Introduction and Problem Statement
As the American population continues to live longer, the number of elderly persons remaining in their own residence, whether by intent or necessity, tends to increase. The financial, physical, and emotional burden that occurs when a person falls places accountability on healthcare providers to aid individuals in developing strategies to decrease incidence of falls. A major risk for older adults desiring health and independence are falls (Centers for Disease Control and Prevention [CDC], 2008). Falls often end a person's ability to remain in his or her own home, as 1 in 10 falls results in hospitalization due to a serious injury such as hip fracture or a head injury (CDC, 2008). Thirty-five percent to 40% of adults living in the community older than 65 years of age fall every year; this number increases to 50% over age 75 years (CDC, 2008).
There are several accrediting agencies for home healthcare; though driven by different processes, common goals are to improve safety and quality in the home setting. These agencies include Community Health Accreditation Program, Accreditation Commission for Healthcare (ACHC), and The Joint Commission (TJC). Each of these agencies is an alternative for one another and is a not-for-profit independent, nationally recognized accrediting body. Reducing risk of falls has been identified as a National Safety Goal by TJC and continues to be listed among the 2013 National Safety Goals (TJC, 2013). A main purpose for The Joint Commission survey is to spotlight areas that critically influence patient safety, including adherence to prominent National Safety Goals such as falls risk reduction (TJC, 2011). Home healthcare workers need to know if the assessment done for the risk of falls for homebound clients is accurate.
The purpose of this research study was to determine if the score from the Falls Risk Assessment (FRA) accurately identifies the risk of falls in a homebound client. This research study also assesses whether individual items on the FRA have a higher predictive power with the incidence of falls. A fall is defined as an event resulting in a person coming to rest unintentionally on the ground or other lower level, and not as a result of a major intrinsic event (e.g., stroke, syncope) or overwhelming hazard. An overwhelming hazard was defined as a hazard that could have resulted in a fall by the youngest, healthiest people (Tinetti et al., 1988). Falls can result in injuries and psychological difficulty as well as social isolation and hospitalization. According to the CDC (2011), in the age group of 65 years and older there were 2,403,146 unintentional falls in 2011 treated in the hospital emergency rooms in the United States. Injuries often lead to nursing home placement, decline in status of activities of daily living, higher costs of healthcare, and decreased quality of life (Chen et al., 2007). At the time of our research, in the area of FRA, whereas a number of assessments of gait, balance, and mobility had been studied for validity, sensitivity, and specificity, no single multifactorial tool has been examined and validated.
Review of Literature
There is adequate research showing that falls occurring in the older adult increase healthcare costs and further debilitates this population. Injuries occurring from falls are often categorized as fatal and nonfatal falls among older adults (Stevens et al., 2006). There has been an abundance of literature reviewing falls: risk factors, depression and falls, and falls in the elderly related to medications that are prescribed. There is a dearth of research looking at home care episodes with fall occurrences, and how the incidence of falls relates to risk assessed at admission.
Several tools have been used to identify patients with a fall risk in the hospital setting; not all have been validated or shown to transfer accurately in a different setting. Vassallo et al. (2005) chose four different assessment tools to evaluate their differences in performance when used in the same setting. The tools chosen were identified as tools familiar to staff and previously used in the older adult population: STRATIFY, the Tullamore, a Tinetti-based assessment, and the Downton (Vassallo et al., 2005). This study did make an effort to use each tool in the same environment under the same conditions to evaluate differences in the efficiency of each of these tools (Vassallo et al., 2005). The study did identify significant differences between each tool, and although some tools were missing data when completed, their results were still used in the analysis (Vassallo et al., 2005).
Eagle et al. (1999) compared the Functional Reach (FR), the Morse Falls Scale (MFS) and the nurse's clinical judgment to predict falls on an inpatient rehabilitation unit and a geriatric medical floor. This Eagle et al. (1999) study failed to demonstrate that use of a tool that was more predictive and less time consuming than nurse's clinical judgment. The MFS and FR were often inconvenient and time consuming to perform. The benefit of the MFS was that it could be done by a chart audit as well as direct observation. A study published by a home care agency in Delaware (Bucher et al., 2007) explored factors contributing to falls in the design of the content and scoring of their FRA. The researchers considered the variable impact of modifiable extrinsic factors that could cause preventable falls and intrinsic physiologic factors that could lead to a higher risk. Risk-specific interventions were developed, including guidelines for referral to other disciplines based on risk factors.
Fabre et al. (2010) identified that some tools are positive predictors of falls and can identify an older adult at risk. Tools that are simple to use may not adequately identify fall risk factors or cause of falls, whereas tools that are more comprehensive may be too complicated to deliver in the community. This study concluded that few tools have been in use to measure multiple domains of risk. It is important to examine multiple domains to identify those older adults at risk for falling so that programs can be initiated for falls prevention and fall risk reduction (Fabre et al., 2010).
Beginning in 2010, the Outcome Assessment Information Set (OASIS) used at admission by home healthcare agencies in the United States was revised by the Centers for Medicare & Medicaid Services (CMS) to include a question concerning assessment for risk of falls in the homebound population served by home care agencies. Before this revision, little information was gathered regarding the number of homebound individuals at risk for falls. The OASIS question specifies use of a multifactorial, standardized assessment to determine level of risk for falls (CMS, 2009). Potentially Avoidable Events are one of the Home Health Quality measures that are based on the OASIS-C data and are a specific focus of OASIS outcomes, which includes the use of risk assessment for falls (CMS, 2013a, 2013b). The Home Health Quality Measure is Emergent Care for Injury Caused by Fall, identified on OASIS-C by items M2300 and M 2310.
The tool chosen by the software provider for the agency for the start of care OASIS is an adaptation of an FRA initially created by Delmarva, the Quality Improvement Organization from Maryland, and later adapted by the Nebraska Quality Improvement Organizations (QIO) (CIMRO) (Table 1). In an effort to improve quality of healthcare outcomes for Medicare beneficiaries, thereby effecting others as well, the CMS has contracted with various organizations in each state to serve as QIO contractor. Legally, the mission of each contractor is to improve the quality of care for each of his or her beneficiaries (CMS, 2013a, 2013b).
The original tool from our software provider reflects common risk factors identified in many studies, including level of consciousness, history of falls, ambulation, elimination status, visual deficits, orthostatic changes, and medications. To this broad range of questions, the software provider added the Timed Up and Go (TUG), an easily administered screening assessment for mobility, which had been studied and reported extensively in professional literature. The TUG evaluates a client's ability to stand up from a chair unassisted, walk 10 feet, turn, walk an additional 10 feet to return to his or her chair, and sit down (Murphy & Lowe, 2013). The TUG provides a quick assessment tool, evaluating strength, ability to ambulate, and dynamic balance (Podsiadlo & Richardson, 1991). The TUG, as demonstrated in reliability studies, indicates that medically stable patients vary minimally in their TUG scores (Podsiadlo & Richardson, 1991). This demonstrates that this scoring can be a useful objective measure of clinical change. According to Podsiadlo and Richardson (1991), the TUG has content validity evaluating movement used in everyday life and concurrent validity correlating more specific measures of balance, gait speed, and functional mobility. In a recent study, Murphy and Lowe (2013) demonstrated the importance of maximizing reliability and accuracy of TUG results when home healthcare interdisciplinary clinicians completed competency training upon hiring and annually to maintain competent best practice standards.
The hypothesis states that there is a positive relationship between an elevated score of combined factors and falls occurring in the home. In this directional hypothesis the dependent variable is the score as defined by the Fall Risk Assessment. The independent variable is the incidence of falls.
The aim of this study is to assess if there is a strong relationship between an aggregate score of 10 or higher on the initial FRA done on admission to home healthcare. A fall prevention and intervention program cannot be effectively developed until patients with the highest likelihood of falling are identified.
Population and Sample
The subjects were patients of a large not-for-profit home healthcare agency serving Philadelphia and the surrounding suburbs. Two groups were randomly selected. Members of the first group of 100 had sustained falls while under care; subjects were identified from event reports completed by clinical staff. Selection criteria were defined as an age range of 65 to 90 years and completion of an FRA at admission to home care. Exclusion criteria included instance of an unavoidable fall (e.g., a fall resulting from sudden onset of cardiac event or stroke). A second group of 25 participants who had not fallen were selected for comparison. Subjects in the same age range who did not fall were identified via record review by the coesearchers; the exclusion criterion was lack of a complete FRA.
This study took place at a not-for-profit home healthcare agency that services clients with varied socioeconomic, racial, and cultural backgrounds in Philadelphia and surrounding suburbs, covering a large geographical area. Referrals are initiated from a variety of sources including area hospitals, physicians' offices, caregivers, and assisted living units or group homes that serve as an individual's primary residence.
Charts were reviewed by the coresearchers, and data were collected in a separate research database. No personal identifiers were collected; all individual information used in the study was coded and disclosed in aggregate with no links or identifiers to individual subjects.
When fallers were compared with nonfallers, the difference in the mean FRA scores among the two groups was found to be statistically significant using a two-sample t-test. Those who fell had a significantly higher fall risk assessment score than those who did not fall (Table 2).
A t-test for difference of two means was carried out on these data. The p-value obtained was .035, which is less than .05 significance level. It can be concluded that there is a statistically significant difference between scores of fallers and nonfallers.
To discover any trends in predicting a fall, linear regression modeling was used. Both single variable and multiple variable models were analyzed. The response variable was fall score and the explanatory variables were factors that could influence a fall. In a single variable or simple regression model, the single variable is responsible for all variation in the response. In a model with multiple variables, the effect of each individual variable is assessed by holding all other variables as fixed, but still including them in the model. The magnitude of an explanatory variable's coefficient is interpreted as the change in the response due to a one unit increase in that explanatory variable. Therefore, the greater the coefficient, the greater effect it will have on the response. For any statistical testing, a p-value of less than.05 is considered statistically significant.
When all explanatory variables are included in the model, together they significantly predict the risk for a fall with an overall regression p-value of <.0001. They also explain 99.7% of the variability in the likelihood of a fall. The coefficients for all explanatory variables were significant with p-values of <.0001. The table below shows the coefficients and p-values for each variable within the multiple regression model. All coefficients are relatively close to 1 meaning for every one unit increase in the variable there will be a one unit increase in the response, all other variables being held fixed (Table 3).
In the simple linear regression models, individual items in the assessment were not shown to accurately predict risk of falls as well as they were in the multiple variable model. A history of falling most significantly predicts the risk for a fall when considered on its own. Gait and balance, level of consciousness, and ambulation/elimination predict the risk for a fall with slightly less predicting power when considered on their own. Both TUG and orthostatic changes do not significantly predict risk for a fall when considered on their own (p>0.05). This information is critical, in that a single approach to decreasing falls is likely to be less effective than a multipronged approach. Caregivers and providers are advised to consider the entirety of the falls risk and direct comprehensive interventions to address the multiple factors that lead to falls.
Figure 1 shows the results from the single variable regression models. The lower the p-value (orange/left axis), the more significant the variable is to the total score when all other variables are held fixed; the further to the right, the less significant the explanatory variable is to the model. As the regression coefficients decrease (blue/right axis), the effect of that explanatory variable on the response also decreases. Therefore, the variables in the chart are ranked from left to right based on importance and significance (Figure 1).
The results suggest that the FRA as a whole is an appropriate tool that significantly predicts the likelihood of a fall. When all variables in the assessment are examined together, they possess more predictive power than when each variable is considered on its own. However, the variables can be ranked in order of influence on the falls score by examining single variable regression model coefficients and p-values. From this, it was found that history of falls was the single most influential variable to the fall score and orthostatic changes were the least influential.
All of the individuals who completed the assessments used in the study have completed orientation to OASIS requirements and FRA. According to Murphy and Lowe (2013), as part of annual competencies, home healthcare clinicians that are providing TUG testing should complete and pass TUG education based on best practice standards of administration. The variances in training between therapists as well as nurses vary between agencies as well as by disciplines (Murphy & Lowe, 2013). The accuracy of results would be stronger having clinicians trained in a uniform manner with consistent verbal instruction, timing guidelines, and control of the physical setting (consistent chair height, floor surfaces) to capitalize on reliability (Murphy & Lowe, 2013).
Discussion and Conclusions
The results support the hypothesis. The total score on the multifactorial risk assessment tool was shown to have a high correlation with incidence of falls. The average scores on admission of individuals who fell after their initial assessment were significantly higher than the scores of individuals who did not fall.
The finding of the most significant single variable, a history of falls, is consistent with many previous studies, and is widely recognized as a strong predictor. According to Rubenstein and Josephson (2002), those having a fall increase their risk of falling again by threefold. Rubenstein and Josephson (2006, p. 811) state, “Although recurrent falls in an individual frequently are due to the same underlying cause (e.g., gait disorder, orthostatic hypotension), they also can be an indication of disease progression (e.g., Parkinsonism, dementia) or a new acute problem (e.g., infection, dehydration).” Targeted interventions in this area of recurrent falls may include outpatient or home healthcare physical therapy several times per week for modalities such as mobility training and instructing in a home exercise program. If caregivers are present in the home, it would be beneficial to involve them in the exercise program to encourage compliance and consistency. Individuals in this category of repeat fallers may also benefit from consistent use of an assistive device.
The findings identified gait and balance, level of consciousness, and ambulation/elimination as the next most likely predictors after history of falls. The scores for medications, predisposing diseases, equipment issues, and vision status were the least associated with a high total score. Gait and balance impairment were noteworthy as they have demonstrated a threefold increase in risk for falls (Rubenstein & Josephson, 2006). As stated in the Vision Health Initiative, the CDC maintains, “In the United States the leading cause of blindness and low vision are age related eye diseases, including macular degeneration, glaucoma and cataracts and diabetic retinopathy” (CDC, 2013).
History of falls includes the number of falls within the previous 3 months. The gait and balance section assesses aspects of mobility in various positions, though objective measures are not defined. Level of consciousness defines mental status (orientation). Ambulation/elimination identifies the presence of continence issues coupled with mobility level.
The results suggest that the FRA as a whole is an appropriate tool that significantly predicts the likelihood of a fall. However, the variables can also be identified in order of influence on the fall risk score by examining single variable regression model coefficients and p-values. Understanding the highest contributors to risk can help lead to effective interventions to reduce that risk by targeting those high-risk factors.
Risk reduction may include either home or group exercise classes, such as those found at local senior centers, two to three times per week focusing on strengthening, increased flexibility, and improved balance (Rubenstein & Josephson, 2006). Balance has also shown improvement with the addition of tai chi classes, and following and adhering to a walking program regularly. Individuals may also benefit from training with an appropriate assistive device.
Additional risk reduction may be as simple as providing a home evaluation and pointing out common themes of home modification, such as decreasing clutter, removing throw rugs, and increasing lighting in dimly lit areas of the home. Providing suggestions for bathroom safety may also be included; frequent suggestions include grab bars in the shower, tub benches to make transfers easier and safer, and the use of either a commode that can also be moved to the bedside at night or an elevated toilet seat.
Based on this information, targeted interventions focusing on the four critical variables: history of falls, gait and balance, level of consciousness, and ambulation/elimination should be emphasized to decrease the actual falls risk. Home healthcare clinicians are in a prime position to educate clients as well as families and caregivers on fall precautions. Clinicians and caregivers would benefit from education on evidence-based approaches to address these four higher areas of risk. The multifactorial fall risk assessments that are readily available for use at present come in many sizes and shapes. There are those assessments that have been validated for people not living in the community setting, and those that have never been validated for the individuals living in the community (Anemaet & Krulish, 2011). A validated tool for community dwellers seems of utmost importance to provide and implement care focused on an individual's specific care needs. The tool to be used needs to be accurate, easy to use, and readily implementable so that it is not viewed as burden by staff clinicians.
CMS does not require the completion of a fall risk assessment currently. Anemaet and Krulish (2011) article states the following:
The public reporting of this process measure, allowing comparison between the rate of best practice use by one agency and the rate of best practice by other agencies, is expected to encourage home health providers to voluntarily implement or standardize the use of such favorable care practices. (p. 126)
Since our study to validate a falls risk assessment was completed, another multifactorial tool has been developed and validated by the Missouri Alliance for Home Care (MAHC). According to MAHC, this new tool—known as the MAHC-10—may be used as an initial screening assessment and if indicated a more targeted tool may be necessary (MAHC, 2012). The MAHC-10 assesses and evaluates 10 items; most of which are on the current FRA that our study evaluated, although the MAHC-10 considers a score or 4 or more at risk for fall. Kornetti and Miller (2012) point out that although the MAHC-10 has a very striking sensitivity of 96.9%, the specificity is not so remarkable, coming in at a low score of 13.3%. Although the MAHC-10 can be assessed on bedbound and nonambulatory clients, we must ask ourselves what is the true goal of a fall risk assessment? Is the goal to be able to document that a completed assessment was provided to every client we evaluated, or is our goal to have an accurate appraisal using a multifactorial assessment to identify those clients at risk for a fall so that an appropriate implementation of an individualized plan may be set in place (Kornetti & Miller, 2012)?
More studies are planned to assess the effectiveness of specific interventions to prevent falls. Recommendations for additional research include examining the effect of targeted interventions toward the four most critical risk areas. Additionally, examining subjects with a high fall risk score who did not fall may yield valuable insight into prevention strategies. Considering the potential impact on patient safety, this area is particularly worthy of additional research. Families in need of home care services now can access this additional information as they shop for their home healthcare needs.