Risk factors for falls in HIV-infected persons

Abstract

Background: The incidence of and risk factors for falls in HIV-1-infected persons are unknown.

Methods: Fall history during the prior 12 months, medical diagnoses, and functional assessments were collected on HIV-infected persons 45-65 years of age receiving effective antiretroviral therapy. Fall risk was evaluated using univariate and multivariate regression analyses.

Results: Of 359 subjects, 250 persons (70%) reported no falls, 109 (30%) had ≥1 fall; and 66 (18%) were recurrent fallers. Females, whites, and smokers were more likely to be recurrent fallers (P ≤ 0.05). HIV-related characteristics including current and nadir CD4 T-cell count, estimated HIV duration, and Veterans Aging Cohort Study Index scores were not predictors of falls (all P ≥ 0.09); didanosine recipients were more likely to be recurrent fallers (P = 0.04). The odds of falling increased 1.7 for each comorbidity and 1.4 for each medication (P < 0.001) and were higher in persons with cardiovascular disease, hypertension, dementia, neuropathy, arthritis, chronic pain, psychiatric disease, frailty, or disability [all odds ratio (OR) ≥ 1.8; P ≤ 0.05]. Beta-blockers, antidepressants, antipsychotics, sedatives, and opiates were independently associated with falling (all OR ≥ 2.7; P ≤ 0.01). Female gender, diabetes, antidepressants, sedatives, opiates, didanosine, exhaustion, weight loss, and difficulty with balance were the most significant predictors of falls in logistic regression (all OR ≥ 2.5; P ≤ 0.05).

Conclusions: Middle-aged HIV-infected adults have high fall risk. Multiple comorbidities, medications, and functional impairment were predictive of falls, but surrogate markers of HIV infection or an HIV-specific multimorbidity index were not. Fall risk should be assessed routinely as part of the care of HIV-infected persons.

Figure 1

Risk of recurrent falls estimated by regression model. A) Fall risk shown by area under the receiver-operator-characteristic (ROC) curve, as estimated from the logistic regression model. B) Probability, sensitivity, and specificity of each additional fall risk factor. The risk factors identified through regression analyses were: difficulty with tandem stand, antidepressant use, exhaustion, diabetes, female gender, shrinking, opiate use, current/prior didanosine, and sedative use.