Relationship of Blood Lead Levels to Incident Nonspine Fractures and Falls in Older Women: The Study of Osteoporotic Fractures

Abstract

Lead is stored in the skeleton and can serve as an endogenous source for many years. Lead may influence the risk of fracture, through direct effects on bone strength or indirectly by disturbing neuromuscular function and increasing the risk of falls. The objective of this analysis is to test the hypothesis that women with higher blood lead levels experience higher rates of falls and fracture. This was a prospective cohort study of 533 women 65–87 yr of age enrolled in the Study of Osteoporotic Fractures at two U.S. research centers (Baltimore, MD; Monongahela Valley, PA) from 1986 to 1988. Blood lead levels (in μg/dl) were measured in 1990–1991 by atomic absorption spectrophotometry and classified as “low” (≤3; lower 15th percentile, referent); “medium” (4–7); or “high” (≥8; upper 15th percentile). Total hip BMD was measured by DXA twice, 3.55 yr apart. Information on falls was collected every 4 mo for 4 yr. Incident nonspine fractures were identified and confirmed over 10 yr. Cox proportional hazards models were used to estimate the hazard ratio (HR) and 95% CI of fracture. Generalized estimating equations were used to calculate the incident rate ratio of falls (95% CI). The mean blood lead level was 5.3 ± 2.3 (SD) μg/dl (range, 1–21 μg/dl). Baseline BMD was 7% lower in total hip and 5% lower in femoral neck in the highest compared with lowest blood lead group (p < 0.02). Hip bone loss tended to be greater in the high lead group, but differences were not significant. In multivariable adjusted models, women with high blood lead levels had an increased risk of nonspine fracture (HR = 2.50; 95% CI = 1.25, 5.03; p trend = 0.016) and higher risk of falls (incident rate ratio = 1.62; 95% CI = 1.07, 2.45; p trend = 0.014) compared with women with lowest lead level. Blood lead levels are associated with an increased risk of falls and fractures, extending the negative health consequences of lead to include osteoporotic fractures.

INTRODUCTION

The skeleton is the major repository for lead, sequestering up to 95% of lead, and can serve as an endogenous source for many years after exposure.(1) Lead is contained in the mineral matrix of bone in close association with calcium, sharing many physical and chemical characteristics and intracellular pathways.(2,3) Lead can compete with calcium in intestinal absorption and deposition into bone. Analogous to calcium, lead absorption from the intestine is regulated by PTH, 1,25-dihydroxyvitamin D, and other factors.(4) Lead may exert both direct (osteoblast and osteoclast function) and indirect (kidney dysfunction) actions on bone turnover.(5)

Lead may be mobilized from the skeleton during conditions of high bone turnover, such as pregnancy, lactation, menopause, and aging.(4) The U.S. population-based NHANES II dataset (n = 20,322), compiled between 1976 and 1980, showed that there was a highly significant increase in both whole blood and calculated plasma lead after menopause independent of age,(4) providing evidence that bone lead is mobilized into blood during conditions of bone demineralization. Menopausal women had adjusted median blood lead levels that were 25% higher than those of premenopausal women. Current use of hormone replacement therapy was associated with significantly lower adjusted median blood lead levels.(6)

Elevated blood lead levels may have a direct causative role in the pathogenesis of osteoporosis.(3,7) Animal studies have shown an association between increased bone lead concentration and decreased mechanical strength.(8–12) In animal models, lead inhibits axial bone development and decreases bone mass because of enhanced resorption.(9) An inverse relationship has also been documented between elevated blood lead levels and skeletal development, chest circumference, and stature in children.(13–15) Lead exposure has been reported to accelerate bone maturation by inhibiting the effects of PTH-related peptide.(16) Accelerated maturation of bone may ultimately result in a lower peak BMD, thus predisposing to osteoporosis in later life.(16) A hypothetical “lead-exposed female” may never reach her peak BMD, and thus, be at risk for fracture.

In older adults, data are sparse. In NHANES III, there was a significant inverse association between blood lead levels and BMD.(6) Factors related to bone turnover were significant predictors of blood lead level.(6) In a study of former workers from a lead smelting factory, there was a significant inverse association between spine BMD and blood lead levels in women.(17) In contrast, a recent study of 112 postmenopausal women reported higher lumbar spine BMD with higher tibia lead levels. There was no relationship between BMD and blood lead or patella lead levels.(18)

In addition to direct effects on bone, lead exposure could predispose to fractures because of increased risk of falls. In adults, environmental lead exposure had been associated with cognitive decline,(19) essential tremors,(20,21) slower response time,(22,23) sensorimotor delay,(24) decreased motor speed and muscle strength,(25,26) depression and mood changes,(25,27) coordination,(22) impaired balance,(20,21) and decreased visual acuity.(28) Compromise in these complex neuromuscular, neurological, and sensory characteristics may disturb coordination(29) and predispose to falls.(30)

In this study, we examined the association between blood lead levels and BMD, changes in BMD, falls, and fractures in a subset of 533 women enrolled in the Study of Osteoporotic Fractures (SOF). We hypothesized that women with higher blood lead concentrations will have a greater risk of falls and fractures.

MATERIALS AND METHODS

Study population

The SOF is a longitudinal cohort study that enrolled 9704 white women from 1986 to 1988 using population-based listings in Baltimore, MD; Minneapolis, MN; Portland, OR; and the Monongahela Valley near Pittsburgh, PA. To be eligible to participate, women had to be ≥65 yr of age and ambulatory.

An ancillary study of blood lead levels was conducted in 1990–1991 in a consecutive sample of 533 white women 65–87 yr of age enrolled in SOF at either the University of Pittsburgh (Monongahela Valley near Pittsburgh) or University of Maryland (Baltimore) clinics. Initially, we examined the correlates of blood lead(22) and the association of blood lead to cognitive function.(31) We found a relationship between higher blood lead and worse cognitive function as measured by the Part B Trailmaking Test, but this association was confined to the rural (Monongahela Valley) clinic.(22,31) In this paper, we extend the lead study to outcomes of BMD, rates of decline in BMD, and incident fractures and falls. The protocol and consent forms were approved by the institutional review boards at the participating institutions. All women provided written informed consent.

Blood lead measurements

A 5.0-ml sample of whole blood was drawn into vacutainer tubes (BD Vacutainer Systems, Rutherford, NJ, USA). Blood samples were analyzed at the Clinical Chemistry Laboratory of the University of Maryland, certified for the analysis of lead in blood by the Occupational Safety and Health Administration and Centers for Disease Control and Prevention, and documents a lower limit of detection for lead of 1 μg/dl. Blood lead was determined by graphite furnace atomic absorption spectrometry (AAS model 5100; HGA with Zeeman Effect Background Correction; Perkin Elmer, Norwalk, CT, USA). To study intra-laboratory variability in both the measurement of lead and the stability of lead over 1 yr, additional blood samples were collected from a random sample of participants in SOF.(22) Two tubes of blood were drawn from each of the 100 randomly selected women (50 women from each clinic) during a later clinic visit. The laboratory was blind with respect to the second blood sample and therefore the intra-laboratory variability in both the measurement of lead and the stability of lead over time was determined. The intraclass correlation coefficient for the duplicates was 0.88. Mean values of 4.76 (range, 1–13 μg/dl) and 4.67 μg/dl (range, 1–12 μg/dl) were obtained for the first and second determinations, respectively.

BMD

BMD of total hip and femoral neck were measured at the second (1988–1990) and at the fourth examination (1993–1994) by DXA using Hologic QDR 1000 scanners (Bedford, MA, USA) an average of 3.55 yr apart. Calcaneal BMD was measured at the baseline visit (1986–1988) and at the fourth visit (1993–1994), with OsteoAnalyzers (Siemens-Osteon, Wahiawa, HI, USA) using single-photon absorptiometry at the baseline examination and single X-ray absorptiometry (Osteoanalyzer; Dove Medical Systems) at the fourth examination (mean follow-up = 5.7 ± 0.33 yr). Annual percent decline in BMD was estimated for the femoral neck, total hip, and calcaneus. A random sample of scans was reviewed by the UCSF Coordinating Center. To assess longitudinal performance of the scanners, a spine phantom was scanned daily, and a hip phantom was scanned once per week at each clinic. Details of these measurement methods and densitometry quality control procedures have been published.(32,33) The mean CV for the femoral neck between centers was 1.2% for two research staff who visited all centers. The intraindividual region-specific correlation coefficients between hip measurements at the second and fourth examinations (3.55 ± 0.29 [SD] yr between examinations) were 0.96 (total hip) and 0.94 (femoral neck). The in vivo mean CV for the calcaneus was 1.2% between centers; the intraindividual correlation coefficient between calcaneus measurement at baseline and fourth examinations was 0.94%.

Other measurements

Potential confounders of the association between blood lead and outcomes of interest were identified based on previous lead(22,31) and BMD and fracture analyses(34–37): sociodemographic factors (age, study site/clinic), education in years, smoking history, alcohol consumption, medication use (including hormone therapy [HT], calcium and vitamin D supplements, central nervous system [CNS] active medications [benzodiazepines, antidepressants, anticonvulsants, and narcotics], health status compared with others, history of fractures after age 50, and falls over the previous year). Height and weight were obtained using a Harpenden stadiometer (Holtain, Crymych, UK) and a standard balance beam, respectively, and body mass index (BMI) was calculated as weight divided by height squared (kg/m²).

Cognitive function was tested with a modified form of Mini-Mental (MMSE) status examination,(38,39) a global measurement of cognitive function, with components for orientation, concentration, language, praxis, and immediate and delayed memory. The modified MMSE score ranges from 0 to 26, with higher scores representing better cognitive functioning.(40) The Trails B is a timed test that measures attention, sequencing, visual scanning, and executive function. A faster time for completion (in seconds) represents better cognitive functioning.(41)

Neuromuscular assessment was comprised of muscle strength, balance, gait, coordination, and reaction time. Grip strength was recorded as the average of the right and left sides using an adjustable hand grip dynameter (Preston Grip Dynameter; Takei Kiki Kogyo, Tokyo, Japan).(42) The chair stand test determined whether the participant could stand up five times from a chair without using the arms.(43) Static balance was measured as the ability to hold tandem stands with eyes open for 10 s.(44) Gait speed was assessed as average of two trials of walking speed at usual pace (m/s).

The groove pegboard test is a measure of manual dexterity and the speed and accuracy of hand-eye coordination. Time to insert and remove pegs in a board with the dominant hand was recorded in seconds. Reaction time was tested on the foot and hand of the dominant side with a two clock response and movement time device (Lafayette Instrument) that measures the speed of reactions and the speed of movement of hands and feet on the dominant side in 1/1000 of a second.(45) Hand and foot reaction time were highly related (r = 0.83). Corrected vision was measured as visual acuity 20/40 or worse (both eyes) on Bailey-Lovie acuity targets.(46,47)

Vibration threshold was measured with a Vibratron-II (Sensortek) to quantify the ability to detect vibratory stimuli at right big toe. The Vibration threshold was calculated as the average of the vibration settings of the five errors and five lowest scores.(43) Light touch discrimination was measured twice on both big toes using a Von-Frey type esthesiometer probe. The thinnest filament that was identified correctly was recorded both times.(43) An increase in these performance measures indicates a decline in function.

Fracture/falls ascertainment

After the initial enrollment visit, all women were contacted by either postcard or telephone every 4 mo to determine whether any fractures or falls had occurred in the preceding 4 mo. After >10 yr, follow-up remains >95% complete. Women who reported a fracture were interviewed by telephone about the circumstances under which the fracture occurred. Fractures occurring from major trauma (e.g., motor vehicle crash) were excluded. All fractures were confirmed by radiographic report or review of the X-rays. Data were prospectively collected on the number of falls for 4 yr after baseline visit. On each postcard, participants were asked if they had fallen in the past 4 mo and, if so, how many times. Participants were contacted by phone when postcards were not returned.

Statistical analyses

Analyses were performed by categorizing the study participants into three groups of blood lead corresponding to the upper and lower 15th percentiles of the distribution of blood lead. Thus, the three groups were low (≤3 μg/dl, median = 3 μg/dl; n = 122); medium (4–7 μg/dl, median = 5 μg/dl; n = 332); and high (≥8 μg/dl, median = 9 μg/dl n = 79). This categorization was determined a priori based on our previous study of blood lead and cognitive function.(22)

The χ² test was used to compare baseline characteristics for categorical variables by blood lead status; ANOVA was used for continuous variables. Two-tailed p values were used for all tests at 5% statistical significance. We plotted the cumulative incidence of fracture in the three groups by Kaplan-Meier survival curves. We used Cox proportional hazard regression analysis to estimate the hazard ratio (HR) of fracture and 95% CIs across the three blood lead groups. The relative risk (RR) of falls was computed using Poisson regression models with generalized estimating equations (GEEs) to adjust SE for correlated data points at 4-mo intervals. A Huber White Sandwich estimator of variance was used with GEE to construct valid standard errors. The incident rate ratio (IRR) was computed as fall rate (average number of falls per 4 mo) in a specific category of blood lead compared with the lowest (referent) category. In multivariate models, we simultaneously analyzed blood lead levels and other potential risk factors. A forward stepwise selection process was used to add or remove potential covariates from the multivariate regression models (exit criteria: p ≥ 0.15) until a final model of significant variables (p < 0.05) was selected. We initially adjusted for age and clinic. In multivariate fracture models, we adjusted for age, clinic, education, BMI, alcohol use, baseline total hip BMD, prior fracture, health status, falls in past year, smoking, estrogen use, and calcium and vitamin D use. The initial multivariable model for falls included age, clinic, education, BMI, alcohol use, health status, smoking, falls in past year, CNS active medications, estrogen use, and calcium and vitamin D use. To test whether these associations were mediated by muscle strength, neuromuscular function, balance sensory, or cognitive function, we adjusted for grip strength, lower limb reaction time, fine motor coordination, visual acuity, balance, vibration and touch sensitivity, chair stands, walking speed, and Trial B. In secondary analysis, we determined the HR of fracture and falls per SD increase in blood lead. The proportionality assumptions of the Cox models were confirmed with Schoenfeld residuals. Data were analyzed with Stata (edition 9; StataCorp, College Station, TX, USA).

RESULTS

The mean blood lead was 5.3 ± 2.3 (SD) μg/dl (range, 1–21 μg/dl). Women in the highest blood lead category (≥8 μg/dl) had lived more years after menopause, were less likely to use estrogen, and were more likely to smoke and drink alcohol and less likely to take vitamin D supplements (Table 1). Blood lead levels were significantly lower in women who reported current use of estrogen (4.6 μg/dl) compared with past users (4.8 μg/dl) or nonusers (5.3 μg/dl; p < 0.05). Cognitive function, health status, muscle strength, balance, gait speed, vibration or touch sensitivity, reaction time, fine motor coordination, or visual acuity did not differ across blood lead categories. Women with highest blood lead had lower mean baseline total hip (7%; p < 0.02) and femoral neck (5%; p < 0.03) BMD compared with women with lowest blood lead category (Table 1). Baseline calcaneal BMD did not differ across the three groups.

Table 1.

Baseline Characteristics of Women in SOF Across Blood Lead Level

	Low	Medium	High	p
Level (μg/dl)	≤3	4–7	≥8
N = 533	N = 122	N = 332	N = 79
Age [yr; mean (SD)]	70.3 (4.3)	70.4 (4.4)	71 (4.5)	0.411
Education [yr; mean (SD)]	12 (3)	12.5 (3)	12.6 (3)	0.360
Years since menopause [mean (SD)]	21 (8)	23 (7)	25 (8)	0.017
Age at menopause [yr; mean (SD)]	49 (6)	48 (6)	46 (6)	0.011
Ever breastfed [n (%)]	66 (24)	181 (66)	29 (11)	0.014
Current estrogen use [n (%)]	22 (18)	33 (10)	4 (5)	0.030
Body mass index [kg/m2; mean (SD)]	27 (5)	27 (5)	26 (4)	0.125
Current smoker [n (%)]	7 (11)	39 (59)	20 (30)	<0.001
Alcohol [drinks/wk; mean (SD)]	1 (3)	2 (4)	3 (6)	0.002
Calcium intake [mg/d; mean (SD)]	674 (357)	654 (399)	653 (513)	0.886
Vitamin D supplement [n (%)]	50 (42)	114 (35)	18 (23)	0.052
No of years Vit D use [mean (SD)]	13 (15)	10 (11)	5 (7)	0.019
Self reported health [n (%) excellent/good]	99 (81)	282 (85)	66 (83)	0.493
Any fracture after age 50 [n (%)]	40 (33)	109 (33)	22 (28)	0.683
Fallen in past year (≥1) [n (%)]	29 (24)	75 (23)	20 (25)	0.873
CNS medication use [n (%)]	31 (25)	52 (16)	14 (18)	0.058
Cognitive function
m MMSE [mean (SD)]	25 (1.6)	25 (1.4)	25 (1)	0.185
Trails B [s; mean (SD]	116 (36)	121 (39)	126 (44)	0.192
Muscle strength
Average grip strength both sides [kg; mean (SD)]	20.6 (3.8)	20.9 (4.2)	20.7 (4.1)	0.756
Use of arm for chair stand five times [n (%)]	1 (11)	8 (89)	0 (0)	0.230
Balance: poor /very poor [n (%)]	40 (23)	111 (63)	25 (14)	0.953
Gait: walk speed [m/s; mean (SD)]	1.01 (0.2)	1.04 (0.2)	1.03 (0.2)	0.483
Somatosensory discrimination
Light touch sensitivity [mm; mean (SD)]*	4.1 (0.5)	4.1 (0.4)	4.0 (0.4)	0.752
Vibration threshold [units; mean (SD)]*	21 (19)	22 (18)	23 (24)	0.837
Reaction time (number of seconds)
Hand [mean (SD)]*	0.63 (0.1)	0.65 (0.2)	0.65 (0.2)	0.723
Foot [mean (SD)]*	0.64 (0.1)	0.66 (0.2)	0.66 (0.2)	0.499
Coordination
Groove peg board [number of seconds; mean (SD)]*	91 (27)	93 (30)	95 (25)	0.679
Vision acuity 20/40 or worse both eyes [n (%)]	13 (15)	57 (66)	48 (7)	0.100
BMD (g/cm2)
Total hip [mean (SD)]	0.77 (0.13)	0.76 (0.13)	0.72 (0.12)	0.021
Femoral neck [mean (SD)]	0.65 (0.11)	0.66 (0.12)	0.62 (0.09)	0.033
Calcaneus [mean (SD)]	0.41 (0.09)	0.42 (0.09)	0.39 (0.09)	0.145

* An increase in these performance measures indicates a decline in function.

Change in BMD

Women with highest blood lead experienced a greater rate of decline in hip BMD (Table 2), but the association was not statistically significant. The annualized rate of decline in calcaneal BMD was higher for women in the higher blood lead category compared with those in lower categories in age- and multivariable-adjusted models.

Table 2.

Age and Multivariate-Adjusted Annualized Percentage Rate of Bone Loss Across Blood Lead Levels

BMD decline (%/yr)	Blood lead level						p trend
	Low (≤3 μg/dl) (n = 122)		Medium (4–7 μg/dl) (n = 332)		High (≥8 μg/dl) (n = 79)
	Mean	95% CI	Mean	95% CI	Mean	95% CI
Total hip
Age, clinic adjusted	−0.31	−0.61, −0.01	−0.46	−0.64, −0.28	−0.81	−1.21, −0.41	0.14
MV adjusted*	−0.34	−0.61, −0.06	−0.41	−0.58, −0.26	−0.63	−1.00, −0.25	0.50
Femoral neck
Age, clinic adjusted	−0.11	−0.50, −0.27	−0.57	−0.80, −0.34	−0.76	−1.27, −0.25	0.07
MV adjusted*	−0.21	−0.60, 0.18	−0.59	−0.82 −0.36	−0.59	−1.12, −0.05	0.25
Calcaneus
Age, clinic adjusted	−1.02	−1.31, −0.73	−1.50	−1.68, −1.32	−1.65	−2.04, −1.26	0.01
MV adjusted*	−1.01	−1.27, −0.74	−1.41	−1.57, −1.24	−1.49	−1.86, −1.10	0.03

* The multivariate model included the following: age, clinic, BMI, weight change between visit 2 and visit 4, smoking, chair stands, fracture history, estrogen use, and baseline BMD.

Fractures

After an average follow-up of 10.5 yr, 163 (31%) women reported an incident fracture. The cumulative incidence of fracture increased over time and differed by blood lead. Women with the highest lead had the highest cumulative incidence of fractures (log-rank test = 0.038; Fig. 1). In age- and clinic-adjusted models, women with the highest blood lead had an increased risk of fracture (HR = 1.61; 95% CI = 0.97–2.66; Table 3). In multivariable-adjusted models, women with the highest blood lead had a significant increased risk of fracture in comparison with women with the lowest blood lead (HR = 1.94; 95% CI = 1.13, 3.32; p trend = 0.024). Further adjustment for neurological and neuromuscular variables increased the HR (HR = 2.50, 95% CI = 1.25, 5.03; p trend = 0.016). Women in the medium blood lead category were not at an increased risk of fracture. Using blood lead as a continuous variable in multivariable-adjusted models, the RH of fracture per SD increase in blood lead was 1.17 (95% CI = 1.00, 1.37; p = 0.039).

FIG. 1.

Cumulative incidence of fractures in women in SOF across blood lead levels.

Table 3.

Age and Multivariate-Adjusted HRs (95% CIs) of Incident Fractures and Incidence Rate Ratios of Falls by Blood Lead Levels

	Blood lead levels (μg/dl)
	Low (≤3) (N = 122)	Medium (4–7) (N = 332)		High (≥8) (N = 79)
HR of incident nonspine fracture	Referent	HR	95% CI	HR	95% CI	ptrend
Age, clinic adjusted	1.0	1.10	0.74–1.64	1.61	0.97–2.66	0.080
MV adjusted*	1.0	1.16	0.76–1.75	1.94	1.13–3.32	0.024
MV adjusted†	1.0	1.13	0.65–1.95	2.50	1.25–5.03	0.016

Incidence rate ratio (RR) of falls	Referent	IRR	95% CI	IRR	95% CI	ptrend
Age, clinic adjusted	1.0	1.14	0.87–1.57	1.40	0.97–2.02	0.073
MV adjusted‡	1.0	1.34	1.00–1.79	1.72	1.17–2.52	0.005
MV adjusted§	1.0	1.41	1.00–2.01	1.62	1.07–2.45	0.014

Total N = 533.

* The fracture model included age, clinic, education, BMI, alcohol use, baseline total hip BMD, fractures after age 50, health status, fall in past year, smoking, estrogen use, and calcium and vitamin D use.

^† Further adjusted for grip strength, reaction time lower limb, groove pegboard score, vibration and touch, visual acuity, walk speed, chair stands, and trail making B score.

^‡ The falls model included age, clinic, education, BMI, alcohol use, health status, smoking, falls in past year, CNS active medication, estrogen use, and calcium and vitamin D use.

^§ Further adjusted for grip strength, reaction time lower limb, groove pegboard score, vibration and touch, visual acuity, walk speed, chair stands, and trail making B score.

Falls

During the 4-yr follow-up period, the risk of falls was higher among women with higher blood lead (Table 3). In age- and clinic-adjusted models, women with the highest blood lead had a 40% increased risk of fall. The trend across groups was statistically significant in the multivariable-adjusted model where women in the highest blood lead category were at 72% increased risk of falling compared with women with the lowest lead levels (IRR = 1.72; 95% CI = 1.17–2.52; p trend = 0.005). Women in the medium blood lead category were at 34% increased risk of falls (IRR = 1.34; 95% CI = 1.00, 1.79; p = 0.050). Further adjustment for neurological and neuromuscular variables did not change the significance of our results (IRR = 1.62; 95% CI: 1.07, 2.45; p trend = 0.014). In multivariable-adjusted models, the risk of falls increased by 9% per SD increase in blood lead (IRR = 1.09; 95% CI = 1.00, 1.21; p = 0.075).

DISCUSSION

In this cohort of older women, circulating levels of blood lead ≥8 μg/dl were associated with lower total hip BMD, and an increased risk of falls and nonspine fractures. We found an association with fractures and falls at blood levels ≥8 μg/dl (median = 9 μg/dl), a level previously thought to be safe. The association with nonspine fractures and falls was not explained by poor cognitive or physical function, tests of vibration or touch sensitivity, balance, or visual acuity. Indeed, fractures and falls are multifactorial, and adjustment for other factors that could confound the association tended to strengthen the association with lead.

Although mean U.S. blood lead levels have decreased dramatically over the past 30 yr, as documented by repeated National Health and Nutrition Examination Surveys from 1988 to 2002, lead toxicity continues to be a public health problem for older individuals with higher lifetime environmental lead exposure.(48,49) More specifically, older women, who grew up when environmental exposures were higher because of leaded gasoline, paint, and unregulated industrial lead use, may have absorbed more lead from the environment and thus have a higher overall body burden of stored lead.(1)

We studied blood lead levels that reflect both recirculation from endogenous sources as well as more recent exposures. Studies report that skeletal contribution to blood lead levels ranges from 40% to 70% in adults.(50–54) In the Normative Aging Study of older men, blood lead was moderately correlated to bone lead levels for both trabecular and cortical bone (rho = 0.43, p < 0.01 for patella and rho = 0.34, p < 0.01 for tibia).(55) It has been proposed that lead released from the skeleton is preferentially partitioned into plasma and maybe more readily bio-available.(56) In middle-aged to elderly women (47–74 yr), an increase of 19 μg/g of lead in tibia was associated with an increase in blood lead level of 1.7 μg/dl, which corresponds to a 0.09-μg/dl increase in blood lead per 1 μg/g of bone mineral.(57)

In our study, baseline hip BMD levels but not calcaneal BMD were lower among women with higher blood lead levels. BMD was also significantly inversely associated with blood lead in the general population after adjusting for other factors associated with blood lead.(6,58,59) We found that greater lead levels were associated with faster rates of hip bone loss, but the results were not statistically significant. On the other hand, the association between blood lead and calcaneal bone loss was statistically significant whereby women in the high blood lead group experienced a greater rate of calcaneal bone loss than women in the low blood lead group. The calcaneus contains almost exclusively trabecular bone, and our findings are consistent with a higher rate of bone turnover in trabecular bone, leading to higher blood lead levels.(60)

Previous research, although limited, have not shown consistent relationships between lead and skeletal outcomes. Longitudinal analyses of postmenopausal women with occupational lead exposure showed that biomarkers for bone resorption were significantly higher in women who had higher blood and calcaneus bone lead.(61,62) Vertebral BMD was significantly inversely associated with blood lead even 10–18 yr after occupational exposure.(5,61–63) A more recent finding showed higher lead with higher bone lead but no relationship to blood lead.(18) Taken together, more research is clearly needed to further our understanding of the relationships of lead to skeletal outcomes.

High bone turnover has been associated with increased risk of fracture.(64) The association between lead and fracture may be caused by a direct effect of lead on bone strength or may simply be a marker of high bone turnover. Small studies of bisphosphonates(50) and hormone therapy(5) showed that treatments that reduce bone turnover resulted in a lower skeletal lead release.(50) In our study, women who reported using estrogen had lower lead levels. Thus, antiresorptive treatments could reduce lead exposure in older adults.

A genetic polymorphism has been identified in the vitamin D receptor gene (VDR) that can influence the accumulation of lead in bone.(65) A negative correlation has also been observed between lead and vitamin D.(3) Low vitamin D has been linked to hip and nonspine fractures and falls.(66,67) In our study, women with the highest blood lead were less likely to take vitamin D supplements at baseline in comparison with women with lower blood lead, suggesting that circulating vitamin D levels may also be lower in this group. Moreover, we observed a significant inverse association in blood lead level and number of years women took vitamin D supplements (p < 0.019). We were, however, unable to adjust for circulating vitamin D levels.

Lead exposure could predispose to fractures because of increased risk of falls because it affects neuromuscular and sensory function and balance.(68,69) Blood lead levels as low as 8 μg/dl were associated with poorer performance on tests of visual perception,(70) psychomotor speed, and manual dexterity.(22,70) Neuromotor dysfunction in lead workers has been reported at average blood lead levels of 14.4 μg/dl.(71) Using computerized static posturography, the effect of lead on postural balance was examined in 49 male lead workers, with mean (range) lead levels of 18 (7–36) μg/100 g. Compared with healthy non–lead-exposed controls, lead-exposed workers had significantly greater postural sway.(72) Thus, our finding of an association between blood lead and falls is consistent with these biological mechanisms. Although our relatively crude measures of balance and sensory and motor function did not differ across lead groups, adjustment for these factors tended to strengthen these associations. Future research should consider more sensitive tests (e.g., of postural sway rather than balance [by a tandem stand]) to further evaluate the effects of lead.

There are a number of strengths to our study. We used a prospective study design and studied a well-characterized cohort of community-dwelling older women. We categorized blood lead levels based on previous published results.(22) Follow-up was nearly complete, and all fractures were validated by review of radiology reports and/or medical records. However, this study does have some limitations. Our participants were likely to be healthier than average community-dwelling older women because they were volunteers. We measured blood lead only once but our approach is similar to other prospective studies of sex hormones predicting fractures.(73) We had no information on occupational exposures but we adjusted for a number of important covariates, including cognitive, neuromuscular function, and sensory. BMD measurements could seem artificially higher in lead-exposed individuals because of co-contamination with cadmium or strontium.(74) BMD increased with increasing bone lead levels, especially when BMD was measured on old pencil beam scanners, similar to the scanners used in our study.(75,76) Thus, our ability to adjust for BMD and test whether the association with fracture is independent of BMD may be limited. Finally, we used blood lead as the measure of lead exposure variable, but blood lead reflects only 40–70% of the skeletal lead burden. Future studies should include measures of skeletal lead, a better measure of the cumulative dose of lead. Finally, our results may not be generalizable to older nonwhite women or to men.

In conclusion, high blood lead levels were associated with an increased risk of falls and fractures in older women extending the negative health consequences of lead to include osteoporotic fractures.