Systematic review for the prevention and management of falls and fear of falling in patients with Parkinson's disease

Abstract Objective To synthesize recent empirical evidence for the prevention and management of falls and fear of falling in patients with Parkinson's disease (PD). Data source Database from PubMed, Cochrane Library, and EMBASE. Study design Systematic review. Data collection We searched the PubMed, Cochrane Library, and EMBASE databases for studies published from inception to February 27, 2021. Inclusion criteria were nonreview articles on prevention and management measures related to falls and fall prevention in Parkinson's disease patients. Principal findings We selected 45 articles and conducted in‐depth research and discussion. According to the causes of falls in PD patients, they were divided into five directions, namely physical status, pre‐existing conditions, environment, medical care, and cognition. In the cognitive domain, we focused on the fear of falling. On the above basis, we constructed a fall prevention model, which is a tertiary prevention health care network, based on The Johns Hopkins Fall Risk Assessment Tool to provide ideas for the prevention and management of falling and fear of falling in PD patients in clinical practice Conclusions Falls and fear of falls in patients with Parkinson's disease can be reduced by effective clinical prevention and management. Future studies are needed to explore the efficacy of treatment and prevention of falls and fear of falls.

falls had decreased quality of life, even causing injury and disability (Michałowska et al., 2005;Pickering et al., 2007;Rahman et al., 2008;Soh et al., 2013). Moreover, falls may bring about fear of fall (FOF), which affects the movement of PD patients in turn and leads to the exacerbation of PD symptoms (Adkin et al., 2003;Chaudhuri et al., 2011;Rahman et al., 2011).
Some experts have suggested that assessing the joint effect of potential falls in PD patients may be useful for fall prediction (Gazibara et al., 2016a).

Literature search
Reports included within PubMed, Cochrane Library, and EMBASE databases published on and before February 27, 2021, were considered. The following search terms were used: fall/fear of falling, Parkinson's disease, and prevention/management, with no language limitations ( Figure 1). No manual search approach was applied. The protocol of this systematic review was registered in PROSPERO (number 285709).

Study selection
Cross-sectional, case-control, prospective, and retrospective cohort studies of falls and FOF in PD were preliminarily screened. Secondary outcomes of interest included consequences of falling including injury, death, and material consumption. Reviews, case studies, and conference abstracts were excluded. Studies that summarized falls or made comparisons between falls and FOF were included. After identifying relevant articles, duplicate studies were removed. A detailed description of the search strategy and studies selected is illustrated in

Data extraction and quality assessment
We screened the full content to evaluate whether the information was potentially related. Two authors selected relevant studies independently, with disagreements resolved via discussion with a third senior author. The Newcastle-Ottawa Scale (NOS) was used to assess the quality of selected studies (Stang, 2010). Any disagreement between the two authors was also resolved by discussion with a third author. The NOS approach includes three domains (selection of study groups, comparability, and outcome assessment) to assess the quality of eligible studies. A study could be awarded up to one star for each item within the selection and outcome domains and up to two stars for comparability. If seven or more stars were awarded, we considered the study to be of high quality (Stang, 2010).

Data synthesis and analysis
Two broad outcome variables were considered, as follows: Risk factors for falls and FOF in PD patients (physical status, underlying health problems, environmental factors, medical care, and cognition function); and fall and FOF clinical practices (traditional measurement scales, fall prevention model, drug therapy, and surgery, and exercise). After selected studies were assessed, reasonable prevention-management strategies for falls and FOF in PD were created.

Study selected characteristics
We obtained 162, 228, and 198 studies from the three databases, respectively. After duplicate removal, 108 studies remained. After all selection criteria were applied, 45 articles remained. Characteristics of included studies, and data regarding falls, FOF, fall-or FOF-associated complications, prevention, and management are summarized in Table 1.

Classification of falls
Falls are common symptoms of PD. Fall events were defined as "some unexpected events that caused the person to unintentionally land on any lower surface such as an object, flood, or ground." (Canning et al., 2015;Del Din et al., 2020;Foongsathaporn et al., 2016;Gazibara et al., 2016;Hoskovcová et al., 2015;Lima et al., 2020;Maki et al., 1994;Martin et al., 2015;Moreno Catalá et al., 2015;Parry et al., 2016;Paul et al., 2014) Some studies classified falls according to their frequency of occurrence, as follows: Never, few, or less; every month or year; and disabled. (Grimbergen et al., 2013) Several studies distinguished falls based on whether they occurred indoors or outdoors (Gazibara et al., 2014(Gazibara et al., , 2016a(Gazibara et al., , 2016bPaul et al., 2017). Falls were also classified based on whether they occurred in a forward or non-forward direction (Youn et al., 2017). Despite the fact that studies classified falls using different methods, all aimed to reduce falls or assess risk factors for falls.
The factors that may cause the fall of PD patients are summarized in Figure 3. We broke these risk factors down into five components: Physical status, pre-existing conditions, environment, medical care, and cognition. These five parts complement each other and generally do not constitute falls in PD patients by a single factor. PD is a neurodegenerative disease, and its disease mechanism is likely to lead to falls.
Therefore, we pay more attention to factors other than PD that may affect falls. Next, we will explain each of these five aspects.

Pre-existing conditions
A pre-existing poor condition is known as an underlying health problem. Age can be seen as a variable that describes health status, so old age can also be seen as a potential underlying disease. Some risk factors, include seriously injured (

Environment factors
The external physical environment is an important reason for PD patients to fall, such as wet or uneven ground, high obstacles, and insufficient lighting, causing majority of the tripping or slipping (Grimbergen et al., 2004;Hely et al., 2005;Olanow et al., 2009). An observational study of fall conditions included 300 PD participants. Of the 180 people who reported falling, the conditions associated with falling included the following characteristics: Outdoors, early morning, daytime, tripping, slipping, and unsteady posture (Gazibara et al., 2014). Preventive actions that can reduce the likelihood of a fall in a complex environment include using crutches, elevating feet higher when crossing obstacles, or using armrests or pads (Gazibara et al., 2017). These external environmental factors are noteworthy in the management of PD patients, and they are also one of the factors that can most reduce the occurrence of falls.

Medical care
In terms of medications, long-term and high-dose levodopa use reduced complications associated with falls (Hauser et al., 2007;Holloway et al., 2004;Parkinson Study Group, 2000;Parkinson Study Group CALM Cohort Investigators, 2009;Rascol et al., 2000). Continuous carbidopa, levodopa enteral suspension, or continuous subcutaneous apomorphine injection reduced pain levels due to complications in PD (Antonini & Nitu, 2018;Katzenschlager et al., 2018;Olanow et al., 2014). In a previously published randomized controlled trial (RCT), effects of levodopa, a dopamine agonist, and a MAO-B inhibitor were compared, revealing that levodopa had a better activity score than the other therapeutics considered (Gray et al., 2014).
People with PD show visual dependence because visual-motor cues are disordered when they combine with the brain's vestibule (Bertolini et al., 2015;Yakubovich et al., 2020). Perceptual overconfidence is evident, not only in vision but also in other senses such as smell (Almeida & Lebold, 2010;Azulay et al., 1999;Azulay et al., 2002;Bronstein et al., 1990;Cooke et al., 1978;Cowie et al., 2010). Visual overconfidence is associated with gait and balance and can predict the risk of falls (Curtze et al., 2016;. A controlled study of 20 PD patients and 21 healthy people found that both groups had a high level of confidence in the correct target, but PD patients were more confident for the wrong reasons compared to normal people . Most falls can have minor or severe consequences, including physical and psychological injuries. One of the most important psychological harms, as opposed to overconfidence, is FOF, which is also a cognitive disorder (Adkin et al., 2003;Rahman et al., 2011). FOF has been defined as "a continuous concern about falling that contributed to individual refraining from activities" (Tinetti & Powell, 1993). The etiology and clinical symptoms of FOF show a big difference and it needs joint methods to measure. Some experts used questionnaires to estimate FOF such as the question "In general, are you afraid of falling over?" (Yardley & Smith, 2002). FOF was used to be called "basiphobia," which is a specific type of phobia, and it manifests itself as a severe fear of standing or walking (Bhala et al., 1982;Gai et al., 2009). Although falling has been illustrated to cause FOF, FOF can also cause falls in reverse. FOF may induce falls through the change of gaits or restriction of movement (Grimbergen et al., 2013;Peto et al., 1995). Reduced activity may also lead to a reduction in the number of falls, but not the probability of falls. This suggests that FOF may indicate a decline in function due to reduced activity, leading to an increased risk of falling (Mak & Pang, 2010).
A study of the elderly in the community suggested that FOF was a dynamic process, in which the fearful stage alternated with the non-fearful stage (Oh-Park et al., 2011). Parkinson's disease, as a neurodegenerative disease, presents a more complex FOF than the common elderly person. Compared with healthy elderly individuals, PD patients show a higher tendency for emotional disorders (Hadjistavropoulos et al., 2011;Trollor et al., 2006), and about one-third of PD patients suffered from anxiety disorders (Broen et al., 2016). FOF was an essential factor affecting the quality of life of PD patients (Brozova et al., 2009;Grimbergen et al., 2013). Several studies have shown that the degree of fear in recurrent fallers and frequent fallers was higher than in single fallers and low-frequency fallers (Mak & Pang, 2010;Rahman et al., 2011;Thomas et al., 2010). A more individualized treatment approach in PD patients with FOF will bring more healthy and economic benefits to patients (Winter et al., 2010).

3.3.1
Traditional measurement scales and fall prevention model Some traditional scales are used to predict falls and FOF, mainly including the following three scales: Consequences of falling (COF) (Yardley & Smith, 2002), falls efficacy scale (FES) (Tinetti et al., 1990), and survey of activities and fear of falling in the elderly (SAFFE) (Lachman et al., 1998;Yardley & Smith, 2002). COF rated 12 questions that described falling, with higher scores indicating greater fear of falling. FES is a self-efficacy rating of 10 activities of daily living without falling, with higher scores indicating less confidence or a high fear of falling. And Saffe is an improved survey of activity and fear of falling in the elderly, where higher scores are associated with greater avoidance of activity.
Besides, Beck Depression Inventory (BDI) (Beck et al., 1961), Beck Anxiety Inventory (BAI) (Beck et al., 1988), and assessment of quality of life scale (QOLS) (Peto et al., 1995) are measurements of mood, and can be auxiliary diagnosis methods of FOF. The scales mentioned earlier are given in Table 2.

Drug therapy and surgery methods
Drug therapy is the most common clinical treatment for patients with PD. Clinically, physicians are more likely to observe complications related to motion. Complications related to motion, such as bradykinesia, stiffness, and tremor (Postuma et al., 2015), can be treated with medication and can improve the patient's motor function and prevent falls. Dyskinesia, represented by a decrease in the velocity and amplitude of repeated movements, and stiffness, represented by an increase in auxiliary joint resistance, can be treated with levodopa (Chou et al., 2018;Hauser et al., 2000;Jankovic, 2005). Several randomized controlled trials have shown that levodopa or dopamine agonists are effective in treating motor symptoms in patients with early PD (Hauser et al., 2007;Holloway et al., 2004;Parkinson Study Group, 2000;Parkinson Study Group CALM Cohort Investigators, 2009;Rascol et al., 2000). Since the pathological feature of PD patients is insufficient secretion of dopamine, some studies have confirmed that continuous infusion of dopamine can reduce motor complications in advanced PD patients (Antonini & Nitu, 2018;Katzenschlager et al., 2018;Olanow et al., 2014). DAs currently in use include ropinirole and pramipexole, and many RCTs have been evaluated (Lieberman et al., 1998;Lieberman et al., 1997;Möller et al., 2005;Pahwa et al., 2007;Rascol et al., 1996;Schapira et al., 2011). In terms of clinical management, early PD drugs are recommended for treatment daily with 25-100 mg treatment of carbidopa-levodopa immediate release to relieve motor symptoms and prevent falls (Freitas et al., 2016;Grosset, 2010Grosset, , 2009).
If the patient fluctuates during activity, the frequency of dosing may need to be increased. Adjuvant drugs can be MAO-B inhibitors, mixed selective MAO-B inhibitors, or ion channel inhibitors (Aradi & Hauser, 2020).
At present, there is no satisfactory drug therapy program for the clinical treatment of exercise complications. There are many individual differences in these strategies for alleviating PD symptoms. There are surgical methods of deep brain stimulation targeting at globus pallidus and subthalamic nucleus (Follett et al., 2010;Odekerken et al., 2013). Relevant RCTs show that deep brain stimulation of the subthalamic nucleus can improve motor symptoms and prevent falls (Lhommée et al., 2018). But on the other hand, subthalamic nucleus deep brain stimulation (STN-DBS) surgery does not have an active effect on nonmotor symptoms (Amami et al., 2015;Gratwicke et al., 2018). A cohort study showed that 6 years after STN-DBS surgery, dopamine addiction and impulse control disorders decreased, but non-operative mental fluctuations decreased, and apathy increased (Abbes et al., 2018

Physical exercise
Although levodopa has shown promising results in the treatment of PD patients, there are serious limitations to long-term levodopa therapy.
In addition to traditional medical therapy, exercise or physical therapy is more helpful to improve the patient's motion and non-motion status, and has benefits in increasing confidence, preventing falls, and improving quality of life (Cruise et al., 2011;Mak et al., 2017;Reynolds et al., 2016). Exercise is defined as any physical activity resulting from the

F I G U R E 4 Detailed rules of The Johns Hopkins Fall Risk Assessment Tool scores and Falls Prevention Pyramid
expenditure of energy to contract skeletal muscles. In a large cohort, moderate-to-heavy exercise participants were found to have a lower risk of PD (LaHue et al., 2016), with even about two-thirds of risk decreasing in men (Corcos et al., 2012;Mak et al., 2017). Some studies have found that physical exercise can slow down the onset and progression of PD (Cheng et al., 2016;Mak et al., 2017). Several large RCTs also found that exercise can improve symptoms of cognitive decline and bradykinesia and can effectively prevent the occurrence of falls (Corcos et al., 2012;Mak et al., 2017). Some studies can also prove that exercise intervention can alleviate the non-motor symptoms of PD, which can relieve the mental stress of PD patients, enhance their self-confidence, and reduce FOF (Aarsland et al., 2005;Cruise et al., 2011). Drugs do not perform well in improving the mood of PD patients, and some drugs may produce many side effects in the treatment of PD patients (Reynolds et al., 2016). This highlights the huge two-sided benefits of exercise compared to drugs (Reynolds et al., 2016;Sacheli et al., 2018). Progressive exercise intervention for patients can restore partly behavioral ability in the physical function and prevent falls. An RCT involving 130 PD patients who were assigned treadmill-based training therapy and physical training via biosensors showed a 55% reduction in falls within 6 months (Mirelman et al., 2016). Another study on treadmills showed that treadmills improved baroreflex sensitivity, significantly improved blood pressure, and reduced some hypotensive falls (Ganesan et al., 2014). A large study reported that low-intensity treadmill exercise in PD patients can be as effective as medication, suggesting that aerobic exercise may improve cardiopulmonary function (Schenkman et al., 2018). Not only that, but other forms of exercise, such as Taijiquan, also have the effects of improving the movement status of patients (Song et al., 2017).

DISCUSSION
To our knowledge, this is the first review of clinical interventions and management of falls and FOF and the first presentation of a novel fall prevention theory. We focus on summarizing the existing association between falls and FOF in PD patients and attempt to summarize clinically feasible prevention and management schemes. The etiology of PD is complex, and the mechanism of this degenerative disease still needs to be studied, but this does not conflict with using existing research findings to find ways to prevent falls and the fear of falling. Because fall is a kind of subjectively unwilling but physically irresistible behavior, and the duration of falls is short, the complications of falls are extremely harmful, so it is particularly important to find a way to prevent falls. FOF is likely to occur in PD patients with or without previous falls. On the one hand, this may be the result of the neurodegenerative lesion, and on the other hand, it may be the side effect caused by PD patients' subjective perception of a different gait from the previous gait (Maki et al., 1994). FOF may lead to falls through gait changes or active restriction. If left uncontrolled, falls and FOF will interact with each other, leading to a vicious circle, and patient's quality of life will suffer unprecedented impacts. We summarized the predictive methods and influencing factors of fall and FOF, as well as the relevant methods of clinical prevention and management, which all consistently indicate that the control and management of PD are urgent.
According to our conclusion, the fall prevention model regards physical exercise as a necessary part of each stage, not only because of its good efficacy and its strong effect on improving patients' physical fitness and reducing falls, but also because it is a controllable and individualized treatment method. In primary prevention, we recommend that moderate-intensity exercise and rehabilitation activities be used as the primary means of preventing falls and FOF, and, if necessary, take medications such as levodopa as recommended by the clinician. With the increase of JHFRAT score, we recommend that the symptoms of PD should be treated primarily, and that medication and low-intensity exercise should be used to assist in preventing falls. Exercises are mainly low-risk and low-intensity exercises such as walking and tai chi. In tertiary prevention, exercise is listed as a non-essential management item. Exercise as much as possible under the premise that patients can accept, do not cause injury and have protection, and stimulate motor nerve and cell metabolism can also reduce the risk of falling due to muscle atrophy to a certain extent (Allcock et al., 2009;Bloem et al., 2001;Duncan et al., 2012;Foreman et al., 2011;Kerr et al., 2010;Robinson et al., 2005;Wood et al., 2002).
We also looked at a state of near-fall that could be reduced by physical exercise. A near-fall is a transient state before a fall, which leads to two outcomes, one that leads directly to the fall, and the other that occurs shortly before the fall called "a fall initiated but arrested by support from a wall, railing, other person and so on" (Gray & Hildebrand, 2000). Near-falls are common in people with PD. Someone will often fall if he or she loses balance and have nothing to cling to (Maidan et al., 2014). In a prospective cohort study of 120 PD patients, the association between near-falls and falls was explored by questionnaire collection and exercise scale scores. FES was also used to assess the extent of FOF (Gazibara et al., 2017). The results showed that there was a significant association between non-falls and falls or near-falls in the scale scores, but no statistical association between falls and near-falls was found. This suggests that near-falls should be listed as a complication of PD along with falls. Medical costs for fall-related consequences also increase as falls have more serious consequences (Pressley et al., 2003). It illustrated that taking some physical measures to reduce nearfalls or prevent falls in the state of near-falls will bring great benefits (Gazibara et al., 2017). A multicenter randomized controlled trial also reported the risk of falls and near-falls. After excluding the nonconforming population, 474 PD patients were randomly assigned to the experimental group for exercise and strategic intervention (PDSAFE), and for a period of 3 months for random monitoring of falls and economic evaluation (Ashburn et al., 2019). PDSAFE is a home-based training program under the guidance of physiotherapists that includes postural control training, gait freeze reduction training, and learning feedback (Hulbert et al., 2019). This study found that the intervention significantly reduced the severity of falls and near-falls, and the decline increased in PD patients with cognitive impairment.
The collection of fall information plays an important role in the prevention of falls. The use of retrospective self-report may cause recall bias and may underestimate the frequency of falls (Hauer et al., 2006;Lamb et al., 2005). It is very important during intervention and management of fall prevention to correctly identify people at high risk of falls (Allen et al., 2013). A cohort study used diary data to explore its feasibility, and found that despite the high rate of loss of followup, the characteristics of the people lost to follow-up were similar to the baseline characteristics of the people who kept a diary, and the diary data was also feasible (Hunter et al., 2018). In addition, postural monitoring of patients with Parkinson's disease is crucial for the prediction and prevention of falls. With the continuous development of science and technology, more and more high-tech equipment is being applied to the field of disease monitoring and prevention. Wearable devices, such as watches and wristbands, can already correctly identify the types of activities that occur in everyday life (Pärkkä et al., 2006).
Different from traditional data collection methods, sensor devices can capture the existence of micro-data and can use machine learning to develop relevant algorithms, which can not only understand the subtle changes in patients' falls but also objectively monitor PD symptoms and daily changes in a remote place and home (Arora et al., 2015;Lakshminarayana et al., 2014;Weiss et al., 2010). There is already a large study using wearable sensors to collect data on falls in Parkinson's patients (Silva de Lima et al., 2016). If PD patients can be screened out early and primary prevention can be carried out in time for people with suspected PD, the number of PD patients can be fundamentally reduced. For example, a large recent case-control study with 274 participants used liquid chromatography-mass spectrometry (LCMS) to separate and detect the presence of lipids and small molecules in the cortex of patients with PD to identify serum biomarkers of PD (Sinclair et al., 2021). The results showed that ceramides, triacylglycerol, and fatty acyl classes in PD patients decreased, while glycosphingolipid and fatty acyl metabolites increased, which is helpful for the development of skin test paper for PD patients, and has an important role in the field of public health.

CONCLUSION
Falls and FOF in PD patients can be reduced by effective clinical prevention and management. Although PD patients have a high rate of falls, FOF is common. More research is needed to explore the treatment and prevention of falls and FOF. The key is to identify these PD patients who are at high risk for falls or FOF, identify and reduce the occurrence of falls and FOF through monitoring, medication, physical exercise, and other means, in order to detect and reduce the occurrence of motor and non-motor complications. The fall prevention model established by us expands the treatment methods of clinicians for PD patients and adopts a comprehensive prevention approach to reduce the incidence of falls and FOF. PD patients would benefit from this integrated prevention and management approach.

CONFLICT OF INTEREST
The authors declare no potential conflicts of interest.

AUTHOR CONTRIBUTIONS
Wen-Yi Liu, Tao-Hsin Tung, Chencheng Zhang, and Leiyu Shi conducted the study and drafted the manuscript. Wen-Yi Liu and Tao-Hsin Tung participated in the design of the study and performed data synthesis.
Chencheng Zhang and Leiyu Shi conceived the study and participated in its design and coordination. All of the authors read and approved the final manuscript.

DATA AVAILABILITY STATEMENT
All data underlying the findings are within the paper.