°C (Vulcan, 3–550; NEY Dental International, Bloomfield, CT, USA), dissolved in 5 ml of a 5% HCl solution, and analyzed for calcium by atomic absorption spectroscopy (AAS; model 3300; Perkin-Elmer, Norwalk, CT, USA) after further dilution with 0.1% lanthanum chloride. For the determination of sodium by atomic emission spectroscopy (AES; model 3300; Perkin-Elmer) further dilutions were carried out using 18-Ω water. All samples were analyzed in duplicate, and a certified typical diet reference material (Typical Diet 1548a; NIST, Gaithersburg, MD, USA) was analyzed with each batch of samples.

Urine analysis

On receipt, all urines were weighed, mixed, and aliquots stored. Sodium and potassium were measured by AES and AAS, after appropriate dilution, respectively (model 3300; Perkin-Elmer). All samples were analyzed in duplicate together with certified urine controls with each batch (Lyphochek, Quantitative Urine Control; Bio-Rad). Calcium isotopes and total calcium were measured by ICP-MS (IsoProbe; GV Instruments, Manchester UK) using the isotope dilution technique (48Ca spike), UV digestion (Model 707; Metrohm), and oxalate precipitation. Phosphorus was measured using the ABX Diagnostics Phosphorous Kit (Shefford, Bedfordshire, UK) on a COBAS MIRA automated analyzer. All samples were analyzed in triplicate, and ABX Diagnostics Human Control N was run for quality control purposes. The intra-assay variation was 3.7%. Interassay variation was avoided by analyzing all samples from each individual in the same run.

Fecal analysis

Individual fecal samples collected for 10 days after dosing were weighed, autoclaved, freeze-dried, thoroughly homogenized, and ashed in a muffle furnace. When more than one fecal sample was collected on the same day, the individual's collection was pooled for that day and processed as described above. Again the samples were spiked with Ca-48 isotope solution, but this time were digested with 0.1 M HNO3 at 80°C. After centrifugation of the digest, the supernatant was subjected to the oxalate precipitation method as used for urine and serum samples.

Biochemical markers of bone metabolism

Urinary Pyr and Dpyr were measured in triplicate using a three-step procedure. Urine was first hydrolyzed with an equal volume of 12 M HCl at 110°C for 18 h, and the cross-links were extracted by CF1 cellulose chromatography with the use of an internal standard (acetylated pyridinoline; MetraBiosystems, Wheatley, Oxon, UK) and measured using a reverse-phase HPLC method with fluorescent detection.(9) The acetylated pyridinoline was used in accordance with the method described by Calabresi et al.(10) and Robins et al.(11) The cross-links content of urine was quantified by external standardization using a commercially available pyridinoline/deoxypyridinoline HPLC calibrator (MetraBiosystems). The intra-assay CVs for Pyr and Dpyr, measured as the variation between 10 chromatograms obtained between column regenerations,(9) were 6% and 7%, respectively. Interassay variation was avoided by analyzing all samples from each individual in the same run.

Urinary NTX was measured in urine samples by an ELISA (Osteomark; Ostex International). The intra-assay CV was 1.6%. Interassay variation was avoided by analyzing all samples from each individual in the same run.

Blood analysis

Total calcium and calcium stable isotopes were quantified in serum. After an overnight fast, blood was collected into sterile trace element free tubes (Vacutainer; Becton Dickinson, Rutherford, NJ, USA) and allowed to clot for a minimum of 30 min. The serum was removed after centrifugation at 1500g for 10 min and stored at −18°C before analysis. After UV digestion (Model 707; Metrohm) and oxalate precipitation, calcium isotopes were determined by inductively coupled plasma mass spectrometry (ICP-MS; IsoProbe, GV Instruments, Manchester, UK).(12) Total calcium was determined by isotope dilution.

Figure Figure 1.

Compartmental model for calcium metabolism. Circles represent compartments; arrows represent movement of calcium in and out of compartments. Vi is the rate of calcium intake, Va is the rate of calcium absorbed from the diet, Vu is the rate of urinary calcium excretion, Vf is the rate of endogenous calcium loss in feces, VF is the rate of dietary calcium loss in feces, V0+ is the rate of calcium deposited in the bone, and V0− is the rate of calcium resorbed from bone.

PTH was measured in plasma from blood collected in EDTA tubes, the plasma having been removed after centrifugation and stored at −18°C. Intact PTH was measured using a chemiluminescent immunometric assay (IMMULITE; Diagnostic Products, Los Angeles, CA, USA). Quality control was monitored through participation in an External Quality Assessment Scheme (DEQAS).

1,25(OH)2D and 25(OH)D was measured in serum. 25(OH)D was identified and quantified using HPLC equipped with a PDA-detector, and a radioimmunoassay (65100E; DiaSorin, Stillwater, MN, USA) was used for the quantification of 1,25(OH)2D as described elsewhere.(13) The accuracy of the analysis was monitored by participation in the Vitamin D External Quality Assessment Scheme (DEQAS, Charing Cross Hospital, London, UK).

Serum OC and B-ALP were measured by an ELISA (BRI-Diagnostics, Dublin, Ireland and MetraBiosystems, respectively). The intra-assay CVs were 11% and 4.5%, respectively. All samples from each individual were analyzed in the same run.

Kinetic modeling

A compartmental model, developed in SAAMII (SAAM Institute, Seattle, WA, USA), was used to fit the experimental data and estimate parameters of interest (Fig. 1), similar to that published by Neer et al.(14) and Wastney et al.(15) but containing just a single soft tissue/bone compartment. The main calcium fluxes were from bone (V0−), to bone (V0+), to urine (Vu), to feces (Vf), and calcium absorption from the diet (Va). Several of these fluxes were used to estimate bone calcium balance, Vbal = Va – Vu – Vf. The process of bone absorption of calcium is represented by the calcium flux (V0+). Because this occurs during the experimental period, it is directly measurable, but this is not the case with bone calcium resorption (V0−). Any stable isotope label that is taken up into the bone will not be reused during the measurement period, and therefore, V0− is estimated by “difference” because V0− = V0+ − Va + Vu + Vf.

Statistical analysis

Statistical analyses were performed using the R data analysis software.(16) Linear models (ANOVA) and paired t-tests were used to determine effects of the different diets on a range of physiological responses. Because the four intervention diets were characterized by combinations of “low” and “high” dietary intake of sodium and calcium, the input variables to the linear models were two-level factors describing sodium and calcium intake and the interaction of these factors. Where appropriate, these factors had a third level, corresponding to the baseline intakes. Volunteer identification was included as a factor variable in all models so that each volunteer could act as her own control. The appropriateness of all final models was checked for outliers (including leverage and influence) and normal errors. Where necessary, outliers were excluded and/or data transformations performed, and the models were refitted. Statistical analysis of the effect of diet on bone biomarkers at the end of each intervention was performed on the calculated percent change from baseline. At no point was it necessary to use nonparametric models. Tukey's honest significant differences were calculated to determine the differences in the levels of sodium or calcium in the linear models. All results were considered significant if p < 0.05.

The kinetic data from the low and high calcium dietary intervention periods was separated and analyzed independently. Within these two treatments, the effect of low versus high salt intake was assessed by paired t-tests. All results were considered significant if p < 0.05.

For some of the outcome measures, baseline values were obtained before each of the four diets. The assumption that all the baseline values were identically distributed could be tested in these cases. Where this assumption was not met, the data were tested to see whether the sequence of the diets was significant. The possibility that sequence of diet may affect outcome was also investigated for calcitropic hormones and bone biomarkers.


Characteristics of the subjects

The mean age was 64 yr (range, 59–73 yr), and the mean body mass index (BMI) was 24.7 kg/m2 (range, 20.9–32.1 kg/m2). According to the WHO definition, 4 of the 11 volunteers were overweight (BMI > 25 kg/m2). Menopause was defined as at least 12 mo since the last menstrual cycle. All participating women reported to be at least 5 yr postmenopausal. The average systolic and diastolic blood pressure for all four intervention periods was 115 and 72 mm Hg, respectively (data not shown). There was no significant difference in blood pressure when consuming the high and low salt diets.

Table Table 1.. Mean (±SD) Urinary Excretion of Sodium, Calcium, Phosphorus, and Potassium According to Dietary Intervention
original image

Intervention diet

The calculated mean energy intake was 1979 kcal, and the percentage of food energy from fat, carbohydrates, and protein was 34%, 54%, and 15%, respectively. The diet met the UK recommended nutrient intake (RNI)(17) for all vitamins except retinol (intake = 384 μg/d; RNI = 600 μg/d) and vitamin D (intake = 2.1 μg/d; RNI for women <60 yr = 0 μg/d and >60 yr =10 μg/d); the mean intake of these two nutrients was also less than the RNI in women 50–64 yr of age in the latest National Diet and Nutrition Survey in Great Britain (retinol = 449 μg/d and vitamin D = 3.5 μg/d).(18) The mean analyzed calcium content of the 12 menus devised for the low calcium diet was 518 ± 49 (SD) mg and for the moderately high calcium diet the intake was increased to 1284 mg/d by the addition of calcium supplements. The mean sodium content of the menus offered for the low and high salt diets was 1557 ± 237 and 4422 ± 253 mg sodium, respectively. The low salt diets provided a mean of 3.9 g salt (1.6 g sodium) and the high salt diets a mean of 11.2 g salt (4.422 g sodium) per day. The high salt interventions required the ingestion of additional salt in six capsules per day, each capsule containing 342.3 mg sodium, and several subjects complained of nausea and vomiting during the first 2 days of the high salt intervention. In response, advice was given on methods of supplement ingestion that would avoid any adverse reactions (e.g., mixing of the salt supplement into the food and dissolving the capsules in hot drinks before consumption). Once adopted, these guidelines prevented further episodes of nausea, and none of the subjects withdrew from the study because of persisting intolerance to the salt supplements. Following Good Clinical Practice (GCP) guidelines, all adverse events were reported to the local Ethics Committee.

Calcitropic hormones

Because this cross-over trial spanned a minimum of 36 wk, vitamin D supplementation of the subjects was essential to avoid any effect of seasonal variation that might have masked changes in intraindividual absorption efficiency. The mean serum 25(OH)D3 concentration was 69.6 ± 11.5 nM (range, 47.8–92.5 nM; data not shown). There were no significant differences in 25(OH)D3 levels at the start and end of each intervention and no differences in 1,25(OH)2D3 at the end of each intervention period. Differences in these parameters were also nonsignificant when analyzed according to sequence of diet consumed. Although subjects received identical vitamin D supplements, the 25(OH)D3 concentration ranged from 48–93 nM, which suggests that interindividual differences in absorption may have been affected by differences in vitamin D status; however, because each subject acted as her own control, this would have had no bearing on the overall findings from the study.

The overall mean serum PTH concentration was 3.5 ± 1.4 pM, which compared well with that reported by others for this age group.(19,20) In response to sodium, average PTH levels (3.9 pM) in subjects consuming the high sodium diets were statistically significantly different (p = 0.007; data not shown).

Urinary sodium, calcium, potassium, and phosphorus excretion

Results for all four controlled dietary interventions and for the habitual diet are summarized in Table 1. Urinary sodium excretion was significantly higher on the high sodium diets compared with the low sodium diets (p < 0.0001) and the habitual diet (p < 0.0001). The difference in sodium excretion between the low and high salt diet was key to the success of the study and reflects compliance. There was no significant difference in sodium excretion between the two low sodium diets (p > 0.05) and the two high sodium diets (p > 0.05). The average sodium excretion from the low (−Ca−Na and +Ca−Na combined) and high (−Ca+Na and +Ca+Na combined) sodium diets was 1402 ± 161 (equivalent to 3.6 g salt) and 3778 ± 460 mg/d (equivalent to 9.6 g salt), respectively, which represents a 2.7-fold difference between the low and high salt diets.

The mean calcium urinary excretion in subjects consuming their habitual diet was 164 ± 91 mg/d (Table 1). Two of 11 subjects excreted calcium in excess of 250 mg/d (>4 mg/kg/d; data not shown). The average calcium excretion for diets −Ca−Na, +Ca−Na, −Ca+Na, and +Ca+Na was 123, 159, 141, and 192 mg/d, respectively (Table 1). Both high sodium and high calcium intake independently provoked a significant increase in urinary calcium excretion (p < 0.0001). The mean increase in urinary calcium excretion of 36 mg/d on the two low salt diets (low versus high calcium) and 51 mg/d on high salt diets (low versus high calcium) presumably reflects an increase in the quantity of calcium absorbed. The mean increase in urinary calcium excretion in response to a high salt intake was 18 and 33 mg/d in subjects consuming the low (−Ca−Na versus −Ca+Na) and high calcium diets (+Ca−Na versus +Ca+Na), respectively. A quarter of the volunteers did not respond to the sodium challenge by increasing their urinary calcium output.

There was no significant difference in potassium excretion between the four dietary treatments, but there was a significant effect of calcium on phosphorus excretion (Table 1). Both low calcium diets were associated with a significantly higher phosphorus excretion; the mean difference between low and high calcium diets was 92 mg.

Table Table 2.. Summary of calcium kinetic data for the Low and High Ca diets
original image
Table Table 3.. Biochemical Markers of Bone Resorption (NTX, Dpyr, Pyr) and Bone Formation (OC, B-ALP) After Each Dietary Intervention [Mean (±SD)]
original image

True calcium absorption and kinetics

Mean calcium absorption after 30 days adaptation to the −Ca−Na, +Ca−Na, −Ca+Na, and +Ca+Na diets was 26 ± 6%, 22 ± 3%, 24 ± 5%, and 23 ± 6%, respectively. The efficiency of absorption from the standard test meal containing 183 mg calcium was higher after a period of adaptation to the moderately low calcium diet (low Ca: 25 ± 5%, high Ca: 23 ± 5%; p < 0.05), but the level of salt in the diet had no effect on calcium absorption (low Na: 23 ± 5%, high Na: 24 ± 5%; p = not significant).

When the data generated from the two low calcium interventions were compared, there was no significant difference between the low and high salt diets for any of the kinetic parameters (Table 2), but there were some apparent trends. Urinary calcium excretion (Vu) was slightly higher as a result of the high salt diet (low Na: 102 ± 71 mg/d; high Na: 123 ± 81 mg/d), and the resulting bone calcium balance (Vbal) was more negative (low Na: −71 ± 93 mg/d; high Na: −115 ± 70 mg/d), but these differences were not statistically significant. There was no effect on bone calcium uptake, V0+ (low Na: 455 ± 129 mg/d; high Na: 469 ± 155 mg/d), bone calcium resorption, V0− (low Na: 565 ± 156 mg/d; high Na: 585 ± 185 mg/d), and endogenous calcium loss in feces, Vf (low Na: 112 ± 133 mg/d; high Na: 115 ± 57 mg/d).

In comparison, the data from the two high calcium interventions showed significant differences in some of the kinetic parameters in response to changes in dietary salt intake (Table 2). Urinary calcium excretion was significantly higher (p < 0.001) as a result of the change from a low salt (Vu = 112 ± 143 mg/d) to a high salt diet (Vu = 153 ± 149 mg/d). The bone calcium balance was significantly impaired (p < 0.05), going from a positive balance on the low salt diet (Vbal = 90 ± 46 mg/d) to a negative balance on the high salt diet (Vbal = −12 ± 46 mg/d). There were nonsignificant changes in V0+ (low Na: 437 ± 139 mg/d; high Na: 361 ± 144 mg/d), V0− (low Na: 346 ± 88 mg/d; high Na: 372 ± 121 mg/d), and Vf (low Na: 82 ± 168 mg/d; high Na: 146 ± 111 mg/d).

Biochemical markers of bone turnover

The average level of biochemical markers of bone resorption measured in 48-h urine pools at the end of each dietary intervention is given in Table 3. Statistical analysis of percentage change from baseline of Pyr and Dpyr measured at the end of each dietary intervention showed a significant increase after the low calcium interventions (Fig. 2). There was a (nonsignificant) decrease in NTX with the high calcium diet, whereas Pyr and Dpyr excretion increased with the low calcium diets. The resorption markers differed in their sensitivity to the changes in sodium intake. There was a significant change in levels of NTX in response to the high compared with the low sodium diets (p = 0.031), but this was not observed for Dpy or Pyr.

Figure Figure 2.

Mean percent change in bone resorption markers after 4 wk of diets low (−Ca) or high (+Ca) in calcium and low (−Na) or high (+Na) in sodium. *Pyr and Dpyr were significantly increased on low calcium diets (p < 0.05). All other changes were not significant.

Although there was no effect of diet on the bone formation marker OC, there was a significant increase in percent change from baseline for B-ALP on the −Ca+Na diet (+14 ± 12%, p = 0.011), suggesting an increase in bone formation activity and overall bone turnover after the increase in salt and decrease in calcium intake.


A marginally negative bone balance is observed in postmenopausal women because of the dominance of bone resorption over bone formation, but large interindividual differences have been reported in women adapted to low and high calcium intakes.(21) The particular strength of the data presented here is the randomized cross-over study design that used four very-well-controlled diets that provided different intakes of calcium and sodium within the range that is commonly observed in Western-style diets. This study investigated the effect of sodium on calcium excretion at two levels of calcium intake in the same individuals using a cross-over design, thereby eliminating the confounding effect of genotype. The calcium and salt levels reflect those consumed by a significant proportion of the UK postmenopausal population. This is an important feature of the study design because many others have used higher salt loads that were likely to elicit a response that would only be seen at the extremes of intake.

The responses to sodium-induced calciuria are similar to those reported by others and summarized by Cohen and Roe,(22) but we observed a significantly more pronounced calciuria on the moderately high (an increase of 33 mg/d) compared with the moderately low calcium diet (an increase of 18 mg/d; p < 0.001), which shows greater conservation on the lower calcium diet. One previously published study based on a parallel group design also reported a trend toward higher calcium excretion on a high calcium diet, but it was not statistically significant.(23)

A significantly positive association between urinary calcium and sodium at baseline (habitual diet) was observed (p < 0.05), which supports observations from cross-sectional and cohort studies.(24–26) However, one quarter of the volunteers did not respond to the sodium challenge by increasing their urinary calcium output. Shortt et al.(27) reported that some individuals do not respond to a sodium challenge by increasing urinary calcium excretion and suggested that interindividual differences may be related to salt sensitivity. Furthermore, it is well documented that the magnitude of the estimated increase in calcium excretion per 100-mmol rise in urinary sodium can vary considerably; 2-fold differences in sodium-induced calciuria at low(24,28) and high calcium intakes(29) are common.

The low calcium diets also induced a significantly higher excretion of phosphorus (mean difference, 92 mg; p < 0.001), which is indicative of increased bone turnover because bone resorption is associated with a loss of both calcium and phosphorus in the urine. Potassium can modify the renal handling of calcium and sodium, and therefore it is an important covariate in metabolic studies on calcium. The average potassium excretion in this study was similar to that obtained by the Dietary Approaches to Stop Hypertension (DASH) trial(30) (i.e., ∼3000 mg/d). The DASH trial observed no effect on sodium-induced calciuria (∼8 g salt/d) between diets low in potassium (1700 mg/d) and calcium (450 mg/d) compared with diets high in potassium (4700 mg/d) and calcium (1250 mg/d). This contradicts results obtained by Sellmeyer et al.,(5) who found that a doubling of potassium excretion from 2720 to 5500 mg/d was effective in reducing the sodium-induced calciuria in postmenopausal women consuming a rather high mean intake of 12 g salt/d. The basal diet in our study contained up to eight portions of fruits and vegetables (range: 6.9–9.1), which are a good dietary source of potassium, but the limited intake of dairy products (another good source of potassium) during the intervention meant that the total potassium excretion (2957 mg/d) was only on average 100 mg higher than that estimated for the habitual diet (2860 mg/d). This explains why potassium can be excluded as a modulator of calciuria in this study.

In this study, we chose to administer the same calcium test dose after each of the 4-wk adaptation periods to avoid the well-documented dose–response effects that could mask any response to the dietary intervention. However, it could be argued that the calcium test loads should have represented the calcium content of the intervention diets. This would mean that our test dose of 183 mg was appropriate for the low calcium diet but that 450 mg calcium should have been given as the test dose for the high calcium diet. According to work by Heaney et al.,(31) calcium absorption from a diet containing 1250 mg/d of calcium (our high calcium intervention diet) should be ∼20%, rather than the 23% that we observed with our test dose of 183 mg. Taking the 20% absorption figure obtained from the work of Heaney et al. and applying it to our model would, theoretically, cause minor reductions in the calcium fluxes shown in Table 2 (high calcium diet). Given this possibly lower absorption on the high calcium diet and the subsequent effect on the calcium fluxes, it would be inappropriate to examine the changes in the kinetic parameters that may be caused by calcium intake between the low and high calcium diets. Therefore, resulting kinetic data were separated into the two dietary calcium groups (Table 2) and analyzed separately to determine the effect of the salt intervention. This is a valid comparison because the difference in salt intake did not significantly alter calcium absorption (low Na: 23 ± 5%, high Na: 24 ± 5%; p = not significant), and therefore the only variable that changes within either the low or high calcium intervention groups is salt intake, which must be responsible for any changes in the kinetic parameters.

Our kinetic data suggest that salt has a significant effect on calcium urinary excretion (p < 0.001) and bone calcium balance (p < 0.05) only on the high calcium diets. More calcium is excreted in the urine as a result of the increased salt intake, and bone calcium balance changes from being positive on the low salt diet to negative on the high salt diet. There was a trend for more endogenous calcium to be excreted into the feces, but this was not significant (p = 0.082). The low calcium diets show similar trends, but they are not significant; more calcium is excreted in the urine as a result of the increased salt intake (p = 0.057), and bone calcium balance becomes more negative when moving from a low to high salt diet (p = 0.196). Extended periods of low calcium intake, as used in this study, have been shown to maximize mechanisms to conserve calcium, which may explain why the salt challenge on the low calcium diet seems to be less pronounced and did not reach significance in terms of calcium kinetics.

An earlier study in postmenopausal women consuming a moderate intake of calcium (816 mg/d) showed no detectable effect of high or low salt diets on the rate of bone resorption, but this is not surprising given the fact that the study was only an 8-day duration.(29) Longer-term studies are needed to evaluate effects on bone turnover in response to diet(30) as well as drug therapy(32) when assessed by bone biomarkers. Notwithstanding this requirement, the bone calcium balance predicted by our kinetic model is supported by the findings of reduced levels of markers of bone resorption in response to the high calcium diets. For each of the three markers of bone resorption, the observed significant changes from baseline in response to calcium were associated with a power of >90%. If we assume a minimum 10% difference from baseline in response to sodium to be significant, the power of identifying such a difference in our sample was 20% for NTX, 80% for Pyr, and 57% for Dpyr. The observed percent change for NTX in response to sodium was actually 19%, with an associated power of 57%. In a longer-term study (6 mo) of postmenopausal black women, a low sodium diet (2 g/d) reduced bone turnover, as indicated by lower serum concentrations of aminoterminal propeptide of type 1 collagen.(33) This was accompanied by a lower urinary excretion of both sodium and calcium, but there were no changes in calcitropic hormones. Conversely, in younger men and women (21–39 yr of age) given a low salt diet for 7 wk, there were no detectable changes in markers of bone metabolism.(34) However, this was a parallel design study, which is not as robust as the cross-over design used in this study.

In conclusion, it seems that both dietary calcium and sodium play a major role in the maintenance of bone health in postmenopausal women. Low calcium intake (518 mg/d) was associated with negative bone calcium balance with both high and low salt diets, but with a moderately high calcium intake (1284 mg/d), the bone balance was positive when the salt intake was low (3.9 g/d) but not when it was moderately high (11.2 g/d).


The authors thank Angela Twaite and Veronica Kellas for assistance with food deliveries, sample collections, and processing, and John Eagles for mass spectrometry analysis. We are also grateful for the care and supervision of the volunteers provided by the staff of the IFR Human Nutrition Unit (Aliceon Blair, Lesley Maloney, Linda Oram, Yvonne Clements, and Nicola Hewitt). SFT and BT were responsible for the study design and manuscript preparation (with JRD) and the conception and funding of the study in collaboration with AF and KDC; BT supervised the study; CAAS, GM-N and DJB undertook the human intervention and the majority of sample preparation and biochemical analysis; JAH was responsible for the mass spectrometric analysis; JRD undertook the modelling and stable isotope calculations; JJ carried out the vitamin D and PTH analysis; KDC and AF supervised the bone biomarker analysis; RJF provided statistical support. This project was funded by the UK Food Standards Agency, the European Commission Quality of Life Fifth Framework Programme QLK1-CT 1999-00752 and the Biotechnology and Biological Sciences Research Council.