Development and Validation of a Novel and HIV-Relevant Prediction Model for Fracture

摘要 艾滋病毒感染者（PWH）的脆弱性骨折（骨折）每年造成9900万美元的超额成本 在美国我们之前的研究表明，PWH中骨折的年发生率明显较高 - 其中大多数是中年人-与未感染者相比（2.0%对1.4%）。这种骨折是 与住院率和疗养院入院率增加，长期身体残疾， 社会孤立和死亡因为中年是人们在经济上和社会上 生产力，在这个生命阶段的骨折为经济困难和健康状况较差的结果奠定了基础， 这些个体会变老存在预防脆性骨折的高效方法， 百分之二十五到五十。然而，没有风险预测模型可以可靠地识别PWH的最大风险。 骨折风险评估工具（FRAX）是用于预测骨折的最广泛使用的模型， 一般人口。它包括人口统计学（年龄，性别，种族）和关键风险因素（体重指数， 既往骨折，父母骨折，当前吸烟，糖皮质激素使用，类风湿性关节炎，继发性 骨质疏松症和危险的酒精使用）。对于双能X射线吸收测定法可用的设置， FRAX的灵敏度可以通过将骨矿物质密度（BMD）添加到算法中来提高。但我们的 先前的研究表明，FRAX在PWH中表现不佳，表现出较差的区分度（C统计量0.59） 灵敏度（0.06）。我们认为，这是因为FRAX没有考虑到重要的风险因素， PWH。例如，FRAX不考虑福尔斯、丙型肝炎和B病毒、物质使用障碍， 多药、多种药物、使用其他与骨折相关的药物或生理虚弱-所有 与PWH中的骨折有关。最后，FRAX不考虑抗逆转录病毒药物 治疗（ART）。个别ART以及ART类别与骨折独立相关。 利用优秀的合作者团队和研究平台，我们将开发，验证和 传播基于FRAX的骨折风险评估计算器，但其中包含艾滋病毒相关风险 我们亦会评估骨折风险因素（FRAC-HIV），以计算威尔斯亲王医院在来年发生骨折的可能性。我们有丰富 经验与退伍军人老龄化队列研究（VACS），最大的队列PWH在北 美国，这将成为我们的发展队列。为了评估可推广性，我们将从外部验证 模型使用耶鲁纽黑文医疗系统和波士顿医疗系统的电子健康记录数据 中心我们的最终目标是通过开发一个 基于网络的骨折风险预测计算器，将直接从提供者那里获得信息， 关注PWH FRAC-HIV将重新概念化并促进骨折预防的临床决策， 强调临床上可改变的危险因素的作用。我们的发现将为未来的研究奠定基础 干预试验将FRAC-HIV纳入临床护理算法，以预防PWH中的骨折。

ABSTRACT Fragility fractures (fractures) among persons with HIV (PWH) account for $99 million in excess costs each year in the United States. Our prior research has shown a significantly higher annual rate of fractures among PWH – most of whom are middle-aged – compared with uninfected individuals (2.0% vs 1.4%). Such fractures are associated with increased rates of hospitalization and nursing home admissions, long-term physical disability, social isolation, and death. Because middle-age is a time when people are most economically and socially productive, fractures at this time of life set the stage for economic hardship and poorer health outcomes as these individuals grow older. Highly effective methods exist for preventing fragility fractures, reducing rates by 25%-50%. However, there are no risk prediction models that reliably identify PWH at greatest risk. The Fracture Risk Assessment Tool (FRAX) is the most widely used model for predicting fractures in the general population. It incorporates demographics (age, sex, race) and key risk factors (body mass index, previous fracture, parental fracture, current smoking, glucocorticoid use, rheumatoid arthritis, secondary osteoporosis, and hazardous alcohol use). For settings where dual-energy x-ray absorptiometry is available, sensitivity of FRAX can be improved by adding bone mineral density (BMD) to the algorithm. However, our prior studies suggest that FRAX underperforms among PWH, exhibiting poor discrimination (C statistic 0.59) and sensitivity (0.06). We believe this is because FRAX does not account for risk factors that are important to PWH. For example, FRAX does not account for falls, hepatitis C and B viruses, substance use disorder, multimorbidity, polypharmacy, use of other medications associated with fractures, or physiologic frailty – all of which have been associated with fractures among PWH. Finally, FRAX does not account for antiretroviral therapy (ART) use. Individual ART as well as ART classes have been independently associated with fracture. Leveraging an outstanding team of collaborators and research platforms, we will develop, validate and disseminate a Fracture Risk Assessment Calculator based on FRAX but that incorporates HIV-relevant risk factors (FRAC-HIV) to calculate the probability of a fracture among PWH in the next year. We have extensive experience working with the Veterans Aging Cohort Study (VACS), the largest cohort of PWH in North America, which will serve as our development cohort. To assess generalizability, we will externally validate the model using electronic health record data from Yale New Haven Healthcare System and Boston Medical Center. Our final aim sets the stage for the implementation of this predictive model via the development of a web-based fracture risk prediction calculator, which will be directly informed by insights from providers who care for PWH. FRAC-HIV will reconceptualize and facilitate clinical decision-making for fracture prevention in PWH by emphasizing the role of clinically modifiable risk factors. Our findings will lay the foundation for future intervention trials that will incorporate FRAC-HIV into clinical care algorithms to prevent fractures among PWH.