Higher serum uric acid as a protective factor against incident osteoporotic fractures in Korean men: a longitudinal study using the National Claim Registry
- First Online:
- Cite this article as:
- Kim, B., Baek, S., Ahn, S.H. et al. Osteoporos Int (2014) 25: 1837. doi:10.1007/s00198-014-2697-2
In this large longitudinal study of 16,078 Korean men aged 50 years or older, we observed that baseline elevation of serum uric acid level significantly associated with a lower risk of incident fractures at osteoporosis-related sites during an average follow-up period of 3 years.
Male osteoporosis and related fractures are becoming recognized as important public health concerns. Oxidative stress has detrimental effects on bone metabolism, and serum uric acid (UA) is known to be a strong endogenous antioxidant. In the present study, we performed a large longitudinal study with an average follow-up period of 3 years to clarify the role of UA on the risk of incident osteoporotic fractures (OFs).
A total of 16,078 Korean men aged 50 years or older who had undergone comprehensive routine health examinations were enrolled. Incident fractures at osteoporosis-related sites (e.g., hip, spine, distal radius, and proximal humerus) that occurred after the baseline examinations were identified from the nationwide claims database of the Health Insurance Review and Assessment Service of Korea by using selected International Classification of Diseases, 10th revision codes.
In total, 158 (1.0 %) men developed incident OFs. The event rate was 33.1 per 10,000 person-years. Subjects without incident OFs had 6.0 % higher serum UA levels than subjects with OFs (P = 0.001). Multivariable-adjusted Cox proportional hazard analyses adjusted for age, body mass index, glomerular filtration rate, lifestyle factors, medical and drug histories, and the presence of baseline radiological vertebral fractures revealed that the hazard ratio per standard deviation increase of baseline UA levels for the development of incident OFs was 0.829 (95 % CI = 0.695–0.989, P = 0.038).
These data provide the epidemiological evidence that serum UA may act as a protective factor against the development of incident OFs in Korean men.
KeywordsAntioxidantMenOsteoporosisOsteoporotic fractureUric acid
Male osteoporosis and related fractures are becoming recognized as important public health concerns because 20–35 % of all femoral fractures and 33–50 % of all vertebral fractures (VFs) occur in men [1, 2]. Moreover, mortality from femoral fractures is two- to threefold higher in men when compared to women , and men are twice as likely as women to die in a hospital after a hip fracture . Therefore, more effort for elucidating the risk factors and the underlying pathogenic mechanism of osteoporotic fractures (OFs) in men would be required to effectively prevent them.
Oxidative stress has been demonstrated to have detrimental effects on bone metabolism. In vitro studies show that reactive oxygen species inhibit the differentiation, proliferation, and activity of osteoblasts by modulating redox-sensitive signaling pathways , whereas they promote bone resorption either by directly stimulating osteoclast differentiation or by indirectly increasing the expression of receptor activator of NF-κB ligand in osteoblasts [6, 7]. Clinical studies have also linked bone disease development to either oxidative stress or low circulating levels of antioxidants [8–10].
Although uric acid (UA), an end product of purine metabolism in humans, is traditionally regarded as a risk factor for a wide variety of diseases, including metabolic syndrome and cardiovascular disorders [11, 12], there is increasing evidence that it may increase life expectancy by functioning as an antioxidant through the free radical scavenging capacity as well [13–17]. Given this, it has been hypothesized that the antioxidant effect of UA may protect against metabolic bone diseases such as osteoporosis. Supporting this is the recent study of Nabipour et al.  on a cross-sectional cohort of older men, which showed for the first time that higher serum UA levels are associated with higher bone mineral density (BMD) values and a lower prevalence of vertebral and nonvertebral fractures. The three other epidemiological studies that have been performed to date show similar results [19–21]. These studies could possess the important implication in that they provide clinical evidence about the protective role of UA on bone metabolism. However, despite the fact that the ultimate goal of bone biology research is to reduce the risk and incidence of fractures, observational studies that designate incident fractures as a primary endpoint in relation to serum UA levels have been limited.
In the present study, we performed a large longitudinal study of Korean men to investigate the association between baseline serum UA levels and the development of incident fractures at osteoporosis-related sites, using the data from the National Claim Registry.
Materials and methods
Baseline study participants
The baseline demographic data of the subjects were acquired from a database that is maintained by the Health Promotion Center at Asan Medical Center. All participants were interviewed and examined by physicians in the health promotion center. Information on history of previous medical or surgical diseases for each subject was obtained. Smoking was categorized into three levels (never, past, or current), as was alcohol drinking (never, moderate, or heavy). Dairy product consumption and physical exercise were categorized into two levels according to frequency per week (< or ≥3 times/week). Height (cm) and weight (kg) were measured by using standardized protocols while the subjects were dressed in light clothing and had their shoes off. The body mass index (BMI, kg/m2) was calculated from the height and weight.
For biochemical measurements, morning blood samples were obtained after 12 h of fasting and subsequently analyzed at a central, certified laboratory at the Asan Medical Center. The serum levels of UA, calcium, and phosphorus were measured by using the colorimetric, cresolphthaleincomplexone, and phosphomolybdate ultraviolet methods, respectively. For this, a Toshiba 200FR autoanalyzer (Tokyo, Japan) was employed. The glomerular filtration rate (GFR; milliliters per minute per 1.73 m2), an indicator of renal function, was calculated using the Cockcroft–Gault formula . The intra- and interassay coefficients of variations of these analyses were consistently <3.5 %.
At the baseline health examinations, the subjects were examined for the presence of morphological VFs by lateral thoracolumbar (T4-L4) radiography, which was performed in accordance with the recommendations of the Working Group on Vertebral Fractures . Radiographs were assessed at Asan Medical Center by expert radiologists who were blinded to the study. A VF was diagnosed if any of the measured vertebral heights (anterior, middle, or posterior) were reduced by >20 % .
Ascertainment of incident fractures
The primary endpoint of our study was the development of incident OFs after the baseline examinations. This information was obtained from the nationwide claims database of the Health Insurance Review and Assessment Service (HIRA) of Korea. In Korea, 97.0 % of the population is obliged to enroll in the Korean National Health Insurance Program. Patients pay around 30 % of the total medical costs to clinics or hospitals, though some services are not covered by insurance (e.g., cosmetic surgery and some unproven therapies). To be reimbursed for the remaining 70 % of the total medical costs, the clinics and hospitals then submit claims for inpatient and outpatient care that document the diagnoses [as determined by the International Classification of Diseases, 10th revision (ICD-10)], procedures, prescription records, demographic information, and direct medical costs. The remaining 3 % of the population that is not insured by the Korean National Health Insurance Program is either covered by another Medical Aid Program or includes temporary or illegal residents. However, since these claims are also reviewed by HIRA, virtually all information about patients and their medical records is available from the Korean HIRA database. This database has been used on several occasions for epidemiological studies [25–27]. For the current study, we used the HIRA database from January 2007 to December 2011. This study was conducted with the approval of the National Strategic Coordinating Center of Clinical Research and HIRA in Korea.
Incident fractures at osteoporosis-related sites were identified on the basis of selected ICD-10 codes [28–30]. The ICD-10 codes for hip fractures were S72.0 (fracture of the femoral neck) and S72.1 (pertrochanteric fracture). The codes for spine fractures were S22.0 (fracture of the thoracic spine), S22.1 (multiple fractures of the thoracic spine), S32.0 (fracture of the lumbar spine), S32.7 (multiple fractures of the lumbar spine), S32.8 (other fractures of the lumbar spine), M48.4 (fatigue fracture of vertebra), and M48.5 (collapsed vertebra, NEC). The codes for distal radius fractures were S52.5 (fracture of the distal radius) and S52.6 (combined fracture of the distal radius/ulna). The codes for proximal humerus fractures were S42.2 (fracture of the proximal humerus) and S42.3 (fracture of shaft of the humerus). In subjects with multiple events that occurred on different dates, only the first event was included in the analysis. In addition, multiple ICD-10 codes occurring on the same date were considered as a single event.
The information regarding the use of drugs that can affect bone metabolism, such as bisphosphonates, vitamin D, thyroid hormone, or glucocorticoids, during the study period was also obtained from the HIRA database.
Continuous and categorical variables are reported as means ± standard deviations (SDs) and percentages, respectively, unless otherwise specified. The baseline characteristics of the study population according to the status of incident fractures were compared by using the Mann–Whitney U tests for continuous variables and chi-square tests for categorical variables. The incidence rate and its 95 % confidence interval (CI) were estimated under the Poisson distribution. To test the hypothesis that higher baseline serum UA levels associated independently with a lower risk of incident fractures, multivariable Cox proportional hazards models to estimate adjusted hazard ratios (HRs) and 95 % CIs were used. Confounding independent variables were selected on the basis of being clinically applicable. Consequently, the base adjustment model included age, BMI, GFR, smoking and alcohol drinking habits, physical exercise, and dairy product consumption. The multivariable adjustment model included the factors from the base adjustment model plus the history of stroke and diabetes, the use of drugs that can affect bone metabolism including bisphosphonates, vitamin D, thyroid hormone, or glucocorticoids during the study period, and the presence of baseline radiological VFs. Crude and adjusted HRs and their 95 % CIs from the two models described above are presented. All analyses were checked for multicollinearity between covariates but significant multicollinearity was not found in any of the analyses. All reported P values were two-sided. A P value of <0.05 was considered statistically significant. All data analyses were performed by using SAS 9.3 (SAS Institute, Inc., Cary, NC, USA).
In the present study, 16,078 men aged 50 years or older were enrolled in the final analysis and observed for the development of incident fractures at osteoporosis-related sites. The total follow-up period was 48,201.8 person-years, and the average follow-up period was 3.0 (SD 1.7) years. Of the participants, 158 (1.0 %) developed incident OFs during the study period (Fig. 1). The event rate was 33.1 (95 % CI = 28.3–38.7) per 10,000 person-years.
Baseline characteristics of the study population according to incident osteoporotic fracture status
Subjects without incident OFs (n = 15,920)
Subjects with incident OFs (n = 158)
Serum UA (mg/dL)
5.84 ± 1.25
5.51 ± 1.47
57.1 ± 6.2
59.2 ± 8.2
70.3 ± 8.8
68.3 ± 9.3
169.1 ± 5.6
168.9 ± 5.9
Body mass index (kg/m2)
24.5 ± 2.6
23.9 ± 2.8
Alcohol drinking habits (%)
Moderate (2–3 times/week)
Heavy (≥4 times/week)
Smoking habits (%)
Physical exercise (%)
Intermittent (≤2 times/week)
Regular (≥3 times/week)
Dairy product consumption (%)
Intermittent (≤2 times/week)
Regular (≥3 times/week)
History of high glucose level (%)
History of stroke (%)
Drug uses during study periods
Thyroid hormone (%)
Vitamin D (%)
Radiological vertebral fracture at baseline (%)
Serum calcium (mg/dL)
9.07 ± 0.31
9.11 ± 0.28
Serum phosphorus (mg/dL)
3.37 ± 0.45
3.40 ± 0.45
GFR (mL/min/1.73 m2)
84.9 ± 17.1
82.7 ± 19.8
Multivariable adjusted Cox proportional hazards models for the incidence of osteoporotic fractures according to baseline serum uric acid concentration
Hazard ratio (95 % CIs) per SD increment of baseline serum UA levels
Age, BMI, and GFR
Independent and significant determinants of incident osteoporotic fractures from a multivariable adjusted Cox proportional hazards model
Independent determinants of incident OFs
Hazard ratio (95 % CIs)
Alcohol drinking habits
Moderate (2–3 times/week)
Heavy (≥4 times/week)
Drug uses during study periods
Radiological vertebral fracture at baseline
Serum uric acid
Finally, we assessed the risk of incident fractures according to serum UA levels after additionally excluding the users of bisphosphonates, vitamin D, thyroid hormone, or glucocorticoids. Among 15,082 eligible men, 131 (0.9 %) developed incident osteoporotic fractures during the study period. In the multivariable adjustment model, the HR per SD increase of baseline UA levels for the development of incident fractures was 0.792 (95 % CI = 0.653–0.960, P = 0.018; Supplementary Table 2).
In this large longitudinal study of 16,078 Korean men aged 50 years or older, we observed that baseline elevation of serum UA level within the physiologic range significantly associated with a lower risk of incident fractures at osteoporosis-related sites during an average follow-up period of 3 years. These data provide the epidemiological evidence for an independent association between serum UA and the future development of OFs, suggesting that higher serum UA level may play a beneficial role against the development of incident fractures.
A strong correlation between elevated UA levels and increased risk for myocardial infarction, stroke, and cardiovascular mortality has been well-known [11, 12], and these observations are mainly explained by the stimulated smooth muscle cell proliferation, the increased inflammation, and the increased endothelial dysfunction by UA [31, 32]. However, many lines of evidence now indicate that UA may also play a physiological role as an antioxidant and cytoprotectant. UA accounts for a substantial part of the antioxidative capacity of the plasma  and is an important intracellular free radical scavenger during metabolic stress, including nitric oxide, peroxyl radicals, and hydroxyl radicals [34, 35]. Indeed, the systemic infusion of UA not only increases plasma antioxidant capacity at rest , but also reduces the exercise-associated oxidative stress in healthy subjects . Higher serum UA levels correlate with the slower progression of many diseases such as Parkinson’s disease, Huntington’s disease, and mild cognitive impairment [37, 38]. Given these observations, it is possible that UA may also function as a protective factor in bone metabolism. This hypothesis is supported by a recent in vitro study showing that UA treatment decreased osteoclastogenesis in a dose-dependent manner and reduced the production of reactive oxygen radicals in osteoclast precursors .
Since the first clinical report by Nabipour et al.  that showed an association between UA levels and bone health, several other epidemiological studies on this subject have been performed. Sritara et al.  showed that serum UA levels correlated positively with BMD in 1,320 young and middle-aged males. Makovey et al.  reported that in 356 peri- and postmenopausal women, the women with higher UA levels were less likely to have bone loss at the lumbar spine, forearm, and total body than the women with lower UA levels. Very recently, our research group observed that in 7,502 healthy postmenopausal women, higher serum UA levels also associated with higher bone mass, lower bone turnover, and lower prevalence of VF . Thus, these studies show consistently that UA may have beneficial clinical effects on bone metabolism. However, because most of these studies were cross-sectional in design and the primary endpoint was BMD or prevalent fracture, it was not clear whether there was a causal relationship between higher serum UA levels and lower frequencies of OFs in the future. Therefore, the present observational cohort study was designed to answer this question. Meanwhile, our present paper was in preparation, Lane et al.  reported that higher serum UA levels were associated with a reduction in risk of incident nonspine fractures in elderly men. These results are consistent with ours and further support the hypothesis that UA may act as a protective factor against metabolic bone diseases.
OF is a heterogeneous entity with multiple underlying causes in any given individual. In particular, fracture risk is strongly affected by ethnicity , which means that fracture risk factors should be evaluated in each ethnic group. To our knowledge, this is the first epidemiological study that reports the clinical risk factors of incident OF in Korean men by using the nationwide claims database. In this cohort, older age, higher alcohol consumption, and the presence of baseline VF were found to contribute independently to the development of OFs. These risk factors have also been identified in other population-specific cohorts [41–43]. Of particular interest in our analysis, however, was that vitamin D and bisphosphonate use associated with a markedly higher risk of incident fractures, contrary to common belief. In Korea, vitamin D supplements and medications for osteoporosis, including bisphosphonates, are only covered by insurance in subjects whose T-scores are lower than −1.0 and −2.5, respectively. Therefore, it is possible that the increased development of OFs in vitamin D and bisphosphonate users actually reflects their lower BMD values.
The present study has several strengths. In addition to the benefits of a large sample size and a longitudinal design, its participants were consecutively enrolled and the primary endpoint was detected by using the National Statistical Office and HIRA databases in Korea—these governmental and quasigovernmental organizations provide quality-controlled and reliable data. Furthermore, to investigate whether there is a pathophysiological link between serum UA levels and incident fractures, strict exclusion criteria based on medical history and routine laboratory findings were intentionally applied, and the data were carefully adjusted for factors that affect bone metabolism.
However, this study also has some notable limitations. Most importantly, the use of the ICD-10 coding system meant that high-energy fractures could not be distinguished from low-energy fractures. Thus, it is possible that fractures due to high-energy trauma were included in this study. However, the ICD-10 coding system, which was established in 1989 by the World Health Organization includes great detail, standardized terminology, and expanded concepts for injuries and other related factors . Therefore, it is more useful in epidemiological studies that use codes than the previous version of codes. In addition, to exclude non-OFs, the criterion of “aged 50 years or more” was used. This approach was also adopted in other studies [27, 29, 45]. The second limitation is that not all patients with osteoporosis-related fractures could be included in this nationwide database. For example, many patients with asymptomatic spine fracture may not visit medical institutions in Korea. Thus, incidence rates based on the HIRA database may be underestimated. Third, since the participants spontaneously visited the health promotion center, there may be some participant-selection bias. Finally, although we considered as many confounding factors as possible, the possibility that the observed association was due to uncontrolled factors that affect UA and/or OFs could not be excluded. For example, because vitamin D is simultaneously classified as both an over-the-counter and ethical drug in Korea, many people take vitamin D supplements without a prescription. This information cannot be obtained from the HIRA database, and thus adjustments for this potential confounder could not be made during our analyses of the HIRA-based data.
In conclusion, analysis of the data of a large number of Korean men in the National Claim Registry demonstrated that the development of incident fractures at osteoporosis-related sites dropped significantly in a dose-responsive fashion as baseline serum UA concentrations rose. These findings support previous in vitro and epidemiological studies showing that higher UA levels within the physiologic range associate with favorable effects on bone metabolism. Further interventional studies are needed to confirm the protective role of UA on osteoporosis-related phenotypes.
This study was supported by grants from the Korea Health Technology R&D Project, Ministry of Health & Welfare, Republic of Korea (Project No. A110536) and by grants from the Asan Institute for Life Sciences, Seoul, Republic of Korea (Project No. 2008-026 and 2013-347).
Conflicts of interest