Comparative Risk of Fracture in Men and Women with HIV

Abstract

Context:

Men and women with HIV have an increased risk of fracture compared with individuals without HIV; however, it is unknown if women with HIV fracture at higher rates than men.

Objective:

We aimed to compare the incidence rates (IR) of fractures between men and women with HIV.

Design:

The study was designed as a cohort study, examining medical records from November 2001 to August 2012.

Setting:

The study was performed using records from two tertiary-care hospitals in Boston, Massachusetts.

Patients:

The study patients were adults with HIV: this was defined by diagnosis codes for HIV on two visits, at least one prescription for antiretroviral therapy, and at least 18 years of age.

Intervention:

IRs per 1000 person-years of all fractures and fractures at osteoporotic sites were calculated. We calculated IRs within age and gender strata and estimated IR ratios (IRR) between men and women.

Main Outcome Measure:

The main outcome measure was fracture at any site.

Results:

We identified a cohort of 3161 HIV-infected patients (869 women and 2292 men) with a total of 587 fractures. The IRR of all fractures was 1.00 (95% confidence interval [CI] 0.83–1.19) between men and women. The IR of fractures at osteoporotic sites among men was 15.2 (95% CI 12.7–17.6) per 1000 person-years compared with 12.1 (95% CI 8.6–15.6) in women, with IRR of 1.26 (95% CI 0.90–1.75). Men had similar or higher IRs than women for osteoporotic site fractures across most age groups.

Conclusions:

This study found similar rates of fracture in men and women with HIV. Further studies validating these findings are required to determine whether men with HIV should be screened for osteoporosis.

The Centers of Disease Control and Prevention estimates that by 2015 over half of the population of people living with HIV in the United States will be over 50 years of age (1). As the HIV population ages, chronic comorbid diseases, such as osteoporosis, are being identified at increasing rates. Emerging data show that people living with HIV are more likely to have low bone mineral density (BMD) than the general population; these data were summarized in a meta-analysis suggesting that people with HIV are over three times more likely to have osteoporosis than people without HIV (2).

Many potential etiologies could explain the increased risk of low BMD in individuals with HIV. Prior research finds higher prevalence of certain risk factors for osteoporosis, such as smoking, among patients living with HIV (3, 4). Other studies suggest certain antiretroviral therapies, such as tenofovir, may decrease BMD particularly during initiation of treatment (5, 6). In addition, the virus itself may alter bone metabolism (7).

Among the HIV population, high rates of osteoporosis are found not only in women but also in men as well. Several of the initial cross-sectional studies finding a higher risk of low BMD among people with HIV were performed in predominantly male populations (8, 9). These findings could potentially be explained by a greater prevalence of risk factors for osteoporosis among men with HIV, such as low body mass index. However, a study that controlled for some of these potential risk factors still found that men with HIV have lower BMD than uninfected men (10).

These reductions in BMD among persons with HIV appear to translate into increased rates of fracture. An early study of fractures among patients with HIV used the same electronic medical records (EMR) database as used in the present study and found a significantly higher prevalence of fractures among HIV-infected individuals (2.87 fractures per 100 persons) compared with uninfected patients (1.77 per 100 persons) in a large urban health care system (11). This increased risk of fracture was observed among both women and men with HIV. Other investigators have replicated these findings of higher fracture rates among individuals with HIV in many different cohorts (12–15).

Although several studies show that both men and women living with HIV have higher rates of osteoporosis and fracture as compared with those without HIV, we find no prior studies specifically comparing fracture rates between men and women with HIV. It is unclear whether the aging HIV population mirrors the general aging population, in which women fracture more frequently than men. This study aims to compare fracture rates between men and women with HIV, at both osteoporotic and nonosteoporotic sites and by age strata.

Materials and Methods

Study design

We performed a cohort study examining fracture rates among men and women with HIV. The fractures rates were compared between men and women and stratified by age at time of entry into cohort. A subject contributed follow-up time from cohort entry until the patient's last visit in the study period. To ensure completeness of follow-up, if there was a gap of greater than 18 months between visits, this gap period did not contribute to follow-up time for that particular patient.

The Partners Human Resource Committee institutional review board located in Boston, Massachusetts approved this study (protocol no.: 2011-P-001949/2).

Study population

We selected patients through use of an EMR database (the Research Patient Data Registry, RPDR) that collects patient data from two large tertiary care hospitals in Boston, Massachusetts. Patient data from November 1, 2001 to August 29, 2012 were searched for potentially eligible subjects: individuals who had received a diagnosis of HIV (International Classification of Diseases 9 = 042.xx, 043.xx, 044.xx) on two separate visits, had at least one prescription for antiretroviral therapy, and were at least 18 years of age at time of the second visit that included an International Classification of Diseases code for HIV. This RPDR definition of HIV diagnosis was validated by comparing it with a detailed chart review of 50 sample patients (performed by L.G.). The positive predictive value (PPV) of the definition of HIV within RDPR was 100% (95% confidence interval [CI] 93%–100%).

Outcome and variables of interest

The primary outcome of interest was fracture at any site, defined through the RPDR by either one diagnostic code or one notation of fracture on a patient's problem list during the study period (please see Supplemental data, published on The Endocrine Society's Journals Online web site at http://jcem.endojournals.org, for diagnostic codes used to define fracture). The secondary outcomes were 1) fractures occurring at sites typical of osteoporosis (hip, wrist, spine) and 2) fractures at all other sites. Only the patient's first fracture at each site was recorded for purposes of this study. Both age and gender of each subject were collected, and age was determined at time of entry into the cohort. We validated the RPDR-based definition of fracture by comparing it to a detailed chart review of 50 sample patients who had fractures as determined by RPDR (performed by L.G.). The PPV of the definition of fracture in our electronic database was 90% (95% CI 78%–97%).

Statistical analysis

The incidence rate (IR) of fracture per 1000 person-years was calculated for the primary and secondary outcomes. The IR for each of the fracture outcomes was then stratified by gender and age categories (≤35 y, 36–45 y, 46–55 y, 56–65 y, and ≥66 y). The follow-up times for each age category were used to calculate the age-stratified IR. The relative risk of fracture between genders in all strata was estimated by the IR ratios (IRR), with 95% CI calculated for all IR and IRR. Although data were not available to adjust for all confounders, a detailed review of a subset of 100 male and 100 female subjects was performed to determine whether there were significant gender differences in the prevalence of select potential risk factors for fracture, including ever-use of protease inhibitors, nadir CD4 less than 200, hepatitis C coinfection, tobacco use, iv drug use, ever-use of steroids, chronic renal failure, and osteoporosis. We performed the analysis with SAS software (v9.3; SAS Institute, Inc).

Results

The study cohort consisted of 3161 patients with HIV who met eligibility criteria, including 2292 men (72.5%) and 869 women (27.5%). A total of 13 506 person-years of follow-up were observed, with mean follow-up time per patient of 4.27 years (range 0–10.7 y). The mean age of the patients was 44.3 years for men and 40.9 for women.

Within the study cohort, we identified 587 total fractures, with 421 occurring in men (71.7%) and 166 in women (28.3%). The IR per 1000 person-years of all fractures was 43.4 (95% CI 39.3–47.6) for men and 43.6 (95% CI 36.9–50.2) for women. The IRR of all fractures between men and women was 1.00 (95% CI 0.83–1.19).

Men sustained 147 fractures at osteoporotic sites, as compared with 46 among women. The IR per 1000 person-years was 15.2 (95% CI 12.7–17.6) for men and 12.1 (95% CI 8.6–15.6) for women, with an IRR of men as compared with women of 1.26 (95% CI 0.90–1.75). The IRRs across age strata demonstrated modestly higher risk of fractures among men than among women less than 46 years of age (IRR ranging from 1.26 to 1.54), and slightly lower risk of fractures among men aged 46–55 (IRR = 0.88). The IRRs become more similar across genders in the oldest age groups. However none of these age-stratified IRRs comparing men and women was statistically significant (Table 1).

Table 1.

IR of Fractures Occurring at Osteoporotic Sites (hip/femur, wrist/forearm, and vertebrae/spine)

	Men, Number of Fractures	Men, Follow-Up Time, y	Men, IR of Fracture/1000 Person-Years (95% CI)	Women, Number of Fractures	Women, Follow-Up Time, y	Women, IR of Fracture/1000 Person-Years (95% CI)	IRR (men/women) (95% CI)
All age groups	147	9696.4	15.2 (12.7–17.6)	46	3809.5	12.1 (8.6–15.6)	1.26 (0.90–1.75)
Age ≤35	9	943.7	9.5 (3.3–15.8)	5	792.9	6.3 (0.8–11.8)	1.51 (0.51–4.51)
Age 36–45	38	3555.6	10.7 (7.3–14.1)	10	1444.4	6.9 (2.6–11.2)	1.54 (0.77–3.10)
Age 46–55	56	3577.4	15.7 (11.6–19.8)	21	1179.5	17.8 (10.2–25.4)	0.88 (0.53–1.45)
Age 56–65	33	1340.6	24.6 (16.2–33.0)	7	311.3	22.5 (5.8–39.2)	1.09 (0.48–2.47)
Age ≥66	11	279.1	39.4 (16.1–62.7)	3	81.5	36.8 (0–78.5)	1.07 (0.30–3.84)

We found that 274 men and 120 women experienced fractures occurring at nonosteoporotic sites. The IR per 1000 person-years was lower for men than women across most age groups, although the IRRs comparing men and women across all age groups were not statistically significant (Table 2).

Table 2.

IR of Fractures Occurring at Sites Not Associated With Osteoporosis

	Men, Number of Fractures	Men, Follow-Up Time, y	Men, IR of Fracture/1000 Person-Years (95% CI)	Women, Number of Fractures	Women, Follow-Up Time, y	Women, IR of Fracture/1000 Person-Years (95% CI)	IRR (men/women) (95% CI)
All age groups	274	9696.4	28.3 (24.9–31.6)	120	3809.5	31.5 (25.9–37.1)	0.90 (0.72–1.11)
Age ≤35	12	943.7	12.7 (5.5–19.9)	15	792.9	18.9 (9.3–28.5)	0.67 (0.31–1.44)
Age 36–45	85	3555.6	23.9 (18.8–29.0)	35	1444.4	24.2 (16.2–32.3)	0.99 (0.67–1.46)
Age 46–55	1254	3577.4	34.9 (28.8–41.1)	54	1179.5	45.8 (33.6–58.0)	0.76 (0.55–1.05)
Age 56–65	39	1340.6	29.1 (20.0–38.2)	10	311.3	32.1 (12.2–52.0)	0.91 (0.45–1.81)
Age ≥66	13	279.1	46.6 (21.3–71.9)	6	81.5	73.7 (14.7–132.6)	0.63 (0.24–1.66)

On comparison of select potential risk factors for fracture among a subset of 100 men and 100 women in our dataset, we found that the distribution of these risk factors was similar across men and women with the exceptions of “nadir CD4 < 200” (P = .008) and “ever-use of steroids” (P = .013), both of which were observed more often in women.

Discussion

As men and women with HIV live longer, they are developing a range of chronic health conditions that may be present at higher rates than in the general population. Osteoporosis appears to be more frequent in patients with HIV and may predispose them to fracture. We examined the EMR data from a large health care system to determine the age- and gender-stratified IRs of fracture among patients with HIV. We found that men and women with HIV sustain fracture at similar rates across most age categories. This finding persisted for all fracture types, fractures occurring at osteoporotic sites, as well as for fractures occurring at sites not classically associated with osteoporosis.

When examining only those anatomic sites associated with osteoporotic fractures, we found that men tended to have a higher IR of fracture as compared with women until 46 years of age, at which point fracture rates were higher in women. We note, however, that the IRRs were not statistically significant, suggesting that the observed associations between age, sex, and fracture rates may have occurred by chance alone. Our data do not permit us to explain this particular age- and gender-stratified pattern of fractures. It is possible, for instance, that young men may have a greater incidence of traumatic fracture, whereas women's risk of osteoporosis may increase rapidly at the time of menopause. The fracture rate in women therefore may “catch up” to men's fracture rate during middle age when bone loss is most rapid in women (16). However, without individual data on the mechanism of injury and several other potential risk factors for fracture in our dataset, our study was not able to determine the reason for our finding that men and women fracture at similar rates, and was unable to estimate to what extent HIV itself modifies these fracture rates. It is also important to note that our study was not adequately powered to examine the differences between age categories.

The rates of fracture in this study are higher than those found in other cohorts examining fracture in patients with HIV. For example, two study cohorts found fracture rates ranging from 3.3–4.0 per 1000 patient-years (17, 18) vs 43.4–43.6 in this study. These cohort studies required patients to volunteer for follow-up and thus may have a slightly healthier population of patients with HIV as compared with our hospital-based EMR database. In addition, the two prior studies used self-reported fracture, which could underestimate true fracture rate (19). A large population-based cohort in Denmark found fracture rates of 21 per 1000 patient-years (14), again lower rates than the present study. It is possible that differences in the population characteristics, such as age and gender, may contribute to the different rates of fracture in the Danish cohort as compared with our study. It is also possible that the present study has overestimated fractures; however, the PPV of the fracture definition was relatively high when compared with manual chart review.

A limitation of the present study is the potential for misclassification of osteoporotic fracture. Our dataset does not contain information about whether fractures were related to trauma, nor do we have BMD data for our subjects. Several prior studies among patients with HIV using large databases have also used wrist, hip, and spine as possible as markers for osteoporotic fracture (11, 20). The growing literature concerning the “premature aging” of the HIV population, in particular in relation to bone disease (21), also provides theoretical support that this definition of osteoporotic fractures may be applied reasonably to some persons within the HIV population. More direct validation of this classification in younger HIV-infected patients would be useful for the field; we encourage further research on this issue. We are also limited in this study by not having data of potential confounders that could contribute to fracture, such as CD4 nadir or hepatitis C coinfection. Although it is possible that HIV is such a strong predictor of fracture that the role of other traditional risk factors for fracture, such as gender, are weakened, as is suggested by a recent paper (22), we were unable to evaluate this in our study.

In summary, we found that men and women with HIV fracture at similar rates across most age categories. Because of our large database and epidemiologic design, we were able to obtain a population large enough to be able to suggest there is no significant difference in fracture between men and women. However, our analysis cannot determine whether such fractures are due to trauma or metabolic bone disease linked to HIV. Implications of our study for bone disease screening will remain unknown until studies further clarify the potential link between HIV and nontraumatic fracture in particular.