Identification of Patients at Risk for Falls in an Inpatient Rehabilitation Program


  • Lisa A. Salamon MSN RN GCNS-BC WOCN,

  • Mary Victory MSN RN,

  • Kathleen Bobay PHD RN NEA-BC



To determine if there is a more sensitive method to identify inpatient rehabilitation patients at high risk for falls rather than the Morse fall scale.


Retrospective analysis of falls occurring during 6-month period in 2009. Age and diagnosis were used to create comparison groups between patients who fell and those who did not. T-tests were used to determine differences between the two groups in FIM scores and Morse fall scores.


Patients who had stroke as a primary diagnosis were more likely to fall than other patients. Length of stay was greater for patients who fell (p = .008). The positive predictive value of the Morse fall scale for patients who fell was 57%, suggesting that it is not a sensitive predictor of falls in rehabilitation patients. Patients who fell had significantly lower FIM expression scores (p = .02).


Unintended falls are common on inpatient rehabilitation units, occurring at a rate of 9.5%–12.5%, and as high as 37% in stroke rehabilitation populations (Lee & Stoic, 2008; Saverino, Venevolo, Ottonello, Esirai, & Sessarego, 2006; Teasell, McRae, Foley, & Bhardwaj, 2002). In comparison, an evaluation of a large academic hospital revealed a 3.1% fall rate for acute inpatient units (Fischer et al., 2005). Patients participating in rehabilitation may experience more falls due to the promotion of independence and mobility, which challenges multiple systems of balance and can increase the risk for falling (Saverino et al., 2006; Zbobysz, Boradia, Ennis, & Miller, 2005).

Consequences of patient falls may include: patient injury causing further disability and deconditioning, prolonged recovery times, increased length of stay, greater healthcare costs, and legal consequences (Lee & Stoic, 2008; Saverino et al., 2006; Suzuki, Sonodo, & Misaa, 2005). A study of elderly rehabilitation patients revealed that patients who fell during their hospital stay were more likely to be discharged to nursing homes and experienced higher mortality rates (Vassallo et al., 2009). To prevent negative outcomes associated with falls, appropriate fall prevention programs should be designed to benefit the rehabilitation patient.

A majority of research related to fall prevention has been conducted on acute inpatient populations and is not sensitive to the uniqueness of rehabilitation patients (Zbobysz et al., 2005). A prospective study conducted by Morse, Black, Oberle and Donahue (1989) found the Morse Fall Scale (Appendix) to be an effective predictor of the fall prone patient in the rehabilitation setting. Scales that are used by nursing to identify patients as high fall risk in acute care settings, such as the Morse Fall Scale, can classify 75% to 90% of patients in inpatient rehabilitation as “high fall risk” (Gilewski, Roberts, Hirala, & Riggs, 2007). This is not specific enough. Schwendimann, DeGeest and Milisen (2006) suggest that the Morse Fall scale should be validated locally to determine the best cutoff scores in a given setting. Unfortunately, at this point in time a valid method to identify the inpatient rehabilitation patients that are at the highest risk for falling has yet to be identified.

The Functional Independence Measure (FIM) (Appendix) is a measurement of disability specific to patients in rehabilitation. The FIM has 18 categories subdivided into motor and cognitive components. Each category is scored on a scale of 1 (total dependence) to 7 (total independence). FIM scores have been shown to predict discharge outcomes and may have value in identifying patients at high fall risk (Gilewski et al., 2007; Teasell et al., 2002). Limited research has been conducted involving FIM scores in association with the identification of patients in inpatient rehabilitation who are high fall risk. Gilewski et al. (2007) found inpatient rehabilitation participants who fell had lower admission and discharge FIM scores than those who did not fall. In addition, mobility and problem solving FIM scores were most predictive of falls, especially when the scores were used in combination. Lee and Stoic (2008) also found that patients who fell had lower total and motor FIM scores upon admission and discharge than those who did not fall. Lower FIM scores were also found in studies of stroke rehabilitation fall patients in comparison to non-fallers (Teasell et al., 2002; Zbobysz et al., 2005). Suzuki et al. (2005) found patients in inpatient stroke rehabilitation with FIM motor scores between 26 and 28 had the highest rate of falls. They also found that patients with low cognitive FIM scores had significantly more falls than patients with cognitive FIM scores higher than 29. It appears that FIM scores may have value in predicting patients that have the highest risk of falls on an inpatient rehabilitation unit. If it can be determined which FIM scores in combination with the Morse Scale score identify the highest risk patients more appropropriate, targeted interventions can be implemented.

The Pilot Study


A 45-bed inpatient rehabilitation program within a Midwest tertiary care hospital with an average daily census of 34 was the setting for this study. The program specializes in stroke rehabilitation. There were five full-time physiatrists, two full-time social workers, a full-time rehabilitation psychologist, and the full complement of therapy services (physical therapy, occupational therapy, speech therapy, therapeutic recreation). Nursing care is provided by registered nurses and nursing assistants. Nursing leadership is represented by a partnership between the patient care manager and the clinical nurse specialist. Ancillary services are provided by staff of the medical center that houses the program. All patient rooms are single occupancy. Therapy sessions occur in the patient's room, in the gym, or in a room set up as an apartment. Caregivers are encouraged to participate in therapy and teaching sessions provided by the nurse in preparation for discharge. The overall patient population for the program ranges in age from 18 to greater than 90.

The pilot study included all inpatient rehabilitation patients who had a fall from January 1, 2007, through June 30, 2007. The definition of a fall as provided by the National Quality Forum (2004) is “an unplanned descent to the floor.” The sample size was 67 patients. The mean age of patients was 66.34 (range = 39–89, SD = 14.08). A retrospective chart review was used to compare the Morse Fall Scale scores and specific components of the FIM. Data were collected from the electronic medical record for that patient encounter. Patients have scores for both the Morse scale and the FIMs each day as part of the routine assessment of patients in the inpatient rehabilitation program. The Morse scoring is performed solely by the nurse, FIM scores are entered by both therapists and nurses. Inter-rater reliability of FIM scoring is met by completing a bi-annual competence test. The Morse score before the fall is taken from the last score entered within 24 hours before the fall. The components assessed include history of falls, secondary diagnosis, use of an ambulatory aid, IV therapy in process, gait, and mental status. The FIM scores are taken from the first 24 hours of the patient's stay. The FIM scores that were included were Comprehension, Expression, Problem Solving, and Memory. Correlation analysis and confidence intervals were performed. Median scores were used to determine the cutoff points for critical FIM values.

Findings from the pilot study (Salamon & Bobay, 2008) showed that there were three significant correlations between the Morse Fall Scale before fall score and FIM scores recorded in the first 3 days of admission. The first was Problem Solving (r = .90, p < .000). The second score was Memory (r = .772, p < .000) and the third was Expression (r = .883, p < .000). These results are intuitively coherent. Patients with problem solving deficits and memory deficits might be expected to be more likely to fall due to cognitive impairments. Those with decreased expression are likely to fall because of the decreased ability to express their needs. There were no significant correlations with comprehension and the Morse Fall Scale. Critical cut points were explored for the significant FIM scores by comparing median scores with Morse Fall Scale scores >45. The critical cut point for Problem Solving was 6, Expression was 5, and Memory was 4.

Interventions as a Result of the Pilot Study

Upon completion of the pilot study there were attempts to utilize the Morse Scale Score in combination with the 3 FIM scores of problem solving, memory, and expression to identify the patients at highest risk for fall. Pocket cards with the information from the pilot study were developed and shared with nurses. The nurses were asked to look at the scores, place a note on the “kardex” to state why the patient was at risk (i.e., Morse Scale 65 and Problem Solving FIM 3), and implement appropriate interventions based on these assessments. In addition, these patients would have a fall leaf sign outside their door to signal the need for increased surveillance. There was very little follow-through with completing these tasks from the initial education sessions. Staff indicated that it was difficult to remember four different scores and it did not appear that the number of patients identified as being at high risk changed at all. More work was needed and a follow-up study was completed in 2009.

Replication Study—Research Questions

Two questions were posed in the replication study: (1) Do inpatient rehabilitation patients with low admission problem solving, expression, and memory FIM scores fall more than patients who have higher admission FIM scores, and (2) Do admission Morse fall scale scores differ between patients who fell and those who did not fall?


This study was a retrospective descriptive comparison of inpatient rehabilitation patients who fell with those patients who did not fall. IRB approval was granted from the hospital and university review boards.

Sample Description/Population

A convenience sample was drawn from a pool of 872 patients admitted to three Midwestern hospital-based inpatient rehabilitation programs between January 1, 2009, and June 30, 2009. All programs accept a wide variety of rehabilitation patients. The smallest program had 10 beds, the largest was 45, and the third was midsized at 20 beds.

The study group consisted of 70 patients who had fallen a total of 82 times. There was no distinction between those patients who were found on the floor versus those who were assisted to the floor. Fifty-nine patients fell once; 11 patients fell more than once. There were 31 women (44.3%) and 39 men (55.7%) in the study group with an average age of 69.8 (range = 21–92, SD = 14.3). Age and diagnosis were used to create a matched comparison group from the pool of 872 patients. The comparison group consisted of 35 women (50%) and 35 men (50%) with an average age of 69.6 (range = 40–90, SD = 12.5) (Table 1). Of the 872 study participants the most common diagnosis was stroke (n = 226). Additional diagnoses included malaise and fatigue (n = 173), amputation and joint care (n = 102), Parkinson's disease (n = 78), encephalopathy (n = 55), back injuries (n = 46), and miscellaneous other (n = 94). Falls occurred at a consistent rate across the three sites. The 45-bed unit had 522 (59%) admissions within the 6-month time frame, and 44 (8%) documented falls. The 10 bed unit had 137 (16%) admissions and 14 (10%) falls. The 20-bed unit had 213 (24%) admissions and 24 (11%) falls. These results indicate that although the 10-bed unit had a slightly higher percentage of falls in relation to percentage of admissions, there was a fairly even distribution of falls across the three units.

Table 1. Gender Characteristics of Patients Admitted to Inpatient Rehabilitation
 Fall GroupComparison GroupSig.
No. patients31393535.459


Patients with certain diagnoses experienced more falls than others. Patients who had a stroke fell the most often (n = 26, 31.7%), followed by malaise and fatigue (n = 22, 26.8%), amputation (n = 7, 8.5%), other (n = 4, 4.9%), Parkinson's disease (n = 3, 3.6%), encephalopathy (n = 11, 13.4%), gait issues (n = 5, 6%), rheumatoid arthritis (n = 3, 3.7%), and pain (n = 1, 1.2%).

Falls occurred 26 times on the day shift (32%), 35 times on the evening shift (43%), and 18 times on the night shift (22%). Three falls occurred during change of shift.

Independent t-tests were used to determine if there were differences between groups on length of stay. The average length of stay was 17.9 days (range = 1–44, SD = 9.8) for the study group and 12.7 days (range = 3–55, SD = 8.4) for the control group (p = .00) (Table 2).

Table 2. Demographic Data of Patients Admitted to Inpatient Rehabilitation
DemographicsFall GroupComparison GroupSig


  1. LOS = Length of stay.


Independent t-tests were used to determine differences in Morse Fall Scale scores upon admission between the two groups. The patients who fell had higher Morse Scale scores (M = 65.1, SD = 19.3) compared to those who did not fall (M = 55.2, SD = 24.4, t(138) = −2.67, p = .008) (Table 3). A three-way ANOVA was used to determine if there were differences between those who had one fall versus patients with multiple falls. A significant difference was found between those who did not fall, those patients that fell once (M = 64.6, SD = 19.3) and those who fell multiple times (M = 68.2, SD = 19.8, F(2) = 3.47, p = .03). There was a significant difference in Morse Scale scores between those who did not fall compared to those who fell once, p = .05. There were no significant differences between those who did not fall or fell once with those who had multiple falls. To answer the question of whether or not the Morse score values alone were able to predict falls on patients on the study units, both positive and negative predictive values were calculated using 45 as the hospital assigned cutoff score. Positive predictive value is the probability that a patient will fall and was calculated by the number of true positives divided by the number who tested positive (true positives + false positives). There were 72 true positives and 54 false positives. Therefore, the positive predictive value for patients who fell was: 72/126 = 57%. There were 28 true negatives and 10 false negatives, therefore, the negative predictive value for patients who did not fall was: 10/28 = 35.7%. The interpretation is that the Morse fall score is not a very good prediction tool for this sample of patients.

Table 3. Morse Fall Score Upon Admission in Fall and Comparison Groups
Morse Fall Score Upon AdmissionFallComparisonSig.

Independent t-tests were used to determine differences in the subcomponents of FIM scores. Patients who fell had significantly lower expression FIM scores (M = 4.65, SD = 1.1) on day one compared to those who did not fall (M = 5.52, SD = 1.2, t(36) = 2.235, p = .02). There were no differences in FIM problem solving or memory scores on the first day of admission between the patients who fell and those who did not. The comparison group and fall patients did not differ in cognitive FIM scores (expression, memory, or problem solving) 2 or 3 days after admission (Table 4). There were no differences in expression, memory, or problem solving FIM scores in the first 3 days after admission in patients who did not fall, fell once, and fell multiple times (Table 5).

Table 4. Differences in FIM Scores Between Fall Patients and Non-Fall Patients
FIM ScoresAdmission DayFallComparisonSig.
Problem solving14.641.335.161.34.187
Problem solving23.361.603.88.99.272
Problem solving33.561.333.801.23.683
Table 5. FIM Scores According to Number of Falls
Problem solving1.697
Problem solving2.218
Problem solving3.612


Utilization of current methods for identifying patients at high risk for falls on an inpatient rehabilitation unit were found to be ineffective since most patients were identified at high risk for falling using the Morse fall score. The purpose of this project was to determine if the Morse fall scale was sensitive enough to identify those patients at highest risk for fallings and to determine if FIM scores could better identify those at highest risk so appropriate nursing interventions could be put in place. We found that the Morse scale was not a sensitive enough tool to be utilized in these inpatient rehabilitation settings and that patients with an expression FIM score of 4 or less may be a better predictor of fall risk.


Functional Independence Measure—Uniform Data System for Medical Rehabilitation (2009)

FIM – Motor

  • 1.Eating
  • 2.Grooming
  • 3.Bathing
  • 4.Dressing – Upper Body
  • 5.Dressing – Lower Body
  • 6.Toileting
  • 7.Bladder
  • 8.Bowel
  • 9.Bed, Chair, Wheelchair Transfer
  • 10.Toilet Transfer
  • 11.Tub, Shower Transfer
  • 12.Walk/Wheelchair
  • 13.Stairs

FIM – Cognitive

  • 14.Comprehension
  • 15.Expression
  • 16.Social
  • 17.Problem Solving
  • 18.Memory


  • 7 – Complete Independence (timely, safely)
  • 6 – Modified Independence (device, safety concerns, increased time)
  • 5 – Supervision (supervision, cues, coaxing, set-up)
  • 4 – Minimal Assistance (patient does 75% or more)
  • 3 – Moderate Assistance (patient does 50%–74%)
  • 2 – Maximal Assistance (patient does 25%–49%)
  • 1 – Total Assistance (patient does less than 25%; assist of two or more staff)
  • 0 – Activity does not occur at all

Morse Fall Scale

1. History of falling
2. Secondary Diagnosis
3. Ambulatory Aid
 Bed rest/nurse assist0
4. IV/Heparin Lock
5. Gait/Transferring
6. Mental Status
 Oriented to own ability0
 Forgets limitations15


  • Lisa A. Salamon, MSN RN GCNS-BC WOCN, Clement J. Zablocki Medical Center, Milwaukee, WI. Address correspondence to

  • Mary Victory, MSN RN, is Staff Nurse at Aurora Healthcare, Milwaukee WI.

  • Kathy Bobay, PHD RN NEA-BC, is Associate Professor at Marquette University, Milwaukee WI.