Journal of Gerontological Nursing

Comparison of Three Instruments in Predicting Accidental Falls in Selected Inpatients in a General Teaching Hospital

D Joan Eagle, RN, BSc, MN, MSc; Suzette Salama, PhD; Dee Whitman, MHSc, OTR, OT(Reg); Lois Ann Evans, RN, BScN, BA; Enoch Ho, RPT, MPh; Jan Olde, RN, CIC, BHSc(N)

Abstract

Use of these tools helps identify elderly individuals at the highest risk for falls and aids in intervention planning.

Abstract

Use of these tools helps identify elderly individuals at the highest risk for falls and aids in intervention planning.

Age generally has been accepted ' widely as a major risk factor in patient falls, and falls are among the leading causes of morbidity in this population. Approximately one third of individuals age 65 and older will fall each year, and the 1 -year mortality rate among individuals who suffer a fracture is more than 15% (Nickens, 1985).

Several studies have examined the variables that contribute to falls in elderly individuals. Higher risk of falling was identified in elderly patients who have impaired or decreased mobility of the lower extremities (Janken, Reynolds, & Sweich, 1986), who need assistive ambulatory devices because of impaired gait (Morse, Tylko, & Dixon, 1987), or who walk without the use of assistive aids (Uden, 1985). Several studies found confusion, bladder urgency, fecal incontinence, and diarrhea were significant variables (Hernández & Miller, 1986; Janken et al., 1986; Lund & Sheafor, 1985). Polypharmacy (especially diuretic, hypotensive, and anticonvulsant medications) also were found to be significant among those individuals who fell (Barbieri, 1983; Lund & Sheafor, 1985). Medical and neuropsychiatrie conditions and impaired vision and hearing also have been implicated as high risk factors (Barbieri, 1983; Schested, & Severin-Neilsen, 1977), as well as increased hospital length of stay (Kostopoulous, 1985; Uden, 1985). These studies also have shown that patients who fall once during their hospital stay are more likely to fall again.

Most likely, falls in elderly individuals represent a complex phenomenon and rarely have a single cause. Therefore, attempts to predict who will fall must account for all factors. After identifying the fall-prone patients, measures can be implemented to prevent falls. Before implementing programs to reduce the number of falls, one must have a reliable measurement tool to identify patients with the highest likelihood of falling. Several assessment scales have been reported, and the best ones were compared for reliability and validity in a recent survey (Duncan, Studenski, Chandler, Oc Prescott, 1992).

There are several dynamic balance tests available to screen people at high risk for falls. Expensive pieces of laboratory equipment, such as the biomechanics force platform, attempt to measure balance but are not accessible or feasible in clinical environments (Lichtenstein, Shields, Shiavi, & Burger, 1988). Other clinical tests such as the Postural Stress Test (Hill, Vandervoort, &c Kramer, 1990), the Get Up and Go Test (Mathias, Nayak, & Isaacs, 1986; Podsiadlo & Richardson, 1991), and the TinettiPerformance Oriented Mobility Assessment (Tinetti, Speechley, & Ginter, 1988) lack evidence of reliability or validity.

Recently, the Functional Reach (FR) test has gained popularity as a clinical test of balance. It represents the maximal distance one can reach forward beyond arm's length in the horizontal plane, while maintaining a fixed base of support in the standing position (Duncan, Weiner, Chandler, & Studenski, 1990). It is an objective quantitative measure of balance and can be measured accurately and inexpensively using a leveled yardstick (Thapa, Gideon, Fought, Kormicki, & Ray, 1994). Early test results indicate that people with an average reach of 7.4 inches or less are at high risk for falls. The test-retest reliability and interobserver reliability of the FR test in 128 healthy subjects has been established (Duncan et al., 1990). The validity of the FR test in 45 elderly people with varied levels of frailty living in the community was found to correlate with other physical measures which include instrumental activities of daily living, tandem gait, and timed ambulation (Weiner, Duncan, Chandler, & Studenski, 1992). The FR test also was found to have good predictive validity in a sample of 214 elderly people. Subjects who were limited in FR also suffered from recurrent falls in a 6-month follow up (Duncan et al., 1992).

The Morse Fall Scale (MFS) (Morse, Morse, Si Tylko, 1989) consists of 6 scored items which can be administered easily using a chart audit or direct observation. The items rated are:

* History of falling.

* Presence of a secondary diagnosis.

* Use of an ambulation aid.

* Intravenous therapy.

* Type of gait.

* Mental status.

A score of 45 or higher is considered to predict high risk for falls, and a score of 20 or lower indicates low risk for falls. The sensitivity of the scale was found to be 78%, with a specificity of 83%. Interrater reliability scores were r = .96. The scale was found to be sensitive to different patient conditions and to length of stay.

STATEMENT OF THE PROBLEM

At the time of admission, nurses on rehabilitation and geriatric medical wards attempt to identify patients who are at high risk for falls. This is accomplished subjectively by the primary nurses during the initial assessments and on a recurrent basis when care is provided. The primary nurses are RNs or Registered Practical Nurses (RPNs) who are assigned consistently to their patients, know the patients very well, and become the individuals most responsible for the care delivery to those patients. The determination of risk often is articulated as a "gut feeling" that patients are likely to fall. Prompted to explore and expand these intuitive assessments, nurses usually incorporate their knowledge of the patients' past experiences of falling, ages, diagnoses, previous use of restraints, cognition, side effects of some medications, functional abilities, and nutritional status.

Having decided that some patients are at risk for falls, the nurses may initiate requests for Physiotherapy (PT), Occupational Therapy (OT), or Pharmacy (PH) assessments to confirm the risk determination and to establish collaborative plans for fall prevention. In the interim, the patients may be restrained minimally to promote safety until all disciplines have completed their assessments and can begin to discuss collaborative fall prevention plans.

Because the current fall risk assessment process is subjective and may not be made on a consistent or regular basis, it is important to identify an effective tool for early detection of patients at high risk for falls. This promotes the development of hospital-wide accidental falls prevention programs and lends credence to policies of least restraint.

PURPOSE OF THE STUDY

The purpose of this study was to compare the abilities of the MFS, the FR test, and the primary nurses' clinical judgment (CJ) to predict falls with inpatients on a rehabilitation unit and a geriatric medical ward.

Table

TABLE 1PATIENT DEMOGRAPHICS

TABLE 1

PATIENT DEMOGRAPHICS

METHODS

Patient Population

Inpatients from a rehabilitation ward and a geriatric medical ward (i.e., where patients are awaiting placement in the community) were studied because these individuals tend to be medically stable and to have longer lengths of stay in the hospital. All inpatients on those wards who signed a consent form were eligible for the study, with the exception of terminally Ul patients.

Study Design

Patients entered the study through one of two methods:

* When the study began, all consenting patients already on the two study wards were assessed on the same day for their risk for falls using the MFS, the FR test, and the nurses* CJ.

* All subsequent admissions to the study wards were assessed by the three methods in the same time period (i.e., morning or afternoon) 3 to 5 days after their admission to the ward, allowing for personal adjustment to the new ward environment.

Patients were assessed using all three measurement tools in the same day in a blind manner (i.e., the test results were not known to the three research teams at the time of testing).

For the purpose of this study, accidental patient falls were documented in Patient Incident Reports and defined as when the patients were found on the floor or assisted to the floor when a fall could not be prevented. These Patient Incident Reports were used as documentation that falls had occurred; thus, they were used as the gold standard (i.e., the actual number of falls) for patient falls.

Clinical practice on the two wards was not changed for the purpose of this study. As usual, PT, PH, and OT consultations were completed on identified high-risk patients, and their recommendations were implemented by the staff. Staff also continued to provide input (regarding education and equipment) to reduce the risk for falls.

Study patients were followed for 3 months, until discharge from the ward, or until the time of their first fall, whichever came first. When patients fell after the initial assessments, they were no longer followed, and this incident was included in the study.

Functional Reach Tool

The FR instrument (Duncan, Weiner, Chandler, & Studenski, 1990) was administered by a team of two people (a physiotherapist and an occupational therapist not working with this sample) who were familiar with the test procedure. Consistency of measurements was ensured by:

* Providing a standardized set of instructions to the patients.

* The same two health professionals consistently administering the test, following the same procedure.

* Averaging the results of the last three valid attempts to obtain the final test score. Individuals with scores of O or 1 (i.e., average reach of 6 inches or less) were considered at high risk for falls.

Morse Fall Scale

A retrospective chart audit was used to gather the data for the MFS including the six items listed in the literature review (Morse et al., 1989). For new patients, this occurred 3 to 5 days after admission to the ward (in the same morning time period as the FR and the nurses* CJ). This audit was completed by one of two research assistants who were trained auditors and who had no direct involvement with the study wards. Intrarater reliability was determined for the MFS by having each auditor blindly evaluate an identical set of 10 charts at two different times. The research assistant asked the nurse working with the patient to score the gait component because the nurse had observed this activity. Patients with scores of 45 or higher were considered at high risk of falls.

Clinical Judgment Assessment

On the same mornings when the FR and the MFS were conducted, the primary nurses caring for the patients were asked to make a CJ regarding the likelihood of each patient having a fall. The research assistant asked each nurse, "Is your patient at risk for falls in the near future?". In addition to a yes or no answer, the staff members were asked to give a rationale for their judgments. A standardized form was developed for ease of recording responses. At the time of this study, the care was provided by both RPNs (60%) and RNs (40%). The main differences between the two groups of nurses are that RPNs must practice under the supervision of a RN and RPNs are restricted in which medications they can dispense. Generally, RNs attend a university nursing program for 3 to 4 years, while RJPNs study nursing for 1 to 2 years.

Data Analysis

Using Patient Incident Reports as the gold standard for inpatient falls, the sensitivity, specificity, and predictor values for each test (FR, MFS, and CJ) were calculated using a two-by-two contingency table for each test. The data were analyzed from the time of the initial fall assessments until 3 months later, until discharge, or until the first fall, whichever came first. This overall set of two-by-two tables was calculated based on all patients in the study regardless of length of stay.

RESULTS

A total of 98 patients was entered into the study. Six were from the geriatric medical ward and the remainder were from the rehabilitation ward. The average age for the total population was 69, with a median and mode age of 73, and an age range of 23 to 96. Sixty percent of the total population were women, and the average length of stay was 60 days (median = 45 days; mode = 39 days), with a range from 10 to 346 days (Table 1).

Table

TABLE 2SUMMARY OF DATA ANALYSIS (N = 98)

TABLE 2

SUMMARY OF DATA ANALYSIS (N = 98)

There was a total of 55 accidental falls during the study period: 70% of the population (n = 69) did not fall; 15 people fell once; 8 people fell twice; 4 people fell three times; 1 person fell four times; and 1 person fell eight times. Therefore, the overall incidence of falls among the patients was 30%. The sensitivity, specificity, positive and negative predictive values, and accuracy of each tool are presented in Table 2. There was no statistically significant difference among the three measurements in any of the indexes.

Because nurses' CJ was as effective in predicting the likelihood of falls as the two standardized tests, the nurses' rationale for each CJ was compared with the actual outcomes. The most discriminating reasons provided for correct prédictions of falls were: previous falls, walking with supervision, impulsive behaviors, aphasia and/or cognitive impairment, unwillingness to follow safety techniques, and poor balance on patients* feet. Aphasia was included because the patients did not follow commands or ask for assistance. Examination of the rationale provided when nurses' CJ correctly predicted a fall would not occur (no fall) included some of the above factors, but no consistent pattern could be detected (Table 3).

LIMITATIONS

Because some inpatient falls may be undetected by the hospital health care professionals caring for the patients and because it is conceivable that a patient may fall without a Patient Incident Report being completed, it is anticipated that the number of patient falls was underreported in this study.

Table

TABLE 3RATIONALE GIVEN WHEN CLINICAL JUDGMENT CORRECTLY PREDICTED FALL OR NO FALL

TABLE 3

RATIONALE GIVEN WHEN CLINICAL JUDGMENT CORRECTLY PREDICTED FALL OR NO FALL

DISCUSSION

Use of the FR test as a predictive tool of a single fall in an inpatient population has demonstrated weak specificity (.34) and fair sensitivity (.76). This may relate to the fact that single falls are more likely to be affected by multiple factors including medication changes, progression of disease, or environmental changes (Teasdale, Stelmach, & Breunig, 1991). Maintaining balance in an upright position is a complex neuromuscular mechanism involving vision, peripheral sensibility, lower extremity strength, and vertebrobasilar function. Falling more than once may be more predictable by the FR test because the falls are likely to be associated with balance. While the FR test is a reliable tool to measure balance and instability, it is not applicable universally to every patient group (Weiner et al., 1993). Use of the tool with patients with dementia and restricted arm movement in forward flexion will affect the reliability of the tool. The scores of individuals who are unable to stand without support by any walking aids will affect the specificity because they were assigned a zero score according to the protocol used in this study.

The prevalence rate of hospital falls in this study was 56%, which is a higher rate than that found in other populations recruited for studies using the FR test (Duncan et al., 1992). A possible explanation for the higher fall rate is that the group of patients included in this study were inpatients, while other studies focused on individuals living in the community (Lichtenstein, Shields, Shiavi, & Burger, 1989). This fact also would the superior sensitivity and values demonstrated in studies (Duncan et al., 1992).

The MFS had slightly better test than the FR test. This may be the MFS is based on a multichart review. In this study blind review was assured the chart reviews were comby research assistants who did know the patients. A more accurate score probably would have been if the primary caregivers completed the MFS. If the MFS is completed on a routine basis by primary caregivers, it only requires a matter of minutes at the caregivers' convenience. Therefore, the MFS could be the more efficient of the two standardized tools in terms of predictive ability, overall accuracy, and convenience.

The CJ of the nursing staff was as effective in predicting patient falls as the two standardized tools used in this study. This may be because the nurses, using a holistic approach, have a clear picture of the patients' past and current functioning, including their strengths and weaknesses. For some time, it has been considered that nurses possess intuitive knowledge unique to their practice, although this has not been validated by scientific studies. Some individuals believe this intuition is derived from reflection on practice and from experience. In contrast, the MFS is based primarily on an audit of the patients* charts which may not contain the amount and detail of information which the nurses, who are the primary caregivers, have obtained. In addition, during the process of assimilating and assessing patient information, the nurses may integrate and categorize information to formulate the resultant CJ. Therefore, it is probable that if a standardized patient fall assessment, such as the MFS, was completed by primary nurse caregivers based solely on the nurses' knowledge and intuition about the patients, the sensitivity, specificity, and predictive values would be increased dramatically.

CONCLUSION

It is important to identify those individuals at the highest risk for falls and to plan interventions to prevent further occurrences. The two objective, standardized tests selected to predict accidental falls in hospital inpatients (i.e., FR test, MFS) were time consuming and were no better at prediction than the CJs made by the primary nurse caregivers. If the nurses' CJ, based on current knowledge of the patients' status, and professional intuition were tapped regularly through the use of a standardized format such as the MFS, the sensitivity, specificity, and predictive values may be increased to such an extent that those individuals at highest risk for falls could be correctly identified, and likely further falls could be prevented.

REFERENCES

  • Barbieri, E. (1983). Patients falls are not patient accidents. Journal of Gerontological Nursing, 9(3), 164-173.
  • Duncan, RW., Studenski, S., Chandler, J., & Prescott, B. (1992). Functional reach: Predictive validity in a sample of elderly male veterans. Journal of Gerontology, 47, M93-M97.
  • Duncan, P.W., Weiner, D.K., Chandler, J., & Studenski, S. (1990). Functional reach: A new clinical measure of balance. Journal of Gerontology, 45, M192-M197.
  • Hernandez, M., & Miller, J. (1986). How to reduce falls. Geriatric Nursing 7(2), 97-102.
  • Hill, K., Vandervoort, A-, & Kramer, J. (1990). Reproducibility of performance on test of postural responses in healthy elderly females. Physiotherapy Canada, 42, 61-66.
  • Janken, J., Reynolds, B., & Sweich, K. (1986). Patient falls in the acute care setting identifying risk factors. Nursing Research, 35(4), 215-219.
  • Kostopoulous, M. (1985). Reducing patient falls. Orthopedic Nursing, 4(6), 14-15.
  • Lichtenstein, MJ-, Shields, S.L., Shiavi, R.G., & Burger, M.C. (1988). Clinical determinants of biomechanical platform measures of balance in aged women. Journal of the American Geriatrics Society, 36, 996-1002.
  • Lichtenstein, MJ., Shields, S.L., Shiavi, R.G., & Burger, C. (1989). Exercise and balance in aged women: A pilot controlled clinical trial. Archives of Physical Medicine Rehabilitation, 70, 138-143.
  • Lund, C., & Sheafor, M. (1985). Is your patient about to fall ? Journal of Gerontologkal Nursing, 11(4), 37-41.
  • Mathias, S., Nayak, U.S.L., Isaacs, B. (1986). Balance in elderly patients: The "get-up and go" test. Archives of Physical Medicine and Rehabilitation, 67, 387-389.
  • Morse, J-, Tylko, S., & Dixon, H. (1987). Characteristics of the fall-prone patient. The Geruntologist, 27(4), 516-522.
  • Morse, J.M., Morse, R.M., & Tylko, SJ. (1989). Development of a scale to identify the fall-prone patient. Canadian Journal on Aging, 8(4) 866-876.
  • Nickens, H. (1985). Intrinsic factors in falling among the elderly. Archives of Internal Medicine, 145, 1089-1093.
  • Podsiadlo, D-, Richardson, S. (1991). The timed "up and go": A test of basic functional mobility for frail elderly persons. Journal of the American Geriatrics Society, 39, 142-148.
  • Schested, P., & Severin-Neilsen, T. (1977). Falls by hospitalized elderly patients: Causes, prevention. Geriatrics, 32(4), 101-108.
  • Teasdale, N., Stelmach, G.E., & Breunig, A. (1991). Postural sway characteristics of the elderly under normal and altered visual and support surface conditions. Journal of Gerontology, 46, B238-B244.
  • Thapa, P.E., Gideon, P., Fought, R.L., Kormicki, M., Sc Ray, W. (1994). Comparison of clinical and biomechanical measures of balance and mobility in elderly nursing home residents. Journal of the American Geriatrics Society, 41, 493-500.
  • Tinetti, M.E., Speechley, M., & Ginter, S.F. (1988). Risk factors for falls among elderly persons living in the community. New England Journal of Medicine, 319, 17011707.
  • Uden, G. (1985). Impatient accidents in hospitals. Journal of the American Geriatrics Society, 33, 833-841.
  • Weiner, D.K., Dennis, R., Bongorni, D.R., Studenski, S.A., Duncan, P.W., & Kochersberger, G.G. (1993). Does functional reach improve with rehabilitation? Archives of Physical Medicine Rehabilitation, 74, 796-800.
  • Weiner, O.K., Duncan, P.W., Chandler, J., & Studenski, S.A. (1992). Functional reach: A marker of physical frailty. Journal of the American Geriatrics Society, 40, 203-207.

TABLE 1

PATIENT DEMOGRAPHICS

TABLE 2

SUMMARY OF DATA ANALYSIS (N = 98)

TABLE 3

RATIONALE GIVEN WHEN CLINICAL JUDGMENT CORRECTLY PREDICTED FALL OR NO FALL

10.3928/0098-9134-19990701-14

Sign up to receive

Journal E-contents