Development Of Prediction Model For Accidental Falls Among Elderly Patients With Diabetes Mellitus

Objective:1.To understand the risk factors of falls in elderly patients with diabetes and to establish a risk factor questionnaire through a variety of methods.2.A cohort study was designed to explore the risk factors of falls in elderly patients with diabetes and to develop and evaluate the fall risk prediction model,so as to provide the evidence to predict and manage the accidental falls of elderly patients with diabetes.Methods:1.Extraction of risk factors for accidental falls in elderly patients with diabetes mellitus: Systematic literature review and semi-structured interviews were used to extract the risk factors of accidental falls,and expert consultation on the above risk factors was used to determine the questionnaire for the risk factors of accidental falls.2.Development and evaluation of a fall risk prediction model for elderly patients with diabetes: A cohort study was designed,and a convenient sampling method was used to collect the baseline data of 1007 elderly patients with diabetes in a tertiary specialist diabetes hospital in Tianjin from April to August 2019.All the patients were followed up for 6 months to determine outcomes of accidental falls.The SPSS 21.0 was used to analyze the data.After the univariate analysis and multivariate logistic regression analysis,a fall risk prediction model was development.The Hosmer-Lemeshow test and receiver operating characteristic(ROC)were used to evaluate the calibration and discrimination of model.Sensitivity and specificity were used to evaluate the predictive ability of outcome.Results:1.After systematic literature review,semi-structured interview to patients and medical stuff,and expert consultation,we formed an assessment questionnaire of fall risk among patients with diabetes,which including social demographic factors,general health,diabetes related factors,other diseases related factors,functions and activities,psychological and cognitive,totally 7 parts.2.Among the 950 elderly patients included in the final analysis,a total of 133 falls occurred in 93 patients during the follow-up period,with a fall rate of 9.79%.Results of univariate and multivariate analysis showed that gender,fall history with 1 year,walking aids,depression,fatigue and diabetic peripheral neuropathy were independent predictors of accidental fall in elderly patients with diabetes.We constructed a fall risk assessment model for the elderly diabetic patients,logit(P)=-4.305+0.760× diabetic peripheral neuropathy +0.752× walking aids +0.683× depression +0.621× fall history within one year +0.601× fatigue +0.498× gender.The Hosmer-Lemeshow test was conducted at P>0.05,which demonstrated a good calibration,the area under the ROC curve was 0.687(95%CI:0.630～0.743).According to the results of regression analysis,a fall risk assessment scale with 6 factors,totally 8 scores was constructed When the cut-off value was 3 points,the sensitivity and specificity of the model were 82.80% and 41.77%,respectively.Conclusion:In the present study,several methods were combined to conducted a fall risk questionnaire for elderly patients with diabetes.Through univariate and multivariate analysis,gender,fall history within one year,fatigue,walking aids,depression,and diabetic peripheral neuropathy were identified to be the independent predictors of accidental falls in elderly diabetic patients.This means that the presence of diabetes-specific risk factors is important for predicting the risk of falls in elderly patients with diabetes.Model had strong ability to detect falls in patients and had positive significance for preventing falls in hospital in elderly diabetes patients.