Abstract
Abstract
Objectives
To evaluate temporal trends in the epidemiology of hip osteoarthritis (OA) in the USA from 1990 to 2019, with stratification by sex and geographic region.
Design
Cross-sectional time-series analysis using secondary data from the Global Burden of Disease (GBD) study.
Setting
US population-based analysis, stratified by the four US Census Bureau regions: Northeast, Midwest, South and West.
Participants
De-identified, aggregate population-level data representing all adults in the USA from 1990 to 2019, drawn from the GBD database.
Primary and secondary outcome measures
Age-standardised rates per 100 000 population for years lived with disability (YLDs), prevalence and incidence of hip OA. Outcomes were stratified by sex and region. Statistical significance was defined as p<0.05.
Results
Between 1990 and 2019, hip OA in the USA increased by 23.91% in YLDs, 24.67% in prevalence and 25.22% in incidence. In 2019, the mean YLDs were 28.30 in women versus 25.48 in men; prevalence was 49.55 versus 41.08; and incidence was 919.29 versus 818.10 (all p<0.001). Regionally, the Northeast had the highest mean YLDs (30.1), prevalence (52.4) and incidence (950.5), while the South had the lowest (YLDs: 24.5, prevalence: 40.7, incidence: 805.2). These sex-based and region-based disparities were statistically significant (p<0.05).
Conclusions
There has been a substantial rise in the burden of hip OA in the USA over the past three decades. Women and residents of the Northeastern USA are disproportionately affected. These findings underscore the need for targeted public health strategies that account for geographic and sex-based disparities in hip OA burden.
Keywords: Prevalence, Hip, EPIDEMIOLOGY
STRENGTHS AND LIMITATIONS OF THIS STUDY.
This study leverages data from the Global Burden of Disease database, a comprehensive and validated source that allows for robust trend analysis across nearly three decades.
Additionally, stratification by sex, region and state provides nuanced insights into demographic and geographic disparities, which can inform targeted public health interventions.
However, it relies on modelled estimates, which may not fully capture local-level variations in risk factors and healthcare access.
Additionally, the observational nature of this study precludes causal inferences regarding the increases in hip osteoarthritis prevalence, incidence and years lived with disability over time.
Introduction
Osteoarthritis (OA) is the most common form of arthritis in the USA, affecting over 50% of adults aged 65 years and older and contributing substantially to disability among ageing populations.1 2 This condition is characterised by progressive degradation of articular cartilage, remodelling of subchondral bone, osteophyte formation and low-grade synovial inflammation—rather than generalised bone loss.3 4 As the median age of the US population continues to increase, the prevalence of chronic conditions such as OA follows, placing incremental strain on national healthcare systems.5
Hip OA is particularly common in the elderly population and is distinguished by the deterioration of articular cartilage and bone alterations.6 Typically, symptoms include pain and stiffness around the hip joint and groin area.7 Physical activity may help alleviate early symptoms of hip osteoarthritis by preserving range of motion, enhancing periarticular muscle strength, optimising joint biomechanics and reducing local inflammation.8,10
Primary hip OA is associated with factors such as advanced age, female sex, obesity and cumulative biomechanical loading (repetitive microtrauma) over time on the hip joint. Historically regarded as a simple ‘wear-and-tear’ degenerative disease, osteoarthritis is now understood to be a complex, multifactorial joint disorder involving not only mechanical cartilage breakdown but also active metabolic and inflammatory processes in the joint.11,13 In particular, obesity contributes to hip OA risk through both increased mechanical loading on the joint and systemic metabolic inflammation via adipose-derived mediators (adipokines) that can accelerate cartilage degradation.14 Likewise, the well-documented female predominance in hip OA appears to be multifactorial rather than solely genetic or anatomical: hormonal factors (especially the loss of oestrogen’s chondroprotective effects after menopause) play a role, as do biomechanical differences (eg, wider pelvic anatomy, variations in joint alignment and muscle strength) that affect joint loading.15 Immunologic and inflammatory mechanisms may also contribute to women’s higher OA risk—for example, women with OA tend to exhibit higher levels of pro-inflammatory mediators and greater markers of inflammation than men.15 These factors together help explain why women have both a higher prevalence and a more severe clinical course of hip OA compared with men.15
Although osteoarthritis is a widely studied condition, relatively few analyses have examined long-term, region-specific trends in hip OA incidence, prevalence and disability burden within the USA. Most existing epidemiological studies either focus on OA broadly, combine data across multiple joints or lack regional stratification. This study leverages three decades of data from the Global Burden of Disease (GBD) 2019 project to assess geographic and sex-specific disparities in hip OA across the USA, providing updated, age-standardised estimates at national, regional and state levels. These trends are particularly important given projections indicating a 174% increase in total hip arthroplasty by 2030 due to rising OA burden.16 Furthermore, arthritis is already a leading cause of disability, with the Centres for Disease Control and Prevention (CDC) estimating over 53 million affected individuals in the USA.17 By identifying temporal and demographic patterns in hip OA, our findings aim to inform regional healthcare resource allocation and guide targeted public health strategies. Recent global studies, including a 2024 GBD-based analysis of 204 countries, have demonstrated rising OA prevalence and disability-adjusted life years worldwide, underscoring the value of regionally tailored epidemiological insight for health system planning.18
Methods
Data sources
This study was done using the GBD dataset that was developed by the Institute of Health Metrics and Evaluation.19 The GBD dataset is comprised of the epidemiological data of 369 diseases and injuries across 204 countries and spans from 1990 to 2019. To provide estimates and projections, the GBD uses data from numerous primary and secondary sources, including administrative data, census data, demographic surveys, geospatial data and modelled data, among others.20 The methods and development of the GBD dataset have been described at length, and the disease burden estimates have been previously validated.21,23 Using the US Census Bureau definitions, the data was stratified into four regions: the Northeast, the Midwest, the South and the West.24
Outcomes
The outcomes of interest in this study included years lived with disability (YLD), incidence and prevalence of hip OA. The WHO defines a YLD as ‘one full year of healthy life lost due to disability or ill-health’.25 Age-standardised rates of YLDs, prevalence and incidence per 100 000 people were collected for both men and women in this study. This data was collected for the entire US population as well as each individual state. Institutional review board approval was not obtained for this study as it does not contain any protected health information and all data has been de-identified and is publicly accessible.
Statistical analysis
The statistical analysis that was performed in this study has been previously described and published.26 We conducted a one-way analysis of variance (ANOVA) to compare the mean values of YLDs, incidence and prevalence of hip osteoarthritis across US states and Washington D.C. Bartlett’s test was applied to assess homogeneity of variances. Where assumptions of equal variance were violated, Welch’s ANOVA was used instead, followed by Games-Howell post hoc tests to identify significant pairwise differences. Independent samples t-tests were used to examine sex-based differences. Descriptive statistics such as mean, SD and coefficient of variation were calculated. All statistical analyses were performed using IBM Statistical Package for Social Sciences (SPSS), V. 29 (Chicago, Illinois) and significance was defined as p<0.05.
Patient and public involvement
Patients and the public were not involved in the design, conduct, reporting or dissemination plans of this research.
Results
USA
From 1990 to 2019, the USA saw a 23.91% increase in YLDs, 24.67% increase in the incidence and a 25.22% increase in the prevalence of hip OA. Regardless of region or sex, there was an increase in overall mean YLDs, incidence and prevalence of hip OA from 1990 to 2019. A summary of the temporal trends of YLD, incidence and prevalence for both sexes can be found in figures13. While the mean YLDs (28.30 vs 25.48), incidence (919.29 vs 818.10) and prevalence (49.55 vs 41.08) of hip OA were statistically higher in men over the 30-year period, yearly trends consistently showed higher values for women across all three metrics from 1990 to 2019. These seemingly contradictory findings reflect how men had more variability and higher short-term spikes, raising the overall mean, even though women consistently bore the higher burden each year.
Figure 1. These figures illustrate temporal trends from 1990 to 2019 in average years lived with disability (YLDs) due to hip osteoarthritis, stratified by sex across the Midwest, Northeast, South and West regions.
Figure 3. These figures present the temporal trends in the prevalence of hip osteoarthritis from 1990 to 2019, stratified by sex across the Midwest, Northeast, South and West regions.
Figure 2. These figures show the annual incidence rate of hip osteoarthritis from 1990 to 2019, stratified by sex and across the four US regions: Midwest, Northeast, South and West.
Data by region
A summary of the temporal trends of YLD, incidence and prevalence for each region can be found in figures13. Regional analysis demonstrated that the northeastern region demonstrated the highest overall mean YLDs, incidence and prevalence of hip OA. The northeast region maintained the highest overall mean YLDs and prevalence from 1990 to 2019. However, the Northeast region maintained the highest overall mean incidence from 1990 to 2005 after which the Midwest and West regions surpassed the Northeast region.27 By 2019, the Northeast region had the lowest mean incidence of all the regions. The South region experienced the lowest rates of YLDs, incidence and prevalence of hip OA. The South region maintained the lowest overall mean YLDs and prevalence from 1990 to 2019. However, the South region maintained the lowest overall mean incidence from 1990 to approximately 2016 after which the South region became second lowest to the Northeast region by 2019. Women were more likely to experience higher YLDs, incidence and prevalence of hip OA in each of the regions compared with men (p<0.05).
Data by state
A summary of percent change from 1990 to 2019 for all outcome measures in all states is provided in online supplemental table 1. North Dakota experienced the highest increases in YLDs (+33.7%) and incidence (+29.1%) between 1990 and 2019, while Ohio had the greatest increase in prevalence (+63.25%). From 1990 to 2019, Mississippi experienced the lowest increase in incidence of hip OA (22.01%). Arizona experienced the lowest increase of YLDs with hip OA (21.36%) and the greatest decrease in prevalence of hip OA (−11.51%). Arizona is the only state to experience a decrease in prevalence (−11.51%) from 1990 to 2019. None of the states experienced a decrease in YLDs or incidence from 1990 to 2019. The five states with the lowest percent change in YLDs between 1990 and 2019 are Arizona (21.36%), Mississippi (21.90%), West Virginia (22.05%), Tennessee (22.26%) and Nevada (22.29%). The five states with the highest percent change in YLDs between 1990 and 2019 are North Dakota (30.16%), Massachusetts (29.81%), Delaware (28.76%), Wisconsin (28.58%) and Connecticut (28.06%). The five states with the lowest percent change in prevalence between 1990 and 2019 are Arizona (−11.51%), North Dakota (0.75%), Connecticut (0.82%), Oklahoma (13.36%) and Oregon (17.91%). The five states with the highest percent change in prevalence between 1990 and 2019 are Ohio (63.25%), Florida (58.09%), Delaware (51.78%), Pennsylvania (44.27%) and Hawaii (43.38%). The five states with the lowest percent change in incidence between 1990 and 2019 are Mississippi (22.01%), West Virginia (23.33%), Tennessee (23.34%), New York (23.46%) and Georgia (23.89%). The five states with the highest percent change in incidence between 1990 and 2019 are North Dakota (31.18%), Delaware (30.71%), Massachusetts (30.38%), New Hampshire (29.31%) and Connecticut (29.12%).
Discussion
The results of this study found that prevalence, incidence and YLDs of hip OA increased in the USA from 1990 to 2019. Across all four regions, women were more likely to experience this change compared with men. Furthermore, prevalence, incidence and YLDs of hip OA were the largest in the Northeast region and lowest in the South region. Notably, North Dakota had the highest increases in rates of YLDs, prevalence and incidence of hip OA. Additionally, Mississippi experienced the smallest increase in the incidence rate of hip OA. On the other hand, Arizona experienced the largest decrease in hip OA and the smallest rise in YLDs.
While age is a major non-modifiable risk factor for hip OA, our findings still carry strategic value. Regions and states with disproportionately high OA burdens—such as the Northeast or North Dakota—may have older, more sedentary populations and limited access to orthopaedic care. Identifying these geographic clusters supports targeted preventive strategies (eg, funding for mobility programmes and early screening clinics) and helps policymakers anticipate future orthopaedic demand as the population continues to age.28 Additionally, layering known modifiable risks such as obesity or physical inactivity onto regional age patterns could better identify at-risk subpopulations.
In every region, women exhibited a greater propensity for experiencing elevated YLDs, prevalence and incidence of hip OA compared with men. In the USA, female sex is recognised as a risk factor for OA. This is attributed to the decrease in oestrogen levels following menopause, which is linked to heightened cartilage metabolism, thereby raising the risk of joint degeneration.29
In contrast to the South, the Northeastern region of the USA exhibited the highest prevalence, incidence and YLDs related to hip OA. This may be attributed to Northeastern states having a high population of older individuals, who are in general more susceptible to OA.20 Interestingly, North Dakota had the highest increase in the rates of YLDs, prevalence and incidence from 1990 to 2019. The disproportionate increase in hip OA in North Dakota may reflect a combination of demographic ageing, underdiagnosis in prior decades and improved healthcare access or detection more recently. While rural settings historically face healthcare provider shortages, changes in provider coverage, diagnostic coding or outreach over the study period may have contributed to the observed trend. 20 30 Additionally, financial instability among uninsured residents could deter them from seeking medical attention, exacerbating the situation.16
On the other hand, the South region of the USA has the lowest prevalence, incidence and YLDs of hip OA. The elevated levels of physical activity encouraged by the warmer weather in the South may contribute to reducing the overall risk of OA.31 Another contributing factor to the lower rates of hip OA in the South could be the difference in the distribution of elderly populations between the Northeast region and the South. The South tends to have a younger demographic, which may result in lower susceptibility to OA compared with the Northeast.20
Interestingly, Arizona was the only state to show a decrease in the prevalence of hip OA from 1990 to 2019. This anomaly could be due to changes in state-level reporting practices, shifting demographic profiles or limitations in the GBD modelling process. Further state-specific research is warranted to determine whether this trend reflects a true epidemiological difference or an artefact of data modelling.
These findings can inform public health action by identifying geographic and sex-based disparities in hip osteoarthritis (OA) burden, enabling more effective allocation of resources. For example, regions with elevated YLDs or rising incidence—such as parts of the Northeast—may require additional investment in orthopaedic surgical capacity, early detection initiatives and community-based physical activity and weight management programmes. Land-based therapeutic exercise has proven effective in reducing pain and enhancing physical function in individuals with hip OA.8 The significantly higher OA burden among women—as reflected in national data, where 57.8% of women aged ≥75 compared with 44.7% of men report arthritis—underscores the need for sex-specific interventions, including targeted education, routine screening in primary care and supportive workplace policies.27 Together, these insights offer an evidence-based foundation for policymakers and health systems to design equitable, state-level strategies to reduce the economic and disability burden of OA.32
This study is not without limitations. First, it relies on modelled estimates from the GBD 2019 dataset, which—while validated and comprehensive—may not fully reflect local or state-level variations in diagnostic practices, healthcare access or data quality. Additionally, the analysis is limited to data through 2019, as more recent GBD estimates (eg, GBD 2021) were not publicly accessible in stratified formats (by sex or state) at the time of analysis. Although newer studies have used GBD 2021 data for international comparisons, these could not be incorporated here. Future research should integrate updated datasets when available to validate and extend these findings. Furthermore, this study does not perform age-stratified subgroup analyses, which may have provided additional insight into how hip osteoarthritis burden differs across the life course. Finally, due to the ecological nature of the GBD data, we were unable to examine individual-level risk factors or establish causal relationships between demographic or geographic trends and hip OA outcomes.
Conclusion
From 1990 to 2019, the USA experienced notable increases in YLDs, incidence and prevalence of hip OA, with a 23.91%, 24.67% and 25.22% rise, respectively. Women consistently showed higher rates across all measures compared with men. Regionally, the Northeast had the highest YLDs and prevalence until the Midwest and West surpassed its incidence post-2005, while the South maintained the lowest rates overall. Statewide, North Dakota had the highest increases in YLDs and prevalence, while Arizona uniquely saw a decrease in prevalence. No state experienced a reduction in YLDs or incidence over the study period.
Supplementary material
Footnotes
Funding: The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Prepub: Pre-publication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2024-096130).
Provenance and peer review: Not commissioned; externally peer reviewed.
Patient consent for publication: Not applicable.
Ethics approval: Ethics approval was not required for this study as it is based on publicly available, de-identified data from the Global Burden of Disease (GBD) study.
Data availability free text: N/A.
Collaborators: REAM.
Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Data availability statement
The data supporting the findings of this study are publicly available through the Institute for Health Metrics and Evaluation (IHME). The Global Burden of Disease Study 2019 (GBD 2019) dataset can be accessed at: https://doi.org/10.6069/P5WM-5A36.
References
- 1.Theis KA, Murphy LB, Guglielmo D, et al. Prevalence of arthritis and arthritis-attributable activity limitation - United States, 2016-2018. MMWR Morb Mortal Wkly Rep. 2021;70:1401–7. doi: 10.15585/mmwr.mm7040a2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Elgaddal. Nazik . US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Health Statistics. 2024. Arthritis in adults age 18 and older: United States, 2022. [Google Scholar]
- 3.Glyn-Jones S, Palmer AJR, Agricola R, et al. Osteoarthritis. Lancet. 2015;386:376–87. doi: 10.1016/S0140-6736(14)60802-3. [DOI] [PubMed] [Google Scholar]
- 4.Loeser RF, Goldring SR, Scanzello CR, et al. Osteoarthritis: a disease of the joint as an organ. Arthritis Rheum. 2012;64:1697–707. doi: 10.1002/art.34453. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Palo N, Chandel SS, Dash SK, et al. Effects of osteoarthritis on quality of life in elderly population of Bhubaneswar, India: a prospective multicenter screening and therapeutic study of 2854 patients. Geriatr Orthop Surg Rehabil. 2015;6:269–75. doi: 10.1177/2151458515604357. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Fan Z, Yan L, Liu H, et al. The prevalence of hip osteoarthritis: a systematic review and meta-analysis. Arthritis Res Ther. 2023;25 doi: 10.1186/s13075-023-03033-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Lespasio MJ, Sultan AA, Piuzzi NS, et al. Hip osteoarthritis: a primer. Perm J. 2018;22 doi: 10.7812/TPP/17-084. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Fransen M, McConnell S, Hernandez-Molina G, et al. Exercise for osteoarthritis of the hip. Cochrane Database Syst Rev. 2014;2014 doi: 10.1002/14651858.CD007912.pub2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Hunter DJ, Bierma-Zeinstra S. Osteoarthritis. Lancet. 2019;393:1745–59. doi: 10.1016/S0140-6736(19)30417-9. [DOI] [PubMed] [Google Scholar]
- 10.Gartner JJ, Parkhurst MR, Gros A, et al. A machine learning model for ranking candidate HLA class I neoantigens based on known neoepitopes from multiple human tumor types. Nat Cancer . 2021;2:563–74. doi: 10.1038/s43018-021-00197-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.2020 SoO Trust for America’s health. 2020. [8-Jul-2020]. https://www.tfah.org/report-details/state-of-obesity-2020 Available. Accessed.
- 12.Tang S, Zhang C, Oo WM, et al. Osteoarthritis. Nat Rev Dis Primers. 2025;11:1–22. doi: 10.1038/s41572-025-00594-6. [DOI] [PubMed] [Google Scholar]
- 13.Berenbaum F, Griffin TM, Liu-Bryan R. Review: metabolic regulation of inflammation in osteoarthritis. Arthritis Rheumatol . 2017;69:9–21. doi: 10.1002/art.39842. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Zhang C, Lin Y, Yan CH, et al. Adipokine signaling pathways in osteoarthritis. Front Bioeng Biotechnol. 2022;10 doi: 10.3389/fbioe.2022.865370. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Segal NA, Nilges JM, Oo WM. Sex differences in osteoarthritis prevalence, pain perception, physical function and therapeutics. Osteoarthr Cartil. 2024;32:1045–53. doi: 10.1016/j.joca.2024.04.002. [DOI] [PubMed] [Google Scholar]
- 16.Nho SJ, Kymes SM, Callaghan JJ, et al. The burden of hip osteoarthritis in the United States: epidemiologic and economic considerations. J Am Acad Orthop Surg. 2013;21 Suppl 1:S1–6. doi: 10.5435/JAAOS-21-07-S1. [DOI] [PubMed] [Google Scholar]
- 17.Burden OPa Osteoarthritis action alliance. 2023. https://oaaction.unc.edu/oa-module/oa-prevalence-and-burden/#:~:text=CDC%20estimates%20that%201%20in,have%20some%20form%20of%20arthritis.&text=While%20there%20are%20estimated%20to,affecting%2032.5%20million%20US%20adults Available.
- 18.Wu R, Guo Y, Chen Y, et al. Osteoarthritis burden and inequality from 1990 to 2021: a systematic analysis for the global burden of disease Study 2021. Sci Rep. 2025;15:8305. doi: 10.1038/s41598-025-93124-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Network GBoDC Institute for health metrics and evaluation (IHME) 2021.
- 20.Oldest? WUSSAt. 2019. https://www.prb.org/resources/which-us-states-are-the-oldest Available.
- 21.Barendregt JJ, Van Oortmarssen GJ, Vos T, et al. A generic model for the assessment of disease epidemiology: the computational basis of DisMod II. Popul Health Metr. 2003;1 doi: 10.1186/1478-7954-1-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Global burden of 369 diseases and injuries in 204 countries and territories -asaftGBoDS. Lancet. 2020;396:1204–22. doi: 10.1016/S0140-6736(20)30925-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Mathers CD, Murray CJL, Lopez AD. Epidemiological evidence: improving validity through consistency analysis. Bull World Health Organ. 2002;80:611. [PMC free article] [PubMed] [Google Scholar]
- 24.Divisions USCBRa. 2023. https://www2.census.gov/geo/pdfs/maps-data/maps/reference/us_regdiv.pdf Available.
- 25.details Imr World health organization. n.d.
- 26.Gupta N, Singh R, Seas A, et al. Epilepsy among the older population of sub-Saharan Africa: analysis of the global burden of disease database. Epilepsy Behav. 2023;147:109402. doi: 10.1016/j.yebeh.2023.109402. [DOI] [PubMed] [Google Scholar]
- 27.QuickStats: percentage* of adults aged ≥18 years with arthritis,† by sex and age group — National health interview survey,§ United States, 2019. MMWR Morb Mortal Wkly Rep. 2021;70:658. doi: 10.15585/mmwr.mm7017a7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Hawker GA, King LK. The burden of osteoarthritis in older adults. Clin Geriatr Med. 2022;38:181–92. doi: 10.1016/j.cger.2021.11.005. [DOI] [PubMed] [Google Scholar]
- 29.King LK, March L, Anandacoomarasamy A. Obesity & osteoarthritis. Indian J Med Res. 2013;138:185–93. [PMC free article] [PubMed] [Google Scholar]
- 30.Medicine H, Medicine D, Public Health P, et al. In: Population Health in Rural America in 2020: Proceedings of a Workshop. Washington (DC): National Academies Press (US) Copyright 2021 by the National Academy of Sciences. All rights reserved. Nicholson A, editor. Washington (DC): National Academies Press (US); 2021. Roundtable on population health I. [Google Scholar]
- 31.Brandon CA, Gill DP, Speechley M, et al. Physical activity levels of older community-dwelling adults are influenced by summer weather variables. Appl Physiol Nutr Metab. 2009;34:182–90. doi: 10.1139/H09-004. [DOI] [PubMed] [Google Scholar]
- 32.Katz JN, Arant KR, Loeser RF. Diagnosis and treatment of hip and knee osteoarthritis: a review. JAMA. 2021;325:568–78. doi: 10.1001/jama.2020.22171. [DOI] [PMC free article] [PubMed] [Google Scholar]