Implementation of a Multifactorial Fall Intervention Model in Hospital: a Longitudinal Before-and-after Controlled Study From 2015 to 2018

Background: Falls are serious public health problems associated with irreversible health consequences and substantial economic burden. To effectively reduce the incidence of falls and mitigate fall-related injuries, we designed and veried a multifactorial fall intervention model. Methods: The current study was a longitudinal before-and-after controlled investigation including 3 phases with clinical characteristics of fall patients retrospectively identied in phase 1, a multifactorial fall intervention model designed in phase 2 and prospectively evaluated in phase 3. Phase 1 and 2 were conducted based on 153,601 hospitalized patients between January 2015 and December 2016. Phase 3 was carried out based on 171,776 hospitalized patients between January 2017 and December 2018. The Pearson Chi-squared test was used to compare categorical variables and the Mann-Whitney non-parametric test was utilized for one-way ordered data. Results: In phase 1, baseline characteristics of 491 fall patients revealed that inpatients falls were highly associated with the age, medication and disease. In phase 2, a new multifactorial fall intervention model covering measures for fall prevention, fall-onset management and continuous improvement was developed. Phase 3 recorded a total of 396 falls and demonstrated a remarkably declined fall rate (Reduction in falls by 0.09%, p<0.001) and fall rate per 1000 patient-days (Reduction in falls/1000 patient-days by 0.07 ‰ , p<0.001) as compared with phase 1. The adjusted incident rate ratio of fall was 1.443 (95%CI: 1.263-1.647) (Phase 1 vs. Phase 3). Furthermore, the occurrence and the severity of fall injuries in phase 3 were signicantly lower than that in phase 1 (Z=-4.426, p<0.001). More specically, the number of uninjured falls accounted for 42.42% in phase 3 in comparison of 32.99% in phase 1. Conclusions: This multifactorial fall intervention model exhibited favorable effect on reducing the occurrence of fall and fall injuries. fall intervention model which included measures for pre-fall prevention, fall-onset management, and continuous improvement after falls occur. The multifactorial fall intervention model was proved to be signicantly effective in decreasing the occurrence of falls and the severity of fall injuries.


Introduction
Falls are reported to be the most common adverse events in hospitals, accounting for 20-30% of all incident reports.(1, 2) Due to unfamiliar environment, diseases and treatments, hospitalization increases fall risks. (1,3) In terms of hospital departments, geriatric and rehabilitation wards have higher rates of falls than surgical or acute care wards. (3) Falling can lead to different degrees of injury, prolong hospital stay, increase hospitalization costs, and even lead to legal disputes between the hospital and patients. (4)(5)(6)(7) Approximately 30% of in-hospital falls were associated with various degrees of physical injuries. (8) Even non-injurious falls may result in psychological stress, functional decline and increasing the morbidity of second falls. (9,10) In the past 20 years, considerable attention has been paid to reduce fall incidence and fall-related injuries in hospital settings. In particular, the exploration of fall prevention measures has attracted a large number of investigators. (11,12) Related research has also yielded many meaningful results, such as the progress of a series of fall assessment tools and the improvement of fall prevention measures for use in hospitals. (13)(14)(15)(16)(17)(18)(19) Correspondingly, the importance of interventions when falls occur and after falls are often overlooked. However, interventions for falls should be multiprocessing interventions involving fall prevention, fall-onset management and improvement measures after falls. In addition, many studies are only devoted to exploring the role of single factor changes in the prevention and treatment of falls. (20) Falls are multifactorial accidents, and interventions for falls should also include management of multiple risk factors. (21,22) In order to effectively reduce the occurrence of falls and mitigate the injury caused by falls, we designed and veri ed a multifactorial fall intervention model. To our knowledge, this is the rst systematic multifactorial fall intervention model that includes measures for pre-fall prevention, fall-onset management, and continuous improvement after falls. Our study included three phases investigation with phase 1 identifying fall patients characteristics, phase 2 proposing the multifactorial fall intervention model based on the results of phase 1, and phase 3 implanting and verifying the model in a hospital setting.

Method
The current study was performed following the STROBE checklist.

Study design
Our study was a longitudinal, before-and-after controlled study consisting of three phases of investigation. Phase 1 retrospectively identi ed clinical characteristics of fall patients, phase 2 designed the multifactorial fall intervention model and phase 3 prospectively evaluated the model.

Data source
Fall related information was collected by reviewing relevant medical records, nursing records and reports of adverse events as well as interviewing fall patients, family members of patients, doctors in charge of patients, and medical staff on duty.

Statistical methods
The Pearson Chi-squared test was used to compare categorical variables between phase 1 and phase 3 (Patient characteristics, fall rate and fall rate per 1000 patient-days). Mann-Whitney non-parametric test was utilized for one-way ordered data (Fall-related injuries). Based on the logistic regression analysis, we assessed the role of each factor in increasing the risk of falling and adjusted results. Data analyses were performed on SPSS 26.0. P value < 0.05 was considered statistically signi cant and all tests were twosided.
The medical ethics committee of hospital has approved the project.

Phase 1
Four hundred and ninety-one falls (3.2‰) were recorded. Among them, 329 patients suffered grade 1 and above injuries and 49 patients were assessed as grade 3 injuries. Analysis of baseline characteristics of falling patients revealed that inpatients' falls were highly associated with the age, medication, clinical features and disease of patients. More than half of the falls (57.84%) occurred among patients aged 65 and over. Noticeably, in addition to the elderly, 47 falls of children aged 0-6 were also recorded, which accounted for a relatively high percentage (9.57%). According to medication records, 480 patients with falls were found to take drugs during hospitalization prior to the fall. Antihypertive drugs (39.58%), analgesics (15.63%) and hypoglycemic drugs (15.42%) were the most common used drugs in decreasing order of frequency. In terms of clinical features, the most common symptoms of fall patients were hypodynamia (40.69%), dizziness (23.33%) and anemia (10.31%). As for diseases of fall patients, circulatory system diseases (20.32%) were the most common, followed by metabolic system diseases (17.98%) and nervous system diseases (15.20%). As recorded, fall patients were mainly from the department of neurology (13.24%), respiratory medicine (12.83%) and pediatrics (10.18%).
In addition, in-patient falls were also obviously characterized by time, place and activities. In-patient falls peaked between 2 am and 8 am (36.86%) and between 4 pm and 6 pm. About places of falling, most falls were found at the bedside or in the washroom (73.52%). Correspondingly, most people were toileting or moving around the bedside before they fell down (78.82%).
Baseline characteristics of fall patients were shown in Table 1 and Fig. 1.

Phase 2
The speci c research progress of the fall intervention model is shown in supplementary le 1.

Leadership and work-team
Our fall intervention team led by the head nurse of the department, with doctors, nurses, rehabilitation teachers and care-taking staff as core members.

Fall risk assessment
The original fall risk assessment tool in our hospital refer to the Thomas Fall Risk Assessment Scale and the Johns Hopkins Fall Risk Assessment Tool. Based on our rst phase of clinical data summary and literature reading, we designed and improved the fall assessment tool and further tested its effectiveness in a large base sample (Fig. 2).

Work-ow
Our fall intervention process covered three parts, including fall prevention process, fall management process, and long-term continuous improvement measures (Fig. 3). There was no signi cant difference in fall rates between high risk and low risk groups during phase 1 (Fall rate: 0.12% vs 0.07%, p = 0.104). However, during phase 3, patients in high risk group had higher odds of falls compared with patients in low risk group (Fall rate: 0.05% vs 0.01%, p = 0.015). Patients assessed as high risk received intensive intervention during phase 3. The number of falls decreased in patients treated with the intensive intervention during phase 3 than those without intensive intervention during phase 1 (Reduction in falls by 0.07%, p = 0.029).
In addition, the new intervention model displayed a sensitivity of 64.3% and a speci city of 86.8% with the are under the curve of 0.821 (96%CI: 0.757-0.885).
The comparison of inpatient falls and fall injuries between phase 3 and phase 1 was shown in Table 2. A summary gure highlighting the study ndings was shown in Fig. 4. Table 2 Comparison of Inpatient falls and fall injuries between phase 1 (2015-2016) and phase 3 (2017-2018

Discussion
Overall, the implementation of the multifactorial fall intervention model reduced the incidence of falls in our hospital from 0.31-0.22% and decreased the incidence of falls per 1000 patient-days from 0.29‰ to 0.22‰. In addition, the uninjured rate of falls also increased from 32.99-42.42%.
Compared with our previous fall intervention system, a signi cant improvement of this model is to prevent falls in high risk patients from multiple aspects, which are composed of not only nurses, but actively involving doctors, patients and family members and care taking staff. Nurses have always been the main force in the prevention and control of falls.(23) However, through analyzing the characteristics of fall patients, we found that falls were closely related to many factors, including doctors' treatment, patients' and their families' consciousness of preventing falls and environmental factors. And the control of these factors is far beyond the ability and responsibility of the nurse, it is more necessary for doctors, patients and their families and care-taking staff to perform their respective duties for these factors. It is important for doctors to pay particular attention to medication for patients at high risk of fall, and to adjust medications according to the result of patient's fall assessment. Previous studies showed that up to 85% of falls occur when patients are alone. (8,24) A randomized controlled trial demonstrated that providing education to patients enabled them to conduct safe behaviors in hospital and effectively alert staff that they need assistance, especially when alone.(13) Family members' perception of patients' risk of fall is an important factor affecting patients' fall, especially for children, the elderly and patients with cognitive impairment. (25,26) Hospital falls of elderly patients have been widely reported. As for children, fall is the most common injury mechanism for children, consisting about one-third of accidental injuries. (27,28) Our study identi ed that children between the ages of 0 and 6 as having a high risk of falling, which was consistent with previous reports. (29,30) Falls from furniture and stairs are important causes of children falls in hospital.(30) Therefore, greater awareness of risk factors in family members and medical staffs is required to predict and prevent falls in children.
With regard to environmental factors, many high-quality researches suggested marked reduction in fall risk after physical environmental intervention offered to high risk patients. (31,32) Our analysis of the characteristics of fall patients in phase 1 showed that 34.42% of falls occurred in the washroom, as the wet oor greatly increased the risk of falling. To speci cally reduce the risk factors for falls in the environment, care-taking staff were required to regularly evaluate and improve the patient's physical environment.
Few studies have focused on the importance of continuous improvement measures after falls, but they are effective interventions to further prevent repeated falls. According to the literature report, approximately 10% of elderly patients experienced recurrent falls within a year that posed a subsequent health risk.(33) In our study, 17 out of 474 fall patients had repeated falls at our hospital between 2015 and 2016. Given that falling again will cause more serious physical and psychological harm to the patient, it is crucial to have continuous improvement measures after falls, which organize medical staff to analyze and improve the de ciencies of previous interventions, minimizing the likelihood of a patient falling again. In addition to regular review of fall cases, our continuous improvement measures also include supervision and inspection of fall care by the leading team, aiming at promoting the improvement of fall prevention in hospitals in the long term.
The rate of falls was increased in several departments including D. of TCM, D. of oncology, D. of endocrinology and D. of rehabilitation. Noticeably, the patients in these departments were accompanied by relatively serious diseases, and the burden of medical staff was relatively arduous. To a certain extent, it had increased the di culty of medical staff in preventing falls and managing falls, so it might cause an increase in the incidence of falls. In addition, the majority of patients in these departments were older patients, who were at higher risk of falling than patients in other departments. Moreover, several patients in the department of endocrinology were involved in hypoglycemia with a higher risk of falls. In the rehabilitation department, there were more patients with physical dyskinesia in urgent need of sports rehabilitation, and their risk of falling was relatively high.
This study has limitations. First of all, we used a before and after intervention control rather than a randomized control over the same time period. We believed that the original fall intervention system had many de ciencies and was not suitable for continued use. Therefore, it's dangerous to set the original fall intervention system as a control group. In addition, the comparison of results before and after the new intervention system allows us to nd changes in the fall of speci c departments. To further validate and modify our model, we plan to compare the model with the currently accepted fall intervention systems in subsequent clinical trials. Secondly, our revised system is a multifactorial fall intervention model without specifying the effectiveness of a single intervention. Since falls are associated with many factors, we believe that only by intervening simultaneously from multiple perspectives can we minimize falls. As for its applicability, hospitals can learn from our model and improve upon their existing fall intervention systems. Lastly, the intervention was not conducted blind, which might introduce reporting bias. However, compared with the previous system, the reporting procedure in the revised intervention model was way more stringent. Regular supervision also ensured that falls were reported truthfully.
Our fall intervention model has several advantages. Firstly, compared with the existing fall management systems, our model forms a complete fall intervention work-ow including measures for pre-fall prevention, fall-onset management, and continuous improvement after falls to minimize the incidence of fall. Speci cally, few studies pay attention to the continuous improvement after falls. However, this is quite important for learning lessons to improve measures and prevent second fall. Secondly, the approach we used in developing and applying the intervention were easy to implement. For medical staff and care-taking staff, measures for fall risk factor assessment and management are all familiar medical operations. Our model is to form a work ow to fully ensure that their work is orderly and correct. In addition, involving patients and their families in falling interventions in the form of written and oral education is also a common method in medicine to protect patients' health together with medical staff. Thirdly, our fall intervention model was constructed based on a large sample of more than 320,000 inpatients, which re ects the reliability of its effectiveness to some extent.

Conclusions And Implications
In conclusion, the results of our study indicated that the revised multifactorial fall intervention model was effective in reducing falls and fall injuries. Further study is needed to verify whether our model is functional and determine if the model is effective in a longer period of time. Seven hospitals have introduced our models and are currently undergoing testing.
Declarations ETHICS APPROVAL AND CONSENT TO PARTICIPATE Not applicable.

CONSENT FOR PUBLICATION
Not applicable.

AVAILABILITY OF DATA AND MATERIALS
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

COMPETING INTERESTS
All authors declare no editorial or nancial con ict of interest.