Data collection and variables
Two trained researchers engaged in geriatrics or geriatric nursing collected data by face-to-face interviews using a structured fall risk checklist. The interviewees were the elderly, their principal caregivers and family doctors (nurses). Written informed consent was obtained from the respondents before the investigation.
The survey items included gender, age, history of falls, diseases and symptoms, medications, difficulties with physical activity, psychological status, cognitive status, and abnormal behaviors of the elderly. Among them, diseases, symptoms, and medications were primarily obtained from medical and nursing records.
History of falls: The subjects were asked whether they had fallen within the past year. When the answer was "yes", they were further queried on whether they had been injured and if they had fallen ≥ 2 times. We defined a ‘severe fall’ by the occurence of fall-related injuries or when the number of falls ≥ 2 times.
Diseases and symptoms: The subjects were asked whether they had ischemic heart disease, hypertension, stroke, diabetes, Parkinson’s disease, urinary incontinence/frequent urination, orthostatic hypotension, dizziness while standing, and sensory disturbances. Sensory disturbances primarily included vision/hearing loss and numbness/pain in legs and feet.
Medications: The subjects were asked about their fall-risk related medications, including sedative hypnotics (ATC N05C), antipsychotics (ATC N05A), diuretics (ATC C03), and antihistamines (ATC R06). In addition, the elderly subjects were queried on whether they had a history of multiple-drug use involving more than four types of drugs.
Difficulties with physical activity: The subjects were asked whether they felt unstable while walking; whether hand support was necessary when getting up from a chair; whether crutches, walking aids, and other auxiliary equipment were needed for conducting daily activities.
Psychological status: The subjects were asked whether they suffered from insomnia/early awakening, depression, and the fear of falling.
Cognitive status: The subjects were asked whether they were able to communicate normally, and comprehend and follow the instructions of doctors and nurses. Based on the responses of "always", "sometimes", and "cannot", the cognitive status of the subjects was divided into "normal", "mild disorder", and "moderate/severe disorder".
Abnormal behaviors: The subjects were asked whether or not they followed necessary safety measures and whether they had been loitering.
Statistical analysis
SPSS24.0 statistical software was used for data analysis. The general information and main variables of the study were described by mean value, standard deviation, frequency, and percentage. A quantile-quantile plot (Q-Q plot) was performed to analyze the normality of variables. The chi-square test was used for univariate analysis. Following univariate analysis, a forward binary logistic regression analysis was performed to identify the independent risk factors of falls, recurrent falls, or fall-related injuries of the elderly. The Hosmer-Lemeshow statistics were used to test the goodness-of-fit of the model. Eventually, the ability of the model to identify fall risk in the elderly was quantified by the area under the receiver operating characteristic curve (AUC). p < 0.05 was considered statistically significant.