Accepted: 15 April 2021

DOI: 10.1111/1440-1630.12732

F E AT U R E A RT I C L E

Mealtime behaviours of young children with sensory food

aversions: An observational study

Angela R. Caldwell | Elise K. Krause

Department of Occupational Therapy,

University of Pittsburgh, Pittsburgh, PA,
Abstract
USA Introduction: Children with sensory food aversions (SFAs) are at risk for nutri-
tional problems and occupational dysfunction during daily meals. To facilitate op-
Correspondence
Angela R. Caldwell, Department of timal occupational performance for children with SFAs, it is critical that clinicians
Occupational Therapy, University of understand child behaviour and caregiver strategy use to manage child behaviour
Pittsburgh, 100 Technology Drive, 368
during meals. The purpose of this paper is to examine the reliability of a novel coding
Bridgeside Point, Pittsburgh, PA 15219,
USA. scheme, the Behavioural Mealtime Coding System (BMCS) and identify associa-
Email: [email protected] tions among child acceptance of preferred and targeted foods, child food exploration
and caregiver use of strategies to manage child behaviour.
Funding information
This research was supported by the Methods: Twenty-­one children (aged 18–­60 months) with SFAs and their caregivers
School of Health and Rehabilitation were recruited using a convenience sample. An observational video coding system
Sciences Research Development Fund,
the Department of Occupational Therapy
was developed to code mealtime behaviour in 63 typical mealtime videos recorded
at the University of Pittsburgh and the by caregivers in the home environment. Inter-­rater reliability, descriptive statistics
National Center Medical Rehabilitation and bivariate correlations were calculated.
Research, National Institute of Child
Health and Human Development/National Results: The BMCS demonstrated excellent inter-­rater reliability (ICC = .95). Child
Institute Neurological Disorders and bites of targeted foods were associated with mealtime duration (rs = .51, p = .02)
Stroke, National Institutes of Health (K12
and events of the child licking food (rs = .57, p = .007). Caregiver use of threats was
HD055931).
positively associated with child age (rs = .48, p = .03) and negatively associated with
caregiver education level (rs = −.49, p = .03).
Conclusion: For children with SFAs, increasing mealtime duration by embedding
positive activities to encourage food exploration may improve acceptance of healthy
targeted foods. Future research is needed to better understand the complex relation-
ships among caregiver strategy use, mealtime duration and child mealtime behaviour.

KEYWORDS
behaviour, feeding, observational coding, parenting, preschool children

1 | IN T RO D U C T ION children, dietary selectivity is often based on the sensory

characteristics of food, such as texture, smell or tempera-
Proper nutritional intake is imperative for successful devel- ture. Approximately 50% of children demonstrate caregiver-­
opment in early childhood (Samuel et al., 2018); however, reported selective eating behaviours by the age of two, many
many children are not consuming well-­rounded diets due to of whom (58%) will outgrow this behaviour pattern (Carruth
selective eating (Carruth et al., 2004). Selective, or picky, et al., 2004; Mascola et al., 2010). However, young children
eating has been defined in a variety of ways but includes an (0–­8 years) who demonstrate sensory food aversions (SFA)
element of low dietary variety (Dovey et al., 2008). Among avoid food based on sensory characteristics (e.g., colour,
336 | © 2021 Occupational Therapy Australia wileyonlinelibrary.com/journal/aot Aust Occup Ther J. 2021;68:336–344.
CALDWELL AnD KRAUSE
| 337

texture and temperature) and are likely to demonstrate per- specifically interested in how children with SFA participate
sistent selective eating because they also resist trying new in the essential occupation of feeding when sharing a meal
foods and have strong food preferences (Chatoor, 2009; with family members. Because children with SFA often
Toyama & Agras, 2016). Moderate to severe levels of food avoid any interaction (e.g., touching, playing and smell-
selectivity during early childhood have been shown to be ing) with novel or non-­preferred foods, the purpose of this
associated with greater likelihood of conflicts regarding study was to describe their unique mealtime behaviours and
food, reduced growth, anxiety, depression and attention defi- identify associations among child and caregiver behaviours.
cits (Zucker et al., 2015). It is important to address these Existing mealtime coding schemes, such as the Dyadic
behaviours early in life to ensure proper nutritional intake Interaction Nomenclature for Eating (DINE) and Mealtime
during a critical time of brain development. Interaction Coding Scheme (MICS), did not meet the precise
In addition, selective eating increases stress within needs of our study (Dickstein et al., 1998; Stark et al., 2000).
the family unit and conflicts during mealtime (Luchini Specifically, we were interested in coding specific strategies
et al., 2017). Caregivers may use coercive strategies (e.g., and behaviours, whereas the MICS examines overall family
threats and force feeding) to increase food acceptance, which functioning more broadly. Conversely, the DINE does provide
can create a negative mealtime experience (Ramos-­ Paúl a scheme for coding specific behaviours but does not specify
et al., 2014). Pressuring a child to eat novel foods is asso- between preferred and non-­preferred foods and lacks codes
ciated with poor outcomes, such as (a) lower overall intake, for child food exploration (Poppert et al., 2015). Because tar-
(b) food dislikes and rejections and (c) increased negative geted food acceptance and the ability to interact with and ex-
comments during mealtimes (Galloway et al., 2006; Jansen plore novel food are central to the clinical treatment of SFA,
et al., 2017). Caregivers may also be tempted to use food it was crucial that we used a system that reliably coded these
rewards or bribes to encourage their child to try a novel or behaviours.
non-­preferred food. This strategy undermines a child's ability The primary aim of this study was to describe the meal-
to build food preferences through exposure alone and may time behaviours of young children with SFA and identify as-
lead to negative consequences such as decreased preference sociations among child and caregiver behaviours. Therefore,
for the targeted food and increased consumption of snacks we developed the Behavioural Mealtime Coding System
(Vaughn et al., 2016). (BMCS) to code child food exploration, child food accep-
Evidence-­based strategies to improve dietary variety tance by category and caregiver strategy use. A secondary
without using coercion include repeated exposure to foods aim of this study was to test the interrater reliability of the
(Anzman-­Franca et al., 2012), non-­food-­based positive rein- BMCS. Reliably describing strategies that caregivers are cur-
forcement or praise (Cooke et al., 2011) and modelling of rently using in the home environment may facilitate the de-
healthy eating behaviours (Fries & Van der Horst, 2019). sign of family-­centred interventions that address the needs of
Repeated exposure to novel foods in the absence of pres- caregivers as well as children with selective eating patterns.
sure to try the food has been shown to improve intake and
preference for those foods among toddlers and preschool-
ers (Anzman-­Frasca et al., 2012; Spill et al., 2019). Verbal 2 | M ETHOD
praise for consuming healthy foods is also associated with
higher intake of these foods (Luchini et al., 2017), without 2.1 | Participants and design
the negative effects of food-­based reinforcement strategies.
Access to positive role models during mealtime is also as- This observational study is a secondary analysis of data col-
sociated with increased acceptance of new foods and in- lected as part of a larger prospective pilot study (Caldwell
take of healthier foods in children (Scaglioni et al., 2018). et al., 2020). We recruited a convenience sample of 21
Essentially, caregivers should strive to maintain a responsive children (aged 18–­ 60 months) and their caregivers from
feeding environment that supports the child's hunger cues and southwestern Pennsylvania in the United States of America
ensures a pleasant and structured feeding environment with through primary care offices, early intervention programmes
few distractions (Black & Aboud, 2011; Satter, 2007). These and online (social media) advertisements. Eligible child par-
environments tend to promote healthy eating behaviour and ticipants met all four criteria for SFA as determined by an
are inversely associated with selective eating in children occupational therapist: (a) refuses foods based on sensory
(Finnane et al., 2017). characteristics; (b) onset of food refusal occurs when novel
It is important to understand the relationships among foods are introduced; (c) eats preferred foods without dif-
caregiver mealtime strategies and child behaviours to facili- ficulty and (d) food refusal causes risk for nutritional defi-
tate positive mealtime experiences for the entire family (e.g., ciencies OR oral motor delay (Zero to Three, 2005). Clinical
improved communication, decreased stress and increased confirmation of SFA was determined using a structured
food exploration). From an occupational perspective, we are mealtime observation and an in-­depth caregiver interview.
338
| CALDWELL AnD KRAUSE

TABLE 1 Behavioural mealtime coding system (BMCS)

Code Definition Example

Mealtime duration Begins when child is seated at table or is provided with food and ends Child sits at table during meal for
when the child walks away from food or stops participating 20 min
Child food acceptance
Bites preferred Child (or caregiver) places a bite of food categorised as preferred into Child eats a bite of cracker (preferred)
child's mouth and does not spit out
Bites targeted Child (or caregiver) places a bite of food categorised as targeted into Child eats a bite of broccoli (targeted)
child's mouth and does not spit out
Bites other Child (or caregiver) places a bite of food categorised as other into child's Child eats a bite of bread (other)
mouth and does not spit out
Child food exploration
Touch Any event when the child's hand or finger comes in contact with a food Child pokes a piece of broccoli with
item, except when self-­feeding with hands finger
Food play Any event when the child playfully interacts with their food Child uses broccoli as ‘paintbrush’ to
draw with ranch dressing
Lick Any event when the child's tongue comes in contact with a food item Child licks broccoli
without consumption
Caregiver strategy use
Labelled praise Caregiver provides a positive evaluation of a specific attribute, product ‘You did a great job trying a bite of
or behaviour of the child broccoli!’
Unlabeled praise Caregiver provides a positive evaluation of the child, an attribute of the ‘Great job, buddy!’
child, a nonspecific behaviour or product of the child
Generic positive Caregiver verbalises a response (other than praise) to a desired behaviour ‘Yummy!’ (in response to child taking
reinforcement in a positive manner and tone a bite)
Direct command Caregiver uses a declarative statement that contains an order or direction ‘Go sit down and eat’
to a child to perform a specific mealtime behavior
Threat Caregiver communicates an intent to inflict loss on child if they do not ‘If you don't start eating, I'm turning off
perform a specific behavior the TV’
Bribe Caregiver persuades a child to act in a certain way by means of an ‘I'll let you go outside if you take 3
outside inducement more bites.’
Feeding child Caregiver places food into child's mouth either by hand or using a fork Caregiver places a small piece of
or spoon broccoli into the child's mouth

Eligible caregiver participants were fluent in English, aged in the space where the child ate most frequently. For example,
18 years or older and a biological parent or legal guardian of if the child ate dinner on the floor while watching television,
the child participant. There were no specific exclusion crite- the camera was positioned to capture meals on the floor. To
ria. All participants who completed the baseline phase of the establish a baseline of mealtime behaviour and food accept-
pilot study were included in this analysis. This study was ap- ance under normal circumstances, caregivers were instructed
proved by the University of Pittsburgh Institutional Review to video-­record 10 typical meals in the home over a 2-­week
Board (PRO15060533). period. Instructions included turning the camera on at the be-
ginning of the meal and stopping after the child was no longer
actively participating in mealtime. All video data were stored
2.2 | Data collection procedure locally on microSD cards, and caregivers completed a meal
log of all meals recorded. This meal log included foods offered
Caregivers were provided with a video camera and collabo- to the child during each meal to assist with coding. Caregivers
rated with a member of the research team to set it up in an were educated on the purpose of baseline data collection and
unobtrusive position within their home. Movement patterns were encouraged to continue offering meals to their child as
of the family, daily routines and caregiver preferences were they had prior to entering the study. Demographic data, such
considered when identifying the best camera location for each as gender, race, education and household income, were also
family. The camera was positioned to capture ‘typical’ meals collected during this initial visit.
CALDWELL AnD KRAUSE
| 339

2.3 | Development of a video coding system of novel or non-­preferred foods (Chatoor, 2009). Therefore,
they have limited exposure and lessened ability to learn about
Our research team (principal investigator and two occu- the sensory characteristics of new foods, both of which have
pational therapy graduate student research assistants) de- been shown to improve intake of novel foods (Coulthard
veloped a video coding scheme to quantify outcomes of & Sealy, 2017; Remington et al., 2012). Child behaviours
interest including child food acceptance, child food ex- coded as food exploration included touching, playing with
ploration and caregiver use of behavioural management and licking food (see Table 1).
strategies (Table 1). Development of the coding system
was influenced by a literature review of commonly used
caregiver strategies to manage child mealtime behaviours 2.3.3 | Caregiver strategy use
and established mealtime and child-­caregiver interaction
coding schemes. Behavioural management strategies were coded for all car-
The BMCS was developed within the Observer® XT soft- egivers on the video who consented to participate in our
ware from Noldus. The Observer® XT allowed us to spec- study. Current evidence suggests that frequently used car-
ify subjects, behaviours and modifiers before or during an egiver strategies to improve consumption of novel foods
observation. Each behaviour coded in our system was rated include positive reinforcement, verbal praise, bribes, threats
by frequency of events or number of times coded per meal. and force feeding the child (Cooke et al., 2011; Fries & Van
Additionally, this system allowed us to code observations der Horst, 2019; Podlesak et al., 2017). The Dyadic Parent-­
by means of keystrokes or mouse clicks; each assigned to a Child Interaction Coding System (DPICS: Eyberg, 2013)
timestamp to allow for advanced specificity. The Observer® influenced the way in which we defined caregiver strategy
XT allows coders to keep notes within the software and add use to increase objectivity. Review of this established coding
or suggest new codes in real time based on mealtime ob- system also informed our decision to categorise verbal praise
servation. Mealtime duration was also calculated using the into labelled praise (e.g., great job trying a new food), unla-
Observer® XT software based on the length of the mealtime beled praise (e.g., nice work today) and generic positive re-
video observation. inforcement (e.g., tasty!) as different types of verbal positive
reinforcement. We also decided to add direct commands as
potential caregiver behaviour management strategies based
2.3.1 | Child food acceptance on our review of DPICS (see Table 1).

Bites of food accepted by children during video-­recorded

meals were categorised as preferred, targeted or other prior 2.4 | Video coding procedure
to coding. In the pilot study, caregivers collaborated with an
occupational therapist to develop a list of preferred foods or Using a random number list to minimise selection bias, we
those accepted >90% of the time when offered. They also de- selected three of a possible 10 baseline meals per participant
veloped a list of targeted foods or those the child refused out- to be coded for this study, totaling 63 videos. Each family
right or spit out after one bite per caregiver report. Targeted recorded at least eight baseline meals that demonstrated ad-
foods were healthy foods that were offered frequently dur- equate quality and were included in a list to be randomised
ing mealtimes and prioritised as those caregivers would like for coding. We chose to code three meals (described by
their child to accept more frequently. The therapist asked families as ‘typical’) based on methods employed in simi-
questions about the child's food preferences as well as fam- lar studies (Spieth et al., 2001; Stark et al., 2000). A trained
ily meal preparation and routines to facilitate brainstorming graduate research assistant was responsible for coding all 63
and identify potential foods for each list. Any food that was video-­recorded meals. A second trained rater (undergradu-
offered during video-­recorded meals that did not appear on ate research assistant) coded one third of the videos (21) to
that specific child's targeted or preferred food list was cat- determine inter-­rater reliability of the coding scheme. As part
egorised as other. Bites of food were coded when a piece of of the training procedure, three sample videos were reviewed
food entered the oral cavity of the child and remained inside and coded by both raters; any inconsistencies observed were
the mouth (i.e., the child did not spit it out; see Table 1). discussed by both raters under the guidance of the principal
investigator until 100% agreement was reached for all dis-
puted codes. Additionally, to improve coding transparency,
2.3.2 | Child food exploration each behaviour was followed by a note. For example, if a
caregiver vocalised a direct command, the coder would key
Child food exploration was also coded because children ‘direct command’, followed by a caregiver quotation such as
with SFA are often unable to tolerate sensory exploration ‘Go sit down and eat’.
340
| CALDWELL AnD KRAUSE

2.5 | Data analyses TABLE 2 Demographics

n (%)
Inter-­rater reliability of all behaviours included in the
Child characteristics (n = 21)
BMCS (see Table 1) was determined by calculating the in-
Gender
traclass correlation coefficient (ICC) and 95% confident in-
tervals based on a single-­rating (2,1), absolute agreement, Male 16 (76.1)
two-­way random-­effects model (Koo & Li, 2016). Prior to Female 5 (23.8)
additional analyses, participant data from the three coded Age (months)
meals were collapsed into an average per meal total to en- 18–­35 15 (71.4)
sure independence of observations, so that each participant 36–­60 6 (28.5)
was only represented in the analyses one time. Descriptive Race/Ethnic group
statistics were used to explore and describe observed meal- Non-­Hispanic white 20 (95.2)
time behaviours. Bivariate correlations were calculated to
Diagnosis
identify potential associations between caregiver and child
Autism 1 (4.8)
behaviour to inform hypotheses about how caregiver strat-
egy use may influence child mealtime behaviours. Finally, Gastrointestinal issues 5 (23.8)

we calculated bivariate correlations between demographic Sensory processing 1 (4.8)

variables likely to influence mealtimes (i.e., child age, Caregiver characteristics (n = 39)
household income and caregiver education) and caregiver Gender
and child behaviour to identify potential confounders. All Male 18 (46.2)
statistics were computed using IBM SPSS for Statistics, Female 21 (53.8)
version 26. Age (y)
Range = 27–­56

3 | R E S U LTS
Mean Female
Mean Male
34
37

Child participants (n = 21) ranged in age from 18 to Race/Ethnic group

54 months and were predominantly male (76.1%) and non-­ Non-­Hispanic white 37 (94.9)
Hispanic white (95.2%). Per caregiver report, five children Highest education completed
had gastrointestinal issues, one had a diagnosis of Autism High school 1 (2.6)
and one had Sensory Processing Disorder; all children met Associates/Vocational 2 (5.1)
the diagnostic criteria for SFA. Of 39 caregiver participants, Bachelors 19 (48.7)
21 were females (mean age = 37 years) and 18 were males Graduate 16 (40.9)
(mean age = 34 years). Two caregivers reported that their
Employment status
children were currently receiving occcupational therapy
Employed full-­time 24 (61.5)
services, and one reported their child received occupational
therapy services in the past (all due to concerns with feeding Employed part-­time 3 (7.7)
and sensory processing). No children enrolled in our study Working at home 4 (10.3)
demonstrated swallowing difficulties or motor impairments Retired 1 (2.6)
that would influence mealtime behaviours. One third (7) of Unemployed 2 (5.1)
the children in this study were positioned in a highchair dur- Household Income ($)
ing mealtimes. Caregiver education level varied across fami- 50,000–­75,000 3 (14)
lies with most caregivers earning their bachelor's degree or 75,001–­100,000 6 (29)
higher (87.1%). Most caregiver participants were employed
>100,000 12 (57)
full-­time (61.5%), and most households earned an income
greater than $100,000 (57%; see Table 2).
3.2 | Descriptives

3.1 | Inter-­rater reliability of BMCS Among our sample, average mealtime duration for children
was 17.5 min (SD = 5 min), with a range from 8 to 26 min
Inter-­rater reliability of the BMCS was found to be excellent average mealtime duration per participant. Food categorisa-
for absolute agreement between our two raters with an ICC tion varied greatly by child; however, foods frequently cat-
of .95 (95% confidence interval = .94–­.96). egorised as targeted included proteins, vegetables and fruit
CALDWELL AnD KRAUSE
| 341

(e.g., grilled chicken, carrots and apples), whereas foods caregiver behaviour management strategies (see Table 4).
frequently categorised as preferred included starches and The only behavioural variable significantly correlated with
dairy products (e.g., cheese, crackers and bread). Children demographic characteristics was caregiver use of threats,
accepted, on average, 26 bites of food into their mouths per which demonstrated a positive association with child age
meal (SD = 15.7 bites), ranging from an average of 2–­61 bites (rs = .48, p =.03) and a negative association with caregiver
accepted during meals per participant. On average, most bites education status (rs = −.49, p = .03).
were of preferred foods (M = 18.2, SD = 13.4), and children
accepted a much lower number of targeted foods per meal
(M = 1.5, SD = 2.5). Children most frequently explored food 4 | DISCUSSION
using touch (M = 3.2, SD = 3.1; see Table 3). Inappropriate
child behaviour codes of throwing food (M = .4) and spitting This study provides clinicians and researchers with insights
(M = .1) were used infrequently, with each code used for about child food acceptance (targeted and preferred), child
only 8 of the 21 children in our sample. food exploration and caregiver strategy use among children
In terms of strategy use, caregivers used direct commands with SFA. It has also identified mealtime duration and child
more than any other strategy, with an average of nearly nine food exploration as potential predictors of targeted food ac-
direct commands per meal (M = 8.8, SD = 6.6). While care- ceptance within this population. We developed a reliable
givers were observed to use different forms of positive rein- observational coding system to quantify child food accept-
forcement, they used generic positive reinforcement (M = 1.3, ance, child food exploration and caregiver behaviour man-
SD = 1.1) and unlabeled praise (M = 1.2, SD = 1.1) more fre- agement strategy use. Most food accepted during meals fell
quently than labelled praise (M = .4, SD = .5). As it relates to into the category of preferred, while children accepted very
coercive caregiver strategies, bribes (e.g., if you eat one bite few, if any, bites of targeted foods. Children in our sample
of chicken, you can have a cookie; M = .8, SD = 1.0) were were most likely to explore food during meals by touching it,
used more frequently than threats (e.g., no hockey tonight if but unlikely to engage in food play. Lack of food play repre-
you don't finish your dinner; M = .5, SD = .9). Finally, al- sents a meaningful target for occupational therapy interven-
though all children within this sample demonstrated the abil- tion; sensory play, in combination with repeated exposure,
ity to self-­feed, caregivers were observed to feed their child has been linked to improved acceptance among preschoolers
(by placing food into the child's mouth) 3.5 times per meal, (Coulthard & Sealy., 2017).
on average (SD = 7.0; see Table 3). Caregivers most frequently used direct commands and
feeding their child to encourage food acceptance and appro-
priate mealtime behaviour. Mealtime duration was positively
3.3 | Associations associated with food acceptance in all categories, indicating
that longer mealtimes may influence child acceptance of
Child bites of preferred and other food were positively and preferred and targeted foods. A child exploring food by lick-
significantly associated with mealtime duration (rs = .61, ing it was also correlated with acceptance of targeted foods.
p =.004; rs = .44, p =.05) and total bites of food consumed It is important to note that in this study, targeted foods were
(rs = .81, p <.001; rs = .59, p =.005). Child bites of targeted foods that caregivers offered frequently, so children were
foods were positively and significantly associated with meal- likely exposed to these foods in the past. Although we did
time duration (rs = .51, p =.02) and events of the child licking not measure prior exposure to targeted foods, repeated ex-
food (rs = .57, p =.007). Notably, there were no significant posure may have influenced child willingness to try these
associations observed between child food acceptance and foods in our sample.

T A B L E 3 Average frequency of
Child (n = 21) Caregiver (n = 39)
caregiver and child behaviours during
Behaviour Mean SD Behaviour Mean SD mealtime

Total bites 25.65 15.67 Labelled praise .40 .49

Bites preferred 18.21 13.35 Unlabelled praise 1.17 1.08
Bites targeted 1.48 2.52 Positive reinforcement 1.33 1.14
Bites other 7.00 8.47 Direct command 8.76 6.58
Touch 3.22 3.09 Threat .49 .85
Food play .16 .27 Bribe .75 1.03
Lick 1.44 2.31 Feeding child 3.46 6.99
342
| CALDWELL AnD KRAUSE

TABLE 4 Associations with child food acceptance of healthy targeted foods (e.g., fruits, vegetables and lean
Preferred Targeted Other proteins) to facilitate wider dietary variety and not simply
bites bites bites increased consumption of preferred foods. We would sug-
gest that occupational therapy clinicians increase mealtime
Mealtime duration .60** .51* .44*
duration by incorporating positive strategies such as a regu-
Child food exploration
lar mealtime routines and play-­based exploration to mitigate
Touch .30 .21 .31
negative mealtime experiences (Caldwell et al., 2020). Future
Food play −.09 .09 .18 experimental research with larger samples is warranted to
Lick .10 .57** .25 parse out the complex relationships among mealtime dura-
**
Total bites .81 .13 .59** tion, child behaviour and child food acceptance during early
Caregiver strategy use childhood.
Labelled praise −.13 .09 −.39 Our analyses also revealed a seemingly obvious associa-
Unlabelled praise .06 .19 .08 tion between total bites accepted by the child and the number
Generic positive .08 .13 .19
of bites accepted of preferred and other foods. Interestingly,
reinforcement the number of accepted bites of targeted foods was not sig-
Direct command −.03 .33 .21
nificantly correlated with the number of total bites. This
signals that intervention to increase acceptance of targeted
Threat −.28 .31 .18
food may not impact overall food acceptance. This is a nota-
Bribe −.09 .32 .06
ble finding because it suggests that intake of targeted foods
Feeding child −.03 .20 .28
may not have a significant impact on total consumption, and
Significant associations are highlighted in bold text. consequently caloric intake. Therefore, interventions actively
* p <.05.; ** p <.001. seeking to alter a child's weight (either due to an underweight
or overweight status) may be more likely to have an effect
The relationship between mealtime duration and food if the focus is placed on altering intake of foods considered
acceptance, specifically for bites of targeted foods, is likely preferred or neutral. However, the importance of increasing
complex. The average mealtime duration within this sample acceptance of healthy foods and promoting a wide dietary
(17.5 min per meal) was notably shorter than those reported variety early in a child's life cannot be understated, as prefer-
among young children (ranging from 19 to 26 min; Adamson ences established during this sensitive period of development
et al., 2015; Crist & Napier-­Phillips, 2001). This is likely due are known to persist (Mascola et al., 2010).
to unique manifestations of SFA, including the presence of The observational nature of this study with data collection
inappropriate mealtime behaviours, inability to tolerate sen- in the home environment strengthens the internal validity of
sory characteristics of foods and overall food refusal when our findings. The development of an objective coding system
not offered preferred foods (Chatoor, 2009). Young children with excellent inter-­rater reliability that includes child food
with feeding disorders have been observed to demonstrate exploration and the specification of preferred versus targeted
the extremes of mealtime duration among peers, falling ei- food acceptance is a notable contribution of this work. Future
ther well below or well above average mealtime duration of studies are needed to establish content validity and determine
children without feeding disorders (Adamson et al., 2015). concurrent validity of the BMCS with more established cod-
Therefore, it is likely that both meals that are too short (hence ing systems. Our ability to average data over multiple meals
limiting opportunities for exposure) and those that are too for each participant represents another strength of this study
long (with the potential to increase caregiver stress) are prob- and enhances the reliability of our results. The potential for
lematic. This is important to consider, as caregivers of young social desirability bias is important to acknowledge, as care-
children with feeding disorders already demonstrate higher givers may have acted differently due to the presence of the
levels of stress than caregivers of children without feeding camera and videorecording of meals. We limited this bias
disorders (Fishbein et al., 2016). through unobtrusive camera positioning and limiting contact
Based on existing evidence and the results of our analy- between the research team and participants during this base-
ses, we recommend tracking mealtime duration for children line period. This study provides novel insights to clinicians
with SFAs to inform treatment planning. If a child's meal- about the mealtime practices and behaviours of young chil-
times are consistently falling below average, examining po- dren with SFA and their caregivers.
tential causes of short duration (such as inappropriate child This study also had several limitations that are import-
behaviour) and intervening to address problems and increase ant to acknowledge. To facilitate the generalisability of our
mealtime duration may be a promising strategy to improve results to the real world, we did not recommend any stan-
acceptance of targeted foods. When increasing mealtime du- dardisation of mealtime procedures across families, which
ration, it is important to incorporate exposure and exploration presents a threat to internal validity. Choosing not to exclude
CALDWELL AnD KRAUSE
| 343

children who had received feeding therapy may have influ- CONFLICT OF INTER EST
enced both child and caregiver behaviour and is therefore a The authors have no conflict of interest to declare.
limitation of this study. Because mealtime routines are in-
fluenced by income, race and cultural values (Horodynski AUTHOR CONTRIBUTIONS
et al., 2010), our small and homogenous sample limits ex- All authors listed meet the four criteria of the International
ternal validity. It is promising that sociodemographic factors Committee of Medical Journal Editors (ICMJE), and no qual-
were not found to be associated with coded child behaviours, ified authors have been omitted.
but additional research with more diverse samples is needed.
Caregiver ratings of child mood and targeted food acceptance DATA AVAILA BILIT Y STATEMEN T
over time are also recommended measures for future trials The data that support the findings of this study are available
to examine potential confounders and the robustness of the from the corresponding author upon reasonable request and
BMCS. Finally, the results reported are correlational and do with approved data use agreement.
not support causal inferences; they should be used to generate
hypotheses and inform the design of future clinical trials. ORCID
Angela R. Caldwell https://orcid.
org/0000-0003-2057-2260
5 | CO NC LU S ION
T WITTER
Using a reliable observational coding system, we determined Angela R. Caldwell @AReneeCaldwell
that mealtime duration and child food exploration were asso-
ciated with child food acceptance within our sample. This de- R E F E R E NC E S
scriptive study provides a glimpse into mealtimes within the Adamson, M., Morawska, A., & Wigginton, B. (2015). Mealtime du-
home environment for young children who demonstrate sen- ration in problem and non-­problem eaters. Appetite, 84, 228–­234.
sory aversions to food. The BMCS is a novel coding scheme https://doi.org/10.1016/j.appet.2014.10.019
Anzman-­Frasca, S., Savage, J. S., Marini, M. E., Fisher, J. O., & Birch,
that may be used to code child and caregiver mealtime be-
L. L. (2012). Repeated exposure and associative conditioning pro-
haviours for children with limited dietary variety in future
mote preschool children’s liking of vegetables. Appetite, 58(2), 543–­
studies. We are currently examining the effects of enhancing 553. https://doi.org/10.1016/j.appet.2011.11.012
child meals by systematically coaching caregivers to incor- Black, M. M., & Aboud, F. E. (2011). Responsive feeding is em-
porate structured routines, positive reinforcement and food bedded in a theoretical framework of responsive parenting. The
exploration and play daily within the home environment. Journal of Nutrition, 141(3), 490–­494. https://doi.org/10.3945/
Future studies are needed to specify the observed relation- jn.110.129973
ship between mealtime duration and food acceptance and to Caldwell, A. R., Skidmore, E. R., Bendixen, R. M., & Terhorst, L.
(2020). Examining child mealtime behavior as parents are coached
determine the optimal range of mealtime duration to promote
to implement the Mealtime PREP intervention in the home: Findings
targeted food acceptance among children with SFA.
from a pilot study. British Journal of Occupational Therapy, 83(10),
631–­637. https://doi.org/10.1177/03080​22620​920086
Carruth, B. R., Ziegler, P. J., Gordon, A., & Barr, S. I. (2004). Prevalence
K E Y P O IN TS FO R OC C U PAT IONAL of picky eaters among infants and toddlers and their caregivers’ de-
TH ER A P Y cisions about offering a new food. Journal of the American Dietetic
Association, 104, 57–­64. https://doi.org/10.1016/j.jada.2003.10.024
• The BMCS is a reliable tool for describing mealtime Chatoor, I. (2009). Diagnosis and treatment of feeding disorders in in-
fants, toddlers, and young children. Zero to three.
behaviours.
Cooke, L. J., Chambers, L. C., Añez, E. V., & Wardle, J. (2011).
• Mealtime duration and food exploration are correlated Facilitating or undermining? The effect of reward on food accep-
with targeted food acceptance among children with SFA. tance. A Narrative Review. Appetite, 57(2), 493–­497. https://doi.
• Targeted food acceptance was not correlated with overall org/10.1016/j.appet.2011.06.016
food acceptance. Coulthard, H., & Sealy, A. (2017). Play with your food! Sensory play is
associated with tasting of fruits and vegetables in preschool children.
ACKNOWLEDGEMENTS Appetite, 113, 84–­90. https://doi.org/10.1016/j.appet.2017.02.003
Crist, W., & Napier-­Phillips, A. (2001). Mealtime behaviors of young
The authors would like to acknowledge Alison Juris for her
children: A comparison of normative and clinical data. Journal of
early contribution to the development of the coding system
Developmental and Behavioral Pediatrics, 22(5), 279–­286. https://
and Emily Haus for her efforts in video coding to determine doi.org/10.1097/00004​703-­20011​0000-­00001
inter-­rater reliability. We would also like to thank the support Dickstein, S., Seifer, R., Hayden, L. C., Schiller, M., Sameroff, A. J.,
staff at Noldus for their early assistance in designing our cod- Keitner, G., & Magee, K. D. (1998). Levels of family assessment: II.
ing scheme within the Observer® XT software. Impact of maternal psychopathology on family functioning. Journal
344
| CALDWELL AnD KRAUSE

of Family Psychology, 12(1), 23. https://psycn​et.apa.org/doi/10.103 the ages of 3 and 6 years. The Open Nutrition Journal, 8(1), 13–­18.
7/0893-­3200.12.1.23 https://doi.org/10.2174/18742​88201​40801​0013.
Dovey, T. M., Staples, P. A., Gibson, E. L., & Halford, J. C. (2008). Remington, A., Anez, E., Croker, H., Wardle, J., & Cooke, L. (2012).
Food neophobia and ‘picky/fussy’eating in children: A re- Increasing food acceptance in the home setting: A randomized con-
view. Appetite, 50(2–­3), 181–­193. https://doi.org/10.1016/j. trolled trial of parent-­administered taste exposure with incentives.
appet.2007.09.009 The American Journal of Clinical Nutrition, 95(1), 72–­77. https://
Eyberg, S. (2013). Dyadic Parent-­ Child Interaction Coding System doi.org/10.3945/ajcn.111.024596
(DPICS): Comprehensive manual for research and training, 4th ed. Samuel, T. M., Musa-­Veloso, K., Ho, M., Venditti, C., & Shahkhalili-­
PCIT International. Dulloo, Y. (2018). A narrative review of childhood picky eating
Finnane, J. M., Jansen, E., Mallan, K. M., & Daniels, L. A. (2017). and its relationship to food intakes, nutritional status, and growth.
Mealtime structure and responsive feeding practices are associ- Nutrients, 10(12), 1992. https://doi.org/10.3390/nu101​21992
ated with less food fussiness and more food enjoyment in children. Satter, E. (2007). Eating competence: Definition and evidence for the
Journal of Nutrition Education and Behavior, 49(1), 11–­18.e1. Satter Eating Competence model. Journal of Nutrition Education
https://doi.org/10.1016/j.jneb.2016.08.007 and Behavior, 39(5), S142–­ S153. https://doi.org/10.1016/j.
Fishbein, M., Benton, K., & Struthers, W. (2016). Mealtime disrup- jneb.2007.01.006
tion and caregiver stress in referrals to an outpatient feeding clinic. Scaglioni, S., De Cosmi, V., Ciappolino, V., Parazzini, F., Brambilla,
Journal of Parenteral and Enteral Nutrition, 40(5), 636–­645. https:// P., & Agostoni, C. (2018). Factors influencing children's eating be-
doi.org/10.1177/01486​07114​543832 haviours. Nutrients, 10(6), 706. https://doi.org/10.3390/nu100​60706
Fries, L. R., & van der Horst, K. (2019). Parental feeding practices Spieth, L. E., Stark, L. J., Mitchell, M. J., Schiller, M., Cohen, L. L.,
and associations with children's food acceptance and picky eating. Mulvihill, M., & Hovell, M. F. (2001). Observational assessment
Nestle Nutrition Institute Workshop Series, 91, 31–­39. https://doi. of family functioning at mealtime in preschool children with cystic
org/10.1159/00049​3676 fibrosis. Journal of Pediatric Psychology, 26(4), 215–­224. https://
Galloway, A. T., Fiorito, L. M., Francis, L. A., & Birch, L. L. (2006). doi.org/10.1093/jpeps​y/26.4.215
'Finish your soup': Counterproductive effects of pressuring children Spill, M. K., Johns, K., Callahan, E. H., Shapiro, M. J., Wong, Y. P.,
to eat on intake and affect. Appetite, 46(3), 318–­323. https://doi. Benjamin-­Neelon, S. E., Birch, L., Black, M.M., Cook, J. T., Faith,
org/10.1016/j.appet.2006.01.019 M. S., & Casavale, K. O. (2019). Repeated exposure to food and
Horodynski, M. A., Stommel, M., Brophy-­Herb, H. E., & Weatherspoon, food acceptability in infants and toddlers: A systematic review. The
L. (2010). Mealtime television viewing and dietary quality in low-­ American Journal of Clinical Nutrition, 109, 978S–­989S. https://
income African American and Caucasian mother–­toddler dyads. doi.org/10.1093/ajcn/nqy308
Maternal and Child Health Journal, 14(4), 548–­556. https://doi. Stark, L. J., Jelalian, E., Powers, S. W., Mulvihill, M. M., Opipari, L.
org/10.1007/s1099​5-­009-­0501-­2 C., Bowen, A., Harwood, I., Passero, M. A., Lapey, A., Light, M., &
Jansen, P. W., de Barse, L. M., Jaddoe, V. W., Verhulst, F. C., Franco, O. Hovell, M. F. (2000). Parent and child mealtime behavior in families
H., & Tiemeier, H. (2017). Bi-­directional associations between child of children with cystic fibrosis. The Journal of Pediatrics, 136(2),
fussy eating and parents' pressure to eat: Who influences whom? 195–­200. https://doi.org/10.1016/S0022​-­3476(00)70101​-­6
Physiology & Behavior, 176, 101–­106. https://doi.org/10.1016/j. Toyama, H., & Agras, W. S. (2016). A test to identify persistent picky
physb​eh.2017.02.015 eaters. Eating Behaviors, 23, 66–­ 69. https://doi.org/10.1016/j.
Koo, T. K., & Li, M. Y. (2016). A guideline of selecting and reporting eatbeh.2016.07.003
intraclass correlation coefficients for reliability research. Journal of Vaughn, A. E., Ward, D. S., Fisher, J. O., Faith, M. S., Hughes, S.
Chiropractic Medicine, 15(2), 155–­163. https://doi.org/10.1016/j. O., Kremers, S. P. J., Musher-­Eizenman, D. R., O’Connor, T. M.,
jcm.2016.02.012 Patrick, H., & Power, T. G. (2016). Fundamental constructs in
Luchini, V., Musaad, S. M., Donovan, S. M., & Lee, S. Y. (2017). food parenting practices: A content map to guide future research.
Differences and agreement in perception of child picky eating Nutrition Reviews, 74(2), 98–­117. https://doi.org/10.1093/nutri​t/
among center-­and home-­based childcare providers and parents and nuv061
its impact on utilized mealtime strategies. Nutrition and Metabolic Zero to Three (2005). DC: 0–­3R. Diagnostic classification of mental
Insights, 10. https://doi.org/10.1177/11786​38816​684830 health and development disorders of infancy and early childhood.
Mascola, A. J., Bryson, S. W., & Agras, W. S. (2010). Picky eating during Zero to Three Press.
childhood: A longitudinal study to age 11years. Eating Behaviors, Zucker, N., Copeland, W., Franz, L., Carpenter, K., Keeling, L., Angold,
11(4), 253–­257. https://doi.org/10.1016/j.eatbeh.2010.05.006 A., & Egger, H. (2015). Psychological and psychosocial impairment
Podlesak, A. K., Mozer, M. E., Smith-­Simpson, S., Lee, S. Y., & in preschoolers with selective eating. Pediatrics, 136(3), e582–­e590.
Donovan, S. M. (2017). Associations between parenting style and https://doi.org/10.1542/peds.2014-­2386
parent and toddler mealtime behaviors. Current Developments in
Nutrition, 1(6), e000570. https://doi.org/10.3945/cdn.117.000570
Poppert, K. M., Patton, S. R., Borner, K. B., Davis, A. M., & Dreyer How to cite this article: Caldwell AR, Krause EK.
Gillette, M. L. (2015). Systematic review: Mealtime behavior mea- Mealtime behaviours of young children with sensory
sures used in pediatric chronic illness populations. Journal of Pediatric food aversions: An observational study. Aust Occup
Psychology, 40(5), 475–­486. https://doi.org/10.1093/jpeps​y/jsu117
Ther J. 2021;68:336–344. https://doi.
Ramos-­Paúl, R., J. Marriage, B., Ruiz Debeza, R., Oliveros Leal, L.,
Ros Mar, L., Torres Cardona, L., & Williams, J. A. (2014). Impact
org/10.1111/1440-­1630.12732
of picky eating on level of family stress in healthy children between