Professional Documents
Culture Documents
A Multidimensional Approach To Understanding The Potential Risk Factors and Covariates of Adult Picky Eating
A Multidimensional Approach To Understanding The Potential Risk Factors and Covariates of Adult Picky Eating
A Multidimensional Approach To Understanding The Potential Risk Factors and Covariates of Adult Picky Eating
Jordan M. Ellis, Rebecca R. Schenk, Amy T. Galloway, Hana F. Zickgraf, Rose Mary
Webb, Denise M. Martz
PII: S0195-6663(17)30836-X
DOI: 10.1016/j.appet.2018.01.016
Reference: APPET 3755
Please cite this article as: Ellis J.M., Schenk R.R., Galloway A.T., Zickgraf H.F., Webb R.M. & Martz
D.M., A multidimensional approach to understanding the potential risk factors and covariates of adult
picky eating, Appetite (2018), doi: 10.1016/j.appet.2018.01.016.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
PT
A Multidimensional Approach to Understanding the Potential Risk Factors and Covariates of
RI
Adult Picky Eating
SC
Jordan M. Ellisa, Rebecca R. Schenkb, Amy T. Gallowayb, Hana F. Zickgrafc, Rose Mary Webbb
U
AN
a
Department of Psychology, East Carolina University, East Fifth Street, 104 Rawl Building,
b
Department of Psychology, Appalachian State University, P.O. Box 32109, 222 Joyce Lawrence
D
c
Department of Psychology, University of Pennsylvania, 425 S. University Ave
This research was funded by the Creating a Healthy, Just and Sustainable Society Student
Research Grant awarded to the first author by the Appalachian State University Office of Student
Research.
ACCEPTED MANUSCRIPT
PT
RI
U SC
AN
M
D
TE
EP
C
AC
ACCEPTED MANUSCRIPT
Running Head: POTENTIAL RISK FACTORS OF ADULT PICKY EATING 1
PT
RI
U SC
AN
M
D
TE
C EP
AC
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 2
Abstract
Objective: Adult picky eating (PE) has received increased attention in the eating behavior
literature due to its important association with adult avoidant-restrictive food intake disorder
(ARFID). The current study tested a model of potential risk factors of adult PE behavior,
PT
including perceived early parental feeding practices. An exploratory model was also utilized to
RI
understand associations with different aspects of adult PE behaviors. Methods: A sample of
1,339 US adults recruited through Amazon’s MTurk completed an online survey that included
SC
the recently developed Adult Picky Eating Questionnaire (APEQ), retrospective reports of
parental feeding practices, and other measures of eating behavior and demographic variables. A
U
structural equation modeling procedure tested a series of regression models that included BMI
AN
and disordered eating behaviors as covariates. Results: SEM modeling indicated that
retrospective reports of greater parental pressure to eat, higher disgust sensitivity, lower PE age
M
of onset, and experiencing an aversive food event were associated with general adult PE
D
behavior. Results also indicated parental encouragement of healthy eating may be a protective
TE
factor, and that men endorsed higher levels of adult PE. Exploratory analyses indicated that
understanding of adult PE is important for the prevention and treatment of severe PE behaviors,
AC
through the avoidance or rejection of both familiar and unfamiliar foods (Dovey, Staples,
PT
Gibson, & Halford, 2008; Taylor, Wernimont, Northstone, & Emmett, 2015). PE that persists
RI
and causes significant distress or impairment related to the consequences of PE can now be
diagnosed as Avoidant/Restrictive Food Intake Disorder (ARFID), a new entry to the Diagnostic
SC
and Statistical Manual of Mental Disorders, fifth addition (DSM-5, APA, 2013). Most PE
research has focused on children, but with the recent addition of ARFID to DSM-5, there has
U
been a shift in focus to the understudied concern of adult PE. The ARFID diagnosis is intended
AN
to identify individuals from infancy to adulthood with clinically significant restrictive eating
problems that result in persistent failure to meet nutritional and/or energy needs and are not
M
driven by a desire to lose weight or associated with distorted body image. The DSM-5 describes
D
three types of eating disturbances that can lead to ARFID symptoms: PE, low appetite/interest in
TE
eating, and fear of negative consequences from eating, although other eating restrictions can lead
to an ARFID diagnosis (APA, 2013). Research relating to PE in adults and potential implications
EP
suggesting that some degree of adult PE is fairly common (Ellis, Galloway, Webb, & Martz,
C
2016; Zickgraf, Franklin, & Rozin, 2016). The purpose of this paper is to use a cross-sectional
AC
design to identify potential predictors of general PE and to explore differential predictors of four
aversion to bitter and sour tastes) of PE behavior in adulthood using a large non-clinical sample.
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 4
There has been extremely little research on the relationship between PE and the formal
ARFID diagnosis. However, PE has consistently been linked to nutritional, weight, and
psychosocial outcomes across the lifespan. PE is associated with slow growth, underweight, and
delayed development in children, and has been shown to have relationships with reduced fruit
PT
and vegetable consumption, functional constipation, internalizing and externalizing
RI
psychopathology symptoms, and increased family stress and conflict (Cano et al., 2016; Cooke,
Carnell, & Wardle, 2006; Coulthard & Blissett, 2009; Ekstein, Laniado, & Glick, 2010;
SC
Galloway, Fiorito, Lee, & Birch, 2005; Goh & Jacob, 2012; Jacobi, Schmitz, & Agras, 2008;
Mascola, Bryson, & Agras, 2010; Micali et al., 2011; Ramos-Paúl, Marriage, Debeza, Leal, &
U
Ros, 2014; Tharner et al., 2014; Zucker et al., 2015). In adulthood, PE has been associated with
AN
symptoms of depression, social eating anxiety, obsessive-compulsive disorder (OCD) symptoms,
psychological inflexibility, lower eating-related quality of life, reduced fruit and vegetable
M
consumption, and report of lower dietary variety. (Ellis et al., 2016; Kauer, Pelchat, Rozin, &
D
Zickgraf, 2015; Wildes, Zucker, & Marcus, 2012; Zickgraf et al., 2016; Zickgraf & Schepps,
TE
2016). PE has not been linked to clinically significant underweight in school-aged children or
adults, but PE appears to be a common precursor to ARFID symptoms. In eating disorder clinics
EP
ARFID is associated with underweight to the same extent as AN in these settings (e.g., Forman
et al., 2014). When PE is measured continuously, it is often negatively associated with BMI,
C
leading some to suggest that it might be protective against the development of overweight
AC
(Berge, Tate, Trofholz, Conger, & Neumark-Sztainer, 2016; Galloway et al., 2005).
Picky eating is common in both childhood and adulthood, and although it has consistently
been associated with negative nutritional, developmental, and psychosocial correlates, it is likely
that only higher levels of PE are associated with negative outcomes and ARFID symptoms. A
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 5
limited body of research in adolescents and adults indicates that individuals with severe PE or
comparable to those of individuals diagnosed with or screened at risk for an eating disorder
(Strandjord, Sieke, Richmond, & Rome, 2015; Wildes et al., 2012; Zickgraf et al., 2016).
PT
PE behavior includes food neophobia, sensory food aversions, and rejection of familiar
RI
foods, and is closely associated with rigidity around food preparation/presentation. Given the
SC
construct may shed more light on the relationships between PE behaviors and clinical outcomes.
Previous research has described some of the social, dietary, and clinical implications of adult PE
U
but has not investigated potential predictors of specific PE behaviors. The current study
AN
attempted to bridge the gap in the literature by exploring common and specific predictors of
Previous research suggests that there are several demographic variables that may relate to
D
the development of PE. Picky eaters are known to have younger mothers (Dubois, Farmer,
TE
Girard, Peterson, & Tatone-Tokuda, 2007; Hafstad, Abebe, Torgersen, & Soest, 2013; Tharner et
al., 2014), lower socioeconomic status as children (Dubois et al., 2007; Tharner et al., 2014), and
EP
to be less-educated as adults (Wildes et al., 2012). In two recent studies of children, parent-
reported picky eaters were more likely to come from non-Western backgrounds (Cano et al.,
C
2015; Tharner et al., 2014), although other studies have not found demographic differences
AC
among picky eaters (Carruth, Ziegler, Gordon, & Barr, 2004). Sex findings have been
inconsistent within the literature, some studies reporting no sex differences in PE among children
and adults (Dovey et al., 2008; Fisher et al., 2014; Jacobi et al., 2008; Kauer, Pelchat, Rozin, &
Zickgraf, 2015), but other findings suggest a higher prevalence of male picky eaters (Cano et al.,
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 6
2016; Tharner et al., 2014). In addition to demographics, there are other likely predictors of PE,
such as the extent to which picky eaters were pressured or forced to eat disliked food (Tharner et
al., 2014), or the degree to which they were encouraged to eat a variety of healthy food during
childhood (Steinsbekk, Sveen, Fildes, Llewllyn & Wichstrom, 2017; Vaughn et al., 2016).
PT
Moreover, negative experiences with food, such as choking or vomiting after eating, have been
RI
linked to ARFID symptoms (Fisher et al., 2014), and individual differences in sensitivity to
disgust may be associated with domains of adult PE (Dovey et al., 2008; Kauer et al., 2015).
SC
Given the mixed findings linking PE to outcomes, it is possible that various domains of
U
and ARFID, a more refined measure of PE that enables the identification of the common and
AN
unique predictors of distinct aspects of adult PE is needed (Ellis et al., 2016). Previous research
supports the expectation that lower income, lower educational levels, and younger maternal age,
M
will likely predict adult PE. In support of previous research, it was hypothesized that
D
retrospective reports of higher levels of parental pressure to eat and lower levels of parental
TE
encouragement to eat certain foods would also be associated with adult PE. Finally, it was
hypothesized that disgust sensitivity and recollections of early negative experience with food
EP
would be associated with overall PE. The model described below also investigated maternal and
paternal age, sex, race, and self-reported age of PE onset as potential predictors. In addition to
C
predictors of different aspects of PE behaviors. It should be emphasized that the aim of the study
was to examine potential predictors of adult PE, not ARFID. The models and measures
employed do not clearly distinguish between subclinical PE and eating behaviors that would
Methods
Participants
A sample of Mechanical Turk (MTurk) workers, from the US and over the age of 18,
were recruited to complete an online survey about adult development, with no indication the
PT
survey would be assessing eating behaviors. No exclusion criteria beyond age and US residence
RI
was employed. Amazon’s MTurk is an online interface for recruiting and paying participants to
complete tasks, and can be more representative of the US population than in-person convenience
SC
samples (Berinsky, Huber, & Lenz, 2012). The sample included 1,339 MTurk workers (804
women, 535 men; 80% White, 10% Black, 5% Asian, 3% Hispanic; mean age 40.39, SD =
U
13.39). The mean BMI of the sample was 27.67 (SD = 7.18) and 53.0% of the sample had
AN
completed at least a 4-year college degree (see Table 1). The university’s Institutional Review
Measures
D
factors, including age, sex, race, educational attainment, family income, and their mother’s and
father's ages at their birth. Participants were asked to self-report their height and weight, from
EP
which Body Mass Index (BMI) was calculated. Given that having an aversive food experience
has been associated with childhood PE (Fisher et al., 2014), and PE attitudes and behaviors
C
appear to manifest early in life (Mascola et al., 2010), a series of specific questions about PE
AC
were also presented. If participants answered "yes" to being unusually picky/selective about the
foods they eat, they were also asked, “At what age did you become a picky eater?” They were
also asked to respond “yes,” “maybe,” or “not applicable” to the question, “Before you became a
picky/selective eater, did you have a bad experience with food?” Participants were then asked if
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 8
they did have a bad experience involving food to provide an open-ended response describing the
experience.
measure that assesses multiple aspects of adult PE, and its factor structure was confirmed in the
PT
present sample (Ellis et al., 2016). The APEQ has fours subscales: Meal Presentation, which
RI
assesses rigid preferences around food preparation and presentation; Food Variety, which
assesses a restricted dietary range across foods and food groups; Meal Disengagement, which
SC
assesses avoidant behaviors around mealtimes; and Taste Aversion, which assesses the rejection
of bitter and sour foods. A composite score can also be calculated by averaging the 15-items, and
U
the measure is scored on a five-point Likert scale ranging from 1 = “Never” to 5 = “Always.”
AN
Parental Feeding Practices. A modified version of the "Pressure" and "Encourage
Balance and Variety" subscales from the Comprehensive Feeding Practices Questionnaire
M
(CFPQ; Musher-Eizenman & Holub, 2007) was utilized to assess retrospective parental feeding
D
practices. The CFPQ was developed as a multi-factor parent self-report measure that assesses a
TE
variety of practices used to feed their children, and has shown validity and adequate
psychometric properties across a range of age and ethnic groups (Melbye, Øgaard, & Øverby,
EP
parents feeding practices during childhood, items from the CFPQ were slightly modified for use
AC
with a college sample. Previous researchers have modified another measure of parental feeding,
the Child Feeding Questionnaire (CFQ; Francis, Hofer, & Birch, 2001), to assess child and
college student perceptions of parental pressure, but the CFQ does not assess encouragement for
healthful eating (Carper, Orlet Fisher, & Birch, 2000; Galloway, Farrow, & Martz, 2010);
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 9
therefore, we similarly modified the CPFQ for the current study. For example, one CPFQ item
asks parents to respond to “My child should always eat all of the food on his/her plate,” on a 5-
point Likert scale (1 = “Disagree” to 5 = “Agree”). The item in the current study was modified to
state, “Your parents believed you should eat all of the food on your plate.” The Pressure subscale
PT
measures the extent to which parents pressured the college student, as a child, to consume more
RI
food at meals, with higher scores indicating greater pressure. The Encourage Balance and
Variety Subscale measures the extent to which college students recall their parents promoting a
SC
well-balanced food intake, including encouraging the consumption of a variety of foods and
healthy foods. Higher scores indicate greater parent encouragement. Both the Pressure (α = .86)
U
and Encourage Balance and Variety (α = .87) subscales showed good internal consistency in the
AN
present sample.
Disgust Sensitivity. The Core Disgust subscale from the Disgust Scale was utilized to
M
measure disgust sensitivity (Rozin, Haidt, & McCauley, 2008; Olatunji et al., 2007). The Core
D
Disgust subscale is a 12-item subscale that is based on a general sense offensiveness and
TE
contamination towards a variety of disgust eliciting stimuli. The scale is rated on a 5-point Likert
repugnance.” The subscale has demonstrated good psychometric properties (Olatunji et al., 2007;
Van Overveld, de Jong, Peters, & Schouten, 2011). This subscale demonstrated satisfactory
C
Disordered Eating Symptoms. The Eating Disorder Diagnostic Scale (EDDS) was
utilized to assess disordered eating behaviors. It is a 22-item measure that has demonstrated
evidence for good reliability and validity and can be used to support clinical diagnoses, as it can
provide diagnostic cutoffs for specific eating disorders (Stice, Fisher, & Martinez, 2004). Given
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 10
that the present study did not include a clinical population, our analytic approach included
calculating a continuous eating disorder symptom composite score by standardizing item scores
and averaging them together to assess symptom severity, as opposed to using diagnostic cutoffs
(Stice, Telch, & Rizvi, 2000). The EDDS composite score demonstrated satisfactory internal
PT
consistency in the present study (α = .71).
RI
Data Analysis
SC
hypothesized predictors of adult PE. The first model assessed predictors of the composite APEQ
latent variable, which was identified within a second order model with 15-item indicators and the
U
four primary latent subscale factors. Latent variable predictors included pressure to eat (4-item
AN
indicators), encouragement to eat (4-item indicators), and core disgust (12-item indicators).
Manifest predictors included sex, race, BMI, the age at which PE developed, aversive experience
M
related to PE onset, family income, education, mother’s age at birth, and father’s age at birth.
D
Race was coded as a binary variable, representing majority (i.e., white) and minority subsets of
TE
the sample. Family income was coded as a binary variable to categorize participants into groups
with family income less than or greater than $50,000. Education was also coded as a
EP
dichotomous variable, capturing whether or not participants had attained a four-year degree or
greater, or less than a four-year degree. Having a bad experience related to PE onset was also
C
“maybe” versus those who replied, “not applicable.” Open ended responses from participants
who endorsed “yes” to having had a bad experience with food were coded as: 1) pressure to eat
weight concerns. While a causal model could not be proposed because the present data set is
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 11
cross-sectional and not longitudinal, the current literature suggests that treating most of the
predictor variables as temporally distinct is justified in that they likely occurred before the
current adult PE behaviors. BMI was included as a covariate to control for weight related
influences, but is not necessarily temporally distinct from PE. Finally, the EDDS was added as a
PT
stepwise covariate to examine and control for possible distortions caused by more traditional
RI
eating disorder symptomatology.
For the SEM procedure, the measurement model, including all of the latent variables, was
SC
estimated using confirmatory factor analysis procedures (CFA). A robust maximum likelihood
(MLR) estimator was employed for the measurement model, as all of the latent variables were
U
continuous data. Because ordinal predictors were included in the measurement model, a diagonal
AN
weighted least squares (DWLS) estimator was employed. The DWLS is a robust weighted least
squares estimator that can analyze both continuous and non-continuous variables and has
M
performed well in computer simulations (Muthén, du Toit, & Spisic, 1997). If the measurement
D
model demonstrated acceptable fit, the structural model was then estimated. Non-significant
TE
Model fit was assessed using the Comparative Fit Index (CFI), the Tucker Lewis Index
EP
(TLI), the root mean square residual (RMSEA) and the standardized root mean square residual
(SRMR). Values at or above .90 on the CFI and TLI have been advised as a cutoff that represents
C
an acceptable fitting model, and values at or above .95 represent a good fitting model. Values at
AC
or below .06 for the RMSEA and .08 for the SRMR further represent a good model fit (Hu &
Bentler, 1999).
The final SEM procedure included an exploratory approach to examine the relationship
between the predictors from the trimmed model and each of the APEQ latent subscale factors, to
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 12
assess the relationship between predictors and specific aspects of adult PE. The subscales were
Results
Correlations between the APEQ subscales and continuous predictor variables are
PT
presented in Table 2. The measurement model demonstrated acceptable model fit, χ2 (584) =
RI
1950.93, p < .001, CFI = .91, TLI = .90, RMSEA = .04, SRMR = 0.05, and loadings across all
latent factors were in an acceptable range (.36 - .92). Modification indices and residuals from the
SC
model were examined to determine if minor adjustments could further improve model fit. They
suggested that two items from the pressure subscale and two items from the encouragement
U
subscale were highly correlated. Item content within each item pair was similar, thus two
AN
residual covariance pathways were added to the measurement model. Model fit improved
following the modifications, χ2 (582) = 1733.90, p < .001, CFI = .92, TLI = .92, RMSEA = .04,
M
SRMR = 0.05. The structural model with all predictors of PE was then estimated and resulted in
D
a mediocre fit, χ2 (987) = 2336.74, p < .001, CFI = .89, TLI = .88, RMSEA = .05, SRMR = 0.07.
TE
Figure 1 presents the conceptual progression of models 1 through 3. As shown in Table 3 (see
Model 1), all of the statistically significant pathways were in the expected direction. However,
EP
race, education, income, mother’s age, and father’s age all fell short of significance as predictors
acceptable model fit, χ2 (722) = 1671.60, p < .001, CFI = .92, TLI = .91, RMSEA = .05, SRMR
AC
= 0.06. These results indicate that adult PE is positively associated with parental pressure to eat
in childhood, disgust sensitivity, the presence of a negative experience with food, and BMI. The
results also suggest that the earlier PE behaviors start, the more likely they are to persist into
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 13
adulthood, and encouragement to eat during childhood was negatively related to adult PE.
The EDDS was added in Model 3 as a covariate to control for possible distortions caused
by more traditional eating disorder symptomatology, χ2 (757) = 1793.74, p < .001, CFI = .92,
PT
TLI = .91, RMSEA = .05, SRMR = 0.06. All of the pathways except for BMI remained
RI
significant and of similar magnitudes. BMI was no longer significant with the addition of the
EDDS, but BMI was retained in the exploratory model to determine if different facets of PE
SC
were related to BMI.
U
model examining predictor relationships with each of the APEQ subscales also demonstrated
AN
acceptable model fit, χ2 (699) = 1364.14, p < .001, CFI = .92, TLI = .94, RMSEA = .05, SRMR
= 0.06. Standardized beta weights are presented in Table 4, Model 4. Results indicate that all
M
variables, except for BMI, were significant in predicting the Meal Presentation subscale, and in
D
the expected direction. All of independent variables significantly predicted the Food Variety
TE
subscale, in the expected directions. Results also showed that disgust sensitivity, age of PE onset,
and BMI were not significant predictors of the Meal Disengagement subscale. Finally, parental
EP
pressure to eat, sex, and age of PE onset were not significant predictors of the Taste Aversion
subscale. Parental encouragement to eat and the presence of a negative food experience were
C
Again, the EDDS was added into the model as a covariate to control for possible
distortions caused by more traditional eating disorder symptomatology, χ2 (731) = 1452.23, p <
.001, CFI = .94, TLI = .93, RMSEA = .04, SRMR = 0.06 (see Model 5 in Table 4). All of the
pathways remained significant and of similar magnitudes, except the relationship between Taste
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 14
Aversion and BMI became nonsignificant, suggesting that the only subscale significantly related
to BMI when controlling for disordered eating was the Food Variety subscale. Overall, these
exploratory results suggest that while certain aspects of PE share predictors, some predictors are
PT
Discussion
RI
Currently, there is a paucity of research on adult PE. Given recent findings that suggest
PE is related to significant psychosocial impairment and distress in adulthood (Ellis et al., 2016,
SC
Kauer et al., 2015, Zickgraf et al., 2016), it is important that researchers continue to work toward
a better understanding of how these eating behaviors may affect physical and mental well-being.
U
Moreover, it is important that researchers explore demographic and environmental factors that
AN
may contribute to the maintenance of PE from childhood to adulthood. The purpose of the
current study was to identify self-reported potential risk factors of adult PE and to examine
M
whether there are different predictors for specific PE behaviors and attitudes. Findings from this
D
study may also aid in understanding the etiology and maintenance of ARFID, an understudied
TE
eating disorder new to the DSM-5, as severe PE can be a core feature of the disorder in both
In this study, the predictive value of several demographic variables was tested, including
race, education, income, parent age, and participant sex for general PE and specific PE
C
behaviors. Of these, only sex was related to general adult PE, in that males were more likely to
AC
report higher PE behaviors related to Food Variety, Meal Presentation, and Meal
Disengagement. This conflicts with one previous study in adults that found women were more
likely than men to identify as severely picky (Zickgraf et al., 2016); although, many participants
in that study were recruited from an online support group for adult picky eaters and may not be
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 15
representative of the demographics of severe adult picky eaters. Interestingly, other studies of
children and adults report no sex differences in PE (Kauer et al., 2015; Fisher et al., 2014; Jacobi
et al., 2008); however, several reports have found that school-aged and younger boys are more
likely than girls to be picky eaters (Cano et al., 2015; Tharner et al., 2014). These inconsistent
PT
findings in the past may be attributed to differences in measurement of PE (i.e., maternal vs. self-
RI
report, single item measurement), particularly considering that different dimensions of PE have
not been previously considered. Surprisingly, several other demographic variables including
SC
race, education, income, and parent age were not associated with adult PE. It is possible that the
measure of participants’ current family income did not consistently represent childhood income
U
status, when this variable is presumed to have had its effect on adult eating behavior. Although
AN
some previous studies report that older maternal age and higher educational attainment may be
protective factors against the development of PE in childhood (Hafstad et al., 2013; Tharner et
M
al., 2014), results from this study did not replicate that finding for adults.
D
Body size as measured by BMI was included in each model as a covariate to account for
TE
weight-related influences on the APEQ subscales. In this study, BMI was not directly correlated
with overall PE or any of the aspects of PE; however, when entered in the multiple regression,
EP
higher BMI was a significant covariate with overall PE, suggesting a potential suppressor effect.
However, it was no longer significantly related when the EDDS was entered into the model. This
C
difference may be due to the higher prevalence of eating disorders in high-weight individuals
AC
(Duncan, Ziobrowski, & Nicol, 2017). BMI was differentially associated with the various aspects
of PE; higher BMI was positively associated with PE related to low Food Variety and negatively
associated with Meal Disengagement. In other words, participants with higher weight status were
more likely to eat from a limited selection of foods and participants with lower weight status
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 16
were more likely to report being disengaged with food. The relationship between BMI and Food
Variety may reflect the tendency for picky eaters to consume more palatable, energy-dense
foods; adult and child picky eaters have been found to be more likely to restrict their variety of
fruits and vegetables more than other food groups, and adult picky eaters are more likely to omit
PT
fruits and vegetables completely (Galloway et al., 2005; Zickgraf & Schepps, 2016). The
RI
negative relationship between Meal Disengagement and BMI might suggest that individuals who
are disengaged from meals consume fewer overall calories because they are avoiding all types of
SC
food rather than specific types of food. Future research should examine the dietary patterns of
individuals who endorse the various aspects of PE. It should be noted that the causal direction of
U
the relationship between adult PE and BMI is currently unclear. PE behavior could influence
AN
BMI, but it is also possible that early experiences, such as childhood food allergies, a history of
vomiting/choking, or gastrointestinal reflux (Fisher et al., 2014) might influence weight as well
M
Results from this study showed clear patterns in how parental pressure to eat and
TE
encouragement to eat certain foods in childhood is related to picky eating in adulthood. The
recollection of being pressured to eat as a child was associated with higher levels of PE across
EP
subtypes and the recollection of being encouraged to eat as a child was associated with lower
levels of PE across subtypes. Results from this study align with a recent prospective study
C
showing that parental feeding practices, such as the use of pressure and encouragement, may
AC
have long-term implications for the development of children’s eating behavior (Steinsbekk,
Belsky, & Wichstrom, 2016). However, because the current data are cross-sectional and
retrospectively reported, it cannot be concluded that parental feeding practices caused or even
preceded the development of PE. In fact, it is likely that the relationship between PE and parental
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 17
use of pressure and encouragement is bidirectional (Berge et al., 2016). Children who are picky
eaters are probably more challenging for parents to feed than children who are open to eating a
variety of foods, hence shaping parental behaviors over time. One cross-sectional study has
suggested that parents use pressure or encouragement in response to their child’s eating behavior
PT
and/or weight (Francis, Hofer, & Birch, 2001). Other findings indicate that the use of pressure
RI
has the potential to exacerbate PE (Galloway, Fiorito, Francis, & Birch, 2006) whereas the use of
encouragement has been shown to predict more positive eating behaviors in children (Ventura &
SC
Birch, 2008).
In a recent study, adult picky eaters scored higher on disgust sensitivity than non-picky
U
eaters (Kauer et al., 2015). Results from the current study corroborated this finding in general,
AN
but also indicated that disgust sensitivity is associated with particular aspects of PE. Perhaps this
represents a response to the characteristics of the food itself that may be explained by an
M
underlying biological predisposition of disgust sensitivity (Sherlock, Zietsch, Tybur, & Jern,
D
2016); however, more research is needed to understand these differential relationships. Early
TE
experiences are also thought to be important for shaping the development of eating behavior
(Dovey et al., 2008; Fisher et al., 2014). Results indicated that recollections of early negative
EP
experiences with food consistently predicted adult PE for all of the subtypes. It is also possible
that factors such as childhood PE, disgust sensitivity, and early aversive experiences with food
C
could be part of a single phenomenon associated with adult eating behavior. The effect of early
AC
aversive food experiences was stronger for Meal Presentation and Food Variety, suggesting that
early experiences, whereas others might more strongly reflect biological predisposition (e.g.,
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 18
Corroborating this hypothesis is that the age of PE onset was not associated with these
same two subtypes. It should be noted that the mean age of onset was 14 years of age, which was
PT
unexpected given that PE incidence has been reported as less than 5% per year after age six, and
RI
most studies that explore PE longitudinally, or cross-sectionally across age groups in childhood,
suggest that PE prevalence and severity decrease with age (Mascola et al., 2010; Pelchat &
SC
Pliner, 1995). It may be the case that many of the negative eating experiences reported by
participants were associated with pre-existing PE rather than causal (e.g., forced consumption of
U
an already-disliked food). It is possible that our participants were more able to remember
AN
experiences in early adolescence than experiences earlier in childhood, and were reporting the
first incident they could clearly remember rather than the first incident they experienced. The
M
salience of this time period could be due to the increasing importance of peer relationships at this
D
age, greater autonomy about eating choices, or the development of cognitively sophisticated
TE
decision-making around eating. Of course, this discrepant age-related finding may represent a
limitation to the generalizability of findings between age of onset and other variables, but also
EP
highlights the need for future studies that explore PE incidence in late adolescence and
adulthood.
C
Interestingly, being pressured to eat was not associated with PE related to Taste
AC
Aversion. It is possible that this facet of PE is the result of more biologically-based taste aversion
and that is less related to negative social experiences with food. This possibility suggests parental
pressure as a potential causal or maintenance factor. If the current findings were driven by PE
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 19
leading to parental pressure, the different aspects of PE may not be predicted to differ in their
Strengths of this study include a relatively large and representative sample. In addition,
PT
more comprehensive understanding of this behavior. The study has several limitations including
RI
the use of self-report measures, some of which were retrospective in nature. Furthermore, all of
the variables in the current study were measured cross-sectionally, and some had stronger
SC
evidence for temporal precedence in the predictive model than others (e.g., disgust sensitivity vs.
current family income). The finding that 8.7% of individuals who reported aversive events
U
associated with their PE described the aversive event as being related to weight concerns limits
AN
the generalizability of some results, given that PE is typically not thought to be a response to
weight concerns. However, this finding also stresses the importance of continued refinement in
M
the comprehensive measurement and operationalization of PE. It should also be noted that the
D
APEQ was developed for the purpose of understanding a range of PE behaviors in adulthood, but
TE
it is not yet validated at a diagnostic screener for the more severe pathology related to individuals
The current study provided evidence for several potential risk factors of adult PE, using a
series of SEM regression models. Importantly, potential risk factors included parental feeding
C
practices and disgust sensitivity, which could serve as future intervention targets. In addition,
AC
models examining multiple aspects of PE showed that some potential risk factors consistently
related across all facets (e.g., variety of food intake), whereas other variables were more strongly
related to a specific aspect of PE. These findings suggest that understanding the nuanced
attitudes, aversions, and behavioral sets that may have different developmental trajectories is
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 20
important for generating effective clinical treatments for severe PE behaviors and ARFID. These
findings also suggest that early experiences may be important predictors for whether PE
continues into adulthood. Specifically, the results suggest the importance of developing and
testing interventions to help parents moderate their negative responses to PE behavior that may
PT
have unintended consequences. Specifically, parents could learn positive approaches to child
RI
feeding (e.g., encouragement, modeling, and repeated exposure to foods) that have been
U SC
AN
M
D
TE
C EP
AC
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 21
References
Berge, J. M., Tate, A. D., Trofholz, A., Conger, K., & Neumark-Sztainer, D. (2016). Sibling
PT
eating behaviours and parental feeding practices with siblings: Similar or different?
RI
Public Health Nutrition, 19, 2415-2423. doi:10.1017/s1368980016000860
Berinsky, A. J., Huber, G. A., & Lenz, G. S. (2012). Evaluating Online Labor Markets for
SC
Experimental Research: Amazon.com’s Mechanical Turk. Political Analysis, 20, 351–368.
https://doi.org/10.1093/pan/mpr057
U
Cardona, S. C., Tiemeier, H., Van Hoeken, D., Tharner, A., Jaddoe, V. W. V., Hofman, A., …
AN
Hoek, H. W. (2015). Trajectories of picky eating during childhood: A general population
https://doi.org/10.1002/eat.22384
D
Cardona, S. C., Hoek, H. W., Hoeken, D. V., Barse, L. M., Jaddoe, V. W., Verhulst, F. C., &
TE
study in the general population. Journal of Child Psychology and Psychiatry,57, 1239-1246.
EP
doi:10.1111/jcpp.12530
Carper, J. L., Orlet Fisher, J., & Birch, L. L. (2000). Young girls’ emerging dietary restraint and
C
disinhibition are related to parental control in child feeding. Appetite, 35, 121–129.
AC
https://doi.org/10.1006/appe.2000.0343
Carruth, B. R., Ziegler, P. J., Gordon, A., & Barr, S. I. (2004). Prevalence of picky eaters among
infants and toddlers and their caregivers’ decisions about offering a new food. Journal of
Cooke, L., Carnell, S., & Wardle, J. (2006). Food neophobia and mealtime food consumption in
4–5 year old children. International Journal of Behavioral Nutrition and Physical
Coulthard, H., & Blissett, J. (2009). Fruit and vegetable consumption in children and their
PT
mothers. Moderating effects of child sensory sensitivity. Appetite, 52, 410–415.
RI
https://doi.org/10.1016/j.appet.2008.11.015
Dovey, T. M., Staples, P. A., Gibson, E. L., & Halford, J. C. G. (2008). Food neophobia and
SC
“picky/fussy” eating in children: A review. Appetite, 50, 181–193.
https://doi.org/10.1016/j.appet.2007.09.009
U
Dubois, L., Farmer, A., Girard, M., Peterson, K., & Tatone-Tokuda, F. (2007). Problem eating
AN
behaviors related to social factors and body weight in preschool children: A longitudinal
doi:10.1186/1479-5868-4-9
D
Duncan, A. E., Ziobrowski, H. N., & Nicol, G. (2017). The Prevalence of past 12-month and
TE
lifetime DSM-IV eating disorders by BMI category in US men and women. European
Ekstein, S., Laniado, D., & Glick, B. (2010). Does picky eating affect weight-for-length
https://doi.org/10.1177/0009922809337331
AC
Ellis, J. M., Galloway, A. T., Webb, R. M., & Martz, D. M. (2016). Measuring adult picky
Fisher, M. M., Rosen, D. S., Ornstein, R. M., Mammel, K. A., Katzman, D. K., Rome, E. S., …
children and adolescents: A “new disorder” in DSM-5. Journal of Adolescent Health, 55,
49–52. https://doi.org/10.1016/j.jadohealth.2013.11.013
PT
Forman, S. F., McKenzie, N., Hehn, R., Monge, M. C., Kapphahn, C. J., Mammel, K. A., …
RI
Woods, E. R. (2014). Predictors of outcome at 1 year in adolescents with DSM-5 restrictive
SC
Collaborative. Journal of Adolescent Health, 55, 750–
756. https://doi.org/10.1016/j.jadohealth.2014.06.014
U
Francis, L. A., Hofer, S. M., & Birch, L. L. (2001). Predictors of maternal child-feeding style:
AN
maternal and child characteristics. Appetite, 37, 231–243.
https://doi.org/10.1006/appe.2001.0427
M
Galloway, A. T., Farrow, C. V., & Martz, D. M. (2010). Retrospective reports of child feeding
D
practices, current eating behaviors, and BMI in college students. Obesity, 18, 1330–1335.
TE
https://doi.org/10.1038/oby.2009.393
Galloway, A. T., Fiorito, L., Lee, Y., & Birch, L. L. (2005). Parental pressure, dietary patterns,
EP
and weight status among girls who are “picky eaters”. Journal of the American Dietetic
Galloway, A. T., Fiorito, L. M., Francis, L. A., & Birch, L. L. (2006). “Finish your soup”:
AC
Counterproductive effects of pressuring children to eat on intake and affect. Appetite, 46,
318–323. https://doi.org/10.1016/j.appet.2006.01.019
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 24
Goh, D. Y., & Jacob, A. (2012). Perception of picky eating among children in Singapore and its
https://doi.org/10.1186/1447-056X-11-5
Hafstad, G. S., Abebe, D. S., Torgersen, L., & Soest, T. V. (2013). Picky eating in preschool
PT
children: The predictive role of the child's temperament and mother's negative
RI
affectivity. Eating Behaviors,14, 274-277. doi:10.1016/j.eatbeh.2013.04.001
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis:
SC
Conventional criteria versus new alternatives. Structural Equation Modeling: A
U
Jacobi, C., Schmitz, G., & Agras, W. S. (2008). Is picky eating an eating disorder? International
AN
Journal of Eating Disorders, 41, 626–634. https://doi.org/10.1002/eat.20545
Kauer, J., Pelchat, M. L., Rozin, P., & Zickgraf, H. F. (2015). Adult picky eating.
https://doi.org/10.1016/j.appet.2015.03.001
Kurz, S., van Dyck, Z., Dremmel, D., Munsch, S., & Hilbert, A. (2015). Early-onset restrictive
C
eating disturbances in primary school boys and girls. European Child & Adolescent
AC
Mascola, A. J., Bryson, S. W., & Agras, W. S. (2010). Picky eating during childhood: A
https://doi.org/10.1016/j.eatbeh.2010.05.006
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 25
Melbye, E. L., Øgaard, T., & Øverby, N. C. (2011). Validation of the Comprehensive Feeding
Micali, N., Simonoff, E., Elberling, H., Rask, C. U., Olsen, E. M., & Skovgaard, A. M. (2011).
PT
Eating patterns in a population-based sample of children aged 5 to 7 years: Association with
RI
psychopathology and parentally perceived impairment. Journal of Developmental &
SC
Musher-Eizenman, D., & Holub, S. (2007). Comprehensive Feeding Practices Questionnaire:
U
32, 960–972. https://doi.org/10.1093/jpepsy/jsm037
AN
Muthén, B., du Toit, S.H.C. & Spisic, D. (1997). Robust inference using weighted least squares
and quadratic estimating equations in latent variable modeling with categorical and
M
Olatunji, B. O., Williams, N. L., Tolin, D. F., Abramowitz, J. S., Sawchuk, C. N., Lohr, J. M., &
TE
Elwood, L. S. (2007). The Disgust Scale: Item analysis, factor structure, and suggestions for
3590.19.3.281
Pelchat, M. L., & Pliner, P. (1995). “Try it. You’ll like it”. Effects of information on willingness
C
Ramos-Paúl, R., Marriage, B. J., Debeza, R. R., Leal, L. O., Mar, L. R., Cardona, L. T., &
Williams, J. A. (2014). Impact of picky eating on level of family stress in healthy children
between the ages of 3 and 6 years. The Open Nutrition Journal,8, 13-18.
doi:10.2174/1874288201408010013
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 26
Rozin, P., Haidt, J., & McCauley, C. R. (2008). Disgust. In M. Lewis, J. M. Haviland-Jones, &
L. F. Barrett (Eds.), Handbook of emotions, 3rd ed (pp. 757–776). New York, NY, US:
Guilford Press.
Sherlock, J. M., Zietsch, B. P., Tybur, J. M., & Jern, P. (2016). The quantitative genetics of
PT
disgust sensitivity. Emotion, 16, 43–51. https://doi.org/10.1037/emo0000101
RI
Stice, E., Fisher, M., & Martinez, E. (2004). Eating Disorder Diagnostic Scale: Additional
SC
https://doi.org/10.1037/1040-3590.16.1.60
Stice, E., Telch, C. F., & Rizvi, S. L. (2000). Development and validation of the Eating Disorder
U
Diagnostic Scale: A brief self-report measure of anorexia, bulimia, and binge-eating
AN
disorder. Psychological Assessment, 12, 123–131. https://doi.org/10.1037/1040-
3590.12.2.123
M
Steinsbekk, S., Belsky, J., & Wichstrøm, L. (2016). Parental feeding and child eating: An
D
doi:10.1111/cdev.12546
Steinsbekk, S., Sveen, T. H., Fildes, A., Llewellyn, C., & Wichstrøm, L. (2017). Screening for
EP
Activity,14. doi:10.1186/s12966-016-0458-7
C
Strandjord, S. E., Sieke, E. H., Richmond, M., & Rome, E. S. (2015). Avoidant/restrictive food
AC
intake disorder: Illness and hospital course in patients hospitalized for nutritional
https://doi.org/10.1016/j.jadohealth.2015.08.003
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 27
Taylor, C. M., Wernimont, S. M., Northstone, K., & Emmett, P. M. (2015). Picky/fussy eating in
349-359. doi:10.1016/j.appet.2015.07.026
Tharner, A., Jansen, P. W., Kiefte-de Jong, J. C., Moll, H. A., van der Ende, J., Jaddoe, V. W.,
PT
… Franco, O. H. (2014). Toward an operative diagnosis of fussy/picky eating: a latent
RI
profile approach in a population-based cohort. International Journal of Behavioral
SC
van Overveld, M., de Jong, P. J., Peters, M. L., & Schouten, E. (2011). The Disgust Scale-R: A
valid and reliable index to investigate separate disgust domains? Personality and Individual
U
Differences, 51, 325–330. https://doi.org/10.1016/j.paid.2011.03.023
AN
Vaughn, A. E., Ward, D. S., Fisher, J. O., Faith, M. S., Hughes, S. O., Kremers, S. P. J., …
Ventura, A. K., & Birch, L. L. (2008). Does parenting affect children's eating and weight
TE
doi:10.1186/1479-5868-5-15
EP
Wildes, J. E., Zucker, N. L., & Marcus, M. D. (2012). Picky eating in adults: Results of a web-
https://doi.org/10.1002/eat.20975
AC
Zickgraf, H. F., Franklin, M. E., & Rozin, P. (2016). Adult picky eaters with symptoms of
avoidant/restrictive food intake disorder: comparable distress and comorbidity but different
eating behaviors compared to those with disordered eating symptoms. Journal of Eating
Zickgraf, H. F., & Schepps, K. (2016). Fruit and vegetable intake and dietary variety in adult
Zucker, N., Copeland, W., Franz, L., Carpenter, K., Keeling, L., Angold, A., & Egger, H. (2015).
PT
Pediatrics, 136, e582–e590. https://doi.org/10.1542/peds.2014-2386
RI
U SC
AN
M
D
TE
C EP
AC
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 29
PT
Father’s age 29.34 7.88
EDDS (standardized) 0.00 0.61
APEQ Total 2.30 0.68
APEQ Meal Presentation 2.35 0.74
RI
APEQ Food Variety 2.32 0.85
APEQ Meal Disengagement 2.05 0.86
APEQ Taste Aversion 2.44 1.08
SC
Variable n Percentage
PE Experience
Yes 117 8.7
Maybe 225 16.8
NA 997 74.5
U
PE Experience Code for “Yes”
Pressure to Eat 27 26.0
AN
Choking/Food Sickness 40 38.5
Medical Problem 28 26.9
Weight Concern 9 8.7
Income
M
Education
< High School/GED 5 0.4
TE
School
Race
White 1070 79.9
AC
Table 2 Two-tailed Correlations Between APEQ Subscales and Continuous Predictors (n = 1339)
PE Subscale Pressure Encouragement Disgust PE Age Father’s Age Mother’s Age EDDS BMI
PT
APEQ- Meal Presentation .20** .02 .28** -.11* -.03 -.06* .28** -.02
RI
APEQ- Food Variety .16** -.17** .17** -.11* -.06* -.06* .23** .05
SC
APEQ- Meal Disengagement .12** -.07* .00 -.02 -.02 -.01 .29** -.03
APEQ-Taste Aversion .10** -.08** .20** -.05 -.05 -.03 .19** .05
U
AN
Note: ** = correlation is significant at the 0.01 level; * = correlation is significant at the 0.05 level.
M
D
TE
C EP
AC
ACCEPTED MANUSCRIPT
PT
RI
U SC
AN
M
D
TE
C EP
AC
Figure 1: Conceptual representation of Models 1 through 3. Circles represent latent variables. Boxes represent manifest variables. The
dashed box represents variables trimmed from Model 1 due to insignificant path weights. Dashed arrow from BMI shows
nonsignificance in Model 3 after controlling for EDDS. APEQ = Adult Picky Eating Questionnaire. PE Age = self-reported age of
picky eating onset. PE Experience = Aversive food experience related to picky eating. BMI = Body Mass Index. EDDS = Eating
Disorder Diagnostic Scale. Full reporting of standardized regression weights can be found in Table 3.
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 32
PT
Sex -0.115 -0.114 Sex -0.160
PE Age -0.139 -0.134 PE Age -0.127
PE Experience 0.277 0.281 PE Experience 0.245
BMI 0.109 0.115 BMI 0.015
RI
Race 0.001 EDDS 0.317
Education 0.001
Income -0.001
SC
Mother’s age -0.129
Father’s age 0.052
U
Note: Nonsignificant paths were trimmed from Model 2 and the Eating Disorder Diagnostic
Scale (EDDS) was added as a covariate to Model 3 to control for disordered eating symptoms
AN
related to shape and weight; PE = picky eating; PE Age = self-reported age of picky eating onset;
PE Experience = Aversive food experience related to picky eating; BMI = Body Mass Index; Sex
was coded as 1 = men and 2 = women; Bolded standardized regression paths indicate p < .05.
M
D
TE
C EP
AC
ACCEPTED MANUSCRIPT
PT
RI
U SC
AN
M
D
TE
C EP
AC
Figure 2: Conceptual representation of Models 4 and 5. Circles represent latent variables. Boxes represent manifest variables. The
Eating Disorder Diagnostic Scale (EDDS) was added to Model 5 as a covariate. APEQ = Adult Picky Eating Questionnaire. PE Age =
self-reported age of picky eating onset. PE Experience = Aversive food experience related to picky eating. BMI = Body Mass Index.
Full reporting of standardized regression weights can be found in Table 4.
ACCEPTED MANUSCRIPT
POTENTIAL RISK FACTORS OF ADULT PICKY EATING 34
Table 4
Standardized Regression Weights of Predictors of the APEQ Subscale Scores
Model 4 Model 5 (EDDS Covariate)
Subscale and Predictors Subscale and Predictors
Meal Presentation Loadings Meal Presentation Loadings
Pressure 0.152 Pressure 0.151
Encouragement -0.115 Encouragement -0.115
PT
Disgust 0.369 Disgust 0.368
Sex -0.075 Sex -0.117
PE Age -0.139 PE Age -0.134
RI
PE Experience 0.264 PE Experience 0.231
BMI 0.062 BMI -0.029
EDDS 0.289
SC
Food Variety Food Variety
Pressure 0.171 Pressure 0.171
Encouragement -0.360 Encouragement -0.361
Disgust 0.165 Disgust 0.164
U
Sex -0.149 Sex -0.169
PE Age -0.157 PE Age -0.155
AN
PE Experience 0.264 PE Experience 0.248
BMI 0.187 BMI 0.144
EDDS 0.139
M