Applied Nursing Research 20 (2007) 86 – 93 www.elsevier.com/locate/apnr
Spartanburg Fall Risk Assessment Tool: A simple three-step process Cathy Robey-Williams, MS, MBA, RNa,4, Kathy L. Rush, PhD, RNb, Heather Bendyk, BSa, Laura Michelle Patton, BSNa, Debra Chamberlain, MBA, RNa, Teresa Sparks, MSN, BSNa a Spartanburg, Regional Healthcare System, Spartanburg, SC 29303, USA Mary Black School of Nursing, University of South Carolina Upstate, Spartanburg, SC 29303-4999, USA Received 31 July 2005; revised 13 September 2005; accepted 18 February 2006
b
Abstract
The purpose of this study was to develop a valid, reliable, and user-friendly fall risk assessment tool that is a sensitive predictor for falls in the acute care population. Fall risk factors were determined from extensive review of evidence-based studies available from a PubMed search. Previous falls, medications, and gait were found to be the top three risk factors for predicting a true risk for falls in multiple health care settings. The Spartanburg Fall Risk Assessment Tool (SFRAT) is unique from other fall risk assessment tools in combining intrinsic, patient-related factors, with a direct measure of the patient’s functional status. Interrater reliability of the SFRAT using Cohen’s j was .9008, which reflects almost perfect agreement. The predictability analysis found the SFRAT to be 100% sensitive for falls (27/27) with no false negatives. Specificity was 28% (48/172) with 124 false positives. These false positives may actually reflect patients who were at true risk for fall but were prevented from falling due to effective interventions instituted by the staff providing their care. The SFRAT fall risk assessment is a simple, reliable tool easily incorporated by nurses into their direct care routine. D 2007 Elsevier Inc. All rights reserved.
1. Background From a patient’s perspective, any fall while hospitalized is unacceptable, and fear of falling increases with age and history of fall (Hampton, Kenny, & Newton, 2002; Perell et al., 2001; Rubenstein, Powers, & MacLean, 2001). Consequences of falls include fractures, soft tissue injury, head injury, fear of falling, anxiety, and depression (Gaebler, 1993). Conversely, limitations placed on patients to prevent falls while hospitalized can negatively impact independence, self-reliance, confidence, and pride. Hospital policies and nurses at the bedside must continuously balance patient safety versus patient rights and responsibility. In 2003, this organization’s fall rate per 1,000 patient days had peaked at 4.5, and benchmark comparisons found the rate to be over the 50th percentile. Nursing staff at the bedside were compliant in completing the fall assessments, but the tool was not effectively measuring fall risk.
4 Corresponding author. Tel: +1 864 809 9280. E-mail address:
[email protected] (C. Robey-Williams). 0897-1897/$ – see front matter D 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.apnr.2006.02.002
Ongoing efforts to ensure staff compliance to fall prevention policies and procedures had not made any significant impact on fall events and patient safety. As a result, this organization, a 588-bed tertiary care facility in the Southeast, commissioned a research team to develop an evidence-based fall risk assessment tool and design an evidence-based fall prevention plan. Effective fall prevention begins with assessment of risk (Semin-Goossens, van der Helm, & Bossluyt, 2003). This problem has challenged nurse researchers for decades. Multiple fall assessment tools have been produced, implemented, and studied. Many lack sensitivity and specificity and have not assisted health care providers in significantly preventing the occurrence of falls. When compared to nurses’ clinical judgment, fall risk assessment tools did not show much accuracy in predicting patient falls. One study looked at the Morse Fall Scale and the Functional Reach Test. It found both to be equal to nurses’ clinical judgment (Eagle et al., 1999). A subsequent study (Myers & Nikoletti, 2003) refuted that finding and found neither nurses’ judgment nor the assessment tools themselves to be effective in predicting falls.
C. Robey-Williams et al. / Applied Nursing Research 20 (2007) 86 – 93
Characteristics of existing risk assessment tools may account for their lack of ability to predict falls. Nursing risk assessment tools focus primarily on measuring intrinsic, patient-related factors without direct assessment of the patient’s functional status. A risk assessment tool that combines both elements may be a more sensitive predictor of fall risk. Additionally, nursing fall assessment tools often include multiple risk factors that are scored on the basis of the patient’s self-report or observed behaviors made by the nurse. This subjectivity limits the tools’ reliability and clinical effectiveness. The multi-item summative format of existing fall risk assessment tools makes them complicated and time consuming for busy nurses at the bedside (Dempsey, 2004). The development of an objective, userfriendly instrument that eliminated complicated summative scoring and targeted specific risk factors with immediate interventions appeared warranted. 2. Purpose Therefore, the purpose of this study was to develop a valid, reliable, and user-friendly falls risk assessment tool that is a sensitive predictor for falls in the acute care population. The pilot study and assessment tool development described in this report are part of a larger study that aims to develop and evaluate the efficacy of a fall prevention program in reducing the incidence of falls of hospitalized patients. Details of the fall prevention program exceed the scope of this report. 3. Literature review Development of the Spartanburg Fall Risk Assessment Tool (SFRAT) began with an extensive review of the literature, performed to identify empirically supported risk factors. Six hundred articles were retrieved from PubMed. Articles were screened for studies relevant to nursing fall risk factors and prevention interventions. Of the 185 articles that were selected, 140 were reviewed in detail and included in the summary provided in Table 1. Previous falls, medications, and gait were found to be the top three risk factors for predicting fall events in multiple health care settings. These risk factors are highlighted in guidelines developed by the American Geriatric Society (2001) for the prevention of falls among the elderly. These guidelines—meta-analyses based on summative evidence from randomized controlled trials—recommend that older persons should be asked about history of falls whenever they are assessed. When given a history of falls, older persons should be observed performing the bget-up-and-goQ test, which is a functional assessment of gait. In addition, the guidelines recommend review and modification of medications, especially psychotropic medications. The get-upand-go test consists of asking a patient to demonstrate ambulation. The patient starts in a seated position, then stands, walks a premeasured distance, turns around, and
87
Table 1 Summary of literature review Search criteria: Falls Hospital falls Fall risk Fall injury Fall prevention Research Fall risk assessment Fall predictors Fall scale Fall score 600 Articles were retrieved from PubMed search Articles were screened and 185 articles were selected, of which 140 were reviewed Screened for: Main idea Risk factors: 120 (85.7%) of 140 articles listed risk factors associated with a fall Interventions used: 110 (78.5%) of 140 articles listed interventions to prevent a fall Populations: Acute care Nursing home/long-term care Community programs Major risk factorsa Risk factor
Number of articles
%
Previous falls Medications (includes medications, medication types, and polypharmacy) Gait Ageb Altered mental status Cognitive impairment Environment Impaired mobility Incontinence (includes incontinence and nocturia) Hypotension (includes orthostatic and postural)
45 42
37.5 35
33 30 22 20 17 16 15
27.5 25 18.3 16.7 14.2 13.3 12.5
15
12.5
a
Visual decline and chronic diagnoses were also prevalent in the literature. b Age range varied in the articles and was not specifically mentioned in 8 (26.6%) of 30 articles.
returns to a seated position at the starting point. This activity is timed, and if the patient is unable to complete the task within the established time frame, the patient would be considered at risk for fall. The primary risk factor associated with patient falls in the hospital setting is history of previous fall (American Geriatric Society 2001; Morse, 1977; Rubenstein et al., 2001). Perell et al. (2001) found 10 studies in a systematic review of the literature that cited history of falls to be a characteristic risk factor. Another descriptor, intermittent episodes of falling, has also been used as a predictor of fall risk (Tinetti, 1997). It stands to reason if intrinsic factors are in place to produce one fall, then the risk of future falls should be expected unless the
88
C. Robey-Williams et al. / Applied Nursing Research 20 (2007) 86 – 93
patient’s condition significantly changes/improves or prevention measures are implemented. Functional assessment tools for determining fall risk have typically been used by physical and occupational therapists in community settings. The get-up-and-go functional test has been shown to differentiate between groups at low fall risk and high fall risk and between fallers and nonfallers. In their comparative study of high and low fall risk groups, Wall, Bell, Campbell, and Davis (2000) measured each component task of the get-up-and-go functional test. They compared young adults (mean age = 25.5 F 5.60) to the elderly (mean age = 72.7 F 3.97) and then compared both groups to an bat risk for fallQ group. The at-risk subjects were described as patients receiving physical therapy at a local long-term care facility and either had history of falls within the past 2 years or had been treated for gait pathologies or balance disorders. The results showed no significant differences between the young and elderly control groups in the amount of time it took them to complete the get-up-and-go test. However, both control groups took significantly less time to complete the task than the group at risk for fall. The overall times for completing the timed get-up-and-go test were as follows: young adult, 15.36 F 1.638 s; elderly, 19.095 F 2.112 s; at risk, 34.52 F 10.628 s. In another study of community-dwelling older adults, fallers (30.0 F 25.0) showed significantly longer times to perform the get-up-and-go test than did nonfallers (21.4 F 9.0; Hui-Chi, Gau, Lin, & Kernohan, 2003). The relationship between medications and fall risk has also been demonstrated in multiple studies (Gatti, 2002; Rollins, 2003; Tinetti, 2003; Vassalo et al., 2002). Mendelson (1996), in a hospital study, found that antidepressants, hypnotics, benzodiazepines, and major and minor tranquilizers were associated with falls. Smith (2003) reviewed 169 medication records. Of those reviews, 9.5% (n = 16) reported falls and traumatic injuries. Medication screening has been included as a component of fall risk assessment established by the Joint Commission for Accreditation of Hospitals as part of its 2005 Safety Goals. The Veteran’s Administration (VA) National Center for Patient Safety has a bFall Prevention and Management AidQ available online for other facilities to use as a patient fall reduction guideline (VA Fall Prevention and Management, 2002). This aid includes a list of high-risk medications, which include psychotropics, antidepressants, benzodiazepines, cardiovascular agents, antihypertensives, diuretics, anticoagulants, antihistamines, bowel prep, and medications to treat nocturia. Age ranked fourth as a risk factor, although the variability in age associated with fall risk makes it an inconclusive predictor of falls. Contrary to previous belief, many falls are experienced by patients much younger than the frail elderly (Wallace et al., 2002). Kingma and Ten Duis (2000), in a Netherlands study, looked at all falls treated in an emergency department over 7 years. Out of 19,593 falls, 60% of all falls occurred in patients aged 10–64. The
highest percentage of injuries occurred in the 20- to 24-year age group. Some hospital studies exclude younger patients because of the known high morbidity associated with falls in the elderly. Many of the physiologic factors associated with aging, such as poor vision, postural hypotension, and altered mental status, can be experienced by younger, more debilitated, hospitalized patients. The literature suggests that functional ability or gait is a better measure of identifying fall risk than using an arbitrary age (Premier Incorporated, 2004). Confusion was found to be the top risk factor for falls in a study of hospitalized patients who had fallen compared with a control group of patients (Hendrich, Nyhuis, Kippenbrock, & Soja, 1995). Heindrich et al. (1995) performed a hospital study of 355 fall patients and 780 controls and found that patients who were confused or disoriented were 7 times more likely to fall, that those with dizziness and vertigo were 1.9 times more likely to fall, and that patients with elimination needs were 1.67 times more likely to fall. 4. Methodology The study was submitted and approved by the organization’s internal review board. 4.1. Design A comparative, correlational, predictive design was used to guide the study. 4.2. Sample Four inpatient hospital medical–surgical units (Neurological, Renal, Respiratory, and Oncology) were randomly selected from a total of eight units. These units represented a cross section of adult medical–surgical patients aged 18 years and older. Patients admitted to the selected units by an admitting nurse during the study period (June through August 2004) were included in the pilot study. A convenience sample of patients who were admitted to the four selected units underwent the usual assessment procedures with the addition of a fall risk assessment using the Morse Fall Scale (Appendix B) and the newly developed SFRAT. Patients were excluded if they were unconscious, chronically bedridden, or incapable of independently getting out of bed. Only patients who were assessed during the same day by three research nurses were included in the reliability study. Of the patients included in the reliability study, 65% were female and 75% were Caucasian; their age ranged from 34 to 87 years. 4.3. Measurement The top 10 major risk factors identified in the literature review were previous falls, medications, gait, age, altered mental status, cognitive impairment, environment, impaired mobility, incontinence, and hypertension. Although all top 10 major risk factors identified were used to guide development of the SFRAT, the top three factors ultimately
C. Robey-Williams et al. / Applied Nursing Research 20 (2007) 86 – 93
formed the tool’s major components. The get-up-and-go test, used specifically to assess gait, also served as a direct measure of the remaining factors. For example, cognitive impairment would be assessed in the patient’s inability to follow directions in performing the get-up-and-go test. The SFRAT is unique from other fall risk assessment tools in combining intrinsic, patient-related factors, with a direct measure of the patient’s functional status. The get-up-and-go test (Mathias, Nayak, & Issacs, 1986) provides a functional assessment of fall risk. Functional assessments, typically used by occupational and physical therapists in community-based settings, have been discouraged in the acute care setting because of the time they take to complete. However, the get-up-and-go test takes less than 1 minute to complete and has demonstrated high sensitivity (87%) and specificity (87%; Shumay-Cook, Brauer, & Woollacott, 2000). The SFRAT tool is adapted from the Podsiadlo and Richardson version published in 1991. Functional test results from the study of Podsiadlo and Richardson (1991) showed high interrater reliability with an intraclass correlation score of .99. The validity was demonstrated with correlations to the following functional measures. Log-transformed scores were used, suggesting a curvilinear relationship; Berg Balance Scale, r = .81; gait speed, r = .61; and Barthel Index of the Activities of Daily Living, r = .78. The patient is tasked to rise from a sitting position on the bed, walk a premeasured distance of 8 ft, turn around and walk toward the bed, and return to sitting position. Patients may use their assistive devices for the test. Patients are considered to have failed the test if they require assistance or take longer than 30 s to complete the test (Appendix A). The nurse has the option to score the patient as high risk based on judgment after observing the patient. This provides for patients who are unsteady (swaying gait) but able to complete the test in the allotted time frame. It also allows for judgment in cases where the patient is at high risk due to special equipment (i.e., electroencephalogram monitoring) or special circumstances (i.e., seizure precautions, alcohol withdrawal). Medication regimens as a fall risk factor were incorporated in the SFRAT using a built-in trigger designed for the computerized software program, Medication Manager (McKesson). Medication regimens as a fall risk factor were assessed at two time periods: (i) on admission, the admitting nurse assessed home medications, identified if the patient was at fall risk, and entered this information into the electronic documentation system (Care Manager, McKesson) and (ii) during hospitalization, for patients placed on medications, placing them at high fall risk; a flag message printed out on the Medication Administration Record (MAR) was generated by the software program. To determine the high-risk medications or those medications associated with falls in the study, the institution conducted a 6-month medication audit (January to June 2004) abstracted by a risk manager. All falls that had
89
occurred during this 6-month period were identified, and a review or examination of all medications taken by the patient within 24 hours of the fall was made. Medications appearing with the greatest frequency in conjunction with falls were benzodiazepines, beta-blockers, anticonvulsants, and antipsychotics. 4.4. Procedures/data collection: reliability Data were collected by the hospital’s three admitting nurses who received training in the use of the Morse Fall Scale and the SFRAT. Admitting nurses received instruction and demonstration of the procedure for conducting the getup-and-go test. The nurses utilized all three tools to assess the same patient (hospital fall risk assessment tool, Morse Fall Scale, and SFRAT) during the admission encounter. Retractable measuring tapes were used to demarcate the 8-ft area needed for conducting the test. Admitting nurses have an associate degree in nursing, were 30 to 55 years old, and had 3 to 30 years of experience in nursing practice. Two members of the fall research team were on call, available to the admitting nurses to answer any questions that arose during the data collection period. The fall research team met with the admitting nurses 1 week after data collection was initiated to discuss the process and identify problems encountered. Feedback from this meeting led to revisions to the tool and clarified data collection procedures. To provide additional information about the interrater reliability of the SFRAT, three fall research team members trained in the use of the Morse tool and SFRAT also completed assessments on 20 of the sample patients. Admitting nurses submitted data collection sheets by midafternoon to the team statistician, who subsequently notified the three research nurses. Two of the three research nurses had to coordinate schedules to assess the patients that same day. This plan provided for three RNs to evaluate the same patient within 12 hours of each other, including the admitting nurse and two research nurses. The research team nurses’ age ranged from 38 to 50 years, with 17 to 30 years of work experience. One nurse has a bachelor of science in nursing degree and the other two have a master’s degree in nursing. The SFRAT tool was implemented into actual practice following the initial reliability testing during summer of 2004. Hospital-wide education occurred in October, and the new fall prevention program was initiated in November 2004. The SFRAT tool was incorporated into the electronic documentation system utilized by Nursing, Care Manager (McKesson). Two sets of questions, one for new admissions and the other for ongoing assessments, were adapted for the Care Manager screens. On admission, patients were asked if they had fallen within 3 months. Patients were then asked if they were taking any home medications associated with high risk for falls. The get-up-and-go test was only performed if the first two questions were negative for fall risk. The second set of questions was developed for ongoing assessments. The first question asks if the patient has had any falls since
90
C. Robey-Williams et al. / Applied Nursing Research 20 (2007) 86 – 93
admission, and the second question determines if any medications are flagged on the MAR based on triggers built into the Medication Manager (McKesson) software. 4.5. Analysis Cohen’s j was used to determine the interrater reliability of the SFRAT. For the purposes of evaluating the SFRAT, Spartanburg Regional used a j in the range of .81–1.0 to determine the efficacy of the tool. j has a range from 1.00 to +1.00, with 0 indicating chance agreement, negative values indicating worse than chance agreement, and +1 indicating perfect agreement. 4.6. Validation study 4.6.1. Sampling for the beginning validity analysis Validity testing was accomplished by comparing random samples of two patient populations: admitted patients with reported falls and admitted patients who did not report any fall. Data were collected between November 2004 and March 2005. The SFRAT tool was implemented November 1, 2004, following education sessions in October 2004. The nonfall group was a randomized sample of 172 patients out of 1,468 admissions. Females comprise 58% of this sample, which has a median age of 62.5 years (SD = 19.3). The fall group was a randomized sample of 27 patients out of 196 patients with a reported fall in the variance reporting system. Females comprise 44% of this sample, which has a median age of 72 years (SD = 13.9). 4.6.2. Data collection Data collection for the validation study involved a retrospective computerized chart audit of fall data collected from implementation of the SFRAT in November 2004 until March 2005. The audit was completed by two members of the Falls Research Team, who obtained data from the following sources. Fall Risk data were obtained from Care Manager, a nursing computerized documentation system that included the SFRAT, which was performed by nurses as part of their routine care. Nurses completed the SFRAT on patient admission, once per shift and as changes in the patient’s status warranted. Using Care Manager, research team auditors identified whether the patients who had fallen had been identified by nurses as being at risk of falling. Actual patient falls were obtained from computerized variance reporting (in MIDAS, a software system from Table 2 Cohen’s j Question
j
Patient has fallen in the past 3 months Home medications prior to admission Patient fails get-up-and-go test Total
.9404 .9107 .8512 .9008
Note. b .0, poor agreement; .0–.2, slight agreement; .21–.4, fair agreement; .41–.6, moderate agreement; .61–.8, substantial agreement; .81–1.0, almost perfect agreement.
Table 3 SFRAT components: fall and nonfall comparison Category
p (Pearson’s v 2 test)
p (FET)
Previous falls Home medication prior to admission Medication in hospital Get-up-and-go
.01584 .9915 .00144 .3127
.02124 .5935 .00364 .2121
4 p b .05.
Tucson, AZ), which is a report of a fall incident that nurses are required to complete when one of their patients falls. 4.6.3. Analysis A 2 2 predictability analysis was performed to measure sensitivity and specificity. Pearson’s v 2 test and Fisher’s Exact Test (FET) were applied to each element of the SFRAT. 5. Results Interrater reliability of the SFRAT using Cohen’s j was .9008, which reflects almost perfect agreement. Individual components of the SFRAT had j coefficients ranging from .85 (get-up-and go test) to .94 (patient fall within last 3 months). j values for the entire tool and its individual components are presented in Table 2. The predictability analysis found the SFRAT to be 100% sensitive for falls (27/27) with no false negatives. Specificity was 28% (48/172) with 124 false positives. These false positives may actually reflect patients who were at true risk for fall but were prevented from falling due to effective interventions instituted by the staff providing their care. Analysis of the SFRAT components using Pearson’s v 2 test and FET revealed that history of previous fall and medication prescribed while hospitalized were statistically significant between the fall and nonfall groups at the p b .05 level (Table 3). FET(a = .05) revealed that the fall population had a significantly greater frequency of previous falls ( p = .0254) and a higher frequency of use of high-risk medications ( p = .0036) than did the nonfall group. 6. Discussion Preliminary findings suggest that the SFRAT shows promise as a valid and reliable tool for assessing fall risk of hospitalized patients. The near-perfect interrater agreement reflects strong evidence for the reliability of the SFRAT. An examination of the individual components of the SFRAT reveals j values in the near-perfect agreement. The lowest j found with the get-up-and-go test is not unexpected and reflects the slightly more complex nature of this compared with other components of the SFRAT. Findings support beginning validation of the SFRAT. The SFRAT was found to predict falls with 100% accuracy in patients at high risk. Only one other fall risk assessment
C. Robey-Williams et al. / Applied Nursing Research 20 (2007) 86 – 93
tool, the Fall Prediction Index, has demonstrated 100% sensitivity (Nyberg & Gustafson, 1997). The nominal nature of the SFRAT, rather than its additive nature, may account for its accuracy in predicting those at high fall risk. Unlike other tools in which multiple factors are scored and summed to determine the patients’ fall risk status, the SFRAT requires the presence of only a single factor. The SFRAT demonstrated specificity somewhat lower than more sophisticated yet time-consuming fall risk assessment tools. Again, the format of the SFRAT may account for this outcome. Using a tool in which there is no additive effect of individual components and where the cutoff point for determining fall risk is any one of the four components has the potential of overestimating patients who are at fall risk. Other research findings make this a plausible explanation. A group of investigators found in their fourfactor additive scale that sensitivity was highest and specificity was lowest with the cutoff at only one factor, with a reverse of this pattern occurring as multiple factors were added (Yauk et al., 2005). It may be that specificity was compromised by the collection of data by staff nurses in real time. Unlike most other studies that report the development and testing of new fall risk assessment tools, the current investigation used staff nurses’ documentation of fall risk assessments performed while they were giving routine care. Two other studies in which data were obtained by staff nurses during routine care similarly demonstrated low specificity. Myers and Nikoletti (2003) found specificities in the range of 25% to 27% for their comparison of two fall risk assessment tools and nurses’ clinical judgment in predicting patient falls, and Yauk et al. (2005) reported a specificity of 37% for their Fall Risk Screener. In the present study, all staff nurses had been educated in the use of the SFRAT and had clinical unit educators available to answer any questions related to its completion. Further, the objective nature of the SFRAT and its dichotomous format were expected to simplify the assessment process. Anecdotal data from staff nurses revealed greater ease in the use of the SFRAT compared with the previous institutional tool. Another possible explanation for the low specificity is that patients who were assessed to be at high risk for falls were, in fact, at high risk, but because of the fall prevention interventions employed by nurses, falls were averted. Examination of nursing compliance with implementation and documentation related to the fall prevention program supports this explanation. Nursing compliance as evidenced in documentation audits found assessments including documentation of interventions to be completed correctly in 94% of the fall patient sample and in 98% of the nonfall patient sample. The higher incidence of previous falls in the fall group corroborates a recurring finding across studies regardless of setting or population being studied (Hendrich et al., 1995; Yauk et al., 2005). Patients with a history of falling are more
91
likely to fall again, although a wide variation in repeat falls during hospitalization has been reported (ranging from 16% to 52%). Although research is inconclusive as to age as a major risk factor in patient falls, the current findings revealed a significant 10-year difference between the fall and nonfall groups. Consistent with findings from other research, patients in the fall group had a mean age greater than 65 years (Kerzman, Chetrit, Brin, & Toren, 2004). Certain medications and combinations of medications are known to contribute to falls. Ensrud et al. (2002), in a prospective study of 8,127 women aged 65 or older participating in an osteoporotic fracture study, found that benzodiazepines, antidepressants, and anticonvulsants increased risk for fall. Serotonin reuptake inhibitors and narcotics had no evidence of association with falls. In a study by Smith (2003), adverse effects of medications that could result in fall injury were reported in 92.9% (n = 157) of the cases. The three most common adverse effects encountered in Smith’s study were dizziness, nerve changes, and edema. The system established at this organization to flag highrisk medications has increased the awareness of the staff. The medications identified at high risk had statistical significance ( p = .0014). The home medication question asked on admission did not have significance for correlation with falls ( p = .9915). There are many factors that could explain this finding. Many patients on admission are unable to answer medication history questions due to their condition. The information may have been obtained from family. Patients themselves are sometimes unclear of their medications. It may be that home medications have little to no impact on fall risk in the hospital. 7. Limitations and recommendations for further research The initial literature search in PubMed provided references from a broad range of disciplines in multiple settings. The search was limited, however, by not including the Cumulative Index to Nursing and Allied Health Literature and other evidence-based practice sites specific to nursing. Using computerized entry of SFRAT data by staff nurses provided a more realistic picture of the tool’s validity at the bedside but limited the control of potential confounding variables such as interindividual variation in completing the assessment. Further, introducing the SFRAT in conjunction with the institution-wide fall prevention program made it difficult to separate the impact of the risk assessment tool on reported falls from the effects of the total programmatic package. The SFRAT’s capacity to differentiate between fallers and nonfallers provides beginning but incomplete evidence for the SFRAT’s validity. Further evidence for the validity of the SFRAT must be compiled through additional testing of the tool such as comparison with established risk assessment tools. Additional multisite validity testing is
92
C. Robey-Williams et al. / Applied Nursing Research 20 (2007) 86 – 93
warranted to overcome the lack of generalizable findings from the current study, which involved nurses from only one hospital. 8. Implications for practice The SFRAT offers a simple, user-friendly tool that can be easily incorporated by busy acute care nurses into their routine patient care. No cumbersome scoring is required; rather, nurses use an algorithmic three-step process in which yes or no responses correspond to implementation of one or more of a three-tiered set of fall risk interoventions. The incorporation of the functional get-up-and-go test as a component of the SFRAT requires the training of nurses to ensure correct and consistent implementation of the procedure across patients. However, nurses in the current study did not find it a difficult procedure to learn. Ongoing monitoring of fall rates over time will ultimately determine the effectiveness and positive impact of this tool on patients.
Appendix B. Morse fall scale 1. History of falling (immediate or within 3 months) No = 0 Yes = 25 2. Secondary diagnosis (a patient with two or more admitting diagnoses would correspond to a Yes response) No = 0 Yes = 15 3. Ambulatory aid None, bed rest, wheelchair, nurse = 0 Crutches, cane, walker = 15 Furniture = 30 4. IV/Heparin lock (or order for INT/IV) No = 0 Yes = 20 5. Gait/Transferring (observe gait/transfer) Normal, bed rest, immobile = 0 Weak = 10 Impaired = 20 6. Mental status (e.g., patient forgets to use call light) Oriented to own ability = 0 Forgets own limitations = 15 Total score
9. Conclusion The SFRAT tool for fall risk assessment is a simple, reliable tool easily incorporated by nurses into their direct care routine. Additional study is required to verify validity findings. Additional education is necessary to maintain compliance and ensure accurate measurement. Ongoing monitoring of fall rates over time will ultimately determine the effectiveness and positive impact of this tool on patient safety. Appendix A. SFRAT data collection tool 1. Has the patient fallen within the last 3 months? Yes No 2. On admission, was the patient on any home medications that increase fall risk? Benzodiazepines—lorazepam (Ativan), clonazepam (Klonopin), diazepam (Valium) Beta-blockers—metorpolol (Toprol/Lopressor), carvedilol (Coreg), atenolol (Tenormin) Anticonvulsants—gabapentin (Neurontin) Antipsychotics—haloperidol (Haldol), risperidone (Risperdal), olanzapine (Zyprexa) Yes No 3. Is fall risk identified on the MAR? Yes No 4. Did the patient fail the get-up-and-go test (i.e., if the patient took longer than 30 s to complete the test or if the patient required any assistance)? Yes No 5. Would you place this patient on Fall Precautions based on your clinical judgment? Yes No
References American Geriatric Society. (2001). Guidelines for the prevention of falls in older persons. Journal of the American Geriatrics Society, 49 (5), 664 – 672. Dempsey, J. (2004). Falls prevention revisited: A call for a new approach. Journal of Clinical Nursing, 13, 479 – 485. Eagle, D. J., Salam, S., Whitman, E., Evans, L. A., Ho, E., & Olde, J. (1999). Comparison of three instruments in predicting falls in selected inpatients in a general teaching hospital. Journal of Gerontological Nursing, 40 – 45. Ensrud, K. E., Blackwell, M. A., Mangione, C. M., Bowman, P. J., Whooley, M. A., Bauer, D. C., et al. (2002). Central nervous system active medications and risk for fall in older women. Journal of the American Geriatrics Society, 50(10), 1629 – 1637. Gaebler, S. (1993). Predicting which patient will fall again. . . and again. Journal of Advanced Nursing, 18 1895 – 1902. Gatti, J. C. (2002). Which interventions help to prevent falls in the elderly? American Family Physician 65(11), 2259 – 2260. Hampton, J. L., Kenny, R. A., & Newton, J. L. (2002). Effective interventions to prevent falls in older people. British Journal of General Practice, 52(484), 884 – 886. Hendrich, A., Nyhuis, A., Kippenbrock, T., & Soja, M.E. (1995). Hospital falls: Development of a predictive model for clinical practice. Applied Nursing Research, 8(3), 129 – 139. Hui-Chi, H., Gau, M. -L., Lin, W. -C., & Kernohan, G. (2003). Assessing risk of falling in older adults. Public Health Nursing, 20(5), 399 – 411. Kerzman, H., Chetrit, A., Brin, L., & Toren, O. (2004). Characteristics of falls in hospitalized patients. Journal of Advanced Nursing, 47(2), 223 – 229. Kingma, J., & Ten Duis, H. -J. (2000). Severity of injuries due to accidental fall across the lifespan: A retrospective hospital based study. Perceptual and Motor Skills, 90, 62 – 72. Mathias, S., Nayak, U. S. L., & Issacs, B. (1986). Balance in elderly patients: The get-up and go test. Archives of Physical Medicine and Rehabilitation, 67, 387 – 389. Mendelson, W. B. (1996). The use of sedative/hypnotic medication and its correlation with falling down in the hospital. Sleep, 19(9), 698 – 701. Morse, J. M. (1977). Preventing patient falls. Thousand Oaks USA7 Sage Publications.
C. Robey-Williams et al. / Applied Nursing Research 20 (2007) 86 – 93 Myers, H., & Nikoletti, S. (2003). Fall risk assessment: A prospective investigation of nurses’ clinical judgement and risk assessment tools in predicting patient falls. International Journal of Nursing Practice, 9, 158 – 165. Nyberg L., & Gustafson Y. (1997). Fall prediction index for patients in stroke rehabilitation. Stroke, 28, 716 – 721. Perell, K. L., Nelson, A., Goldman, R. L., Luther, S. L., Prieto-Lewis, N., & Rubenstein, L. Z. (2001). Fall risk assessment measures: An analytic review. Journal of Gerontology, 56A(12), M761–M766. Podsiadlo, D., & Richardson, S. (1991). The timed get up and go: A test of basic functional mobility for frail elderly persons. Journal of Gerontological Society of America, 39(2), 142 – 148. Premier Incorporated. (2004). Fall Prevention. www.premierinc.com/all/ safety/resources/falls/#Risk. Accessed March 1, 2007. Rollins, G. (2003). Pharmacists can help reduce falls in the elderly through medication management. Reports of Medical Guidelines and Outcomes Research, 14(12), 9 – 12. Rubenstein, L. Z., Powers, C. M., & MacLean, C. H. (2001). Quality indicators for the management and prevention of falls and mobility problems in vulnerable elders. Annals of Internal Medicine, 135(8), 686 – 693. Semin-Goossens, A., van der Helm, J. M. J., & Bossluyt, P. M. M. (2003). A failed model-based attempt to implement an evidence-based nursing guideline for fall prevention. Journal of Nursing Care Quality, 18(3), 217 – 225.
93
Shumway-Cook, A., Brauer, S., & Woollacott, M. H. (2000). Predicting the probability for falls in community-dwelling older adults using the Timed Up and Go Test. Physical Therapy, 80, 896 – 903. Smith, R. G. (2003). Fall-contributing adverse effects of the most frequently prescribed drugs. Journal of the American Podiatric Medical Association, 93(1), 42 – 50. Tinetti, M. E. (1997). Falls. In: Calssel C., et al., (Eds.), Geriatric Medicine, (3rd ed. pp. 787–799). New York7 Springer-Verlag. Tinetti, M.E. (2003). Preventing falls in elderly persons. New England Journal of Medicine, 348(1), 42 – 49. VA Fall Prevention and Management. (2002). www.patientsafety.gov/ FallPrev/SafetyTopics/fallstoolkit/index.html. Accessed March 1, 2007. Vassalo, M., Sharma, J. C., & Allen, S. C. (2002). Characteristics of single fallers and recurrent fallers among hospital in-patients. Gerontology, 48, 147 – 150. Wall, J.C., Bell, C., Campbell, S., & Davis, J. (2000). The timed get up and go test revisited: Measurement of the component tasks. Journal of Rehabilitation Research and Development, 37(1), 1 – 7. Wallace, C., Reiber, G. E., LeMaster, J., Smith, D. G., Sullivan, K., Hayes, S., et al. (2002). Incidence of falls, risk factors for falls, and fall related fractures in individuals with diabetes and in presence of foot ulcer. Diabetes Care, 25(11), 1983 – 1986. Yauk, S., Hopkins, B. A., Phillips, C. D., Terrell, S., Bennison, J., & Riggs, M. (2005). Predicting in-hospital falls: Development of the Scott and White Falls Risk Screener. Journal of Nursing Care Quality, 20(2), 128 – 133.