Evaluation of the Morse Fall Scale in
Sir—Several risk factors associated with falls in hospitalised
patients have been identified [1, 2]. Although a substantial
number of assessment instruments for identifying hospital-
ised patients at risk of falling exists , their generalisability
is limited  because only a few [2, 5] have been tested in
settings other than those in which they were originally
developed. The Morse Fall Scale (MFS) has been evaluated
in different hospital settings [6–9] and has been used in a
variety of patient populations [10–16]. In search of an
appropriate tool to identify admitting patients for risk of
falling, the MFS appears to be most elaborate in view of its
extensive development and testing in different hospital
populations compared with others [3, 4]. Its easy application
in clinical practice supported this choice.
However, no investigation to date has reported results
of different cut-off scores of the scale. This study aimed to
evaluate the diagnostic value of different MFS cut-offs to
determine which score would be most useful in identifying
in-hospital patients at risk of falls.
This prospective cohort study utilised baseline data col-
lected during a four-month falls intervention study per-
formed at two units in the department of internal medicine
of a 300-bed urban public hospital in Switzerland. The data
were collected on consecutively admitted adult patients
(18 years and older, >48 hours in hospital) who presented
with a wide range of medical conditions.
Because the study hospital is situated in the German-
speaking part of Switzerland, the MFS was translated into
German (MFS-G) and piloted with six registered nurses to
determine their understanding of wording of items. Inter-
rater reliability was examined and the level of agreement was
84% (K = 0.68). The scale consists of six items reflecting
risk factors of falling such as: (i) history of falling, (ii) sec-
ondary diagnosis, (iii) ambulatory aids, (iv) intravenous
therapy, (v) type of gait and (vi) mental status. The total
score ranges between 0 and 125 [17, 18]. For further details
of the scale, please see Appendix 1 in the supplementary
data on the journal website (http://www.ageing.oxford-
All registered nurses on the designated study units
received a 30-minute group instruction on the use of the
MFS-G as part of the implementation of the in-hospital fall
risk screening programme. The primary nurses completed
the MFS-G for each newly hospitalised patient within
24 hours of admission. Patient falls during hospitalisation
were registered with a standardised fall incident report form
that had been implemented earlier in this hospital .
A fall was defined as ‘an incident in which a patient sud-
denly and involuntarily came to rest upon the ground or sur-
face’ . Patient demographics and clinical characteristics
(i.e. gender, age, length of stay and medical diagnosis) were
extracted from the hospital administrative patient data base.
The study was approved by the local ethics committee.
Descriptive statistics such as frequencies, per cent as well as
mean and standard deviations were calculated for demo-
graphic and clinical characteristics of the patients.
The diagnostic value of the MFS-G scores ranging from 20
to 70 was explored using receiver operating characteristic
(ROC) curves, with an area under the curve (AUC) analysis
based on admission MFS-G scores, and using patients who fell
while hospitalised as the ‘gold standard’. Sensitivity analysis—
including specificity, positive predictive value (PPV) and nega-
tive predictive value (NPV) and accuracy—was performed for
the different cut-off scores of the MFS-G. Chi square statistics
were calculated for the estimation of risk of falling with odd
ratios and 95% confidence intervals. All data were analysed
with SPSS for Windows, version 12.0 (SPSS Inc., Chicago, IL).
A total of 386 patients (female: 59.6%) with a mean age of
70.3 (SD: 18.5) years, and a mean length of stay of 11.3 (SD:
8.9) days were included in the study. Forty-seven (12.2%)
patients experienced a total of 69 falls. For patient demo-
graphics, clinical characteristics including primary medical
diagnosis and risk factors for falls (MFS-G items), please see
Appendix 2 in the supplementary data on the journal web-
site (http://www.ageing.oxfordjournals. org/).
The percentage of the patients identified as at risk of fall-
ing at admission varied with the MFS-G cut-off scores used
and ranged from 89.4% (cut-off score: 20 points) to 20.7%
(cut-off score: 70 points). According to the different cut-off
scores, the sensitivity ranged from 91.5 to 38.3%, the specif-
icity from 81.7 to 10.9%, the PPVs from 12.5 to 22.5% and
the NPVs from 90.2 to 95.7% (Table 1). High false positive
rates (i.e. patients who were classified as at risk of falling but
did not fall) ranging from 87.5% (cut-off score: 20 points)
to 75.9% (cut-off score: 60 points) were observed.
The area under the ROC curve ranged from 0.512 to
0.701, and the accuracy of the MFS-G ranged from 20.7 to
76.4% (Table 1). The most optimal cut-off point for the
MFS-G was found to be 55, which showed a fairly good
sensitivity of 74.5% (95% CI = 60.5–84.7%), an acceptable
specificity of 65.8% (95% CI = 60.1–70.6%) and a high
NPV (94.9%), with an acceptable accuracy of 66.8%. The
ROC curve with an arrow indicating the highest peak with
the cut-off of 55 points for the MFS-G is demonstrated in
Figure 1. With a cut-off score of 55 points, 23.2% of the
patients were screened positive and presented a relative
odds ratio of 5.6 (95% CI = 2.8–11.2) for falling.
This study constitutes a prospective test of the sensitivity,
specificity and predictive value of the MFS-G in hospitalised
patients. The 12.2% proportion of patients who fell in the
present study lies between rates reported in previous studies
of 15.7, 29.6 [6, 9], 5 and 4% [7, 21]. The variation in fall rates
may reflect the different types of settings, sample sizes,
patient characteristics and reporting practices. The MFS-G
demonstrated moderate ability to predict patients’ risk of fall-
ing using a cut-off score of 55 points as evidenced by an AUC
of 0.701 in a sample of internal medicine patients.
at University of Basel/ A284 UPK on November 11, 2015
Using the originally identified cut-off score of 45 points,
only 26% patients in another study  were identified as
being at risk of falling, whereas the same cut-off score iden-
tified 51% patients as being at risk of falling in the present
study. This difference may be explained by the heterogene-
ity of the other sample, with patients enrolled from acute,
rehabilitation and long-term care units, whereas the present
study may reflect a more homogenous sample in relation
to fall risk factors despite a variety of medical diagnoses.
Additionally, in the original prospective study , the fall
risk status of the patients was assessed at different points of
time during their hospital stay, whereas in the present study
all patients were screened for risk of falling at admission.
This and the prospective follow-up during the patient’s hos-
pital stay allowed calculation of the diagnostic value of the
MFS-G in relation to its predictive power.
Only one other study  scored patients at admission and
performed ROC analysis. In that study, an MFS cut-off
score of 45 points identified 75% of the patients as at risk of
falling with a false positive rate of 82%. The same cut-off in
the present study resulted in a false positive rate of 81%, but
decreased slightly to 77% with a cut-off of 55 points. O’Con-
nell and Myers  concluded, based on an AUC of 0.621,
that the MFS had low ability to discriminate patients who fell
and those who did not fall. However, the high positive rate
may reflect a limitation in the present study because the
effects of fall interventions subsequently implemented with
some of the patients identified as being at risk of falling were
not considered. Furthermore, the performance of falls incid-
ent reporting may be inflated by virtue of the study being
conducted (Hawthorne effect). Finally, changes in the
patient’s health condition which may have altered risk fac-
tors for falls were not considered. Although the high NPVs
(e.g. 95% of the non-falling patients were not at risk of fall-
ing) may give appropriate reassurance for patients with low
risk of falling, the scale seems to be of limited operational
value because the PPV is only between 12 and 24%. We
therefore recommend that the MFS undergo local validation
to determine the best cut-off score for a given setting before
it is used clinically. Screening patients for risk of falling
should lead to more targeted assessment and modification of
risk factors using multifactorial interventions [22, 23]. How-
ever, because the effectiveness of hospital fall prevention
programmes that incorporate fall risk assessment leads to a
25% or less reduction in fall rates , the idea of looking at
reversible risk factors in all patients and reassessing their risk
following a fall may be an appropriate approach .
• The MFS should be used to screen hospitalised patients
at risk of falling only after local validation to determine
best cut-off scores in a given setting.
The authors thank Prof. Kathy Dracup and Prof. Sandie
Engberg for editing the manuscript.
Conflicts of interest
The authors have no conflicts of interest to declare.
R. SCHWENDIMANN1,2*, S. DE GEEST1,3 K. MILISEN3
1Institute of Nursing Science, University of Basel,
Table 1. Predictive validity of MFS-G cut-off scores at admission (n = 386)
aPositive predictive value
bNegative predictive value
cArea under the ROC curve
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Sensitivity91.5% 91.5%91.5% 80.9%
Specificity10.9% 13.9%16.5% 53.4%
20253545 50556065 70
Figure 1. ROC curve with AUC of the MFS-G (n = 386). The
arrow indicates the highest peak with the cut-off of 55 points.
0.00.10.20.30.40.5 0.6 0.7 0.8 0.9 1.0
1 - Specificity
at University of Basel/ A284 UPK on November 11, 2015
Research letters Download full-text
2Stadtspital Waid, Zurich, Switzerland
3Center for Health Services and Nursing Research, Catholic
University of Leuven, Leuven, Belgium
*To whom correspondence should be addressed:
1. Evans D, Hodgkinson B, Lambert L, Wood J. Falls risk fac-
tors in the hospital setting: a systematic review. Int J Nurs Prac
2001; 7 (1): 38–45.
2. Oliver D, Daly F, Martin FC, McMurdo ME. Risk factors and
risk assessment tools for falls in hospital in-patients: a system-
atic review. Age Ageing 2004; 33 (2): 122–30.
3. Perell KL, Nelson A, Goldman RL, Luther SL, Prieto-Lewis N,
Rubenstein LZ. Fall risk assessment measures: an analytic
review. J Gerontol 2001; 56 (12): M761–6.
4. Myers H. Hospital fall risk assessment tools: a critique of the
literature. Int J Nurs Prac 2003; 9 (4): 223–35.
5. Morse JM. Preventing Patient Falls, 1st edition. Thousand
Oaks, California: SAGE Publications, Inc., 1997.
6. Eagle DJ, Salama S, Whitman D, Evans LA, Ho E, Olde J.
Comparison of three instruments in predicting accidental falls
in selected inpatients in a general teaching hospital. J Gerontol
Nurs 1999; 25 (7): 40–5.
7. McCollam ME. Evaluation and implementation of a research-
based falls assessment innovation. Nurs Clin North Am 1995;
30 (3): 507–14.
8. McFarlane-Kolb H. Falls risk assessment, multitargeted inter-
ventions and the impact on hospital falls. Int J Nurs Prac
2004; 10 (5): 199–206.
9. O’Connell B, Myers H. The sensitivity and specificity of the
Morse Fall Scale in an acute care setting. J Clin Nurs 2002; 11
10. Barnett K. Reducing patient falls in an acute general hos-
pital. Foundation of Nursing Studies Dissemination Series
2002; 1 (1).
11. Camicioli R, Licis L. Motor impairment predicts falls in spe-
cialized Alzheimer care units. Alzheimer Dis Assoc Disord
2004; 18 (4): 214–18.
12. Cheng GYC, S. Fall prevention. In: 3rd Australasian Joanna
Briggs Institute Colloquium for Evidence Based Nursing and
Midwifery: 2002; Auckland, New Zealand: Contemporary
Nurse, 2002: 61.
13. Lai KCKYS, Wong KST. Validation of the Cantonese version
of the Morse Fall Scale. In: 11 Annual Congress of the Hong
Kong Association of Gerontology: 2003, Hong Kong, 2003.
14. Ledsham RBJ, Beardsall A. Implementing a fall risk assess-
ment strategy for older people: issues and outcomes. Clin
Govern Bull 2002; 3 (3): 2–4.
15. McCarter-Bayer A, Bayer F, Hall K. Preventing falls in acute care:
an innovative approach. J Gerontol Nurs 2005; 31 (3): 25–33.
16. Weber H. Pflegeexpertinnen helfen Stürze verhindern. In:
Kantonsspital Luzern Newsletter, 2003: 3.
17. Morse JM. Computerized evaluation of a scale to identify the
fall-prone patient. Can J Public Health 1986; 77 (Suppl. 1): 21–5.
18. Morse JM, Prowse MD, Morrow N, Federspeil G. A retro-
spective analysis of patient falls. Can J Public Health 1985; 76
19. Schwendimann R. [Frequency and circumstances of falls in
acute care hospitals: a pilot study]. Pflege 1998; 11 (6): 335–41.
20. Gibson MA, RO, Isaacs B, Radebaugh T, Worm-Petersen J.
The prevention of falls in later life. A report of the Kellogg
International Work Group on the Prevention of Falls by the
Elderly. Dan Med Bull 1987; 34 (Suppl. 4): 1–24.
21. Morse JM, Black C, Oberle K, Donahue P. A prospective
study to identify the fall-prone patient. Soc Sci Med 1989; 28
22. Haines TP, Bennell KL, Osborne RH, Hill KD. Effective-
ness of targeted falls prevention programme in subacute
hospital setting: randomised controlled trial. Br Med J
2004; 328 (7441): 676.
23. Healey F, Monro A, Cockram A, Adams V, Heseltine D.
Using targeted risk factor reduction to prevent falls in older in-
patients: a randomised controlled trial. Age Ageing 2004;
33 (4): 390–5.
24. Oliver D, Hopper A, Seed P. Do hospital fall prevention pro-
grams work? A systematic review. J Am Geriatr Soc 2000;
48 (12): 1679–89.
Published electronically 9 March 2006
NHS continuing care: reliable decisions?
SIR—Decisions denying free National Health Service
(NHS) continuing health care have been reversed by the
Health Service Ombudsman from 1994 . From 1996,
health authorities published their individual criteria govern-
ing eligibility for continuing NHS health care . Concern
that criteria were too restrictive led to further guidance .
The ‘Coughlan case’  and others reviewed by the
Ombudsman  revealed people who had been wrongly
denied NHS care. Restitution followed, reimbursing indi-
viduals or their estates if care had been wrongly denied.
Currently, Primary Care Trusts (PCTs) convene panels
of senior staff (doctor, nurse, therapist, social worker) to
determine eligibility using their Strategic Health Authority
(StHA) criteria. Rejected applicants can appeal and another
panel may reverse the decision.
The appeals system, however, does not indicate the level
of inconsistency amongst panels. To our knowledge, this
point had not been explored previously by presenting the
same case to panels applying identical criteria. Here we
describe a small audit undertaken during 2004 with the sup-
port of the Continuing Care Steering Group of the Norfolk,
Suffolk and Cambridgeshire StHA.
The standard was that StHA panels should reach consistent
decisions when determining eligibility for continuing NHS
The authors and the StHA designated officer for continuing
care examined 110 completed restitution cases and selected
10 to reflect the range of conditions frequently giving rise to
applications. The conditions were chronic and usually pro-
gressive. The individuals were likely to need frequent atten-
tion because of unpredictable health need, or management
of challenging behaviour, characteristics often demanded by
the eligibility criteria. The aetiologies were dementia, stroke
at University of Basel/ A284 UPK on November 11, 2015