Bone fracture among people living with HIV: A systematic review and meta-regression of prevalence, incidence, and risk factors

Introduction People living with HIV (PLWH) had a higher prevalence and incidence rate of bone fracture compared to general population. Although several studies have explored this phenomenon, the prevalence and incidence rate of fracture were varied. Objective The aim of the study is to determine and analyze the pooled prevalence, incidence rate of fracture and fracture risk factors among people living with HIV (PLWH). Methods PubMed, Cochrane Library, CINAHL with full Text, and Medline databases for studies published up to August 2019 were searched. Studies reporting the prevalence or incidence of fracture among PLWH were included. Study quality was assessed using the Joanna Briggs Institute (JBI) appraisal tool. A meta-analysis with random-effects model was performed to determine pooled estimates of prevalence and incidence rates of fracture. A meta-regression was performed to determine the source of heterogeneity. Results The pooled estimated prevalence of fracture among PLWH was 6.6% (95% CI: 3.8–11.1) with pooled odds ratio of 1.9 (95%CI: 1.1–3.2) compared to the general population. The pooled estimates of fracture incidence were 11.3 per 1000 person-years (95% CI: 7.9–14.5) with incidence rate ratio (IRR) of 1.5 (95% CI: 1.3–1.8) compared to the general population. Risk factors for fracture incidence were older age (aHR 1.4, 95% CI: 1.3–1.6), smoking (aHR 1.3, 95% CI: 1.1–1.5), HIV/HCV co-infection (aHR 1.6, 95% CI: 1.3–1.9), and osteoporosis (aHR 3.3, 95% CI: 2.2–5.1). Conclusions Our finding highlights a higher risk of fracture among PLWH compared to the general population. Osteoporosis, smoking and HIV/HCV coinfection as the significant modifiable risk factors should be prioritized by the HIV health providers when care for PLWH.


Introduction
The success of highly active antiretroviral therapy (HAART) over the past decades has increased the life expectancy among people living with HIV (PLWH) [1]. PLWH have increased co-morbidities, including bone fractures, which have emerged to become an important healthcare issue [2]. PLWH experience a greater risk of fractures, ranging from 1.6 to 3 times that of the general population [3]. The risk of fracture among PLWH has been associated with an increasing rate of osteoporosis due to accelerated bone loss [4,5], bone microstructure alterations, and a reduction in bone mineral density [6]. The International Osteoporosis Foundation noted that bone fractures result in chronic pain, long-term disability, and even death [7].
A proactive approach in the prevention and treatment program is needed to minimize the consequence of bone loss and morbidity associated with fractures among PLWH [8]. Therefore, identifying the prevalence, incidence, and risk factors for fractures among PLWH is important to evaluate a prevention program. However, the current prevalence and incidence rate data are varied. Previous studies have focused on the high prevalence of osteoporosis among PLWH (15%) but not fracture risk [9]. A cohort study of PLWH identified a fracture incidence rate of 31 per 1000 person-year [10], however, a rate of 13.5 per 1000 person-year was reported in another study [3]. Arnsten's study [10] only included older subjects (� 49 years), while older age was found as a risk factor for fractures among PLWH in a study by Dong et al [11].
The risk factors for fracture among PLWH remain unclear. Previous studies have been limited to select HIV populations or have included osteoporosis, not fractures, as an outcome. Earlier studies on fracture risk factors among PLWH showed a varied result. A study by Bedimo et al. found the exposure to tenofovir (TDF) regimen was the risk factors for fracture [12]. Meanwhile, Arnsten et al. found low bone mineral density (BMD) as a significant risk factor for fracture [10] The variability of conclusive information on fracture among PLWH requires additional study. The study aimed to conduct a systematic review, and meta-regression to determine the pooled prevalence, incidence rate of fracture and fracture risk factors among PLWH.

Search strategy
Using PRISMA guidelines for a systematic review, databases were searched up to 9 August 2019 included PubMed, Cochrane Library, CINAHL with full text, and MEDLINE [13]. The terms used in the searches varied according to the database utilized, thus included HIV, Human Immunodeficiency Virus, Acquired Immunodeficiency Syndrome, bone fracture, bone fragility, broken bone, osteoporosis, bone density, prevalence, incidence, risk factor, and epidemiology.
A study was eligible for inclusion if it included HIV-infected subjects aged 16 years old or above and reported fracture prevalence or incidence rate data. Cross-sectional, cohort, and casecontrol studies with or without a control group were included. Studies were excluded if they were not in English. Two researchers (IP, LL) independently screened all titles, abstracts, and full texts and appraised study quality. The disagreement was resolved by a third researcher (CY).

Data extraction
Data extraction included author, year of publication, study location, study design, demographics, sample size, prevalence and incidence rates of fracture, and method of fracture classification. Prevalence of fracture was expressed as a percentage of fracture among HIV positive patients from the total of HIV positive patients. Incidence was expressed as the number of fractures divided by the period of risk for all included subjects. Prevalence and incidence rate data were obtained from the studies selected. All reported cases of osteoporosis in the included studies were diagnosed by using Dual-Energy X-ray Absorptiometry (DXA). When data was not provided, the prevalence rate was calculated by determining the number of fractures of the total HIV positive sample.

Quality assessment
Study quality was assessed using the Joanna Briggs Institute (JBI) critical appraisal checklist tool for a prevalence study [14] The checklist consists of nine questions with four categories of answers: yes, no, unclear, and not applicable (insufficient data). "Yes" is scored as 1 and "no" is scored as 0, with a total quality score ranging from 0-9.

Data analysis
Pooled estimates used a random-effects model because of heterogeneity indicated by the I 2 statistic (values of <25%, 25-75%, and >75% representing low, medium, and high heterogeneity, respectively) [15]. Subgroup analysis and meta-regression were performed to determine potential sources of heterogeneity. Publication bias test were performed using Egger test and Funnel plot. Statistical analysis was conducted using the Comprehensive Meta-Analysis Software (Version 3.0 Biostat, Englewood, New Jersey, USA).

Results
Four databases provided 843 articles from the year 2000 to February 2017 (Fig 1). After excluding duplicates, and applying inclusion and exclusion criteria, 45 articles remained. After fulltext examination, 21 articles remained for quality appraisal with all receiving a quality score above 7. Twelve studies received a score of 8 out of 9, and the remaining received a score of 9 out of 9 (Supporting information). Overall, the study quality was acceptable.

Bone fracture prevalence
The pooled prevalence of the 21 studies reporting bone fracture among people living with HIV (PLWH) was 6.6% (3.8-11.1) with a high level of heterogeneity ( Table 2). The prevalence of PLWH was higher than in the general population with an odds ratio of 1.9 (95% CI: 1.1-3.2) (Fig 2). Males showed a higher prevalence compared to females (6.2% vs 4.9%, respectively). The injecting drugs users (IDUs) PLWH showed the highest prevalence of fracture (8.6%) among the risk group. PLWH who are receiving HAART showed a higher prevalence of fracture compared to those without HAART (6.7% vs 3.5%, respectively).

Bone fracture incidence
The pooled incidence of the 13 reports bone fracture among people living with HIV (PLWH) was 11.3 per 1000-person years (95% CI: 7.9-14.3) with a high level of heterogeneity ( Table 3). The pooled rate ratio was calculated from six studies reporting fractures of the control group as 1.5 (95% CI: 1.3-1.8) (Fig 3).

Meta-regression analysis and publication bias
A high level of heterogeneity existed across the studies reporting prevalence (I 2 = 99.8%, Table 2). Univariate analysis was conducted to find significant independent covariates as a

PLOS ONE
a systematic review of bone fracture prevalence among people living with HIV potential source of heterogeneity that could be included in a meta-regression model ( Table 5).
The covariates of study design (p < 0.001) and fracture identification method (p < 0.05) showed significant coefficient regression. After these covariates were included in the metaregression analysis, only study design covariates provided a significant coefficient regression (p < 0.05). Publication bias was analyzed by generating a funnel plot and Egger's test with neither demonstrating evidence of asymmetry (Fig 4).

Discussion
The findings of this study revealed that people living with HIV have an increased risk of bone fracture, 1.9 times higher than that observed in the general population. This finding was higher than a systematic review of HIV postmenopausal women that reported a fracture prevalence 1.5 times greater than HIV negative postmenopausal women [36]. The greater percentage of males in the studies included in the current study may explain the difference. People living with HIV experienced an incidence rate of fracture 1.5 times that of the general population. Dong and colleagues' meta-analysis reported a similar incidence rate ratio of fracture among PLWH of 1.8 [11]. However, the analysis only focused on subjects with HIV and HCV co-infection which may have yielded a higher incidence rate ratio. Maalouf et al. have reported a higher risk of fracture among patients with HIV/HCV co-infection compared to HIV mono-infection which was identified as a risk factor in the current study [20].
The four risk factors found significant in this meta-analysis have previously reported including older age, osteoporosis, smoking and HCV co-infection. A meta-analysis by Shiau et al. identified similar results but did not report pooled hazard ratios [37]. Older age as a traditional risk factor in this study was reported in a meta-analysis by Dong et al. who studied risks among HIV and HCV co-infected patients [11]. Osteoporosis as a risk factor for fractures showed a lower pooled adjusted hazard ratio than Short et al. [38] who reported that HIV infected men with osteoporosis had a 3.5-fold risk of fracture compared to those without osteoporosis. However, the limited population of the study may have overestimated the findings. Smoking, a significant and modifiable fracture risk factor among PLWH, is associated with a bone mineral density which is influenced by both dose and duration of smoking [39]. Wu et al. [40] reported that smokers had twice the risk of fracture compared to non-smokers.
Smoking cessation programs among smokers with HIV may reduce fracture risk after just three months [41]. The HCV co-infection risk, with pooled aHR of 1.6, confirms the findings of Dong et al's meta-analytic study. [11]. Another meta-analytic study by Shiau et al. [37] also found that HCV co-infection was an independent risk factor for both fragility and non-fragility fractures. Early intervention for HCV co-infection and PLWH, recommended by Bedimo et al. is critical to reducing bone turnover [42] The current study adds to the literature by reporting both the prevalence and incidence rate of fractures including the HIV-related risk factor for fracture among people living with HIV. The studies included in the meta-analysis included sample sizes exceeding 1,000 and a diverse population of patients with HIV. Previous reviews focused on smaller samples with at-risk populations. Finally, a pooled adjusted Hazard Ratio of fracture risk factors provide more specific analysis than a simple summary of factors.

Limitations
The literature search excluded non-English reports as the studies all originated in developed countries, fracture risk factors of less developed countries may differ. Despite subgroup and meta-regression analysis, the synthesized literature revealed significant heterogeneity.

Conclusions
A comprehensive meta-analysis revealed a higher prevalence and incidence rate of fracture among PLWH compared to the general population. Smoking, older age, osteoporosis, and HCV coinfection are significant risk factors for fracture among PLWH.
Our finding highlights a higher risk of fracture among PLWH compared to the general population. Osteoporosis, smoking and HIV/HCV coinfection as the significant modifiable risk factors should be prioritized by the HIV health provider when managing PLWH. Further investigation of the interactions among risk factors is urgently needed to prevent bone fractures.