Campanini Isabella, Mastrangelo Stefano, Bargellini Annalisa, Bassoli Agnese, Bosi Gabriele, Lombardi Francesco, Tolomelli Stefano, Lusuardi Mirco, Merlo Andrea
Department of Rehabilitation, LAM-Motion Analysis Laboratory, AUSL of Reggio Emilia, S. Sebastiano Hospital, Via Mandriolo Superiore 11, 42015, Correggio, RE, Italy.
Clinical and Experimental Medicine PhD Program, University of Modena and Reggio Emilia, Modena, Italy.
BMC Health Serv Res. 2018 Jan 11;18(1):18. doi: 10.1186/s12913-017-2815-x.
Falls are a common adverse event in both elderly inpatients and patients admitted to rehabilitation units. The Hendrich Fall Risk Model II (HIIFRM) has been already tested in all hospital wards with high fall rates, with the exception of the rehabilitation setting. This study's aim is to address the feasibility and predictive performances of HIIFRM in a hospital rehabilitation department.
A 6 months prospective study in a Italian rehabilitation department with patients from orthopaedic, pulmonary, and neurological rehabilitation wards. All admitted patients were enrolled and assessed within 24 h of admission by means of the HIIFRM. The occurrence of falls was checked and recorded daily. HIIFRM feasibility was assessed as the percentage of successful administrations at admission. HIIFRM predictive performance was determined in terms of area under the Receiver Operating Characteristic (ROC) curve (AUC), best cutoff, sensitivity, specificity, positive and negative predictive values, along with their asymptotic 95% confidence intervals (95% CI).
One hundred ninety-one patents were admitted. HIIFRM was feasible in 147 cases (77%), 11 of which suffered a fall (7.5%). Failures in administration were mainly due to bedridden patients (e.g. minimally conscious state, vegetative state). AUC was 0.779(0.685-0.873). The original HIIFRM cutoff of 5 led to a sensitivity of 100% with a mere specificity of 49%(40-57%), thus suggesting using higher cutoffs. Moreover, the median score for non-fallers at rehabilitation units was higher than that reported in literature for geriatric non fallers. The best trade-off between sensitivity and specificity was obtained by using a cutoff of 8. This lead to sensitivity = 73%(46-99%), specificity = 72%(65-80%), positive predictive value = 17% and negative predictive value = 97%. These results support the use of the HIIFRM as a predictive tool.
The HIIFRM showed satisfactory feasibility and predictive performances in rehabilitation wards. Based on both available literature and these results, the prediction of falls among all hospital wards, with high risk of falling, could be achieved by means of a unique tool and two different cutoffs: a standard cutoff of 5 in geriatric wards and an adjusted higher cutoff in rehabilitation units, with predictive performances similar to those of the best-preforming pathology specific tools for fall-risk assessment.
跌倒在老年住院患者和康复科收治患者中都是常见的不良事件。亨德里希跌倒风险模型II（HIIFRM）已在所有跌倒发生率高的医院病房进行了测试，但康复科除外。本研究旨在探讨HIIFRM在医院康复科的可行性和预测性能。
在意大利一家康复科进行了为期6个月的前瞻性研究，研究对象为来自骨科、肺科和神经科康复病房的患者。所有入院患者在入院24小时内采用HIIFRM进行登记和评估。每天检查并记录跌倒的发生情况。HIIFRM的可行性通过入院时成功评估的百分比来评估。HIIFRM的预测性能根据受试者工作特征（ROC）曲线下面积（AUC）、最佳截断值、敏感性、特异性、阳性和阴性预测值及其渐近95%置信区间（95%CI）来确定。
共收治191例患者。HIIFRM在147例患者中可行（77%），其中11例发生跌倒（7.5%）。评估失败主要是由于卧床患者（如最低意识状态、植物状态）。AUC为0.779（0.685 - 0.873）。HIIFRM最初的截断值5导致敏感性为100%，但特异性仅为49%（40 - 57%），因此建议使用更高的截断值。此外，康复科未跌倒患者的中位数得分高于老年未跌倒患者的文献报道得分。通过使用截断值8可在敏感性和特异性之间获得最佳平衡。这导致敏感性 = 73%（4,6 - 99%），特异性 = 72%（65 - 80%），阳性预测值 = 17%，阴性预测值 = 97%。这些结果支持将HIIFRM用作预测工具。
HIIFRM在康复病房显示出令人满意的可行性和预测性能。基于现有文献和这些结果，对于所有跌倒风险高的医院病房，可通过一种独特工具和两个不同的截断值来实现跌倒预测：老年病房标准截断值为5，康复科调整后的更高截断值，其预测性能与最佳的特定病理跌倒风险评估工具相似。
