Manova analysis is SPSS and Scales Analysis

Research Article

Feeding and Mealtime Correlates of Maternal
Concern About Children’s Weight

Jacqueline M. Branch, MD1; Danielle P. Appugliese, MPH2; Katherine L. Rosenblum, PhD3,4;
Alison L. Miller, PhD4,5; Julie C. Lumeng, MD1,4,6; Katherine W. Bauer, PhD6

1Departm
School,
2
Appugl
3Departm
4Center
5
Departm
Public H
6Departm
Arbor, M
Conflict o
with thi
Address
Develop
MI 4810
�2017 S
reserved
http://dx

490

ABSTRACT

Objective: To examine associations between maternal concern regarding their children becoming
overweight and two domains of weight-related parenting; child feeding practices and family meal charac-
teristics.
Design: Cross-sectional study.
Participants: Low-income mothers (n ¼ 264; 67% non-Hispanic white) and their children (51.5% male,
aged 4.02–8.06 years).
Variables Measured: Maternal concern and feeding practices, using the Child Feeding Questionnaire.
Meal characteristics were assessed using video-recorded meals and meal information collected from mothers.
Analysis: The authors used MANOVA and logistic regression to identify differences in maternal feeding
practices and family meal characteristics across levels of maternal concern (none, some, and high).
Results: Approximately half of mothers were not concerned about their child becoming overweight,
28.4% reported some concern, and 19.0% had high concern. Mothers reporting no concern described
lower restrictive feeding compared with mothers who reported some or high concern (mean [SE],
none ¼ 3.1 [0.1]; some ¼ 3.5 [0.1]; and high ¼ 3.6 [0.1]; P ¼ .004). No differences in other feeding practices
or family meal characteristics were observed by level of concern.
Conclusions and Implications: Concern regarding children becoming overweight was common.
However, concern rarely translated into healthier feeding practices or family meal characteristics. Maternal
concern alone may not be sufficient to motivate action to reduce children’s risk of obesity.
Key Words: childhood obesity, maternal concern, feeding practices, family meals (J Nutr Educ Behav.
2017;49:490-496.)

Accepted March 16, 2017. Published online April 28, 2017.

INTRODUCTION

Despite the heightened attention to
childhood obesity over the pastdecade,
several studies documented that only a
limited proportion ofparents recognize
that their children are overweight, and

ent of Pediatrics and Communicable
Ann Arbor, MI
iese Professional Advisors, North Easto
ent of Psychiatry, University of Mich

for Human Growth and Development
ent of Health Behavior and Health Ed
ealth, Ann Arbor, MI
ent of Nutritional Sciences, University
I
f Interest Disclosure: The authors’ confli
s article on www.jneb.org.
for correspondence: Jacqueline M. Bra
ment, 300 North Ingalls Bldg, Rm 1024
9; Phone: (716) 785-2244; Fax: (734) 93
ociety for Nutrition Education and Beh
.
.doi.org/10.1016/j.jneb.2017.03.011

relatively few parents reported concern
about their children’s current weight or
future risk of becoming overweight.1-6

Parents of young children and lower
socioeconomic status in particular reported
less concern about their children’s current
or future risk of overweight than did

Diseases, University of Michigan Medical

n, MA
igan Medical School, Ann Arbor, MI
, University of Michigan, Ann Arbor, MI
ucation, University of Michigan School of

of Michigan School of Public Health, Ann

ct of interest disclosures can be found online

nch, MD, Center for Human Growth and
NW, University of Michigan, Ann Arbor,
6-6897; E-mail: [email protected]
avior. Published by Elsevier, Inc. All rights

Journal of Nutrition Education and Beh

parents of older children or higher
socioeconomic status.1,2,4 This limited
concern prompted calls for programs
and policies to elevate parents’ concern
about obesity and/or their children’s
weight.2,3 Initiatives designed to increase
parental concern about their children’s
weight included universal body mass
index (BMI) screening during health
care visits,7 BMI report cards,8 and me-
dia campaigns highlighting the health
risks of obesity.9

Interventions to elevate concern
about childhood obesity assumed that
parents who are concerned about their
children’sweightaremorelikely totake
actiontoimprovetheirchildren’sbehavior
and weight status.8 For such interven-
tions to be successful at reducing obesity,
it is essential that concern prompts
parents to participate in evidence-based
approaches to improve children’s en-
ergy balance, and that actions do not
contribute to weight gain or that there
be other negative health problems. Ev-
idence is mixed as to whether parental

avior � Volume 49, Number 6, 2017

Journal of Nutrition Education and Behavior � Volume 49, Number 6, 2017 Branch et al 491

concern about children’s weight is
associated with healthy changes in
child weight6,10 or parents’ participation
in behaviors that promote children’s
healthy weight.1,4,11-14 Some studies
found that parents who were concerned
about their child’s weight were more
likely to limit screen time, encourage
physical activity, and change the family
diet, compared with parents who were
not concerned.4,11 These evidence-
based actions were recommended to
address childhood obesity.7 However,
1 study found that parental concern
about children’s weight was not associ-
ated with healthier food available in
the home.1 Furthermore, parents who
wereconcernedreportedgreaterencour-
agement of skipping meals and diet-
ing, as well as higher use of restrictive
feeding practices.1,11,12,14 These practices
were associated with low body satisfaction,
poorself-regulationofeating,increased
binge eating, and a greater risk of
obesity.15-17 Based on this literature,
evidence is insufficient to determine
whether promoting concern will prompt
engagement in evidence-based parenting
practices to improve children’s weight
status. This lack of evidence is particu-
larly true for parents of young chil-
dren, because the majority of studies
of concern were conducted among
parents of older grade school or
adolescent-aged children.1,4,11,13

Given the need to understand the
association between parental concern
about young children’s weight and
parents’ participation in actions that
promote healthy behavior and weight,
the aim of the current study was to
examine differences within 2 domains
of weight-related parenting: child feeding
practices and family meal characteris-
tics, among mothers of young children,
by mothers’ concern about their chil-
dren becoming overweight. This study
drew from data from ABC Feeding,
which enrolled children eligible for
Head Start and their caregivers. This
study’s population provided a unique
perspective on how low-income mothers
sought to address their children’s risk
for overweight and obesity. This insight
is important given the increasing burden
of childhood obesity among low-
income families18 and the need to
develop interventions that are effec-
tive in this context.19 The authors hy-
pothesized that greater concern by
mothers that their child would become
overweight would be associated with

more restrictive feeding and greater
monitoring of child’s eating, but less
pressuring feeding practices. In addi-
tion, the authors hypothesized that
concern would be associated with fam-
ily meal characteristics that reflected
current clinical guidance for child nutri-
tion promotion and obesity preven-
tion and treatment.7,20

METHODS
Study Design

The current cross-sectional study used
data from the first measurement of
ABC Feeding, a longitudinal study of
maternal feeding practices.

Participants and Recruitment

The current study included a sample of
264 low-income female primary care-
givers (mean age, 31.0 years; SD,
7.06 years; 67% non-Hispanic white;
45% single parent) and their children
(mean age, 5.4 years; SD, 0.75 years;
range, 4.0 to 8.1 years; 153 boys). The
caregivers were predominantly (95%)
biological mothers; therefore, caregivers
will be referred to as mothers. These
mother–child dyads were originally
recruited via participation in Head Start
programs in South Central Michigan
and enrolled in ABC Preschool, a longi-
tudinal study conducted between 2009
and 2011. All mothers enrolled were
fluent in English and had less than a
4-year college degree. Approximately
2 years after participation in ABC Pre-
school, mothers were invited to partic-
ipate in a follow-up study on child
feeding, ABC Feeding. Of the 380 care-
givers invited, 284 participated and an
additional 17 families were newly re-
cruited from Head Start, which result-
ing in a final sample size of 301. Among
these dyads, 5 were excluded because
the primary caregiver was male and
32 had incomplete data; this resulted
in an analytic sample of 264 (87.7%
of the total sample). The study proto-
col was approved by the University of
Michigan Institutional Review Board.

Measures

Mother–child dyads completed activ-
ities over the course of 2 study visits.
Mothers were provided a video cam-
era during the second visit and were
asked to record 3 routine, weeknight,

dinnertime meals within 1 week. After
each recorded meal, mothers received
a telephone call from a trained inter-
viewer to collect information on the
foods available to the child during the
meal. After the meals were recorded,
the camera was collected by study staff.
This protocol was previously described
in detail elsewhere.21

Maternal concern about child over-
weight. Maternalconcernwasmeasured
using 1 item from the Child Feeding
Questionnaire22 that asked: How con-
cernedareyouaboutyourchildbecoming
overweight? Responses were rated on a
5-point scale with the response op-
tions ranging from unconcerned to
concerned. Responses were then cate-
gorized into 3 levels of concern: no
concern for mothers who reported
they were unconcerned, some concern
for mothers who reported the next 2
higher levels of concern, and high
concern for mothers reporting the
highest 2 levels of concern.

Maternal feeding practices. Three
feeding practices were measured using
the Child Feeding Questionnaire: pres-
sure to eat (4 items, Cronbach a ¼ .62),
restriction (8 items, Cronbach a ¼ .75),
and monitoring (3 items, Cronbach
a ¼ .86). Pressure to eat was assessed
using items including: My child should
always eat all of the food on her plate.
Restriction was assessed using items
including: I have to be sure that my
child does not eat too many sweets
(candy, ice cream, cake, or pastries).
For both scales, responses were measured
using a 5-point scale ranging from
disagree to agree. Monitoring was
assessed using items including: How
much do you keep track of the snack
food (potato chips, Doritos, or cheese
puffs) that your child eats? Responses
were measured using a 5-point scale
ranging from never to always.

Meal characteristics. Characteristics
of typical family meals were measured
using video recorded meal observations
and meal information collected from
mothers.Tocollectthemealdata,mothers
were asked to video record 3 dinner-
time family meals over the course of
1 week occurring when the mother
was home and awake, when the meal
occurred at home, and when the meal
was prepared by the primary caregiver.

492 Branch et al Journal of Nutrition Education and Behavior � Volume 49, Number 6, 2017

To record the meals, mothers were
instructed to set up the camera so
that the child’s upper torso, plate,
and drink were always visible, and to
record the entirety of each meal. To
quantify the data collected during the
observations, the study team devel-
oped a coding scheme adapted from
prior approaches23,24 to code each
meal with regard to whether the meal
was pre-plated (vs served family style
or eaten out of the serving package),
the television was audible, the mother
ate with the child for any portion of
the meal, and, if requested by the
child, the mother allowed second serv-
ings. Coders were trained to increase
reliability; 12% of videos were coded
by 2 raters and interrater reliability by
Cohen’s kappa exceeded 0.70 for all
codes. Each family meal characteristic
was coded affirmatively if it was
observed in at least half of meals.

Foods served during family meals.
Information on foods served during
the meal was obtained from the meal
report collected from mothers by inter-
viewers after each recorded meal. Each
meal report was coded into food and
beverage categories determined by the
groupings on ChooseMyPlate.gov in
accordance with the current US Dietary
Guidelines for Americans.25 The pres-
enceorabsenceofeachfoodorbeverage
group for each meal was coded. The
preparation method for meats (ie, deep
frying vs not) was identified by the
food name (ie, chicken nuggets, fish
sticks), and coded accordingly.21

To obtain a composite measure of
food and beverage types served during
family meals, families were coded as
typically serving fruits, vegetables, and
refined grains if these foods were
reported as present in at least half of
meals. Food types that were overall less
prevalent and would not be expected
to be served at every meal were coded
as typically served if they were present
in any of the meals. These included dark
greenvegetables,wholegrains,deepfried
proteins, low-fat or skim milk, diet drinks,
sugar-sweetened beverages, and dessert
(including ice cream, frozen yogurt,
pudding, and other nondairy sweets).

Sociodemographic characteristics.
Mothers reported their child’s sex and
birth date, and maternal education
and race/ethnicity. Child birth date

was used to calculate the child’s age
by subtracting the birth date from the
date of the first study visit. Maternal
education was included as high school
diploma or less or equivalent vs more
than high school diploma, with the
highest educational level in this sample
being less than a 4-year college degree.
Maternal race/ethnicity was included
as non-Hispanic white vs Hispanic and/or
not white.

Maternal and child anthropometrics.
Heights and weights of mothers and
children were measured according to
standardized procedures.26 Body mass
index was calculated as weight in kilo-
grams divided by height in meters
squared. For 12 mothers who were preg-
nant or had given birth within the past
3 months, self-reported pre-pregnancy
weight was used. Body mass index z-
scores and percentiles were calculated
for children, and children were catego-
rized as being underweight or normal
weight (BMI < 85th percentile for
age and sex), or overweight or obese
(BMI $ 85th percentile for age and
sex) based on the US Centers for Dis-
ease Control and Prevention growth
charts.27 Only 3 children were under-
weight; therefore, underweight and
normal weight were combined.

Data Analysis

The researchers examined bivariate
differences in sociodemographic and
anthropometric characteristics by level
of maternal concern using ANOVA and
Pearson chi square tests. They used
MANOVA to identify differences in
mean maternal feeding practices by level
of maternal concern adjusted for child
sex, age, race/ethnicity, BMI z-score,
and maternal education and BMI, and
calculated adjusted means for each
level of maternal concern. For feeding
practices in which overall differences
in means were detected, pairwise com-
parisons wereusedto identifydifferences
between levels of concern. Unadjusted
prevalence of family meal characteris-
tics and foods served were calculated
for each level of concern. Multivari-
able logistic regression was then used
to examine associations between level
of maternal concern and each meal
characteristic or food served, adjusted
for covariates. All analyses were run
for the full sample and were also

limited to the dyads with overweight
andobesechildren.Findingsdidnotdiffer;
therefore, results from the full sample
are presented. All analyses were con-
ducted using SAS software (version 9.3;
SAS Institute, Inc., Cary, NC); P < .05
was used to indicate statistical signifi-
cance.

RESULTS
Characteristics of Maternal
Concern

Among this sample of low-income
mothers, 52.7% reported that they
were not concerned about their child
becoming overweight, 28.4% reported
some concern, and 18.9% reported
high concern (Table 1). Concern about
the child becoming overweight did not
differ by maternal education (P ¼ .89)
or child’s sex (P ¼ .52), race/ethnicity
(P ¼ .21), or age (P ¼ .76). Differences
in maternal concern were observed by
child BMI z-score (P < .001) and weight
status (P < .001), and maternal BMI
(P < .001). Among mothers of under-
weight or normal weight children,
7.2%reportedhighconcernabouttheir
child becoming overweight whereas
34.9% of mothers of children with
overweight or obesity reported high
concern. A total of 29% of mothers of
children with overweight or obesity re-
ported no concern about their child
becoming overweight.

Maternal Concern and Child
Feeding Practices

Maternal concern about the child
becoming overweight was associated
with greater use of restrictive feeding
practices (Table 2). Among mothers re-
porting no concern, mean restrictive
feeding was significantly lower than
that for mothers reporting some or
high concern (mean [SE], 3.1 [0.08],
3.5 [0.11], and 3.6 [0.14], respectively;
P ¼ .004). Mean restrictive feeding
scores did not differ between mothers
reportingsomevshighconcern.Nodif-
ferences were observed by level of
concern in mothers’ report of moni-
toring or pressure to eat.

Maternal Concern and Meal
Characteristics

No differences in characteristics or con-
tent of family meals were observed by

Table 1. Sociodemographic and Weight Characteristics of Children and Mothers, in Total and by Maternal Concern

Child and Maternal
Characteristics Total Sample

Maternal Concern About Child Becoming Overweight

PNone Some High

Total sample (% [n]) 100.0 (264) 52.7 (139) 28.4 (75) 18.9 (50)
Child characteristics
Child gender (% [n]) .52
Male 51.5 (136) 55.9 (76) 25.7 (35) 18.4 (25)
Female 48.5 (128) 49.2 (63) 31.3 (40) 19.5 (25)

Child race/ethnicity (% [n]) .21
Non-Hispanic white 55.3 (146) 57.5 (84) 25.3 (37) 17.1 (25)
Hispanic or not white 44.7 (118) 46.6 (55) 32.2 (38) 21.2 (25)
Child age, mo (mean [SD]) 70.8 (8.4) 70.8 (0.7) 70.3 (1.0) 71.5 (1.2) .76
Child BMI z-score (mean [SD]) 0.9 (1.0) 0.4 (0.1) 1.1 (0.1) 1.7 (0.1) <.001

Child weight status (% [n]) <.001
Underweight/normal weight 58.4 (153) 69.9 (107) 22.9 (35) 7.2 (11)
Overweight/obese 41.6 (109) 29.4 (32) 35.8 (39) 34.9 (38)

Maternal characteristics
Maternal education (% [n]) .89
High school diploma/General
Equivalency Diploma/less

47.0 (124) 51.6 (64) 30.0 (37) 18.6 (23)

At least some college education 53.0 (140) 53.6 (75) 27.1 (38) 19.3 (27)
Maternal BMI (mean [SD]) 33.2 (9.4) 30.8 (0.8) 34.7 (1.0) 37.8 (1.3) <.001

BMI indicates body mass index.
Note: Pearson chi-square and ANOVA were used to examine differences in child and maternal characteristics by level of
maternal concern.

Journal of Nutrition Education and Behavior � Volume 49, Number 6, 2017 Branch et al 493

level of mothers’ concern (Table 3). For
example, the prevalence of mothers
pre-platingtheirchildren’smeals,eating
withchildren,allowingsecondservings,
and serving fruits, vegetables, sugar-
sweetened beverages, and desserts
were similar across levels of concern.
Overall, vegetables, refined grains, and
fried proteins were commonly avail-
able during meals. For example, vege-
tables were served at 89.2% to 94.0%
of meals. Sugar-sweetened beverages
were also typically available during
meals; 60.0% to 68.0% of families

Table 2. Associations Between Maternal C

Maternal Concern
About Child
Becoming Overweight

Restr
Practic

M

None

Some

High

Fdegrees of freedom ¼ 2
P

Notes: Multivariate ANOVA adjusted for
education and BMI was used to examine
significantly different values by level of co

served a sugar-sweetened beverage
during at least 1 meal. Fruit, whole
grains, and low fat or skim milk were
less commonly served during observed
meals.

DISCUSSION

The objectives of the current study were
to examine maternal concern regarding
youngchildren’s risk forbecoming over-
weight and to identify differences in
child feeding practices and family meal

oncern About Child Becoming Overweight an

ictive Feeding
es (Range, 1–5)

Monitoring Ch
Eating (Range, 1

ean (SE) Mean (SE)

3.1 (0.1)a 3.9 (0.1)

3.5 (0.1)b 4.0 (0.1)

3.6 (0.1)b 4.3 (0.2)

5.7 1.6

.004 .20

child gender, age, race/ethnicity, body ma
differences by level of maternal concern. Di
ncern at P < .05.

routines among mothers with differing
levels of concern. Approximately half
of mothers reported some level of
concern about their child becoming
overweight; over 70% of mothers of
currently overweight or obese children
reported at least some concern. These
findings run counter to the prominent
belief that mothers, especially mothers
of young children and those of low so-
cioeconomic status, had limited concern
about obesity among their children.1-6,14

Thisdifferencemaybebecausearelatively
contemporary sample of mothers was

d Maternal Feeding Practices

ild
–5)

Pressuring Feeding
Practices (Range, 1–5)

Mean (SE)

2.7 (0.1)

2.8 (0.1)

2.8 (0.2)

0.04

.96

ss index (BMI) z-score, and maternal
ffering superscripts indicate statistically

Table 3. Associations Between Maternal Concern About Child Becoming Overweight and Family Meal Characteristics

Family Meal
Characteristics

Maternal Concern About
Child Becoming Overweight Adjusted Comparisonsa

Unadjusted Prevalence OR (95% CI)
P

OR (95% CI)
PNone Low High Low vs None High vs None

Meal characteristics
Dinner pre-plated 87.6 80.3 79.1 0.6 (0.2–1.4) .23 0.8 (0.2–2.5) .66
Television audible during dinner 64.6 68.1 72.7 0.9 (0.4–2.1) .78 1.5 (0.5–4.5) .44
Mother eats with child 82.4 84.1 88.1 1.1 (0.4–2.7) .92 1.5 (0.4–5.8) .53
Mother allows second serving 46.8 44.0 46.0 0.93 (0.5–1.7) .82 1.1 (0.5–2.5) .78

Foods served during meals
Fruit 13.0 13.3 18.0 0.9 (0.4–2.2) .78 0.9 (0.3–2.7) .83
Vegetables 89.2 92.0 94.0 1.2 (0.4–3.4) .78 1.5 (0.3–6.4) .61
Dark green vegetables 18.0 24.0 16.0 1.5 (0.7–3.4) .28 0.9 (0.3–2.6) .87
Refined grains 71.9 80.0 72.0 1.1 (0.5–2.4) .75 0.6 (0.3–1.5) .28
Whole grains 12.2 13.3 14.0 1.1 (0.4–2.8) .86 1.5 (0.5–4.5) .51
Fried protein 32.4 44.0 28.0 1.9 (1.0–3.6) .05 1.1 (0.4–2.5) .92
Low fat/skim milk 6.2 9.0 16.3 1.5 (0.5–4.7) .53 2.7 (0.7–10.5) .15
Diet beverages 2.2 2.7 4.0 0.9 (0.1–6.5) .88 1.0 (0.1–10.6) .99
Sugar-sweetened beverages 62.6 68.0 60.0 1.2 (0.6–2.2) .68 0.8 (0.3–1.7) .52
Dessert 19.4 17.3 20.0 0.8 (0.4–1.9) .67 0.8 (0.3–2.3) .73

CI indicates confidence interval; OR, odds ratio.
aLogistic regression models adjusted for child gender, age, race/ethnicity, body mass index (BMI) z-score, and maternal
education and BMI were used to examine the odds of foods or drinks served given the level of maternal concern.

494 Branch et al Journal of Nutrition Education and Behavior � Volume 49, Number 6, 2017

used among whom obesity was discussed.
For example, Head Start regularly provides
parental education regarding child nut-
rition and obesity prevention. Maternal
concern about the child becoming ove-
rweight was also positively associated
with mothers’ own BMI. This heightened
concern may reflect that mothers with
higher BMIs are more likely to have
children with higher BMIs. Mothers with
higher BMIs may recognize that their
children are at risk of overweight and
obesity in the future owing to a family
history of obesity.

Despite the high levels of concern
about future overweight among low-
income mothers, few differences in
maternal behavior were observed by
level of concern. In particular, concern
about children’s risk of becoming over-
weight did not manifest as differences
in family meal practices or food avail-
ability. These findings differed from pre-
vious studies in which maternal concern
about child weight was associated with
parental reports that they engaged in
actionstotrytoimprovetheirchildren’s
diets.4,11 The current study differed from
those previous studies in the use of
observed mealtime characteristics, vs
parent-reported behaviors, which may
explain the difference in findings.

Social desirability may lead parents
with high concern regarding their
child’s weight to report they are en-
gaging in action, even if they are not.
Alternatively, the researchers’ measure-
ment of family meals may not repre-
sent behaviors that occurred outside
thesemeals:forexample,atothermeals
or snacks. Furthermore, the current
study drew from an exclusively low-
income sample, which may explain dif-
ferences in findings. Findings demon-
strated that regardless of maternal
concern, many family meals did not
reflect recommendations to promo-
te healthy weight among children.
Common family meals had sugar-
sweetened beverages, refined grains,
and fried proteins available, with the
television audible, whereas meals with
dark green vegetables, whole grains,
and low-fat or skim milk were relatively
uncommon. These meal characteris-
tics,evenamongmotherswhoreported
high concern that their children would
becomeoverweight,mayhavereflected
unclear guidance regarding what con-
stitutes a healthy meal. These meals
may also have been a product of time
or financial limitations, or competing
food preferences among children or
other family members in the home.

In the current study, mothers who
reported any level of concern about
their child becoming overweight also
reported greater use of restrictive feeding
practices compared with mothers who
reported no concern. Similar associa-
tions between concern about child
weight and restrictive feeding were
demonstrated in other studies.1,12-14

Restrictive feeding practices were
associated with increased disinhibi-
ted eating and weight gain among
children28-31; therefore, current obesity
prevention and treatment guidelines
recommended that parents avoid overly
restrictive feeding practices.7 However,
additional evidence suggested that
mothers’ restrictive feeding is often a
response to concern about children’s
weight and obesogenic eating and
weight gain among children, and not a
cause of these outcomes.13,32 Given the
consistency with which maternal con-
cern about child weight and use of
restrictive feeding practices are assoc-
iated, further research is needed to
understand how mothers can limit
children’s eating effectively without
promoting negative outcomes.

There were several limitations to
the current study. First, only 2–3
meals were observed per family and

Journal of Nutrition Education and Behavior � Volume 49, Number 6, 2017 Branch et al 495

the meals may not be representative
of typical meals. Families may have
served different foods or conducted
the family meal differently than they
would typically because they were be-
ing recorded. In addition, the authors
were not able to capture the portion
sizes that were available or served in
a valid manner. Although parents
with high concern for child weight
may not alter what is served, they
may modify the amount of each
food available to the child. Despite
these limitations, objective observa-
tions of family meals provide unique
information about behavior and food
availability during meals that may not
be captured through self-report. Sec-
ond, the study sample was exclusively
low-income families, who often expe-
rience unique barriers to providing
health-promoting meals; therefore,
findings may not be generalizable to
higher-income families. Additionally,
the pressure to eat sub-scale had only
moderate internal reliability. Finally,
the current measure of maternal concern
about child weight captured concern
about future risk of overweight. This
measure is commonly used to examine
maternal concern about child weight,
but it does not capture concern about
current weight. Parents who are con-
cerned that their child is currently
overweight may be more likely to sup-
port children’s healthy eating and
modify family meals, whereas parents
who are concerned that their child
may become overweight in the future
may see less immediate need to imple-
ment these changes.

IMPLICATIONS FOR
RESEARCH AND
PRACTICE

Among low-income mothers, concern
regarding children becoming over-
weight was common and highly preva-
lent among mothers of children who
werealreadyoverweightorobese.Whereas
restrictive feeding practices were more
likely to be reported by mothers who
were concerned about their children’s
future risk of overweight, maternal
concernwas notassociatedwithgreater
monitoring of child eating or healthier
characteristics of family meals. Further
research is needed to understand the
characteristics of families among whom
concern about child weight prompts

healthy actions to prevent obesity. In
addition, because the existing litera-
ture, including this study, used a vari-
ety of measures of parental concern
about child weight, greater consis-
tency in the use of measures that
validly capture bothconcern aboutcur-
rent weight and future weight may
clarify what types of concern prompt
parental action. Currently, these find-
ings suggest that future family-based
interventions to address childhood
obesity may be more likely to be bene-
ficial if they do not focus merely on
raising parental concern about chil-
dren’s risk of becoming overweight,
but assist …

Empirical Articles

Assessing Burnout in Portuguese Health Care Workers who Care for the
Dying: Validity and Reliability of a Burnout Scale Using Exploratory Factor
Analysis

Carol Gouveia Melo*a, David Oliverab

[a] Centre for Professional Practice, University of Kent, Chatham, United Kingdom. [b] Wisdom Hospice, Rochester, United Kingdom.

Abstract
Aims: The aim of this study was to develop an effective instrument to measure levels of burnout in Health Care Workers (HCWs) who care
for dying patients and confirm the validity and reliability of the scale. The Burnout scale for workers who care for dying patients was created
in 2005, by Gouveia Melo, using items from the Maslach Burnout Inventory (Human Services Survey) (Maslach, Jackson, & Leiter, 1997), the
Burnout Test (Service Fields) (Jerabek, 2001) and items specifically designed for burnout in end-of-life care. Method: The scale was validated
with 280 HCWs working in oncology hospitals and in community home care in different parts of the country. The psychometric methods used
were exploratory factor analysis using principal components analysis (PCA), Cronbach’s α coefficients, and intra-class correlation coefficients.
Results: The initial 40 items were submitted to analysis for suitability of the data and 38 items were chosen for PCA. Results showed 3 main
components with 36 items explaining a total of 34.29% of the variance. These factors were emotional exhaustion (15 items), professional
fulfillment (14 items) and depersonalization (7 items). Cronbach’s α coefficients were .86 for emotional exhaustion, .83 for professional fulfillment
and .63 for depersonalization. Pearson bivariate correlations were performed on the 150 participants, with an interval of 4 months for test-retest
purposes with intra-class correlations from .55 to .59 in each domain. Convergent and divergent validation showed significant correlations.
Conclusions: The validity and reliability of this scale was established, enabling it to be used within the Portuguese population.

Keywords: burnout, scale, validation, palliative care, oncology

Psychology, Community & Health, 2012, Vol. 1(3), 257–272, doi:10.5964/pch.v1i3.21

Received: 2012-06-24. Accepted: 2012-09-07. Published: 2012-11-30.

*Corresponding author at: Rua Gil Vicente 12, Bloco C R/C, 2775-198 Parede, Portugal, email: [email protected]

This is an open access article distributed under the terms of the Creative Commons Attribution License
(http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the
original work is properly cited.

Introduction

Health care workers who care for dying patients are in daily contact with physical degradation, suffering and
inevitable death. This is a source of stress that is often more present in this field of medical action than in other
medical care areas. Research has shown that health care workers (HCWs) who care for terminal patients can
also suffer from death anxiety and that this may lead to burnout affecting the quality of patient and family care
(Bernard & Creux, 2003; Connelly, 2009; Keidel, 2002; Lowry, 1997). The underlying causes may be the HCWs’
own fear of death, feelings of inadequacy, insufficient understanding of the needs of dying patients, and difficulties
in communicating (Keidel, 2002; Lowry, 1997).

In this study, the term Health Care Worker (HCW) will be used to refer to all trained health professionals who care
for patients in their area of expertise. In this particular case, they are nurses, nursing aides, doctors, psychologists,

Psychology, Community & Health
pch.psychopen.eu | 2182-438X

social workers, physiotherapists, nutritionists, occupational therapists, hospital chaplains and unit secretaries who
have direct contact with patients and their families.

Death anxiety has been defined as “a negative emotional reaction provoked by the anticipation of a state in which
the self does not exist” (Tomer & Eliason, 1996, p. 345). It has also been suggested that the fear expressed by
the dying may be the same type of fear that people experience “in everyday life rather than in acute situations
where there are immediate threats to life…[It] has multiple components including: anticipating [oneself’s] dead,
fear of the process of dying and fear about the death of significant others” (Payne, Dean, & Kalus 1998, p. 701).
Other similar existential fears may include: fear of pain and suffering, fear of being alone and of not having close
and fulfilled relationships, fear of living with uncertainty, fear of the unknown, fear of not living a meaningful life,
fear of physical degradation, fear of losing one’s dignity and being judged by others, and fear of what comes after
death (Hennezel & Leloup, 1997). Lack of awareness of the existence of these existential fears in their own lives
may lead HCWs to experience feelings of death anxiety when caring for terminal patients, which can interfere
with the HCW-patient relationship (Lowry, 1997; Meier, Back, & Morrison, 2001), and eventually lead to burnout.

Burnout has been defined as “a syndrome of emotional exhaustion, depersonalization and reduced personal
accomplishment that can occur among individuals who work with other people in some capacity” (Maslach,
Jackson, & Leiter, 1997, p. 192). HCWs’ burnout and death anxiety can affect patients, institutions, and HCWs
themselves in many ways. It can lead to poor quality of care, increased absenteeism and job turnover, and personal
dysfunction – physical exhaustion, insomnia, increased use of alcohol and drugs, and marital and family problems
(Maslach, Jackson, & Leiter, 1997). Emotional exhaustion was referred to by the same authors as “The depletion
or draining of emotional resources” and depersonalization as “the development of negative, callous and cynical
attitudes towards the recipients of one’s services”. This is different from the psychiatric meaning where
depersonalization is used to denote a person’s extreme alienation from the self and the world. In Maslach and
Jackson’s definition, it refers to a “callous or even dehumanized perception of others, rather than to an impersonal
view of the self” (Maslach, Jackson, & Leiter, 1997, p. 192). In this study, the term ‘personal accomplishment’
used by Maslach was referred to as ‘professional fulfillment’ and can be defined as a sense of well-being in relation
to one’s working performance, one’s relationships with colleagues and patients and within the working environment.

When assessing burnout in this specific population, there is a need to take death anxiety into account as one of
the causes towards burnout. This is particularly important, because studies using the Maslach Burnout Inventory
(MBI) have shown that burnout is generally not high in these health care workers. However qualitative studies
have shown that health care workers who care for the dying may have different sources of stress than other HCWs
(Gouveia e Melo & Oliver, 2011; Pereira, 2011; Soares, 2010) and thus, there is a need for an instrument that will
provide a more accurate evaluation of burnout in this population.

Methods

Participants

The participants were 280 health care workers in Portugal, aged 21 to 67 years, comprising 177 (63.2%) nurses,
62 (22.1%) nursing aides, 13 (4.6%) psychologists, 11 (3.9%) doctors, eight (2.9%) social workers, three (1.1%)
physiotherapists, two (0.7%) secretaries, one (0.4%) occupational therapist, two (0.7%) chaplains and one (0.4%)
nutritionist. Overall, 247 (88.2%) were female and 33 (11.8%) male. A total of 124 (44.3%) worked in palliative
care units and 156 (55.7%) worked with dying patients but not within a palliative care unit. Most of the participants

Psychology, Community & Health
2012, Vol. 1(3), 257–272
doi:10.5964/pch.v1i3.21

Burnout Scale for Workers who Care for the Dying 258

worked in oncology hospitals, both in the centre and north of Portugal and others worked in community home
care.

Procedure

The scale was designed as an anonymous self-administered scale using a Likert-type scale from 1 to 6. To
examine the validity of the scale, construct, convergent and divergent validation analyses were performed on the
data from the 280 participants. The reliability of the scale was assessed through analysis of the internal consistency
of each factor, also on the data from the 280 participants. The test-retest reliability of items was performed with
an interval of 4 months on 150 of the 280 participants. In between the two tests, these 150 participants had
participated in an intervention to reduce burnout and death anxiety and improve the quality of their helping
relationship. The statistical package SPSS 19 was used for all analyses.

Development of the Scale — The scale was devised by the corresponding author to measure emotional
exhaustion, depersonalization, and professional fulfillment specifically in HCWs who care for the dying, as opposed
to HCWs in general. It was comprised of items from different origins to assess emotional exhaustion (EE),
depersonalization (D) and professional fulfillment (PF):

A – The Maslach Burnout Inventory (Maslach, Jackson, & Leiter, 1997)

Authorization was received from the Maslach Burnout Inventory’s (MBI) authors to translate questions from the
MBI into Portuguese and to integrate them into the scale. The MBI’s items integrated in the present study are
shown in Table 1.

B – Psychtests Aim Inc. (Jerabek, 2001)

Authorization was received from the authors of PsychTest Aim Inc. to use some of their items, translate them into
Portuguese and to integrate them into the scale. Those items are displayed in Table 2.

C – Questions Developed by the Researcher Specifically for Burnout Related to Working With Dying Patients

A total of 14 items were devised, based on a survey performed with 20 nurses who worked with dying patients,
to understand what their difficulties were. Table 3 presents this information.

Self-reported answers to each item were evaluated using a Likert-type scale, ranging from 1 (disagree completely)
to 6 (agree completely).

Distribution of the Scale — A pilot test was performed among 10 participants who were debriefed to ensure that
all questions were understood as intended. With the ethical consent of the clinical director of each hospital
department, the scale was then distributed to 280 Portuguese HCWs who cared for terminally ill patients and who
agreed to participate in the research project. The study aims at understanding whether an intervention, comprising
both an educational component, as well as a personal introspection component regarding death anxiety, could
reduce burnout and improve the quality of helping relationship skills with patients and their families.

The scale was filled in by HCWs who had not yet received the intervention. They were each given a cover letter
explaining the research project and an identification sheet with demographic data (age, sex, place of work – in or

Psychology, Community & Health
2012, Vol. 1(3), 257–272
doi:10.5964/pch.v1i3.21

Melo & Oliver 259

Table 1

Questions Integrated from the MBI

Emotional exhaustion
1. I feel emotionally drained by my work.
2. I have no strength left at the end of a day’s work.
3. I feel tired when I get up in the morning and have to face another day of work
4. I feel frustrated by my work
5. I feel that I work too hard in my profession
6. Working directly with people causes me a lot of stress
7. I feel worn out
8. Working every day with people is a real burden for me

Depersonalization
9. I feel that I treat many people impersonally, as if they were objects
10. I have become more insensitive to people since I have this job
11. I am afraid this job will make me become emotionally hard
12. I do not pay real attention to what happens to other people
13. I feel that other people censor me because of their own problems

Professional fulfillment (referred to by Maslach as ‘Personal Accomplishment’)
14. I resolve other people’s problems efficiently
15. I can easily understand what other people are experiencing
16. I have a positive influence on the people I coordinate at work
17. I feel energetic
18. It is easy for me to create a relaxed atmosphere with other people
19. I feel fulfilled when I work in close collaboration with others
20. I have accomplished many useful things in this work
21. At work, I deal with emotional problems very calmly

Table 2

Questions Integrated From Psychtest Aim Inc.

Depersonalization
1. I would be incapable of coping with my work if I considered my patients as unique individuals
2. I don’t really care what happens to my patients
3. I cannot afford to answer to the individual needs of my patients.

Professional fulfillment
4. I feel that what I do makes a difference
5. I feel that other people have realistic expectations regarding my working performance

Table 3

Questions Devised by the Researcher

Emotional exhaustion
1. Dealing psychologically with terminally ill patients makes me feel insecure and anxious
2. I feel helpless when faced with the patient’s fragility
3. I am emotionally disturbed by the death of so many patients
4. The relationship with the patient’s family wears me out
5. I am frustrated because I cannot find the time to have a quality relationship with the patient
6. I feel stressed due to lack of debate and support within the team, with regard to our difficulties
7. I ask myself many times if I could have “done more” and this makes me feel anxious

Psychology, Community & Health
2012, Vol. 1(3), 257–272
doi:10.5964/pch.v1i3.21

Burnout Scale for Workers who Care for the Dying 260

Depersonalization
8. I give more importance to the technical part of my work than to the human part
9. I give a lot of importance to treating the illness, but do not have patience for the psychological and
spiritual caring of the patient

Professional fulfillment
10. I feel fulfilled at work because I manage to find time to just “be” with the patient or their family
11. I have moments of sharing with the patients, with no need to hide my feelings
12. My work allows me to value life more
13. I manage to find time in my work to talk to patients and to help them find meaning in their lives
14. I often contribute towards giving my patients quality of life, comfort and dignity at the end of their
life

out of a palliative care unit – level of education, and profession). Anonymity was assured. The scales were collected
for construct, convergent and divergent validation and tests of internal consistency.

Four months later, the scale was distributed again to the 150 participants who had completed the intervention
and test-retest reliability analyses were performed.

Sensitivity — Each item was examined in terms of median, skewness, and kurtosis. Items with a Sk > 3 and Ku
> 8 were eliminated from the scale before performing factor analysis. These are shown in Table 4.

Validation — Construct validity of the 40 items was examined using exploratory factor analysis (principal
components analysis – PCA) with Varimax rotation. Prior to performing the PCA, the suitability of the data for
factor analysis was assessed. Inspection of the correlation matrix revealed the presence of many coefficients of
.3 and above. The Keiser-Meyer-Oklin value was .84, exceeding the recommended value of .6 (Kaiser, 1970,
1974) and Bartlett’s Test of Sphericity (Bartlett, 1954) reached statistical significance, supporting the factorability
of the correlation matrix.

Convergent and divergent validation was performed using Pearson correlations with factors from another
questionnaire measuring death anxiety and the quality of HCWs’ helping relationship with their patients (Gouveia
e Melo, & Oliver, 2011).

Reliability — Once the validity of the domains of the scale had been established, the internal consistency of each
factor was tested using Cronbach’s α coefficients and test-retest reliability was determined by calculating intra-class
correlation coefficients with a 4-month interval (Pearson bivariate correlations).

Calculation of the Cut-Off Point — The cut-off point should be calculated taking into account the sample of
population as follows (see Table 5):

• The sum of each factor (total scores/nr. of participants), is divided by the number of validated items;

• The final score of the two negative factors, is summed and divided by 2;

• The final positive score and negative scores are summed and divided by 2:

Psychology, Community & Health
2012, Vol. 1(3), 257–272
doi:10.5964/pch.v1i3.21

Melo & Oliver 261

Table 4

Medians (Mdn), Skewness (sk) and Kurtosis (ku), Minimum and Maximum Values, to Evaluate Sensitivity of the 40 Items

MaxMinKuSkMdnItemItem ID

Emotional exhaustion (ee)
613.00I feel emotionally drained by my workee01 .95-.08
613.00I feel worn out.ee15 .02-1.03
613.00I feel tired when I get up in the morning and have to face another day of work.ee06 .07-1.33
613.00I have no strength left at the end of a day’s work.ee20 .03-1.27
614.00I feel that I work too hard in my profession.ee23 .76-.32-
614.00I am frustrated because I cannot find the time to have a quality relationship with the patientee13 .66-.54-
613.00Dealing psychologically with terminally ill patients makes me feel insecure and anxiousee21 .03-1.25
614.00I feel stressed due to the lack of debate and support within the team, with regard to our

difficulties.
ee35 .02-1.21-

614.00I feel helpless when faced with the patient’s fragility.ee10 .95-.13-
613.00The relationship with the patient’s family wears me out.ee28 .68-.34
614.00I ask myself many times if I could have “done more” and this makes me feel anxiousee38 .85-.19-
611.00I feel frustrated by my workee08 .601.501
614.00I am emotionally disturbed by the death of so many patients.ee25 .06-1.15-
612.00Working directly with people causes me a lot of stress.ee31 .31.93
612.00I cannot afford to answer to the individual needs of my patients.d27 .45-.76

Professional fulfillment (pf)
614.00I have a positive influence on the people I coordinate at work.pf32 .08.41-
614.00It is easy for me to create a relaxed atmosphere with other people.pf14 .21-.24-
614.00I resolve other people’s problems efficientlypf03 .28.55-
615.00I often contribute towards giving my patients quality of life, comfort and dignity at the end

of their life.
pf40 .03.59-

615.00I feel that what I do makes a differencepf22 .83.85-
614.00At work, I deal with emotional problems very calmly.pf26 .19-.33-
614.00I feel that other people have realistic expectations regarding my working performance.pf29 .61.78-
615.00I have accomplished many useful things in this work.pf19 .504.85-1
614.00I can easily understand what other people are experiencingpf05 .05.53-
615.00I have moments of sharing with the patients, with no need to hid my feelings.pf33 .10-.61-
615.00I feel fulfilled when I work in close collaboration with others.pf17 .653.71-1
614.00I feel energeticpf36 .22-.36-
614.00I manage to find time in my work to talk to patients and to help them find meaning in their

lives.
pf39 .22-.46-

614.00I feel fulfilled at work because I manage to find time to just “be” with the patient and their
family

pf09 .48-.48-

Depersonalization (d)
611.00I don’t really care what happens to my patients

a
d24 .2213.383

611.00I have become more insensitive to people since I have this jobd07 .82.331
612.00I give more importance to the technical part of my work than to the human part.d30 .193.541
612.00I am afraid this job will make me become emotionally hardd12 .76-.70
612.00I do not pay real attention to what happens to other people.d16 .201.191
611.00I give a lot of importance to treating the illness, but do not have patience for the

psychological and spiritual caring of the patient
d34 .824.142

611.00I feel that I treat many people impersonally, as if they were objectsd02 .125.002
611.00Working everyday with people is a real burden for me

a
ee18 .218.572

612.00I would be incapable of coping with my work if I considered my patients as unique
individuals

d04 .32-.90

Note. ee = emotional exhaustion; d = depersonalization; pf = professional fulfillment.
aItems with inadequate sensitivity values.

Psychology, Community & Health
2012, Vol. 1(3), 257–272
doi:10.5964/pch.v1i3.21

Burnout Scale for Workers who Care for the Dying 262

Table 5

Calculation of Cut-Off Point

NegativePositive

MeanTotal scoreFactorsMeanTotal scoreFactor

3.2048.03Nf1: Emotional exhaustion (15 items)4.3560.91Professional fulfillment (14 items)
2.0214.13Nf2: Depersonalization (7 items)
5.22Total negative (Nf1+Nf2)4.35Total positive
2.614.35Mean
3.48Cut-off point = mean of positive and negative factors: (4.35+2.61)/2

Results

Of the 300 scales that were delivered to HCWs who cared for dying patients, 280 were returned (response rate,
93.33%). Of the 280 initial responses, 150 HCWs agreed to take the retest, all of which were returned. Of the 280
scales, 208 were delivered personally and checked immediately for missing data and these were filled in by the
participant at the time. The remaining 72 were checked for missing data at a later date; ten were found having
missing values and were filled in using the mean score of the item.

Sensitivity

Table 4 shows that most items fell within the acceptable range of -3.0 to +3.0 for skewness, and -8.0 to + 8.0 for
kurtosis. Items 18 and 24 were eliminated from the scale before proceeding to the factor analysis. For a Likert
scale of 1 to 6, an acceptable median would be within the range of 3 to 4. This is the case for items relating to
emotional exhaustion. However, results were skewed for items relating to professional fulfillment and
depersonalization. This is to be expected with this sample. Whereas it is feasible for HCWs to suffer from emotional
exhaustion, for example due to excessive contact with death and suffering, and work overload, it is not expected
that they would show attitudes of depersonalization with patients who are fragile, vulnerable and dying. Likewise,
due to the close and caring relationship with patients, this work is professionally rewarding and brings meaning
into the lives of these HCWs. For this reason, these items were maintained for factor analysis.

Construct Validation

Principal components analysis revealed the presence of 11 components with eigenvalues exceeding 1, explaining
60.3% of the variance. An inspection of the scree plot revealed a clear break after the 3rd component. Using
Catell’s Scree test (Cattell, 1966), the researchers decided to retain the three components for further investigation.
Varimax rotation was performed extracting three factors and suppressing absolute values under 0.3 (see Table
6). The three-component solution explained a total of 34.29% of the variance. The rotation sums of squared
loading showed Component 1 contributing 14.99%, Component 2 contributing 11.00% and Component 3 contributing
8.3%. Analysis of the questions of each component show that they correspond to:

• Factor 1: Emotional exhaustion;

• Factor 2: Professional fulfillment;

• Factor 3: Depersonalization.

Two items with coefficients below .3 were eliminated from the scale:

• “My work allows me to value life more”;

Psychology, Community & Health
2012, Vol. 1(3), 257–272
doi:10.5964/pch.v1i3.21

Melo & Oliver 263

• “I feel that other people censor me because of their own problems”.
Table 6

Principal Component Analysis of Burnout Scale: Full Description of the 3 Factors and Coefficients of Each Item, the Alpha C Value and
Variance of Each Factor

BURNOUT (N = 280)

CommunalityFactorLoadingItemItem Id

F1: Emotional exhaustion (α C = 0.86; M = 3.20; SD = .83081; V = 14.99%)
.6940.767I feel emotionally drained by my workee 01
.7380.751I feel worn outee 15
.6620.709I feel tired when I get up in the morning and have to face another day of workee 06
.6200.605I have no strength left at the end of a day’s workee 20
.6690.558Dealing psychologically with terminally ill patients makes me feel insecure and anxiousee 21
.5950.552I feel that I work too hard in my professionee 23
.7040.540I am frustrated because I cannot find the time to have a quality relationship with the patientee 13
.5480.506I feel stressed due to the lack of debate and support within the team, with regard to our difficultiesee 35
.6230.499The relationship with the patient’s family wears me outee 28
.5810.496I feel helpless when faced with the patient’s fragilityee 10
.5990.490I ask myself many times if I could have “done more” and this makes me feel anxiousee 38
.6440.489I am emotionally disturbed by the death of so many patientsee 25
.5560.468I feel frustrated by my workee 08
.5410.460Working directly with people causes me a lot of stressee 31
.5150.307I cannot afford to answer to the individual needs of my patientsd 27

F2: Professional fulfillment (α C = 0.83; M = 4.35; SD = .61583; V = 11.00%)
.5610.729I have a positive influence on the people I coordinate at work.pf 32
.5530.645It is easy for me to create a relaxed atmosphere with other peoplepf 14
.5580.610I resolve other people’s problems efficientlypf 03
.6160.578I often contribute towards giving my patients quality of life, comfort and dignity at the end of their

life
pf 40

.6150.578At work, I deal with emotional problems very calmlypf 26

.6220.549I feel that what I do makes a differencepf 22

.6200.516I feel that other people have realistic expectations regarding my working performancepf 29

.5850.508I can easily understand what other people are experiencingpf 05

.6180.470I have accomplished many useful things in this workpf 19

.6320.423I have moments of sharing with the patients, with no need to hid my feelingspf 33

.6230.414I feel energeticpf 36

.5740.399I feel fulfilled when I work in close collaboration with otherspf 17

.6920.378I manage to find time in my work to talk to patients and to help them find meaning in their livespf 39

.6040.304I feel fulfilled at work because I manage to find time to just “be” with the patient and their familypf 09

F3: Depersonalization (α C = 0.63; M = 2.02; SD = .69603; V = 8.31%)
.7220.618I have become more insensitive to people since I have this jobd 07
.6780.568I am afraid this job will make me become emotionally hardd 12
.7090.471I feel that I treat many people impersonally, as if they were objectsd 02
.6110.454I give more importance to the technical part of my work than to the human partd 30
.5850.417I would be incapable of coping with my work if I considered my patients as unique individualsd 04
.6330.406I do not pay real attention to what happens to other peopled 16
.4880.358I give a lot of importance to treating the illness, but do not have patience for the psychological

and spiritual caring of the patient
d 34

Note. α C = Cronbach’s Alpha; M = mean; SD = standard deviation; V = variance.

Psychology, Community & Health
2012, Vol. 1(3), 257–272
doi:10.5964/pch.v1i3.21

Burnout Scale for Workers who Care for the Dying 264

Convergent and Divergent Validation

The relationships between burnout, death anxiety and quality of helping relationships were investigated using
Pearson’s product-moment correlation coefficient. Preliminary analyses using Q-Q Plots were performed to ensure
no violation of the assumptions of normality, linearity and homoscedasticity had occurred. The results are presented
in the following diagrams. Correlations with an r value > .500 can be considered to be strong, r from .30 to .49
medium, and r from .10 to .29, weak. Figures 1, 2 and 3 display the correlations.

Figure 1. Convergent validation.

Reliability

Cronbach’s alpha coefficients were .86 for factor 1, .83 for factor 2, and .63 for factor 3. In order to establish
test-retest reliability, Pearson bivariate correlations were performed on the 3 factors, with an interval of 4 months.
Intra-class correlation was from .55 to .59 in each domain (Figure 4). We can therefore conclude that all three
factors are reliable both internally and over time.

Psychology, Community & Health
2012, Vol. 1(3), 257–272
doi:10.5964/pch.v1i3.21

Melo & Oliver 265

Figure 2. Divergent validation.

Figure 3. Intra-scale correlations.

Calculation of Cut-Off Point

Justification for Adding Extra Questions to the Scale, Specific for HCWs who Care for the Dying — As
mentioned in the introduction, a review of the literature revealed low levels of burnout in health care workers who

Psychology, Community & Health
2012, Vol. 1(3), 257–272
doi:10.5964/pch.v1i3.21

Burnout Scale for Workers who Care for the Dying 266

Figure 4. Test-retest correlations to test reliability of scale.

care for dying patients when using the MBI. However, qualitative data from these same studies shows that …