ANÁLISE DOS DADOS

Estatística descritiva foi realizada para os dados ordinais da análise visual e para os

dados numéricos da análise colorimétrica (valores VOH). As avaliações dos dois

examinadores foram comparadas por meio de testes não-paramétricos (teste Wilcoxon

Signed-Ranks e Friedman). Concordância intra e inter-examinador para a análise visual foi

mensurada usando os seguintes parâmetros: porcentagem da concordância absoluta;

coeficiente de correlação intraclasse (ICC do inglês Intraclass Correlation Coeficient) para os

dados de categoria ordenada, medindo a concordância absoluta entre as classificações; e o

Kappa ponderado para mensurar diferenças extremas. A análise da concordância

intra-examinador da análise subjetiva foi realizada em uma subamostra aleatória de 30% com base

na avaliação inicial usando a análise visual direta e uma segunda avaliação cega pelos dois

avaliadores das fotografias digitais de ambos os lados da goma mastigada.

A confiabilidade da análise VOH foi avaliada por meio do gráfico Bland-Altman. As

diferenças entre as mensurações duplicadas foram comparadas com a média das respectivas

leituras para determinar o padrão da confiabilidade e para avaliar qualquer diferença

sistemática, através do cálculo da diferença entre a média e os limites de concordância das

mensurações (±1,96 x desvio padrão). Diferenças em VOH de acordo com o número de ciclos

mastigatórios foram analisadas usando o teste Anova para testar os efeitos intra-sujeitos para

um único fator do estudo. Para ter significância estatística foi considerado o valor de p<0,05

para todos os testes de hipóteses. A análise dos dados foi realizada pelo software MedCalc

5 ARTIGO

Validation of a two-colour chewing gum test to assess masticatory

performance in complete denture wearers

Lorrany C. Silva

¹

; Túlio E. Nogueira

¹

; Martin Schimmel

²

; Cláudio R. Leles

¹

1

Department of Oral Prevention and Rehabilitation, School of Dentistry, Federal

University of Goias, Brazil.

2

Division of Gerodontology and Removable Prosthodontics, Dental University

Clinic, University of Geneva, Switzerland. Division of Gerodontology, School of Dental

Medicine, University of Bern, Switzerland.

Abstract

The aim of this clinical study was to test the validity and reliability of a method to

measure the masticatory performance of complete denture wearers employing a

colour-mixing ability test and assessment by visual and opto-electronical colourimetric analysis. A

sample of 75 subjects was selected from patients who received new conventional

tissue-supported complete dentures in a university clinical setting. Masticatory tests were performed

using a two-colour chewing gum that was masticated for 5, 10, 20, 30 and 50 chewing cycles,

performed in a random order. The mixing level of the two colours of the chewed gum was

assessed visually by two independent raters based on a 5-point ordinal scale. Then, the

specimens were flattened into a 1 mm width wafer, scanned and saved as a two-sided digital

image. Each pair of images was submitted to a colourimetric analysis using a custom-made

software (ViewGum©, Dhal Software, Greece). The circular variance of hue (VOH) was used

as a measure of the level of the colour mixture. The lower is the VOH, the greater is the

mixture and, consequently, the masticatory performance. In the current study design, a

varying number of chewing cycles simulated different levels of masticatory performance in

one subject. Overall inter and intra-rater agreement in visual analysis was 64% and 68%,

respectively (almost 99% of scores ranged within ±1 point). Overall weighted kappa was

>0.80 and intraclass correlation coefficient around 0.90. A proportional increase in the level

of mixture occurred with increased number of chewing cycles for both the visual and

colourimetric analysis (p<0.001). Similarly, VOH and the visual analysis were highly

correlated (r= -0.89; p<0.001). Bland-Altman plots revealed excellent agreement and

extremely low systematic error between duplicated VOH measures. It was concluded that the

two-colour chewing gum test is a valid and reliable method for assessment of the masticatory

performance in complete denture wearers using both the visual and electronic colourimetric

Introduction

Mastication is a complex physiological process that aims to promote food

comminution and mixture of the resulting fragments with saliva to subsequent swallowing

(Speksnijder et al., 2009). Tooth loss is one of the main factors that can negatively impair

masticatory function, resulting in significant impacts on oral-health related quality of life and

physical well-being, especially in completely edentulous individuals (Allen & McMillan,

2003).

The assessment of the masticatory performance is helpful for a direct evaluation of

the subject’s functional ability and to establish treatment strategies for the rehabilitation of the

impaired oral function (Prithviraj et al., 2014). Traditionally, the masticatory performance has

been evaluated by food comminution tests associated with the sieving method (Manly &

Braley, 1950; Molenaar et al., 2012), which had been considered the gold standard method for

clinical and experimental purposes.

In the sieving method the comminuted median particle size is calculated for a given

number of chewing strokes as the objective measurement of masticatory efficiency (Eberhard

et al., 2015). However, subjects with impaired oral function are not always able to fragment

test foods, since their maximum bite force can be lower than that required to crush test food

particles (Woda et al., 2011). Partially edentulous subjects or complete denture wearers have

significant reduction in masticatory performance, requiring from 4 to 8 times the number of

chewing strokes than dentate subjects to achieve the same degree of comminution (Prithviraj

et al., 2014). It has also to be considered that dysphagia is highly prevalent in elderly patients

and therefore, the comminution test might render a risk for particle aspiration (Wilkins et al.,

Alternative methods for assessment of masticatory performance have been proposed,

such as the evaluation of the color-mixing ability test, i.e., bolus-kneading test, using

two-coloured chewing gum (Liedberg & Owall, 1995; Prinz, 1999; Schimmel et al, 2007). In this

test, a two-coloured gum is chewed by a certain number of strokes and, subsequently, the

chewed gum is classified based on a visual reference scale, and also assessed by an

opto-electronic method (Schimmel et al., 2015). It has been recommended as the preferred method

to assess masticatory function in subjects with impaired masticatory performance (Weijenberg

et al., 2013), since it is suitable for measuring statistically significant differences when

compared to the sieve method (Speksnijder et al., 2009; van der Bilt et al., 2010). Studies

showed that the mixing ability test with two-coloured chewing gum is a valid and reliable

alternative to the food comminution test in fully dentate subjects (Sato et al., 2003; Van der

Bilt et al, 2010), correlates significantly with the sieving method (Speksnijder et al., 2009),

can be used in dysphagic patients (Schimmel et al., 2013) and the electronic assessment of the

chewed gum is also a discriminative tool to assess chewing efficiency in clinical and research

settings (Schimmel et al., 2015).

The use of an alternative method to assess masticatory function in denture wearers

could be useful as an objective way to measure the effectiveness of the treatment, since it is a

relevant clinical outcome that influences the satisfaction with the dentures and the functional

well-being. Despite being considered a less complex method, its clinical validity and

reliability had not been extensively tested in subjects with impaired oral function, such as

edentulous subjects treated with conventional complete dentures. Hence, the aim of this study

was to test the validity and reliability of a colourimetric technique to assess chewing function

in complete denture wearers, based on the colour-mixing ability of a two-colour chewing

Material and methods

The study sample included fully edentulous subject who were treated to receive a

new set of maxillary and mandibular tissue-supported dentures at the School of Dentistry of

the Federal University of Goias, Goiania, Goias, Brazil. No restrictions related to age, gender

and level of satisfaction with the dentures was considered for inclusion in the study. Subjects

unable to perform the masticatory tests due to poor oral coordination, extremely unstable

dentures or poor condition of the denture bearing mucosa were excluded from the study. The

masticatory tests were performed after the functional adaptation period of the 1 to 3 month

with new dentures, when the patient did not present any complaints regarding the use of the

dentures. The research protocol was previously approved by the local research ethic

committee within the context of a larger study (CAAE 37596114.2.0000.5083), and all

participants signed an informed consent before taking part in the study.

Test food and chewing protocol

A two-coloured chewing gum (Vivident Fruitswing Karpuz/Asai Üzümü, Perfetti

van Melle, Turkey) served as the test food. The chewing gums had the dimensions of 43 mm

x 12 mm x 3 mm, comprising two layers – violet (grape flavor) and green (watermelon

flavor). The hardness of the gum lies between mean depth of indentation 0.8±0.08mm, mean

Durometer 67.4 for the green side and mean depth of indentation 1.1±0.08mm, mean

Durometer 58.4 for the violet side (Shore Scale OO, Ø 2.4 mm, 1.11 N). This chewing gum

was selected because it was considered suitable for complete dentures wearers and because it

did not stick to the denture-resin (Schimmel et al., 2015).

Each subject was asked to sit comfortably in the dental chair and instructed about the

test procedures. Five subsequent tests were performed for 5, 10, 20, 30 and 50 chewing cycles

Between each test, a one-minute interval was used for rest and to avoid muscle fatigue, except

after the 50 cycles test when a 2 minutes interval was respected. The order of the five tests for

each individual was defined randomly using an electronic tool to generate sets of unsorted

random numbers (https://www.randomizer.org/), and the test sequence was disclosed

immediately before the first test for each participant. After each cycle, the chewed gum was

collected from the oral cavity by the operator, photographed with a digital camera with a 100

mm macro lens with 1:3 magnification on both sides of the specimen in a fixed 35 cm focus

distance and saved on JPEG format. Then, the specimens were placed in a transparent plastic

bag, which was labeled with an identification code.

Subjective and electronic colourimetric assessment

Specimens were classified by two independent examiners using as reference an

ordinal rating scale that categorizes the specimens according to the level of mixture of the two

colours of the chewed gum: Score 1 – chewing gum not mixed, impressions of cusps or

folded once; Score 2 – large parts of chewing gum unmixed; Score 3 – bolus slightly mixed,

but bits of unmixed original colour; Score 4 – bolus well mixed, but colour not uniform; and

Score 5 – bolus perfectly mixed with uniform colour (Schimmel et al, 2015). Both examiners

were blind in relation to the participant and the number of chewing cycles relative to the

specimen. The photographed specimens were used to assess the intra-rater agreement of the

visual analysis. To assure the independence of ratings, all specimens were examined in

random order, raters were blinded to their first ratings, and a minimum of two weeks was used

between repeated measurements.

Subsequently, the specimens were flattened to a wafer of 1 mm thickness by pressing

with a glass plate under manual pressure. Then, both sides of the wafer were scanned (HP

dpi resolution. Both sides were joined in a single file and this was placed in a maximum size

of 1000 pixels in the vertical or horizontal direction.

The colourimetric analysis was performed using the freeware ViewGum© software

(dHAL Software, Greece, www.dhal.com), a computer program specifically developed to

evaluate masticatory performance by the two-colour chewing gum test (Halazonetis et al.,

2013). The images are transformed into the HSI colour space using an image processing

application to analyze separately the hue, intensity and saturation. Then, the hue value was

calculated for each pixel in the semiautomatically segmented images to obtain a

representative measurement of the mixture by concentration only on the hue component. If

the colours of the specimen are not mixed, two well-separated peaks on the hue axis are

presented and, conversely, with increasing degree of colour mixing, the two hue peaks of each

colour will converge as the colours are mixed. The circular variance of the hue (VOH) is

considered the measure of mixing. The smaller is the VOH value, the greater is the mixing of

the two-coloured layers of the chewed gum, which in turn means better chewing performance

of the subject (Halazonetis et al., 2013; Schimmel et al., 2015). The step-by-step procedure to

delimitate the outer contours of the gum images and VOH calculation is detailed in the

supplemental material of the study by Schimmel et al. (Schimmel et al., 2015 – supplemental

material).

Data analysis

Descriptive statistics were performed for ordinal data from the subjective analysis

and for numerical data from the electronic analysis (VOH values). The ratings of the two

examiners were compared using non-parametric tests (Wilcoxon Signed Ranks and Friedman

tests). Intra- and inter-rater agreement for the visual analysis was measured using the

following parameters: percent of absolute agreement, intraclass correlation coefficient (ICC)

weighted Kappa (with quadratic weights, to take account of extreme differences). Intra-rater

agreement analysis was performed in a 30% random subsample, based on the initial

assessment using direct visual analysis and a second blind assessment based on a double-face

digital image of the specimens.

Reliability of the VOH analysis was assessed using Bland-Altman plots. The

differences between duplicate measurements were plotted against the mean of the respective

readings to determine the pattern and spread, and to evaluate any systematic differences, by

calculating the mean difference and the limits of agreement between the measurements (±1.96

standard deviations). Differences in VOH according to the number of chewing cycles were

analyzed using repeated measures Anova to test within-subjects effects for a single-factor

study. Statistical significance was set at p < 0.05 for all hypothesis testing. MedCalc Software

version 16.1 (MedCalc Software, Ostend, Belgium) was used for all data analysis.

Results

The study sample comprised 75 complete denture wearers, 51 (68.0%) female, age

ranging from 44.1 to 85.0 years (mean = 67.1±8.5 years). All subjects performed the

masticatory tests with no missing data. Images in Figure 1 show a set of samples of the

specimens that represent the chewed gum scored from 1 to 5 and the correspondent two-side

flattened specimen and their VOH values obtained from the electronic colourimetric analysis.

Subjective (visual) assessment

Data obtained from the subjective analysis is shown in Table 1 for the two raters and

their mean values. Marginal significant differences between raters were found for 10 cycles,

and a progressive and significant increase in scores was observed between cycles for the two

Results of the inter- and intra-rater agreements are detailed in Table 2. For the

different number of chewing cycles, agreement ranged from 56.0 to 72.0% (pooled %

agreement = 64.0%), and 99.5% of cases fall within the limits of agreement of ± 1 score.

Analysis of the reliability of ratings using ICC showed good agreement for all numbers of

chewing cycles (> 0.69) and for overall data (ICC = 0.89; 95% CI = 0.87 – 0.91). Overall

kappa value for inter-rater agreement was considered as “very good” (kappa = 0.81; 95% CI =

0.78 – 0.84). Moderate agreement was achieved for 5 cycles (kappa = 0.52; 95% CI = 0.38 –

0.67) whereas good to very good agreement occurred in all other conditions. Table 2 also

shows results of the intra-rater agreement. There was a satisfactory overall performance of

raters (ICC > 0.90; kappa = 0.83).

Opto-electronic colourimetric assessment

Agreement between duplicated electronic colorimetric measures was assessed using

Bland-Altman plot comprising the overall measures, expressed in Figure 2. The systematic

error was extremely low, near the horizontal zero line (mean = 0.0005) and only a few cases

(n=23; 6.1%) were outside the 95% CI limits, which were also extremely low. This high

agreement was expressed as a perfect correlation between the duplicated electronic

measurements (ICC=1.00; p<0.001).

Subjects’ performance measured by the colourimetric test was checked against the

criteria of number of chewing cycles. Results expressed in Figure 3 show that the test has a

good discriminating ability, since there was a significant decrease in VOH values with

greater number of cycles (lower values mean greater mixture of the chewing gum) (p <

0.001).

Figure 4 shows the correlation between the overall visual and colourimetric

mean values of the duplicated colourimetric assessments. A significant negative correlation

between measurements was observed using the non-parametric Spearman correlation

coefficient (r=-0.89; p<0.001), indicating that the level of the mixture measured by the

electronic VOH measure and visual analysis were highly correlated.

Discussion

This study showed that the method using a two-colour chewing gum test is able to

quantify the levels of masticatory performance in complete denture wearers, by measuring the

bolus-kneading capacity through the ability to mix a two-coloured chewing gum. Both the

visual and electronic colourimetric analysis were sensitive enough to discriminate

intra-individual performance through increasing visual scores and decreasing VOH values with

increasing number of chewing cycles. Inter and intra-rater agreements in visual analysis, as

well as the low magnitude of measurement errors in colourimetric analysis were considered

within acceptable confidence limits.

Interpretation of Kappa coefficients and their correspondent 95% confidence

intervals for inter-rater assessment result in moderate to strong agreement. Reliability was

also reported using the ICC values for the ordered-category data in visual analysis, since

weighted Kappa and ICC are equivalent for the same data under general conditions (Fleiss

and Cohen, 1973). The absolute agreement model of the intra-class correlation coefficient was

considered good agreement (≥0.80) for inter- and intra-rater assessment, except for the 5

chewing cycle analysis.

A slight examiner bias in inter-rater agreement was observed in visual analysis of the

5 and 10 chewing cycle groups, evidenced by the lower kappa values and barely significant

difference between raters. This is coherent with another study in dentate subjects that showed

chewed 20 cycles or longer (Schimmel et al., 2007). It is also expected that masticatory

performance in complete denture wearers is markedly lower than dentate subjects. For

example, virtually no patients achieved a score of 5 after 50 cycles in our study, whereas all

specimens chewed 50 times were judged as score 5 in other studies with dentate subjects

(Schimmel et al., 2007) and dentate and subjects with 2-implant mandibular overdentures

(Schimmel et al., 2015).

However, factors such as prevalence bias (the proportion of cases on which the raters

agree) may influence the magnitude of the kappa coefficient (Sim and Wright, 2005) and,

therefore, comparison across studies may be considered with caution since the proportion of

cases scored from 1 to 5 may vary greatly among subjects with different dental status or level

of masticatory impairment. In addition, the lower values of Kappa for 5 chewing cycles may

not necessarily reflect low rates of agreement, and could be affected by the prevalence bias,

since the agreement on scores 1 and 2 was high and rating the 5 cycle specimens with scores

greater than 2 was such a rare finding. Nevertheless, moderate agreement in visual analysis

for scores 1 and 2 in the 5 chewing cycles may also be due to poor definition of cut-off

criteria for classification of the specimens. More accurate description of classification criteria

may be recommended for the whole subjective scale used for visual analysis.

According to Anastassiadou and Heath (Anastassiadou and Heath, 2001), when

chewing gum tests are used for older subjects, particularly complete denture wearers, the

number of strokes used deserves special consideration. The difficulty to differentiate jaw

movements to position the gums for initial fracture and the first power stroke may lead to an

underrated appraisal of the number of effective chewing strokes (Anastassiadou and Heath,

2001). A wide range of numbers of masticatory cycles, from 5 to 50, was used to encompass

the differences in masticatory efficiency between edentulous subjects with presumed impaired

No documento VALIDAÇÃO DO TESTE DE HABILIDADE DE MISTURA COM GOMA DE MASCAR DE DUAS CORES PARA AVALIAÇÃO DA PERFORMANCE MASTIGATÓRIA EM USUÁRIOS DE PRÓTESES TOTAIS (páginas 42-88)