Heart Failure is an Independent Risk Factor for Incident Hip, Proximal Humerus, and Wrist Fractures in a High-Risk Older Population

This retrospective study included Kaiser Permanente Northern California members aged 65–84 years who initiated oral bisphosphonate therapy in 2010–2019 (index date). Those treated with other osteoporosis drugs and those with multiple myeloma, secondary metastatic cancer, selected bone disorders, or receipt of kidney dialysis/transplant were excluded. Prevalent HF was ascertained using ≥ 1 hospital diagnosis or ≥ 3 outpatient diagnoses within five years before index. Incident hip fracture (hospital diagnosis) and proximal humerus and distal radius/ulna (wrist) fractures (hospital/outpatient diagnosis) were ascertained during three years of follow-up using International Classification of Diseases (ICD-9/10-CM) diagnosis codes (excluding fractures in the first year if a same-site fracture diagnosis occurred before index and within one year of the incident fracture diagnosis). Baseline covariates included age, self-reported race, ethnicity, smoking, diabetes, rheumatoid arthritis, prior fracture, and body mass index (Table 1 ). Continued bisphosphonate therapy was based on ≥ 80% adherence (≥ 50–90% in sensitivity analysis) for each 3-month interval of follow-up. Fracture incidence was compared by HF status using the log-rank test. The association of HF and incident fracture was examined using Cox proportional hazards regression, adjusting for baseline covariates, continued osteoporosis therapy, and an interaction term for baseline HF status and sex. These analyses were repeated with Fine-Gray regression models, accounting for death as a competing risk. Adjusted hazard ratios (aHR) are reported with 95% confidence intervals (CI). Among 60,892 adults (82.5% women, 65.7% non-Hispanic White, 18.1% Asian/Pacific Islander, 11.8% Hispanic, and 3.0% Black) who initiated bisphosphonate therapy, one-third had prior fracture (34.9%). Men were more likely to be age ≥ 75 years (69.6% versus 38.6%) and have diabetes (26.3% vs 20.4%), and less likely to have obesity (17.3% versus 21.1%) than women (Table 1 ), but current smoking prevalence was similar (7.7% vs 8.0%). During up to three years follow-up, the unadjusted incidence (per 1000 person-years, 95% CI) of hip, proximal humerus, and wrist fractures was 6.0 (5.6–6.4), 4.4 (4.1–4.8), and 7.1 (6.6–7.5) for women and 8.4 (7.4–9.5), 2.7 (2.2–3.4), and 3.0 (2.4–3.7) for men, respectively. For each skeletal site, cumulative fracture incidence was higher among those with versus without HF (p < 0.001). In multivariable analyses adjusting for potential confounders, HF was independently associated with increased risk of hip (aHR 1.48 [1.20–1.83]), proximal humerus (aHR 1.46 [1.10–1.92]), and wrist fracture (aHR 1.37 [1.07–1.76]), with no significant interaction by sex (Fig. 1 ). Results were unchanged in sensitivity analyses, using ≥ 50% adherence criteria for bisphosphonate continuation. Multivariable association of heart failure and incident major osteoporotic fracture among adults who initiated osteoporosis therapy. Separate multivariable Cox proportional hazard models were conducted for each fracture outcome, adjusting for age, sex, self-reported race and ethnicity, prior fracture, current smoking status, diabetes, rheumatoid arthritis, body mass index (category), and continued osteoporosis therapy during follow-up. There was no significant interaction between heart failure and sex. Adjusted hazard ratios were unchanged when examined using Fine Gray regression models with death as a competing risk. Results were similar/unchanged when using lower (≥ 50%, ≥ 70%) or higher (≥ 90%) adherence to define treatment continuation.