Abstract was presented at the 2002 Annual Meeting of the International Society for Environmental Epidemiology, Vancouver, Canada, August 8, 2002.
The authors have no conflict of interest
The aim of this study was to analyze the relationship between low-level cadmium exposure and distal forearm fractures. Altogether, 1021 men and women exposed to cadmium in Sweden were included. The study indicates that cadmium exposure is associated with increased risk of forearm fractures in people over the age of 50.
Introduction: Very few studies have been performed on environmental risk factors for fractures. Cadmium is known to cause damage to the kidneys and in high doses to the bone. The aim of this study was to analyze the relationship between low-level cadmium exposure and distal forearm fractures.
Materials and Methods: A total of 479 men and 542 women, 16-81 years of age, that were environmentally or occupationally exposed to cadmium were examined in 1997. Cadmium in urine was used to estimate dose, and information about previous fractures and risk factors for fractures was obtained from questionnaires. Fractures were validated using medical records. The association between cadmium dose and risk of forearm fracture was evaluated using Cox proportional hazard regression analysis.
Results and Conclusion: The mean urinary cadmium in the study population was 0.74 nmol cadmium/mmol creatinine (10% and 90% percentiles are 0.19 and 1.42, respectively). For fractures occurring after the age of 50 years (n = 558, 32 forearm fractures), the fracture hazard ratio, adjusted for gender and other relevant co-variates, increased by 18% (95% CI, 1.0-38%) per unit urinary cadmium (nmol cadmium/mmol creatinine). When subjects were grouped in exposure categories, the hazard ratio reached 3.5 (90% CI, 1.1, 11) in the group of subjects with urinary cadmium between 2 and 4 nmol/mmol creatinine and 8.8 (90% CI, 2.6, 30) in the group of subjects with ≥4 nmol/mmol creatinine. Associations between cadmium and fracture risk were absent before the age of 50. Cadmium exposure is associated with increased risk of forearm fractures in people over 50 years of age.
FRACTURES CAUSED BY osteoporosis are a worldwide problem, and the incidence is increasing.(1) One of the more severe health effects of osteoporosis is hip fracture. However, only one-third of the fractures in the elderly are hip fractures, and a comparable number occur in the arm, most often in the forearm.(2) Arm fractures result in only a modest amount of morbidity compared with hip fractures, but they cause pain and account for considerable health care costs.(3) Persons with antecedent forearm fracture are also more prone to subsequent hip fracture.(4,5)
Menopause,(1) low BMD, height loss, low weight, low dietary calcium intake, and falls(6,7) are generally considered as risk factors for upper limb fractures. It has long been debated whether fluoridation of drinking water could be associated with an excess fracture risk, but this is not supported by recent research.(8,9) Besides water fluoridation, epidemiological studies have rarely investigated the associations between fractures and environmental exposure.
It is well known that cadmium in high concentrations may cause osteoporosis and osteomalacia, the classical example being Itai-itai disease in Japan.(10) Cadmium-contaminated rice was ingested by local farmers, and many hundreds, mostly elderly women, suffered from pathological fractures. A Belgian population study suggested that even low-level cadmium exposure may increase bone fragility and fracture risk.(11) Animal experiments have shown that administration of cadmium reduced the mechanical strength of the bone.(12–14)
Nickel-cadmium and lead batteries have been produced since 1912 in two plants in southeastern Sweden. Cadmium and lead emissions from these plants have been substantial in the past.(15) We have previously showed that low-level cadmium exposure may increase the risk of low BMD.(16,17) Moreover, the association between BMD and cadmium-induced tubular proteinuria, a sign of early kidney damage, suggested pathways for the pathogenesis.(16)
In this study, the population living in the vicinity of one of these plants was examined with respect to the associations between cadmium exposure and forearm fracture and between cadmium-induced early renal damage and forearm fracture. Associations between lead exposure and forearm fracture were also investigated.
MATERIALS AND METHODS
A total of 1465 subjects, 16-80 years of age in 1997, who had resided in the vicinity of one of the battery plants for at least 5 years between 1910 and 1992 were asked to participate in the study, and 904 of them (62%) agreed to do so. Vicinity was defined as people who had lived in the parish where the battery plant was situated. This means up to 10 km away from the battery plant. However, most people lived within 1 km of the battery plant. A number of workers with previous occupational exposure from the two battery plants were also included. Of the 242 occupationally exposed workers, 117 (48%) agreed to take part in the examinations. Thus, a total of 1021 individuals (60%) agreed to participate in the study and gave their informed consent to the investigation. A telephone survey of a random sample (5%) of the nonparticipants was performed to examine whether they differed from the examined group with regard to age, gender, or morbidity.
Each participant received a questionnaire including questions about employment, residences, smoking and food habits, and medical history, including fractures.
All but three of the participants answered the forearm fracture history question. To discriminate between nonadult and adult fractures, only the fractures that had occurred at the age of 20 or later were taken into account. Therefore, all persons younger than 20 years of age were excluded (five men and seven women) from all analyses. In the study area, there is one major hospital, where fractures occurring in that region are treated. The fractures reported in the questionnaires were validated using X-ray and medical records from this hospital. A random sample of 40 participants that had not reported any fractures was also checked against the medical records. Civic registration numbers (unique for each Swede) were used to trace the patient records.
An estimate of dietary calcium intake was calculated from the reported consumption of dairy products (100 ml milk = 120 mg calcium, 1 slice cheese = 87 mg calcium).(18)
Smokers were classified into two categories: (1) never or (2) former/current smokers. Former or current smokers had been smoking regularly for ≥1 year.
In 1997, specially trained nurses collected urine and blood samples and measured height and weight. Morning urine was voided in acid-washed polyethylene bottles. The blood and urine samples were stored frozen (−20°C) until transfer to the analytical laboratory at the Department of Occupational and Environmental Medicine at the Lund University Hospital. Cadmium in urine and lead in blood was determined by inductively coupled plasma mass spectrometry (ICP-MS). The method accuracy was checked using commercial reference samples.
For the workers from the battery plants, measurements of urinary cadmium from 1976 to 1980 were also available. These earlier measurements of cadmium had been performed at the Department of Hygiene at Karolinska Institutet with quality controls fulfilling WHO standards.
Protein HC (human complex-forming glycoprotein, also called α1-microgloubulin) was used as a marker for early renal tubular damage. The analyses were made at the Department of Clinical Chemistry at the University Hospital in Lund. Adjustment for variation in urinary concentrations between individuals was made by dividing the urinary cadmium and protein HC values by the creatinine concentrations.(19) The methods are more fully described in an earlier paper.(16)
The associations between cadmium in urine, blood lead, and tubular proteinuria and the risk of forearm fracture in the present data were analyzed using logistic regression analysis and Cox proportional hazards regression analysis, which may be applied to retrospective incidence data to take full advantage of the information on the dates of fracture occurrence. Current cadmium in urine and lead in blood were used as proxies for the cadmium and lead dose, respectively, at the time before the fracture. The analyses were done with or without adjustment for gender, weight, dietary calcium intake, smoking, and occupational exposure. The follow-up period was defined as the time period of being at risk for an adult forearm fracture (i.e., from age 20 or from age 50 when a constraint on attained age was implemented) and ending with the first reported forearm fracture or in 1997 at the latest if no fracture occurred. All statistical analyses were performed using STATA 7.0 software.
Three study participants did not report whether they had had a fracture or not and were therefore excluded from the analyses. One hundred thirteen study subjects reported a forearm fracture. Sixty-eight of these fractures were found in medical records, and 56 were confirmed to be forearm fractures. The 12 other fractures proved to be 9 fractures of the carpus or metacarpus, 2 of the elbow, and 1 forearm fissure. For 38 of the 45 fractures not found in the medical records, the year of the fracture had been given in the questionnaire. Seven subjects for whom it was not possible to allocate a year of fracture were excluded from the analyses. Thus, it was possible to obtain information about the year of the fracture for a total of 94 fractures. Sixty-three of the subjects had their forearm fracture when they were adults (age at fracture ≥ 20 years), and 43 of these were confirmed in the medical records.
During the course of the study. it was discovered that several (n = 105) environmentally exposed people also had worked in the battery plant. These people were therefore transferred to the occupationally exposed group, which thus increased to 222 workers.
Table 1 shows characteristics of the final study population. The mean age for having a forearm fracture was 46 years (range, 21-69 years). The subpopulation comprised of those ≥50 years of age are presented in Table 2. Thirty-two of the subjects had their forearm fractures when they were 50 years or older, and 28 of these were confirmed by the medical records In this subgroup, the mean age for having a forearm fracture was 58 years. No one who had a forearm fracture at the age of 50 or later had had a forearm fracture earlier.
Table Table 1.. Characteristics of Subjects ≥20 Years of Age Who Had, or Had Not, Experienced a Forearm Fracture Above This Age (Validated by Medical Records or Age for Fracture Given)
Table Table 2.. Characteristics of Subjects ≥50 Years of Age Who Had, or Had Not, Experienced a Forearm Fracture Above This Age (Validated by Medical Records or Age for Fracture Given)
Generally, both men and women with forearm fractures were older and had higher levels of urinary cadmium and protein HC than people without fractures. The mean urinary cadmium in the whole study population was 0.74 nmol cadmium/mmol creatinine (10% and 90% percentiles are 0.19 and 1.42, respectively), and in the subgroup 50 years and older, the corresponding numbers are 0.94 nmol cadmium/mmol creatinine (10% and 90% percentiles are 0.25 and 1.78, respectively).
Table 3 presents the unadjusted and the adjusted hazard ratios for forearm fracture at the age of 50 and over. The adjusted hazard ratio for forearm fracture was 1.18 (95% CI, 1.01, 1.37) per unit nmol cadmium/mmol creatinine. Using body mass index (BMI) instead of body weight did not affect the results. Including age at time of cadmium measurement did not change the adjusted fracture hazard ratio in relation to urinary cadmium. If the analyses were made with only the verified fractures included, the results only changed marginally.
Table Table 3.. Hazard Ratio for Forearm Fracture in Relation to Various Characteristics Among Subjects ≥50 Years of Age
For forearm fractures in people <50 years of age, no associations were observed in relation to cadmium dose. Therefore, the cadmium related fracture hazard ratios for different levels of cadmium dose would not be constant over the follow-up time, as assumed in Cox regression analyses, without implementing a restriction on age at risk. For age at risk in people >50 years of age, the proportional hazard assumption was not violated.
Hazard ratios for forearm fractures for different urine cadmium groups, adjusted for gender, are shown in Fig. 1. The cut-off points used were <0.5, 1, 2, and 4 nmol cadmium/mmol creatinine, resulting in five groups with mean urine cadmium 0.30, 0.69, 1.4, 2.9, and 6.0 nmol/mmol creatinine, respectively. The number of subjects in each group was 238, 211, 91, 33, and 17, and the number of fractures in each group was 6, 14, 6, 3, and 3, respectively. A steep increase in fracture risk was observed with a hazard ratio of 3.5 (90% CI, 1.1, 11) for the group of subjects with urinary cadmium between 2 and 4 nmol cadmium/mmol creatinine, and 8.8 (90% CI, 2.6, 30) for the group of subjects with ≥4 nmol cadmium/mmol creatinine.
The hazard fracture ratio in relation to urinary protein HC, adjusted for the same variables as in Table 3, increased by 22% (95% CI, 0.5%, 40%) per increase of 1 mg protein HC/mmol creatinine in urine.
No similar relationships were found between lead levels in blood and the risk of forearm fracture.
The results remained the same when logistic regression analysis was used. Analyzing the occurrence of forearm fracture after the age of 50, including the variables listed in Table 3, as well as age in the model, the odds ratio was 1.20 (95% CI, 0.97, 1.48) per unit nmol cadmium/mmol creatinine.
Earlier measurements from the workers were used to study the relationship between past measurements, made in 1976-1980, and contemporary measurements, made in 1997, used in this study. The Spearman's rank correlation coefficient and the Pearson correlation coefficient were similar, with values close to 0.65. However, urinary cadmium levels were, for most of the workers, twice as high 20 years earlier.(20)
The telephone survey of the random sample including 5% of the nonparticipants gave no indication that they differed from the examined group in a systematic way with regard to age, gender, or fracture incidence.
The random sample of 40 participants who had not reported any forearm fractures and whose information was checked against medical records had no fractures in this location. However, two shoulder fractures and one index finger fracture were found.
Our results show that low-level cadmium exposure may increase the risk of forearm fractures. For the population ≥50 of age, the fracture hazard ratio increased by 18% (95% CI, 1.0%, 37%) per nmol cadmium/mmol creatinine. There was also an increased risk of forearm fractures with increasing urinary protein HC, a marker of cadmium-induced tubular dysfunction.
Bone mass tends to decrease after the fourth or fifth decade of life. Osteoporosis predisposes to fractures in the distal forearm as well as other fractures, especially in the hip and vertebra. The incidence of forearm fractures rises sharply around age 50.(2) Therefore, separate analyses were conducted according to whether age at risk attained 50 years or not.
In the general population in Sweden, nonsmokers have urinary cadmium concentrations of 0.02-0.7 nmol cadmium/mmol creatinine,(21) which is the range of normal cadmium concentrations in most industrialized countries without specific cadmium exposure. Including both environmentally and occupationally cadmium exposed people in the study yielded a wide range of exposure (range, 0.06-18 nmol cadmium/mmol creatinine; mean, 0.74 nmol cadmium/mmol creatinine). Japanese patients with Itai-itai disease had much higher levels, with mean urinary cadmium excretions of nearly 30 nmol cadmium/mmol creatinine.(22)
Cadmium has a very long half-life in the human body (10-20 years), and urinary cadmium is commonly used to assess body burden.(23) In this study, comparisons between measurements of cadmium in the late 1970s and those in 1997 among workers indicated that individual cadmium concentrations were correlated, although the earlier cadmium levels were roughly twice as high as the 1997 levels. The decrease in urinary cadmium in the occupationally exposed workers is most probably due both to the fact that some of the workers have stopped working at the battery plants and that the exposure at the battery plants has decreased substantially over time because of tighter environmental safeguards. Although this applies primarily to the occupationally exposed subjects, it is also possible that the environmentally exposed subjects had somewhat higher cadmium levels when the battery plant still was operating compared with their current levels. The current urinary cadmium concentrations must be regarded as proxies for the concentrations the study participants had when they developed their bone lesions.
It is well known that women have a greater risk for forearm fractures than men,(2) and this was also the case in this study, with a hazard ratio for forearm fracture of 2.9 for women compared with men. Smoking is known to increase cadmium dose(24) and is also regarded as a possible risk factor for osteoporosis and fractures.(2,25–27) A recent meta-analysis(28) showed that smoking increased the risk of all types of fractures, hip and vertebral fractures in particular, whereas the risk increase for forearm fractures was only moderate. Smoking did not show any clear relationships with the risk of forearm fractures in this study. Being overweight is another factor that is known to influence BMD, and in earlier studies, this has been proposed as a protective factor for forearm fractures.(7) Low calcium intake has also been shown to increase the risk of upper limb fractures.(6,7) Other studies have not shown any relationships between weight, dietary calcium intake, and the risk of forearm fractures,(18) and this was also the case in this study.
Age is clearly an important risk factor for fractures. The time scale for the time of fracture in this study was given by the subjects' age, that is, all subjects constituting the risk sets (the subjects with fracture occurrence and all other subjects who had not yet experienced fractures) had the same age at each fracture event. Nonetheless, because the cadmium body burden is likely to increase with age, it may be necessary to account for differences in subjects' ages at the time of urinary cadmium assessment. However, the fracture hazard ratio estimate was not affected by the inclusion of a variable in the model representing the subject's age at the time of urinary cadmium assessment. For a given age at fracture event, this variable also allows a representation of the time elapsed since the time that would have been directly relevant for cadmium assessment and corresponding adjustment. Longer time elapsed may imply an overestimation of relevant cadmium level caused by aging, but high cadmium levels may decrease with time. This adjustment obviously does not correct for uncertain measurements but increases comparability between subjects.
Cut-off points are always arbitrarily chosen. For the dose-response analyses, the lowest cut-off point in this study was set to 0.5 nmol cadmium/mmol creatinine, the lowest level where we have earlier seen effects on bone from cadmium exposure(16); the other cut-off points were doubled in each step. The results did not majorly change when other cut-off points were used.
Very few studies have been performed on environmental risk factors for fractures. A Belgian study(11) showed that when urinary cadmium levels doubled, the relative risk of a fracture in women increased to 1.73 (95% CI, 1.16, 2.57). No similar associations were observed in men, but in both men and women, external exposure to cadmium was also a significant predictor of the occurrence of fractures. In that study, the mean urinary cadmium excretion was ∼1 nmol cadmium/mmol creatinine,(11) which is slightly higher than in our study but in the same range.
Different possible mechanisms for cadmium effects on bone have been proposed.(29) Cadmium may inhibit the hydroxylation in the kidneys of 25-hydroxycholecalciferol to 1,25-dihydroxycholecalciferol, increase urinary excretion of calcium, decrease gastrointestinal calcium absorption, decrease parathyroid hormone (PTH) stimulation of adenylcyclase, or have a direct effect on bone mineralization.
Several experimental and some human studies have shown that lead may affect bone. Studies on children have shown negative correlations between lead in blood and the levels of 1,25-dihydroxyvitamin D.(30,31) Experimental studies have shown different possible mechanisms for lead effects on bone.(32) However, in contrast to our findings on cadmium, no associations were found between blood lead and the risk of forearm fracture. Lead in blood is a commonly used biomarker of lead exposure,(33) although the half-life of lead in blood is short (∼5 weeks),(33) and it typically represents mostly, but not only, relatively recent exposure. There is a possibility that the results would have differed if the total body burden could have been measured in a better way. The lack of association between blood lead and fractures nevertheless speaks against a hypothesis that the associations we see for cadmium are merely incidental and explained by uncontrolled confounding rather than being causal.
In conclusion, this study indicates an association between low-level cadmium exposure and risk of forearm fractures. This adds to the increasing evidence that cadmium in the environment is not only a historical environmental health problem in Japan but may be of current concern for the general population in many other countries.
We thank Ann-Christin Palmqvist and Ann-Kristin Thunberg, who collected the data. This study was supported by a grant from the Swedish Environmental Protection Agency. This work was performed at the Division of Environmental Epidemiology, Institute of Environmental Medicine at Karolinska Institutet; Department of Environmental Health, Stockholm County Council; Department of Renal Medicine at Huddinge University Hospital; Department of Medical Epidemiology at Karolinska Institutet, Stockholm, Sweden; and the Department of Epidemiology and Public Health, Imperial College, London, UK.