Viscosity and Non-Newtonian Features of Thickened Fluids Used for Dysphagia Therapy

E:FoodEngineering

&PhysicalProperties

Viscosity and Non-Newtonian Featuresof Thickened Fluids Used for Dysphagia TherapyMark O’Leary, Ben Hanson, and Christina Smith

Abstract: Thickening agents based primarily on granulated maize starch are widely used in the care of patients withswallowing difficulties, increasing viscosity of consumed fluids. This slows bolus flow during swallowing, allowing airwayprotection to be more properly engaged. Thickened fluids have been shown to exhibit time-varying behavior and are non-Newtonian, complicating assessment of fluid thickness, potentially compromising efficacy of therapy. This work aimed toquantify the flow properties of fluids produced with commercial thickeners at shear rates representative of slow tipping ina beaker to fast swallowing. Results were presented as indices calculated using a power-law model representing apparentviscosity (consistency index) and non-Newtonian nature of flow (flow behavior index). Immediately following mixing,3 fluid thicknesses showed distinct consistency indices and decreasing flow behavior index with increasing thickenerconcentration. An increase in consistency index over 30 min was observed, but only for samples that were repeatedlysheared during acquisition. Three-hour measurements showed changes in consistency index across fluids with the largestbeing a 25% rise from initial value. This may have implications for efficacy of treatment, as fluids are not always consumedimmediately upon mixing. Flow behavior indices were comparable across thickeners exhibiting similar rises over time.The indices were a more complete method of quantifying flow properties compared with single viscosity measurements,allowing an increased depth of analysis. The non-Newtonian nature of fluids perhaps renders them particularly suitable foruse as dysphagia therapies, and such analysis may allow the possibility of altering these properties to optimize therapeuticefficacy to be explored.

Keywords: dysphagia, rheology, swallowing, thickener, viscosity

Practical Application: Effective treatment of swallowing disorders relies upon the appropriate choice and subsequentreproduction of drinks thickened to one of a number of predetermined levels. Currently there are no agreed methods ofmeasuring the thickness of these drinks in use and the specifications are subjective, relying on descriptions such as “syrup”thick. This research aims to further understanding of the flow properties of thickened drinks and bring a quantifiedmeasure of thickness closer to being a practical reality.

IntroductionDysphagia is a common health problem describing a range of

mechanical disorders that affect the safety, efficiency, or quality ofeating and drinking (Whelan 2001). Affecting approximately 20%of the adult population in primary care (Wilkins and others 2007),dysphagia is even more widely diagnosed in cerebrovascular pa-tients (Mann and others 2000) and is a frequent consequence ofhead and neck cancer (Garcıa-Peris and others 2007). The con-dition occurs chronically as a result of various neurodegenerativediseases and has been found to affect greater than half of elderlyindividuals living in care and nursing homes (Layne and others1989). Dysphagia that is not adequately managed can cause dehy-dration, malnutrition, and inhalation of food and drink into the

MS 20091039 Submitted 10/19/2009, Accepted 3/2/2010. Authors O’Learyand Hanson are with Dept. of Mechanical Engineering, University College London,Torrington Place, London, WC1E 7JE, U.K. Author Smith is with Div. of Psychol-ogy and Language Science, University College London, Chandler House, 2 Wake-field St., London, WC1N 1PF, U.K. Direct inquiries to author Hanson (E-mail:[email protected]).

lungs (aspiration), which can lead to aspiration pneumonia andother serious respiratory problems (Kayser-Jones and others 1999;Goulding and Bakheit 2000; Finestone and others 2001; Whelan2001; Langmore and others 2002). Normal swallowing occurs as aresult of several distinct physiological phases; the safe transport ofa food or liquid bolus depending on the correct timing of variousmuscular movements as the bolus passes the tongue, pharynx, lar-ynx, and esophagus in sequence. With dysphagia, muscular powerand control is often compromised, for example, there is a dan-ger that watery liquids can flow quickly to the larynx before theentrance to the airway has closed, leading to aspiration. Thus,the safety of an abnormal swallow can depend on the mechanicalproperties of the bolus (Dantas and others 1990; Miller and Watkin1996; Cichero and others 2000) and the use of thickening agentsto decrease the flow speed of fluids has become a widespread andsignificant part of the clinicians’ treatment toolkit (Leonard andKendall 1997).

Thickeners are commercially available as powders that can beadded to any drink; the resulting viscosity of which is a functionof the initial properties of the drink and the brand and quantity ofpowder added (Sopade and others 2007, 2008a, 2008b), tendingto continually thicken following the initial mixing (Mertz Garcia

C© 2010 Institute of Food Technologists R©E330 Journal of Food Science � Vol. 75, Nr. 6, 2010 doi: 10.1111/j.1750-3841.2010.01673.x

Further reproduction without permission is prohibited

E:Fo

odEn

gine

erin

g&

Phys

ical

Prop

ertie

s

Thickened fluids used for dysphagia therapy . . .

and others 2005; Dewar and Joyce 2006a). The magnitude of thischange may have a significant effect on the clinical efficacy offluid thickening treatment. In the following sections of this study,we present the results of a scientific evaluation of the behavior ofthickened drinks, utilizing several experimental protocols, with adiscussion of the clinical relevance and implications. Thickeningpowders are primarily based on granulated maize starch. Thesegranules comprise a microstructure of insoluble semicrystallineregions of amylopectin bound to amorphous regions of amylosemolecules. When mixed with water, the amorphous amylose un-dergoes water imbibition (Dewar and Joyce 2006a); thus, the resultis a composite comprising a continuous phase of amylose gel in-terspersed with swollen granules of amylopectin (Dewar and Joyce2006b). The starch granule production process can be controlledto produce variations of the microstructure, and it is reasoned thatthis would have a significant influence on the fluid absorption andrheology of the subsequent fluid. In addition to starch, manu-facturers are increasingly including quantities of maltodextrin andvarious gums (see Table 1) to their dysphagia therapy products. Inorder to evaluate the time-dependent effects on fluid rheology, weperformed side-by-side comparisons of some commonly availablethickeners with varying constituents and present the results herein.

Previous studies have utilized a variety of methodologies toestablish the variation over time of thickened fluid rheology. Thesehave included protocols that employ a repeatedly tested singlefluid sample (Dewar and Joyce 2006a) and others that use multiplesamples over time, testing each sample only once (Mertz Garciaand others 2005). It is reasonable to assume that repeatedly shearinga sample may either alter the levels of dissolution of the thickener orhave an effect on some other structural or chemical property of thefluid. In clinical use, the fluids may not be consumed immediatelyfollowing mixing and may also be subject to repeated stirring. Anaim of this study was to establish whether viscosity data resultingfrom subsequent testing had a dependence on the previous shearhistory of the sample.

Starch-based thickeners have been shown to produce stronglynon-Newtonian fluids with shear-thinning properties (Germainand others 2006; Sopade and others 2007), the apparent viscosityof which may change over time (Dewar and Joyce 2006a). Thelevel of shear-thinning may also alter over time, an effect that hasnot been comprehensively investigated to date. Mechanically, theshear-thinning behavior may be highly beneficial for the thera-peutic efficacy of the fluids. At low shear rates, such as, when thefluids are sipped from a cup or during the oral preparatory stageupon the tongue, the high-apparent viscosity results in a greatlyreduced rate of flow under gravity, providing more time for mus-cular control of the swallow to occur. Yet, the additional muscularforces required in order to achieve higher shear rates for swallowand esophageal transport are disproportionately lower. However,strong shear-thinning behavior can complicate the issues of theclinical significance of rheology for dysphagia therapy. The shear-thinning behavior of starch-thickened drinks means the subjectiveappearance of fluid motion in a beaker or on a spoon is not nec-essarily indicative of its behavior during swallowing. Assessmentof the viscosity of thickened drinks administered to patients isalmost universally subjective, and unfortunately, the trained med-ical practitioners responsible for prescribing the patient’s diet haveshown poor repeatability in estimation of the viscosity of thera-peutic fluids (Steele and others 2003); the situation amongst carersand institutional caterers is expected to be worse. In the UnitedKingdom, the recently reissued “National Descriptors for TextureModification in Adults” (The British Dietetic Association 2009)

provide guidance on levels of texture modification for both flu-ids and foods using subjective descriptions. Recently, numericalquantification of viscosity ranges has appeared on some thickeningagent labeling (Nutricia Ltd., Trowbridge, Wiltshire, U.K.), how-ever, widespread quantitative measures of viscosity are acknowl-edged to be presently impractical (The British Dietetic Association2009). The behavior exhibited by starch-thickened fluids cannotbe described by a single measure of viscosity without specifyingthe shear rate at which measurements were taken. Previous studieshave suggested a range of shear rates that may be appropriate forstudying the viscosity of dysphagia therapy products (Cichero andothers 2000; Sopade and others 2007) but the actual shear ratesexperienced by the bolus during each stage of swallowing remainpoorly defined. A further aim of this current study was to evaluatethe relationship between the quoted viscosity ranges given belowand the shear-thinning behavior of a range of commercially avail-able thickeners to identify whether a single viscosity value could bean appropriate practical measure for these non-Newtonian fluids.

Materials and Methods

Sample preparationSix commercially available thickeners were analyzed; a summary

of the various thickener properties is included below, Table 1. Notethat the reformulated “Thick & Easy” uses the same constituentingredients, although, the granular structure may be different.

The thickeners’ packaging contained instructions for achieving3 different thickness levels in all cases except for Vitaquick that pro-vided instructions for 4 levels. The UK National Descriptors forTextured Modification in Adults (The British Dietetic Association2009) refers to 3 consistency stages: one brand (new formulation“Thick & Easy”) included direct references to these specificationson the labeling. A previous study has shown that Vitaquick “ThinPuree” is the closest consistency to the National Descriptors stage2 (O’Leary and others 2010) of the 4 levels of thickness describedon the product packaging, whilst “Thick Puree” is considerablymore viscous.

The viscosity of the resultant fluids achieved using thickenershas been shown to be strongly dependent on the properties of themedium in which the thickener was dissolved (pH, temperature,fat content, and density) (Sopade and others 2007, 2008a, 2008b).For this study, only water was considered as the mixing mediumfor reasons of practicality and to maintain clarity of research out-comes. Water, filtered to ISO 3696 Grade III standard, was leftstanding to equilibrate with room temperature and then 200 mLwas dispensed into a flask prior to the weighing and addition ofthe required amount of thickening agent. For comparison betweendifferent brands of thickener, we selected the “medium” level ofthickness in each case (referred to as “custard” on product labeling,stage 2 in the National Descriptors), except for Vitaquick where“Thin Puree” was assessed. The required amount of thickeningagent was added to the water and stirred for 30 s until a smoothconsistency was achieved. The fluids were then left to stand priorto rheometer sample loading for a period of time either as spec-ified by the manufacturer on the packaging or for 3 min wherea nonspecified “thickening time” was required. In addition, eachof the 3 thickness levels of current formulation Nutilis were pro-duced and tested in order to assess the correspondence between themanufacturer’s stated viscosity values and the measured rheologicalproperties.

A Bohlin CVO rotational rheometer system (Malvern Instru-ments Ltd., Malvern, Worcestershire, U.K.) was used to perform

Vol. 75, Nr. 6, 2010 � Journal of Food Science E331

E:FoodEngineering

&PhysicalProperties


Tab

le1–

Sel

ecte

dch

arac

teri

stic

sof

the

thic

ken

ing

agen

ts.

Nutilisa

Thic

k&

Eas

yb

Form

ula

tion

Pre

vious

New

Pre

vious

New

Res

ourc

eth

icke

nup

cV

itaq

uic

kd

Con

stitu

ent

ingr

edie

nts

Mod

ified

star

ch(E

1442

)M

odifi

edm

aize

star

ch(E

1442

),m

alto

dext

rin,

tara

gum

,xa

ntha

ngu

m,g

uar

gum

Mod

ified

mai

zest

arch

(E14

42),

mal

tode

xtri

nM

odifi

edm

aize

star

ch(E

1442

),m

alto

dext

rin

Mod

ified

mai

zest

arch

Mod

ified

mai

zest

arch

Man

ufac

ture

rsp

ecifi

edth

icke

ning

time

‘few

min

’‘fe

wm

in’

‘60

s’‘6

0s’

‘stir

until

desir

edco

nsist

ency

achi

eved

’‘a

ppro

xim

atel

y1

min

’

Des

crip

tor

“Syr

up”

“Syr

up”

“Syr

up”

“Syr

up”

“Syr

up”

“Thi

ckFl

uid”

Quo

ted

visc

osity

rang

e0.

45+/

-0.

2Pa

.sQ

uant

ityre

quir

edpe

r10

0m

Lliq

uid

(equ

ival

ent

wei

ght

ingr

ams)

1–

11 /2

scoo

ps(3

.0–

4.5)

App

roxi

mat

ely

1sc

oop

(3.0

)1

scoo

p(4

.5)

1sc

oop

(4.5

)1

Tbs

p(4

.5)

0.67

scoo

ps(3

.3)

Des

crip

tor

“Cus

tard

”“C

usta

rd”

“Cus

tard

”“C

usta

rd”

“Cus

tard

”“T

hin

Pure

e”Q

uote

dvi

scos

ityra

nge

1.2

+/-

0.4

Pa.s

Qua

ntity

requ

ired

per

100

mL

liqui

d(e

quiv

alen

tw

eigh

tin

gram

s)

2sc

oops

(6.0

)1.

5–

2sc

oops

(4.5

–6.

0)11 /

2sc

oops

(7.0

)11 /

2sc

oops

(7.0

)11 /

2T

bsp

(7.0

)1.

33sc

oops

(6.6

)

Des

crip

tor

“Thi

ckPu

ree”

Quo

ted

visc

osity

rang

eQ

uant

ityre

quir

edpe

r10

0m

Lliq

uid

(equ

ival

ent

wei

ght

ingr

ams)

2.0

scoo

ps(1

0)

Des

crip

tor

“Sem

i-so

lid”

“Pud

ding

”“P

uddi

ng”

“Pud

ding

”“P

uddi

ng”

“Sof

tFo

od”

Quo

ted

visc

osity

rang

e3.

0+/

-1.

0Pa

.sQ

uant

ityre

quir

edpe

r10

0m

Lliq

uid

(equ

ival

ent

wei

ght

ingr

ams)

21 /2

–3

scoo

ps(7

.5–

9.0)

2–2.

5sc

oops

(6.0

–7.

5)2

scoo

ps(9

.0)

2sc

oops

(9.0

)2

Tbs

p(9

.0)

2.67

scoo

ps(1

3.3)

a Nut

rici

aLt

d.,N

ewm

arke

tH

ouse

,New

mar

ket

Ave

nue,

Whi

teH

orse

Bus

ines

sPa

rk,T

row

brid

ge,W

iltsh

ire,

U.K

.bFr

esen

ius

Kab

iLtd

,Ces

tria

nC

ourt

,Eas

tgat

eW

ay,M

anor

Park

,Run

corn

,Che

sire,

U.K

.c N

estle

Nut

ritio

n,St

Geo

rges

Hou

se,C

royd

on,U

.K.

dV

itaflo

Int.,

11–1

6C

entu

ryB

uild

ing,

Bru

nsw

ick

Bus

ines

sPa

rk,L

iver

pool

,U.K

.

E332 Journal of Food Science � Vol. 75, Nr. 6, 2010

E:Fo

odEn

gine

erin

g&

Phys

ical

Prop

ertie

s


the rheological measurements. A cone and plate geometry (40-mm diameter, 4◦ angle) was chosen as this ensured near-uniformstrain across the radius of the sample (sample volume: 1.2 mL).Samples were subjected to a swept range of shear rates, chosento represent the range of flow conditions encountered in clinicalpractice from gentle manipulation in a cup (<0.1 s−1) to the mus-cular, healthy swallow, estimated at rates of up to 100 s−1. Thetemperature of the samples during rheometry was maintained bythe “Peltier Plate” control unit to within ±0.1 ◦C. A summary ofthe rheometer measurement parameters used is included in Table2 below.

Each shear-sweep measurement recorded shear rate, γ , and shearstress, τ , at discrete intervals over a duration of 100 s. Herein, wepresent these results in terms of apparent viscosity, ηa, as a functionof shear rate.

Rheological modelIn previous studies, investigators have successfully described

the behavior of starch-thickened fluids using the Hershel-Bulkleymodel (Germain and others 2006; Sopade and others 2007) thatcomprises a power-law relationship between shear stress and strainrate, plus a small yield stress (τ 0).

τ = τ0 + K γ n (1)

The flow behavior index (n) is equal to 1 for Newtonian fluids;in such cases the consistency index (K) is equivalent to the fluid’s

Table 2–Rheometrical parameters used for data acquisition.

Parameter Value

Measurement type Shear ramp, controlled rateMinimum shear (1/s) 0.1Maximum shear (1/s) 100Sweep time (s) 100Number of samples 25Progression LogarithmicTemperature 25 ◦C, isothermal

viscosity. For shear-thinning liquids n < 1. Results of shear sweepmeasurements were plotted and rheological models were appliedto the data to provide values of model parameters for each shearsweep. These parameters were observed over time as describedbelow.

Effect of fluid sample shear historyA variety of experimental protocols can be utilized to assess the

rheology of thickeners in the period following initial mixing. Inorder to establish the effect of the shear applied as part of themeasurement on the rheological properties of the fluids, studieswere performed of repeated shear sweeps over a 30-min periodusing 2 differing methodologies. The 1st protocol (A) utilized asingle sample (1.2 mL) drawn from 1 batch of thickened fluid thatwas loaded onto the rheometer immediately following the mixingprocedure and repeatedly sheared at 3-min intervals throughoutthe acquisition (see Figure 1, protocol A). Evaporation from thesmall volume sample throughout the duration of the measurementwas prevented by performing the sample shearing in an environ-mental chamber held at room temperature and 100% humidity.To avoid repeatedly shearing the same fluid sample, 2nd proto-col (B) varied from this regime by using a fresh fluid sample foreach shear sweep throughout the measurement. In order to obtaincorrect acquisition timings, protocol B required multiple batchesof thickened fluid to be produced, from each of which a singlesample was drawn after an appropriate delay from finalization ofpreparation (see Figure 1, protocol B).

Protocols A and B were utilized to acquire data from custardconsistency fluids (stage 2) produced using both previous and newformulations of “Thick & Easy” and the time dependence ofconsistency and flow behavior indices obtained using linear re-gression.

Temporal stability of thickened fluid rheology (Protocol C)Thickened drinks, as presented to dysphagic patients, are often

consumed over a long time period. To assess the behavior of fluidsover 3 h, shear-sweep measurements were performed as describedabove at intervals of 15 min (Figure 1, protocol C). The beakerof thickened fluid was initially mixed and then left, unstirred and

Load sample onto plate using modified syringe originating from central beaker bulk fluid volume

Advance cone to desired gap width

Verify fluid was loaded correctly and remove any excess

Commence rotational viscometry measurement

Save parameter and data files

Shear ramp, controlled rate Protocol A:

Repeat at three minute intervals

Withdraw cone. Clean cone and plate

Fresh sample from

fresh batch

Protocol C: Repeat at 15 minute

intervals for long duration measurement

Protocol B:Wait an appropriate interval

Protocol B

Fresh sample from

same batch

Figure 1–Measurement protocol used to obtainthe rheological properties of the thickenedfluids.


E:FoodEngineering

&PhysicalProperties


uncovered throughout the duration of the measurement. A freshsample was drawn from the beaker, via large-bore syringe with anexpanded orifice, and loaded onto the rheometer for each shearsweep. This was intended to replicate the situation of sips takenfrom a cup and to avoid the effects of evaporation from the sam-ple whilst in situ on the rheometer, which could be considerablegiven the sample volume of only 1.2 mL. Protocol C was appliedto custard consistency (stage 2) fluids produced with all thickeningagents shown in Table 1. Rheological models were applied to theshear-sweep measurement results and time dependence of consis-tency and flow behavior indices obtained using linear regression.

Results and Discussion

Comparison of fluids with manufacturer specified viscosityAn individual shear-sweep measurement for each of the 3 fluid

thickness levels produced using new formulation Nutilis is pre-sented below (Figure 2). Logarithmic axes are used to moreclearly illustrate the relative viscosities of the 3 thicknesses. Inorder to compare the observed non-Newtonian shear-thinningbehavior with the stated values of (implied Newtonian) viscosity,Figure 2 presents apparent viscosity compared to shear rate withmanufacturer-specified steady-flow viscosity ranges added (lightgray bands).

Over the range of shear rates tested, the apparent viscosities ofeach fluid were seen to be distinct and were approximately evenlydistributed on the log scale. The non-Newtonian behavior resultedin apparent viscosity values for the 3 thickness levels that variedover a much wider range than the manufacturer’s stated viscosities(illustrated by light gray bands on Figure 2). At the lowest shearrate tested (intended to represent slow tipping/swirling in a cup orflask), all 3 thickness levels have apparent viscosities that are greaterthan the stated viscosity range for the thickest fluid level. As theshear rate of the test increased, the apparent viscosities of eachfluid were seen to enter the ranges specified by the manufacturer;however, the shear rates at which the apparent viscosities matchedthe specified viscosity ranges were different for each fluid. Theseranges are indicated in Figure 2 using dark gray bands.

Rheological modelTo evaluate the quality of fit of the rheological model, shear

stress data from the rheometer were plotted against shear rate for a

single shear sweep measurement from each of the 3 consistenciesof fluid produced using new formulation Nutilis (Figure 3). TheHershel-Bulkley model was applied to these plots but the resultsfor yield stress were found to be inconclusive. Each of the shearsweeps resulted in a yield stress equal or near to zero. Only anegligible improvement in model fit was found when a yield stresswas included. For purposes of simplicity, yield stress was omittedfrom these results. The trend lines on the graph indicate the power-law rheological model (neglecting yield stress) applied to the data.The fluids all showed a good agreement with the model over therange of shear rates observed (R2 values given in Table 3). Shear-thinning behavior was observed in all fluids and this behavior wasmore pronounced with greater thickener concentration.

Effect of fluid sample shear historyProtocols A and B (described previously) were utilized to re-

peatedly acquire shear-sweep measurement data (n = 5) from bothformulations of “Thick and Easy” over a 30-min period followinginitial mixing. Each data point was averaged over the 5 acquisitions,and models utilizing a power-law relationship were applied to theset of averaged results. Consistency and flow behavior indices ob-tained from the model fitting are shown in Figure 4. Inclusion ofa yield stress parameter resulted in nearly half of the averaged shearsweeps having zero-yield stress. The remainder of the averagedmeasurements showed only a negligible improvement in model fitwhen a yield stress was included. Again, for purposes of simplicity,yield stress was omitted from these results. Correlation coeffi-cients (R2) from the power-law model (neglecting yield stress)for data from each shear sweep measurement were considered asan indication of the validity of the model assumptions. R2 valueswere found to range from 0.974 to 0.998, indicating a sufficientlygood correlation for analysis to continue using the power-lawmodel.

Use of single sample protocol “A” resulted in data indicating arise in consistency index over the duration of the measurement forboth thickeners. In contrast, the multiple batch, multiple sampleprotocol “B” indicated a small decrease in consistency index. Anincrease in consistency index over time was only observed forsamples that were repeatedly sheared during the acquisition. Linearregression was utilized to generate trend lines for the data in Figure4 that illustrated the overall changes in viscosity over the durationof the acquisition, see Table 4 for coefficients and percentage

Figure 2–Shear sweep measurements ofinstantaneous viscosity for fluids at 3 thicknesslevels using new formulation Nutilis comparedwith the stated viscosity ranges for newformulation Nutilis.


E:Fo

odEn

gine

erin

g&

Phys

ical

Prop

ertie

s


change. The multiple batch, multiple sample protocol “B” resultedin greater intersample variability than observed using protocol“A” as each data point resulted from a separate batch and sampleof fluid and was subject to the entire variability inherent in theprocesses of thickened fluid production, sampling, and rheologicalmeasurement.

The flow behavior indices, shown in Figure 4B, have values con-sistently above zero, and thus represent the degree of shear thinningobservable over the range of shear rates tested. The shear-thinningbehavior of each thickened fluid was found to be similar regardlessof the protocol used for data acquisition, with values rising slightlyduring the measurement. This contrasted with the different resultsfor consistency index obtained with the 2 protocols.

Time dependency of thickened fluid rheologyModels utilizing a power-law relationship were applied to a

single set of shear-sweep measurements for each thickening agentperformed at 15-min intervals over a 3-h duration following initialmixing (protocol C). Once again, yield stress inclusion producedinconclusive results and only negligible improvements in modelfit, allowing omission from further consideration for purposes ofsimplicity. Consistency and flow-behavior indices obtained fromthe model fitting are shown in Figure 5. The power-law modelcorrelation coefficients (R2) for data from each shear-sweep mea-surement were considered as a positive indication of the validityof the model assumptions ranging from 0.859 to 0.999. Boththe consistency and flow-behavior index were observed to possesslarger point-to-point variability than seen in the 30-min protocol“A” data since each point constituted a single measurement andwas subject to the full inherent intersample variation. Since thesesamples were taken from the same original batch, the variationwas mainly attributed to the inhomogeneity of the fluid mixtures:despite careful mixing according to manufacturers’ instructions,the fluids were observed to contain small bubbles and pockets of

poorly mixed thickener powder. Though small in size, these arti-facts become significant in relation to the rheometer’s minimumgap width (at the centre) of 150 microns. The viscosity of the bulkfluid would not be expected to exhibit the same variation as theresults obtained using these small sample volumes. The statistics inTable 5 were extracted through application of linear regression tothe time dependency of the rheological model parameters. Overthe 3-h duration of testing, the fluids showed varying changes inconsistency index. The smallest of these was just a 1% increaseabove the initial value for new formulation Nutilis, and the maxi-mum change observed was an increase of 25% at 3 h for previousformulation “Thick & Easy.” Two thickeners were observed toexhibit a decrease in consistency index over the duration of acqui-sition, the largest of which was a drop of 23% from the initial valuefor Vitaquick “Thin Puree.” Both of these fluids were producedfrom thickeners made solely from modified maize starch. Use ofthis ingredient alone, however, does not necessarily result in afluid whose consistency index decreases over time: the previousformulation Nutilis shared the same single constituent ingredientbut showed a 12% increase in consistency index over 3 h. Flowbehavior indices were all observed to increase slowly over the 3-hduration of the measurement, with increases ranging from 3% to16%.

Figure 5 demonstrated that thickeners whose instructions pro-vide for 3 levels of thickness produce fluids with consistency in-dices that are similar in magnitude, with the exception of previousformulation “Thick & Easy.” The reformulation of Thick & Easyhas brought this product into closer agreement with the otherstested. Of the 4 thickness described on the Vitaquick product in-structions, “Thin puree” has been shown previously to have theclosest consistency to “National Descriptors” stage 2 (O’Leary andothers 2010). The consistency index, however, was approximatelyhalf that of middle thickness fluids produced using products withinstructions for 3 thickness.

0.1

1

10

100

1000

0010111.0

Sh

ear

stre

ss (

Pa)

Shear rate (1/s)

Syrup (Stage 1) Custard (Stage 2) Pudding (Stage 3)

Figure 3–Rheological model trend lines fitted toshear data for the 3 consistency fluids.

Table 3–Rheological model parameters for 3 levels of thickened fluids from new formulation Nutilis.

Power-Law model indices:National descriptor Strength oftexture level Description Consistency Flow behavior correlation (R2)

Stage 1 “Syrup” 1.89 0.417 0.98Stage 2 “Custard” 12.8 0.368 0.97Stage 3 “Pudding” 70.3 0.273 0.96


E:FoodEngineering

&PhysicalProperties


DiscussionFigure 2 showed the large range of apparent viscosity values ob-

tained for thickened fluids when the shear rate was swept across therange chosen for these measurements. The highly non-Newtoniannature of these fluids increases the difficulty of subjectiveassessment of thickness as the fluid will appear thicker when rest-ing in its containing vessel than when it is manipulated (subject toa shear). Currently available quantitative measurements of thick-ened fluids, such as the line spread test (Paik and others 2004),are also complicated by the non-Newtonian flow properties. Ifthe fluids were to be tested using such a methodology, utilizingthe low shear rates derived from gravity, the result would be ex-pected to be significantly different from that obtained using anextrusion rheometer or viscometer set at the commonly used rateof 50 s−1. In the case of new Nutilis, Figure 2 shows the appar-ent viscosity at low shear rates to be far higher than the viscosityspecified by the manufacturer. It is, therefore, not sufficient tospecify the consistency of these fluids using a single value of vis-cosity, unless the shear rate at which the measurement took placeis also quoted. The measured viscosity data were found to coin-cide with the manufacturer specification at different shear rates foreach consistency of fluid. This apparent variation in measurementconditions across fluid consistencies serves to illustrate the lack of

relevant, quantified data describing the viscosity of these fluids.A single choice of shear rate stated for such measurements wouldsimplify comparison of the fluid and would serve as an acrossthe board standard for product labeling and comparison. Furtherstandardization of fluids delivered to patients may be achieved bya cooperative labeling policy across thickeners, such as already uti-lized by 1 thickener manufacturer, who has added labeling thatcorresponds to the “British National Descriptors” 3 fluid stages. Itwas assumed that the majority of the other manufacturers intendtheir products to be used to produce fluids that correspond tothese specifications, but without this standardization the labelingremains ambiguous. This was particularly true for the manufac-turer that specified 4 levels of thickened fluids and reinforces theneed for objective fluid testing rather than relying on a descriptivemethod of fluid specification. Shear rates involved in the variousstages of swallowing remain poorly defined. These are expectedto vary from person-to-person and with other factors such as ageand the level of impairment to normal swallowing. A functionalmodel of the swallowing process, defining the rates of shear in-volved, would allow fluid-testing to be performed under the mostappropriate conditions.

The non-Newtonian behavior of thickened fluids may have apositive impact upon the efficacy of the fluids as dysphagia texture

50

40

30

10

20

30

40

50

10

20

0 3 6 9 12 15 18 21 24 27 30

Elapsed Time (mins)

Co

nsi

sten

cy In

dex

-0.25

-0.50

0.50

0..25

0.00

-0.25

-0.50

0.00

0..25

0.50

0 3 6 9 12 15 18 21 24 27 30

Elapsed Time (mins)

Po

wer

Ind

ex

New Thick & Easy New Thick & Easy

Previous Thick & Easy

Previous Thick & Easy

A B

A

B

A

B

A

B

A

B

Figure 4–Indices for thickened fluids over a30-min duration obtained using protocol A(open symbols) and protocol B (closed symbols).(Error bars indicate range of indices calculatedusing averaged raw data ± 1 standarddeviation.)

Table 4–Changes in consistency and flow behavior indices over the 30-min period following mixing from linear regression appliedto the experimental data and the linear regression coefficients, slope (m), and intercept (c).

Consistency index Flow behavior index

Protocol A Protocol B Protocol A Protocol B

Pre T&E New T&E Pre T&E New T&E Pre T&E New T&E Pre T&E New T&E

% change +11 +39 −11 −5 +33 +35 +19 +8m 0.39 0.30 −0.11 0.04 0.0072 0.0031 0.0061 0.003c 35 23 32 21 0.25 0.24 0.23 0.25R2 0.84 0.93 0.14 0.09 0.55 0.83 0.52 0.17

T & E = Thick and Easy.


E:Fo

odEn

gine

erin

g&

Phys

ical

Prop

ertie

s


2015

10

5

030

25

20

15

10

30

25

2015

10

4540

35

30

2525

20

15

105

30

2520

15

10

5045

40

35

3025

20

15

10

5

0 0:000:150:300:451:001:151:301:452:002:152:302:453:00

Elapsed Time (h:mm)

Co

nsi

sten

cy In

dex

0.100.150.200.250.300.350.400.450.500.550.600.200.250.300.350.400.200.250.300.350.400.200.250.300.350.400.150.200.250.300.350.200.250.300.350.400.200.250.300.350.40

0:000:150:300:451:001:151:301:452:002:152:302:453:00

Elapsed Time (h:mm)

Po

wer

Ind

ex

Nutilis, new formulation

Thick & Easy, new formulation

Thick & Easy, previous formulation

Resource ThickenUp

Vitaquick Thin Puree

Combined plot

Nutilis, new formulation

Thick & Easy, new formulation

Thick & Easy, previous formulation

Resource ThickenUp

Vitaquick Thin Puree

Combined plot

Nutilis, previous formulation Nutilis, previous formulation

Figure 5–Consistency index (A) and flowbehavior index (B) for thickened fluids over a3-h period.

Table 5–Percentage change in consistency and flow behavior indices from linear regression applied to the time dependency of therheological model parameters and the associated linear regression coefficients, slope (m), intercept (c), and R2.

Nutilis Thick & Easy VitaquickPrevious New Previous New Thicken up thin puree

Consistency index m 0.20 0.0095 0.69 0.38 −0.19 −0.22c 20 18 32 20 22 12R2 0.21 0.004 0.68 0.48 0.12 0.11

% Change +12 +1 +25 +23 −11 −23

Flow behavior index m 0.0009 0.0032 0.0028 0.0030 0.0018 0.0029c 0.30 0.30 0.27 0.22 0.28 0.30R2 0.04 0.48 0.35 0.36 0.12 0.17

% Change +3 +13 +13 +16 +8 +12


E:FoodEngineering

&PhysicalProperties


modification therapies. At low shear rates, such as when the flu-ids are stored, gently manipulated or during the oral preparatorystage upon the tongue, the flow properties are such that the bulkfluid behaves as described by the national descriptors for thickenedfluids. An increase in shear, such as during rapid manipulation orthe transport phase of swallowing, has the effect of thinning thefluids, allowing an increased flow rate to occur. Quantificationof shear-thinning behavior of new formulation Nutilis revealeda small decrease in flow behavior index from 0.42 to 0.27 withincreasing thickener concentration, indicating greater shear thin-ning response. The flow behavior index of the middle thicknessfluid, 0.37, corresponds to a fluid whose apparent viscosity woulddecrease by a factor of 0.65 if the shear rate were doubled.

The power-law model was found to provide a sufficiently goodmatch to the data obtained from all the thickened fluids tested andprovided a more complete characterization of the flow propertiesof the fluids than 1 viscosity measurement at a given shear ratewithout the added complexity of introducing a yield stress. Thisstudy only considered fluid measurements whilst undergoing acertain level of shearing, and this may have rendered the yieldstress negligible.

The shear history of a sample of thickened fluid was shown tosignificantly affect the results of the rheological analysis. This wasestablished using short-duration temporal-stability investigationsperformed with 2 differing protocols. The single sample method-ology (protocol A) resulted in a rise in measured consistency indexthat was not attributable to sample dehydration but the multiplebatch, multiple sample method (protocol B) indicated a small de-crease over time. It is surmised that repeatedly testing the samesample of fluid may increase the level of dissolution of the thick-ener through sample agitation. This consistency index rise couldbe duplicated in a clinical environment by excessive or repeatedstirring of a thickened drink. In contrast to the observed increasein consistency index, the flow behavior indices for both thicken-ers obtained using the differing protocols were not found to besubstantially different.

The temporal stability of the rheological properties of middleconsistency thickened fluids was examined over a duration of 3h using a single batch, multiple sample method. Both rises andfalls in consistency indices were observed over this time with thelargest change being a 25% rise over the initial value. This mayhave implications for the efficacy of patient treatment as not allfluids will be consumed immediately upon mixing. It has not beenpossible to establish whether the changes in consistency index wereas a result of evaporative concentration, changes in the levels ofagent dissolution over time, or possessing a more fundamentalchemical origin. The reformulation of “Thick & Easy” resulted inlower viscosity fluids, broadly in line with the other fluids of “stage2” consistency. This was achieved through manufacturing processmodification, without altering the ingredients, highlighting theimportance of thickener manufacture. The flow behavior indicesmeasured represented the level of non-Newtonian response whenthe fluids were subject to analysis and were found to be comparablein value and exhibit similar small rises over time.

ConclusionThe implications of this study for practical, in situ viscosity mea-

surements are that a single shear rate measurement can be used toquantify the viscosity of thickened fluids that utilize starch-basedthickeners to achieve their purpose as the non-Newtonian re-

sponse of the thickeners was shown to be similar. However, theapparent viscosity was shown to vary very significantly depend-ing on shear rate. In practice, if measurements upon all thickenerswere performed at an agreed shear rate, the viscosity results wouldserve to allow a basic comparison of the fluids produced. A morecomplete method of specifying the flow properties of the flu-ids was utilized and found to increase the depth of analysis thatis possible regarding these fluids. The ability to clearly interpretboth the viscosity and non-Newtonian response of the thickenedfluids may allow an interesting possibility to be explored: it isperhaps the non-Newtonian nature of these fluids that rendersthem particularly suitable for their use as dysphagia therapies andit may be possible to alter the properties of the fluids to vary theirnon-Newtonian response (flow behavior index) to optimize thetherapeutic efficacy.

AcknowledgmentsThis study was funded by a grant from the EPSRC and some

thickeners were kindly provided by Fresenius-Kabi Ltd.

ReferencesCichero JAY, Jackson O, Halley PJ, Murdoch BE. 2000. Which one of these is not like the

others? An inter-hospital study of the viscosity of thickened fluids. J Speech Lang Hear Res43:537–47.

Dantas RO, Kern MK, Massey BT, Dodds WJ, Kahrilas PJ, Brasseur JG, Cook IJ, Lang IM.1990. Effect of swallowed bolus variables on oral and pharyngeal phases of swallowing. Am JPhysiol Gastrointest Liver Physiol 258:G675–81.

Dewar RJ, Joyce MJ. 2006a. Time-Dependent Rheology of Starch Thickeners and the ClinicalImplications for Dysphagia Therapy. Dysphagia 21(4):264–9.

Dewar RJ, Joyce MJ. 2006b. The thixotropic and rheopectic behaviour of maize starch andmaltodextrin thickeners used in dysphagia therapy. Carbohydr Polym 65:296–305.

Finestone HM, Foley NC, Woodbury MG, Greene-Finestone L. 2001. Quantifying fluid intakein dysphagic stroke patients: a preliminary comparison of oral and nonoral strategies. ArchPhys Med Rehabil 82:1744–6.

Garcıa-Peris P, Paron L, Velasco C, de la Cuerda C, Camblor M, Breton I, Herencia H, VerdaguerJ, Navarro C, Clave P. 2007. Long-term prevalence of oropharyngeal dysphagia in head andneck cancer patients: impact on quality of life. Clin Nutr 26:710–17.

Germain I, Dufresne T, Ramaswamy HS. 2006. Rheological characterization of thickenedbeverages used in the treatment of dysphagia. J Food Process Eng 73:64–74.

Goulding R, Bakheit AMO. 2000. Evaluation of the benefits of monitoring fluid thickness inthe dietary management of dysphagic stroke patients. Clin Rehabil 14:119–24.

Kayser-Jones J, Schell ES, Porter C, Barbaccia JC, Shaw H. 1999. Factors contributing todehydration in nursing homes: inadequate staffing and lack of professional supervision. J AmGeriatr Soc 47(10):1187–94.

Langmore SE, Skarupski KA, Park PS, Fries BE. 2002. Predictors of aspiration pneumonia innursing home residents. Dysphagia 17(4):298–307.

Layne KA, Losinski DS, Zenner PM, Ament JA. 1989. Using the fleming index of dysphagia toestablish prevalence. Dysphagia 4(1):39–42.

Leonard R, Kendall K. 1997. Dysphagia assessment and treatment planning. A team approach.San Diego, Calif.: Singular Publishing Group.

Mann G, Hankey GJ, Cameron D. 2000. Swallowing disorders following acute stroke: prevalenceand diagnostic accuracy. Cerebrovasc Dis 10:380–86.

Mertz Garcia J, Chambers E, Matta Z, Clark M. 2005. Viscosity measurements of nectar- andhoney-thick liquids: product, liquid, and time comparisons. Dysphagia 20:325–35.

Miller JL, Watkin KL. 1996. The influence of bolus volume and viscosity on anterior lingualforce during the oral stage of swallowing. Dysphagia 11:117–24.

O’Leary MT, Hanson BM, Smith C. 2010. Variation of the apparent viscosity of thickeneddrinks. Int J Lang Commun Disord (forthcoming).

Paik N-J, Han TR, Park JW, Lee EK, Park MS, Hwang I-K. 2004. Categorization of dysphagiadiets with the line spread test. Arch Phys Med Rehabil 85(5):857–61.

Sopade PA, Halley PJ, Cichero JAY, Ward LC. 2007. Rheological characterization of foodthickeners marketed in Australia in various media for the management of dysphagia. I: waterand cordial. J Food Eng 79:69–82.

Sopade PA, Halley PJ, Cichero JAY, Ward LC, Liu J, Teo KH. 2008a. Rheological charac-terization of food thickeners marketed in Australia in various media for the management ofdysphagia. II. Milk as a dispersing medium. J Food Eng 84(4):553–62.

Sopade PA, Halley PJ, Cichero JAY, Ward LC, Liu J, Varliveli S. 2008b. Rheological charac-terization of food thickeners marketed in Australia in various media for the management ofdysphagia. III. Fruit juice as a dispersing medium. J Food Eng 86(4):604–15.

Steele CM, Lieshout PHHM, Goff HD. 2003. The rheology of liquids: a comparison of clini-cians’ subjective impressions and objective measurement. Dysphagia 18:182–95.

The British Dietetic Association. 2009. National descriptors for texture modification in adults.Birmingham, U.K.: The British Dietetic Association.

Whelan K. 2001. Inadequate fluid intakes in dysphagic acute stroke. Clin Nutr 20(5):423–8.Wilkins T, Gillies RA, Thomas AM, Wagner PJ. 2007. The prevalence of dysphagia in primary

care patients: a HamesNet Research Network study. J Am Board Fam Med. 20(2):144–50.


Documents

Viscosity and Non-Newtonian Features of Thickened Fluids Used for Dysphagia Therapy