You are on page 1of 6

PM R 7 (2015) 593-598

Original Research

National Institutes of Health Stroke Scale (NIHSS) as An Early

Predictor of Poststroke Dysphagia
Rebecca D. Jeyaseelan, MBBS, MD, Mary M. Vargo, MD, John Chae, MD


Background: Despite the availability of multiple comprehensive screening methods to detect dysphagia during acute stroke care,
consensus is lacking as to the best practice. Our previous study demonstrated favorable sensitivity of the Functional Independence
Measure (FIM) compared with a bedside 3-sip test. However, the FIM is challenging to administer during acute stroke care. The
National Institutes of Health Stroke Scale (NIHSS) is administered routinely in the emergency department.
Objective: To evaluate the utility of the NIHSS as a predictor of clinically relevant poststroke dysphagia compared with FIM data in
the same cohort.
Design: Retrospective analysis.
Setting: Academic medical center.
Patients: Individuals with acute stroke who were admitted for acute care and later transferred to acute rehabilitation within the
same institution.
Methods: Clinically relevant dysphagia was defined as aspiration on modified barium swallow or laryngeal penetration on modified
barium swallow requiring diet change, or aspiration pneumonia. NIHSS and FIM scores were compiled for all patients.
Main Outcome Measurements: Sensitivity, specificity, positive predictive value, and negative predictive value were calculated
for NIHSS and FIM. Sensitivity and specificity of different values of NIHSS and FIM were analyzed via receiver operator charac-
teristic curves.
Results: Of 290 patients admitted to acute stroke rehabilitation, 88 (30%) manifested clinically relevant dysphagia during their
rehabilitation stay. Sensitivity analyses suggested cut-off values for the NIHSS and the FIM of >9 and <55, respectively. Sensitivity,
specificity, positive predictive value, and negative predictive value for the NIHSS were 75%, 62%, 46%, and 85%, respectively. For
the FIM, these parameters were 80%, 72%, 55%, and 92%, respectively.
Conclusions: The NIHSS >9 and FIM <55 are moderately predictive of clinically relevant dysphagia. Although the NIHSS clinical
test characteristics are not as favorable as the FIM, NIHSS appears to be more sensitive than some other reported methods such
as a 3-sip water test. Further study into development of paradigms that incorporate NIHSS into initial assessment of dysphagia
risk may be appropriate.

Introduction and death. For patients who survive the acute phase,
dysphagia-related morbidities are associated with a
Stroke is among the 5 leading causes of death for protracted rehabilitation course [2], decreased likeli-
people of all races and ethnicities [1], and dysphagia- hood of home residence [4], poor nutritional status, and
mediated morbidity is one of the major underlying reduced quality of life [5]. Management of dysphagia-
causes for stroke-related mortality [2]. Poststroke dys- mediated poststroke complications comprises a large
phagia consists of impairment in swallowing function economic burden to the health care system, requiring
resulting from a lack of strength or coordination of oral significant use of health care personnel and resources
and pharyngeal muscles. Failure to detect dysphagia in [3]. Timely detection of dysphagia after stroke remains
a timely manner after stroke greatly increases morbidity a major clinical challenge in the field.
and mortality [2], which occurs mainly from aspiration The incidence of dysphagia in stroke rehabilitation
of oral contents into the respiratory tract and can lead patients is estimated to range from 29% to 81% [6-8].
to aspiration pneumonia [3], acute respiratory failure, There is clinical evidence of aspiration pneumonia in

1934-1482/$ - see front matter ª 2015 by the American Academy of Physical Medicine and Rehabilitation
594 NIHSS as a Predictor of Poststroke Dysphagia

7%-29% of patients with dysphagia, whereas only 2%-7% one phase of the examination to the next depending on
develop aspiration pneumonia in the absence of the patient’s performance.
dysphagia [8]. A major concern is that an estimated Kopey et al [15] also reported that combining the
2%-50% of patients with clinically significant dysphagia 3-sip water test with the Functional Independence
elude detection under the current screening methods Measure (FIM) score greatly increased the sensitivity and
and exhibit no overt clinical symptoms of aspiration, specificity of detecting subclinical dysphagia. The FIM
such as choking, coughing, or wet voice [9,10]. Impor- score, however, is not compiled until admission to acute
tantly, there is growing evidence that early intervention stroke rehabilitation, typically several days or more
through dysphagia screening has resulted in positive after the onset of stroke. Holding oral intake until then
outcomes with close to a 3-fold reduction of pneumonia can compromise nutritional status. The FIM score,
in patients with stroke [11]. Thus, there is a need to therefore, is most valuable as an adjunct to identify at-
develop novel screening methods to detect occurrence risk patients, along the continuum of their hospital stay,
of subclinical dysphagia in a timely fashion. rather than as an initial screening device.
In contemporary clinical practice, evaluation for Given the high likelihood of failure of detection of
dysphagia is performed following guidelines outlined by dysphagia with silent aspiration, it would be of immense
the American Heart Association/American Stroke Asso- clinical value to have a cost-effective, standardized,
ciation, which recommends screening of all ischemic timely approach to predict and screen for clinically
and hemorrhagic stroke patients before food, fluids, or relevant dysphagia. With this background, we sought an
medication are given by mouth [12]. Patients who have alternative tool that is readily available and could
undergone and passed dysphagia screening have been potentially be implemented for detection of dysphagia
found to have a lower rate of developing pneumonia at the first point of contact upon presentation to the
than unscreened patients, consistent with dysphagia emergency department (ED). One such tool is the Na-
risk being missed in some unscreened patients [13]. tional Institutes of Health Stroke Scale (NIHSS), which
Clinical documentation of “Dysphagia Screening” had routinely is documented in the ED after the initial
been required by the Joint Commission on Accreditation evaluation of a stroke patient, available within few
of Healthcare Organizations as a performance measure hours of admission. The NIHSS is used to predict major
for Primary Stroke Center certification. However, in clinical outcomes in patients after acute stroke [16].
2010, with no consensus made on the standard method Little is known, however, about the role of the NIHSS in
of assessment, “Dysphagia Screening” was discontinued predicting dysphagia after acute stroke [17].
as a performance measure [12]. We retrospectively evaluated the sensitivity, speci-
More than 35 screening tools for poststroke dysphagia ficity, positive predictive value (PPV), and negative
have been developed [14]. Most of these methods predictive value (NPV) of the NIHSS in predicting clini-
include baseline observations, such as alertness, head cally relevant dysphagia after acute stroke. In addition,
control, respiratory status, tongue strength, manage- we compared the test characteristics of the NIHSS and
ment of saliva, and gag reflex, and dynamic evaluation the admission FIM score in screening for clinically rele-
after bolus (usually water) ingestion. If dysphagia is vant dysphagia in stroke patients.
suspected, a videofluoroscopic swallow study, also
called the modified barium swallow (MBS) study, is Methods
performed as the gold standard confirmatory test.
Fiberoptic endoscopic evaluation of swallowing is Subjects
another diagnostic tool, which involves passing a flex-
ible endoscope across the nasal passage into the phar- Medical records of all consecutive stroke patients
ynx, with direct and dynamic visualization of the admitted between July 2004 and July 2010 to a Joint
pharyngeal phase of swallowing. Commissioneaccredited inpatient acute stroke rehabil-
As reported by Kopey et al [15], the 3-sip water itation unit of a tertiary academic medical center were
test, previously used at our institution to screen for reviewed retrospectively. Inclusion criteria were diag-
dysphagia, has advantages of ease of administration and nosis of an acute stroke, admission through the same
very good specificity (98%), but with sensitivity of only institution’s emergency care and acute medical service,
about 21%, it is an unreliable screening test (see the documentation of NIHSS at the first point of contact,
Methods section for details about the 3-sip testing pro- and documentation of the rehabilitation admission FIM
cess). Because the 3-sip test was found to have such score. Exclusion criteria were rehabilitation admission
poor sensitivity, however, our institution has evolved to from outside facilities, nonacute stroke diagnosis, and
using a more elaborate screening process, developed by lack of documentation of NIHSS or FIM. The institutional
speech pathology staff and administered by nursing review board at the home institution approved the study
staff, including assessment of alertness, oral-motor protocol. Data were gathered from a combination of
function, 3-sip water trial (2 by teaspoon and 1 by sip paper medical charts, minifilms, and electronic medical
from cup), and direct meal observation, advancing from records.
R.D. Jeyaseelan et al. / PM R 7 (2015) 593-598 595

For each patient, routine care process included that oropharyngeal), and severity of dysphagia ranging from
“code stroke” was activated in the ED, which included normal to severe. Identification of aspiration pneumonia
emergent consultation of a neurologist. As part of the was based on diagnostic suspicion per medical record,
initial assessment, NIHSS score was calculated, usually radiologic evidence, treatment with antibiotics, and if
by the attending neurologist and at times by the medical applicable autopsy results. Medical records of every
personnel in emergency care. Patients exhibiting ad- patient with a documented NIHSS were reviewed for a
equate alertness, trunk control, and voice quality were diagnosis of aspiration pneumonia, whether the patient
administered a 3-sip water swallow test by nursing staff, underwent MBS study or not. Clinically relevant dys-
attending physician, or resident physician. This 3-sip phagia was defined as laryngeal penetration or aspira-
water test typically was performed on day 2 of the tion on MBS leading to a modification in diet (ie, either a
acute hospitalization. In performing the 3-sip test, level modified diet or nothing by mouth recommendation), or
of alertness and the presence of dysarthria were occurrence of radiographically documented aspiration
assessed. If the patient was not alert or was significantly pneumonia for which the patient required antibiotics.
dysarthric, then a nonoral feeding route would be
established. If the patient was alert and did not dem-
onstrate dysarthria, then voluntary cough was assessed,
and the patient would be given 3 sips of water through a
The current study was an extension of the previous
straw. If the patient was observed to choke, cough,
work by Kopey et al [15] (spanning admissions from 2004
and/or develop a wet voice, a dysphagia team consult
to 2007) and includes the same patient population plus
was placed and a nonoral feeding route would estab-
additional patients (up to year 2010) that fulfill the in-
lished in the interim. If choking and coughing did not
clusion criteria. The sensitivity, specificity, PPV, and
occur, then a mechanical soft diet with nectar thick
NPV of NIHSS and FIM were calculated. Receiver oper-
liquids would be initiated. Subsequently, diet would be
ator characteristic curves (ROCs) were generated to ex-
advanced if tolerated.
amine the relationship of sensitivity and specificity of
If the patient failed the 3-sip test or if the initial diet
different values of NIHSS and FIM. Test characteristics of
was not tolerated, then a dysphagia team consult for
NIHSS and FIM were compared visually with ROC. The
MBS was placed. During the MBS, the patient underwent
ROC was used to identify threshold values of NIHSS and
clinical and videofluoroscopic evaluation by a speech
FIM in determining dysphagia risk.
pathologist and a physiatrist. A radiologist participated
in the radiographic portion of the evaluation. MBS also
could be performed later in a patient’s course, on the Results
basis of either on follow-up recommendation from
the initial MBS study or a new clinical suspicion of Of 734 cases, all necessary data points were available
dysphagia. On the patient’s admission to the acute for 290 patients. Forty-six patients were outside of the
stroke rehabilitation unit, FIM was compiled and docu- specified date range for the study, 198 came from
mented by trained rehabilitation staff, within 24-48 outside institutions, and 200 did not have an NIHSS score
hours of admission. The speech pathologist also would that could be located. Eighty-eight patients (30.3%)
perform a clinical screening examination for dysphagia were found to have clinically relevant dysphagia per
upon admission to acute rehabilitation, with referral for MBS findings and/or clinical history of aspiration pneu-
MBS generated if deemed warranted. monia. The large majority of patients identified as
having dysphagia received MBS study; only one subject
Measures was identified as having clinically relevant dysphagia
without undergoing MBS. Of patients in the group
Data collection included NIHSS administered at first without clinically relevant dysphagia (ie, those who
point of contact between the stroke patient and had an unremarkable MBS, or those in whom significant
neurologist, admission FIM (total, motor, cognition), dysphagia was not clinically suspected, and did not
age, gender, diet orders, and, when applicable, MBS receive an MBS), 30.6% received MBS study. FIM Motor
results and/or occurrence of aspiration pneumonia. and FIM Cognitive scores were significantly different
NIHSS ranges from 0 to 42, with high scores corre- between those patients with and without clinically
sponding to increased severity of stroke with worse prog- relevant dysphagia (P < .001). There were no significant
nosis. FIM score ranges from 18 to 126, with greater scores differences in age and gender (Table 1). The presence of
corresponding to increased functional independence. clinically relevant dysphagia correlated with stroke
The results of the MBS studies done during the entire severity, increasing with greater NIHSS score and
acute hospital and rehabilitation stay were reviewed, with lower FIM score (P < .001 for both FIM and NIHSS)
noting the presence of aspiration or laryngeal penetra- (Figure 1). Figure 2 shows the actual distribution of
tion, recommendations for alteration of dietary consis- subjects by NIHSS and FIM scores, categorized by pres-
tency, type of dysphagia (whether oral, pharyngeal or ence or absence of clinically relevant dysphagia.
596 NIHSS as a Predictor of Poststroke Dysphagia

Table 1 advantage of being available earlier in the clinical

Patient characteristics course and is more quickly obtained, because the FIM
No Dysphagia Dysphagia requires interdisciplinary evaluation and input.
n 202 88 P Value To be an optimal screening tool, sensitivity should be
Age, y 63.2 (13.4) 67.1 (13.4) .024 high while avoiding false-positive results. With NIHSS 9
Gender, % female 47.5 57.9 NS or greater, sensitivity is 75% and specificity 62%. If NIHSS
MBS performed, % 30.6 98.9 <.001 of 8 or greater is used, sensitivity improves slightly to
FIM-Motor 40.1 (11.3) 26.8 (11.9) <.001 78%, and specificity decreases to 54%. If the bar is raised
FIM-Cognition 22.7 (7.9) 15.0 (7.4) <.001 and threshold set at NIHSS 10 or greater, sensitivity
NIHSS 8.4 (5.5) 12.3 (5.9) <.001
decreases to 67% and specificity improves to 71%.
NS ¼ not significant; MBS ¼ modified barium swallow; FIM ¼ Functional The reader is referred to a recent systematic review
Independence Measure; NIHSS ¼ National Institutes of Health Stroke
that details test characteristics of various clinical dys-
phagia screening tools [14]. Although there are clinical
As the ROC (Figure 3) shows, sensitivity and speci- dysphagia screening protocols reported to have more
ficity of NIHSS of >9 were 75% and 62%, respectively. favorable sensitivity, some greater than 90%, these
PPV and NPV based on NIHSS >9 were 46% and 85%, involve specialized examination, examiner training, and
respectively. The sensitivity and specificity of FIM <55 typically are performed at a later time than NIHSS. In
were 80% and 72%, respectively. PPV and NPV based on addition to test characteristics of sensitivity, specificity,
FIM <55 were 55% and 92%, respectively. PPV, and NPV, other characteristics of an optimal
screening tool include administration at the right time,
ease and speed of administration, and possible contex-
Discussion tual factors specific to the organization [12]. Because
the NIHSS is performed anyway, independent of dys-
Clinically significant dysphagia was seen in approxi- phagia considerations, it serves to minimize the extra
mately 30% of our subjects, which is within the range logistical burden involved with dysphagia screening.
reported by numerous other investigators [2,4,6,8]. Our A major limitation of this study is that, given the
data indicate that both NIHSS and FIM correlate with retrospective design, not all subjects received a modi-
presence of clinically significant dysphagia. This finding fied barium swallow study, and there was reliance on
is not surprising, because both tools measure overall clinical suspicion of dysphagia. Also, NIHSS score was
severity of neurologic impairment and functional limi- not available on all charts; for nearly half of the original
tations, respectively, and dysphagia is presumably more 444 patients who were excluded, the reason was that
likely to be clinically significant in more severe strokes, the score could not be found in available records. For
generally speaking. Test characteristics as displayed the period under review, our facility was in transition to
on the ROC are more favorable for FIM than for NIHSS, becoming a Primary Stroke Center and also was tran-
with greater area under the curve seen with FIM; how- sitioning from paper records to electronic data, both
ever, the difference is not pronounced. NIHSS has the factors possibly contributing to the inconsistency in

Figure 1. Dysphagia as a function of FIM and NIHSS. f, as a function of; FIM, Functional Independence Measure; NIHSS, National Institutes of Health
Stroke Scale.
R.D. Jeyaseelan et al. / PM R 7 (2015) 593-598 597

Figure 2. Histograms showing the actual distribution of subjects by NIHSS and FIM scores, categorized by presence or absence of clinically relevant
dysphagia. FIM, Functional Independence Measure; NIHSS, National Institutes of Health Stroke Scale.

ability to locate the NIHSS data in the medical record for An inherent limitation of using NIHSS may be that it is
a significant number of cases. To address the question of collected so early that the patient’s neurologic status
whether our findings were skewed by the high number of may be very dynamic, with stroke either extending, or
excluded patients, we compared FIM data, which was neurologic findings dramatically improving, in the near
available for all subjects, between the included and term after the NIHSS is compiled, which would pre-
excluded groups. The median FIM total score among our sumably affect the underlying dysphagia risk, and this
included subjects was 59, and among our excluded factor may account for the relatively more favorable
subjects was 57 (nonsignificant difference, P ¼ .25). characteristics of the FIM. In addition, the study ex-
amines a population of patients who underwent acute
rehabilitation, and the test characteristics may be
different for other stroke subpopulations, such as
those with milder stroke. From that standpoint, the
present study may overestimate the prevalence of
clinically significant dysphagia, resulting in possible
decrease in PPV and increase in NPV when NIHSS
screening for dysphagia is applied to a more mildly
affected group.
Although the NIHSS has predictive characteristics for
dysphagia, it is not sensitive enough for use in isolation
for initial dysphagia management. Rather, it serves as a
potential platform for further study of streamlined
decision-making in identifying individuals at risk for
poststroke dysphagia.

Figure 3. Receiver operator characteristic curve for sensitivity and
specificity of NIHSS and FIM for clinically relevant dysphagia. FIM, Func-
tional Independence Measure; NIHSS, National Institutes of Health Stroke NIHSS >9 by itself, as well as low FIM total score
Scale; NPV, negative predictive value; PPV, positive predictive value. (<55) by itself are moderately predictive of the
598 NIHSS as a Predictor of Poststroke Dysphagia

presence of clinically relevant dysphagia in the sub- 6. Barer DH. The natural history and functional consequences of
population of acute stroke patients who require acute dysphagia after hemispheric stroke. J Neurol Neurosurg Psychiatry
rehabilitation. The FIM has slightly better sensitivity 7. Meng NH, Wang TG, Lien IN. Dysphagia in patients with brainstem
than NIHSS, but NIHSS has the advantage of being stroke: Incidence and outcome. Am J Phys Med Rehabil 2000;79:
available earlier in the clinical course and to a broader 170-175.
range of patients than FIM. Clinical suspicion remains 8. Martino R, Foley N, Bhogal S, Diamant N, Speechley M, Teasell R.
warranted, because clinically relevant dysphagia can Dysphagia after stroke: Incidence, diagnosis, and pulmonary
complications. Stroke 2005;36:2756-2763.
be seen in association with milder strokes. Additional 9. Terre R, Mearin F. Oropharyngeal dysphagia after the acute phase
research on NIHSS efficacy in predicting dysphagia of stroke: Predictors of aspiration. Neurogastroenterol Motil 2006;
is needed, including prospective study, examination of 18:200-205.
utility in the broader stroke population, and testing 10. Ramsey D, Smithard D, Kalra L. Silent aspiration: What do we
of NIHSS combined with other dysphagia screening know? Dysphagia 2005;20:218-225.
11. Hinchey JA, Shephard T, Furie K, Smith D, Wang D, Tonn S. Formal
element(s) to determine whether test characteristics dysphagia screening protocols prevent pneumonia. Stroke 2005;
can be further improved. 36:1972-1976.
12. Donovan NJ, Daniels SK, Edmiaston J, Weinhardt J, Summers D,
Mitchell PH. Dysphagia screening: state of the art: Invitational con-
ference proceeding from the state-of-the-art nursing symposium,
References International Stroke Conference 2012. Stroke 2013;44:e24-e31.
13. Lakshminarayan K, Tsai AW, Tong X, et al. Utility of dysphagia
1. Towfighi A, Saver JL. Stroke declines from third to fourth leading screening results in predicting poststroke pneumonia. Stroke 2010;
cause of death in the United States: Historical perspective and 41:2849-2854.
challenges ahead. Stroke 2011;42:2351-2355. 14. Schepp SK, Tirschwell DL, Miller RM, Longstreth WT. Swallowing
2. Smithard DG, O’Neill PA, Parks C, Morris J. Complications and screens after acute stroke: A systematic review. Stroke 2012;43:
outcome after acute stroke. Does dysphagia matter? Stroke 1996; 869-871.
27:1200-1204. 15. Kopey SA, Chae J, Vargo MM. Does a 3-sip test detect dysphagia in
3. Martino R, Pron G, Diamant N. Screening for oropharyngeal acute stroke rehabilitation patients? PM R 2010;2:822-828.
dysphagia in stroke: Insufficient evidence for guidelines. Dysphagia 16. Adams HP, Davis PH, Leira EC, et al. Baseline NIH Stroke Scale
2000;15:19-30. score strongly predicts outcome after stroke: A report of the Trial
4. Odderson IR, Keaton JC, McKenna BS. Swallow management in of Org 10172 in Acute Stroke Treatment (TOAST). Neurology 1999;
patients on an acute stroke pathway: Quality is cost effective. 53:126-131.
Arch Phys Med Rehabil 1995;76:1130-1133. 17. Bravata DM, Daggett VS, Woodward-Hagg H, et al. Comparison of two
5. Kwok T, Lo RS, Wong E, Wai-Kwong T, Mok V, Kai-Sing W. Quality of approaches to screen for dysphagia among acute ischemic stroke
life of stroke survivors: A 1-year follow-up study. Arch Phys Med patients: Nursing admission screening tool versus National Institutes
Rehabil 2006;87:1177-1182; quiz 1287. of Health stroke scale. J Rehabil Res Dev 2009;46:1127-1134.


R.D.J. Department of Physical Medicine and Rehabilitation, Case Western J.C. Department of Physical Medicine and Rehabilitation, Case Western Reserve
Reserve University, MetroHealth Medical Center, Cleveland, OH University, MetroHealth Medical Center, Cleveland, OH
Disclosure: nothing to disclose Disclosure: nothing to disclose

M.M.V. Department of Physical Medicine and Rehabilitation, Case Western This work was presented at the 2013 Association of Academic Physiatrists Annual
Reserve University, MetroHealth Medical Center, 2500 MetroHealth Drive, Assembly as an oral poster presentation.
Cleveland, OH 44109. Address correspondence to: M.M.V.; e-mail: mvargo@
Submitted for publication May 4, 2014; accepted December 17, 2014.
Disclosure: nothing to disclose