Community Dent Oral Epidemiol 2002; 30: 438–48

Printed in Denmark. All rights reserved

David Locker1, Aleksandra Jokovic1,

Family impact of child oral and Marlene Stephens1, David Kenny2,
Bryan Tompson3 and Gordon Guyatt4

oro-facial conditions
1
Community Dental Health Services Research
Unit, Faculty of Dentistry, University of
Toronto, Toronto, 2Department of Dentistry,
Hospital for Sick Children, Toronto,
3
Department of Orthodontics, Faculty of
Locker D, Jokovic A, Stephens M, Kenny D, Tompson B, Guyatt G. Family impact
Dentistry, University of Toronto, Toronto,
of child oral and oro-facial conditions. Community Dent Oral Epidemiol 2002; 30: 4
Department of Clinical Epidemiology and
438–48. ß Blackwell Munksgaard, 2002 Biostatistics, McMaster University, Hamilton,
Canada
Abstract – Objectives: The aim of this study was to develop and evaluate the Family
Impact Scale, a measure of the family impact of child oral and oro-facial disorders. This
formed one component of the Child Oral Health Quality of Life Instrumentß. Methods:
The scale was developed using a process described by Guyatt et al. (1987) and
Juniper et al. (1996). An item pool was developed using a review of existing child
health status and family impact questionnaires, interviews with 41 parents–
caregivers of children with paedodontic, orthodontic and oro-facial conditions and
discussions with dental specialists. The resulting pool of 21 items was used in an item
impact study in which 93 parents–caregivers provided data on the frequency and
importance of these items. The 14 items identified most frequently or rated the most
important were selected for the final questionnaire. The discriminant and construct
validity and internal consistency reliability of this 14-item scale were assessed in a
study of 266 parents–caregivers from the three clinical groups. Seventy-nine of these
participants completed a second copy of the questionnaire to facilitate assessment of
test–retest reliability. Results: Family Impact Scale scores ranged from 0 to 33,
indicating that the measure was sensitive to variations in family impact. Floor effects
were minimal with only 10.2% of subjects having a score of zero and there were no
ceiling effects, that is, subjects with maximum scores. Almost three-quarters of
parents–caregivers reported some family impact ‘sometimes’ or ‘often/everyday’ over the
previous 3 months. Impact on parental or family activities of this frequency
was reported by 53.0%, impact on parental emotions by 44.0%, conflict in the
family by 31.6% and financial difficulties by 31.2%. The measure and its component
items were reasonably good at discriminating between the three clinical groups
included in the study and showed good construct validity. It had excellent internal Key words: family impact; child oral health;
consistency reliability with a Cronbach’s alpha of 0.83 and was reproducible for reliability; validity
parent–caregivers who reported that their child’s condition was stable (ICC ¼ 0.80). Dr David Locker, Faculty of Dentistry,
Conclusions: The study provides some data to suggest that child oral and oro-facial University of Toronto, 124 Edward Street,
conditions have a pervasive impact on the family. The Family Impact Scale had Toronto, Ontario M5 G 1G6, Canada
good technical properties. Its evaluative properties need to be tested in longitudinal Submitted 5 October 2001;
studies. accepted 20 March 2002

In the two decades since Cohen & Jago (1) first
advocated the development of sociodental indica-
tors, considerable effort has been invested in
research into instruments designed to measure what
is now referred to as oral health-related quality of
life. To date, a number of multidimensional indexes
have been constructed, most of which have been
shown to have good technical properties (2). How-
ever, all were developed to assess oral health out-
comes in adult and elderly populations. None were
designed to measure the oral health-related quality
of life of children. This is a significant omission since
children are subjected to numerous oral and oro-
facial conditions, including dental caries, malocclu-
sions, cleft lip and palate and craniofacial anomalies,
all of which have the potential to significantly
impact on the quality of life. Moreover, it is highly
unlikely that any of the measures developed for
adult and elderly populations would be suitable
for children given their conceptual basis, item

438

content and response formats. Measures for children
need to be appropriate to the age-related activities
and roles of the child (3).
To a certain extent, this situation parallels that in
medicine where health status measures for children
are less well developed than those for adults. How-
ever, recent increases in the prevalence of children
living with chronic physical or psychosocial condi-
tions, the growing concern with outcomes in health
services research and the growing demand for well-
designed and validated outcome measures have
spawned several generic and disease-specific instru-
ments for use in paediatric health assessment (4).
There are two broad concepts that need to be add-
ressed with respect to the outcomes of oral and oro-
facial conditions in children. The first is the child’s
oral health-related quality of life and the second is
the impact of the child’s condition on the family (3).
There are a number of reasons why a Family Impact
Scale is an essential component of a child health-
related quality of life measure (5). These are: (i) the
central role played by the family in child health; (ii)
the likelihood that chronic illness in a child will
impact on the family to some degree; (iii) the fact
that health care interventions often address parental
needs and concerns as well as the child’s and (iv) the
fact that parental reports of a child’s health may be
influenced by the degree to which the parent is
physically or emotionally affected by the child’s
condition. Family impact data are then essential in
order to assess this ‘caregiver-burden’ bias.
In addition, contemporary concepts of child health
refer to both the child and the family. The American
Academy of Paediatrics defines child health as ‘the
social, physical and emotional functioning of the
child and, when indicated, his or her family . . .
therefore, measurement of health-related quality
of life must be from the perspective of the child
and the family’ (6). Consequently, a number of
measures of child health-related quality of life con-
tain items pertaining to the family and a limited
number of scales have been developed specifically to
assess family impact.
For example, the Child Health Questionnaire (7),
contains questions that assess the impact of a child’s
illness on parents’ time, the emotional impact on
parents, family cohesion and the effects on the
family’s day-to-day activities. The Impact-on-Family
Scale (8) measures the effects of child illness on
family finances, the quality and quantity of interac-
tion within and without the family unit and the
subjective distress and strain experienced by the pri-
mary caregiver. The Paediatric Asthma Caregiver’s
Child oral and oro-facial conditions
Quality of Life Questionnaire (9) addresses the emo-
tional functioning and activity limitations of parents
and caregivers of children with asthma.
This paper describes the development and evalua-
tion of a scale to measure the family impact of child
oral and oro-facial conditions. The aim was to pro-
duce a generic measure that could be used in asses-
sing the outcomes of clinical interventions for
children with these conditions. The paper also pro-
vides some preliminary data on the nature and
extent of family impact and variations across groups
affected by different oral disorders.
Methods
The Family Impact Scale
The Family Impact Scale was developed as one
component of the Child Oral Health Quality Of Life
Instrument (COHQOL)ß. This instrument was
designed to assess the oral health-related quality
of life of children aged 6–14 years with oral and
oro-facial conditions. It consists of a Parental–Care-
giver Perceptions Questionnaire and two analogous
Child Perceptions Questionnaires, one for children
aged 6–10 years and the other for children aged 11–
14 years. The Parental–Caregiver Perceptions Ques-
tionnaire measures parental or caregiver percep-
tions of a child’s oral health-related quality of life
and the impact of the child’s condition on the family.
The Child Perceptions Questionnaires measure chil-
dren’s perceptions of their oral health-related qual-
ity of life. Both parental–caregiver and child
questionnaires contain items which address oral
symptoms, functional limitations, emotional and
social well-being, i.e. schooling, peer interaction
and leisure activities. However, because of the emo-
tional content of the questions, the Family Impact
Scale is included in the parental–caregiver question-
naire only.
Since the aim was to develop a generic oral health
quality of life instrument that could be used as an
outcome measure in clinical trials or evaluation
research, three clinical groups participated in the
development of the COHQOL. These were paedia-
tric dental patients, most of whom had high levels of
dental decay but also other conditions such as gin-
givitis or amelogenesis imperfecta, orthodontic
patients and patients with cleft lip or palate or
craniofacial anomalies. Parents/caregivers and chil-
dren who participated in the development process
were recruited from the paediatric and orthodontic
clinics at the Faculty of Dentistry, University of
439
Locker et al.
Toronto; dental public health clinics operated by the
city of Toronto and dental and craniofacial clinics at
the Hospital for Sick Children, Toronto. Inclusion
criteria were that both parent–caregiver and child
spoke English, the child had no other systemic or
developmental disorders and the child’s treatment
should not have started or be in its early stages.
All procedures used in developing the question-
naires, including the Family Impact Scale, were
approved by the University of Toronto Human
Subjects Certification Committee and the Hospital
for Sick Children Ethics Board.
Development process
The questionnaires were developed using the pro-
cess described by Guyatt et al. (10) and Juniper et al.
(11). This is summarized in Fig. 1. For the parental–
caregiver questionnaire, a preliminary pool of 54
items was generated by a review of existing oral
health-related quality of life measures, generic and
disease-specific health status questionnaires for chil-
dren which included a parental–caregiver compo-
nent and measures of the family impact of child
chronic conditions. Eight of these items concerned
family impact. The face and content validity of this
preliminary pool was assessed by means of quali-
tative interviews with parents–caregivers of chil-
dren with oral and oro-facial conditions. The main
aim of this stage was to ensure the comprehensive-
ness of the item pool. Participants were asked to
comment on the relevance and clarity of each item
and to identify family impacts not covered by the
item pool. Although there is no consensus on the
440
sample size needed for a content validation study
(12), Guyatt et al. (13) suggest that 50–100 semistruc-
tured interviews be undertaken to create the item
pool. However, we used an iterative process and
terminated data collection when 41 interviews had
been completed and no new information regarding
the item pool was being obtained. The preliminary
pool was also reviewed by 15 clinicians who provide
care to children with the conditions of interest.
Following this process, a revised item pool was
developed consisting of 68 items, of which 21
addressed family impact. Many items were rewrit-
ten to reduce ambiguity and enhance clarity.
Items for the questionnaires were selected using
the item impact method. This method is used to
reduce the number of items, to select items of most
importance to study subjects, to promote the respon-
siveness of the measure, that is, its ability to detect
change, and to transform health status items into a
health-related quality of life instrument (14, 15). For
the Parental–Caregiver Perceptions Questionnaire,
including the Family Impact Scale, a convenience
sample of parents and caregivers of children from
the three clinical groups described above completed
an item impact questionnaire having the following
format. For each item in the revised pool, parents
were asked, ‘During the last three months, how often
. . . (has there been disagreement or conflict in your
family). . . because of your child’s condition’.
Response options were: never, sometimes, often or
all the time. Parents responding sometimes, often or
all the time were then asked, ‘How much has this
bothered you?’. The response options and codes
Fig. 1. Summary of the process used to
develop the Family Impact Scale.

for this question were: not at all ¼ 0; a little ¼ 1; quite
a bit ¼ 2; very much ¼ 3. The first question measured
the frequency of the event described by the item
while the second measured its importance. This
questionnaire was administered in fact-to-face inter-
views.
An impact score was calculated for each item by
multiplying the percentage of subjects responding
sometimes, often or all the time to the first question
by the mean importance rating attributed by subjects
with those responses. Items were then ranked
according to their impact scores. Since our main
aim was to develop an evaluative measure which
would be responsive to change, the top scoring items
were selected for the ‘final’ questionnaire (16). This
questionnaire contained 45 items, 31 of which
assessed parental–caregiver perceptions of the
child’s functional, emotional and social well-being,
and 14 assessed family impact.
In the final questionnaire, questions were asked
about the frequency of events in the previous
3 months; data on importance were not collected.
Based on the comments of parents–caregivers taking
part in the item impact study, the response options
with respect to frequency were modified as follows:
never ¼ 0; once or twice ¼ 1; sometimes ¼ 2;
often ¼ 3; everyday or almost everyday ¼ 4. A ‘Don’t
know’ response was also allowed. Parents and care-
givers were also asked to give overall or global
assessments of the child’s oral health and the extent
to which the oral or oro-facial condition in question
affected the child’s overall well-being. These global
ratings had a 5-point response format from ‘Excel-
lent’ to ‘Poor’ for oral health and from ‘Not at all’ to
‘Very much’ for well-being. This questionnaire was
self-administered.
Assessment of validity and reliability
The technical properties of the ‘final’ questionnaire
were assessed by means of a validity and reliability
study. A new convenience sample of parents–care-
givers recruited from the same clinical locations
completed a copy of the final questionnaire and after
a period of 2 weeks, a subgroup completed a second
copy. Since test–retest reliability analysis requires
that individuals are stable between the two admin-
istrations of the questionnaire (9), the second ques-
tionnaire asked parents–caregivers if the child’s
condition had changed since recruitment. Since pae-
dodontic patients are likely to exhibit short-term
change as a result of dental treatment, only par-
ents–caregivers of children with orthodontic or oro-
facial conditions were included in this component of
Child oral and oro-facial conditions
the study. Clinical data pertaining to the oral and/or
oro-facial conditions in question were abstracted
from the dental and medical charts of children
whose parents–caregivers participated in the study.
A Family Impact Scale score was generated by
summing the numerical response codes for each
item. Subscales were identified using factor analysis
(principal components with varimax rotation and
Kaiser normalization) and subscale scores com-
puted. Discriminant validity was assessed by com-
paring item, overall and subscale scores for the three
clinical groups included in the study. Since the items
were scored using an ordinal scale, nonparametric
statistical procedures (Kruskal–Wallis tests) were
used to examine differences between groups, while
parametric statistical procedures (one-way analysis
of variance tests) were used to examine differences
in scale and subscale scores. The hypotheses tested
were that the oro-facial group would have the high-
est scores, the paedodontic group the lowest with
the orthodontic group in between. Where possible,
within-group analyses were also conducted to
examine the association between family impact
scores and data on the nature and severity of the
child’s condition. Construct validity was assessed by
means of associations between scale scores and the
two global indicators of health and well-being. Here,
we hypothesized that scores would be highest where
the child’s oral health was rated as poor and the
condition assessed as affecting the child’s overall
well-being very much. The correlation between
scores on the Family Impact Scale and a score
obtained by summing the responses to the 31 items
dealing with the child’s oral health-related quality of
life was also computed. We expected a strong posi-
tive correlation.
Internal consistency reliability was assessed by
means of Cronbach’s alpha, and test–retest reliabil-
ity by means of the intraclass correlation coefficient
using a one-way random effects parallel model
(17).
Sample size calculations
Guyatt et al. (13) and Juniper et al. (11) suggest that
between 50 and 100 subjects should be interviewed
for the item impact stage of questionnaire develop-
ment. This estimate is based on a 95% confidence
interval of 10–15% for an observed proportion of
50%. No recommendations are given regarding the
sample size required for discriminant and construct
validity testing. Consequently, we arbitrarily set the
sample size for this phase at a minimum of 180, or
60 parents for each of the three clinical groups.
441
Locker et al.

Table 1. Characteristics of informants and children for the three studies

Item impact (n ¼ 93), Validity and reliability (n ¼ 278), Test–retest reliability (n ¼ 79),
number (%) number (%) number (%)

Informant
Child’s mother 55 (59.1) 197 (70.9) 61 (77.2)
Child’s father 33 (35.5) 72 (25.9) 16 (20.3)
Other 5 (5.4) 9 (3.2) 2 (2.5)
Clinical group
Paedodontic 20 (21.0) 68 (24.8) 0 (0)
Orthodontic 31 (33.3) 124 (45.3) 36 (44.3)
Oro-facial 42 (45.2) 82 (29.9) 43 (54.4)
Gender of child
Boy 44 (47.3) 132 (47.5) 31 (39.2)
Girl 49 (52.7) 146 (52.5) 48 (60.8)
Age of child
6–10 years 26 (28.0) 103 (37.1) 28 (35.4)
11–14 years 67 (72.0) 175 (62.9) 51 (64.6)

A sample size of this magnitude is more than suffi- ‘Don’t know’ or had missing values for three or more
cient (n ¼ 153) to detect a correlation coefficient as items. Seventy-nine parents–caregivers completed
low as 0.20 with a one-tailed test when the null two copies of the questionnaire and provided data
hypothesis is that r ¼ 0; alpha ¼ 0.05 and beta ¼ 0.20 for test–retest reliability assessment. The majority of
(18). Using a commonly applied rule of thumb the informants were the mothers of the child patients
regarding sample sizes for factor analysis, that is, in question (Table 1). The age and gender character-
10 subjects per item, 140 subjects are necessary to istics of these children are shown in Table 1. The
assume that a factor loading of 0.4 is significant (19). clinical characteristics of the child patients whose
The sample size for the test–retest reliability study parents–caregivers took part in the validity and
was derived from Donner & Eliasziw (20) who reliability study are shown in Table 2. The mean
provide contours that display the required number number of decayed teeth among the 62 children in
of subjects and the number of repeated measure- the paedodontic group who had a diagnosis of caries
ments for testing hypotheses concerning intraclass
correlation coefficients with alpha set at 0.05 and
Table 2. Clinical characteristics of children whose parents–
beta at 0.20. In a test–retest reliability study, the caregivers took part in the validity and reliability study
number of measurements is two and the null (n ¼ 278)
hypothesis should be set at ICC ¼ 0.61, correspond-
Group Diagnosis Number (%)
ing to the minimum value indicating substantial
agreement between the two administrations of the Paedodontic Caries only 37 (54.4)
questionnaire (21). Consequently, 42 subjects are Caries and gingivitis 10 (14.7)
Caries and other 15 (22.1)
required to ensure that an ICC of 0.80 is significantly Amelo/dentinogenesis 4 (5.9)
different from the null. These sample sizes were imperfecta
confirmed using formulas derived from Kraemer Trauma 2 (2.9)
& Thiemann (18) as applied by Brent et al. (19). Orthodontic Class I 26 (21.5)
Class I Division I 47 (38.3)
Class II Division II 8 (6.6)
Class III 11 (9.9)
Results Oligodontia 7 (5.8)
Not known 21 (17.3)
Participants
Oro-facial Isolated cleft lip 12 (14.6)
Ninety-three parents–caregivers took part in the Isolated cleft palate 6 (7.3)
item impact study and 278 in the study to assess Unilateral cleft lip/palate 24 (29.3)
discriminant and construct validity and internal Bilateral cleft lip/palate 17 (20.7)
consistency reliability. Twelve of the latter were Craniofacial anomaly 22 (26.8)
Not known 1 (1.2)
excluded from the analysis because they responded

442

was 5.2 (SD ¼ 3.2) and the mean number of decayed
surfaces was 10.2 (SD ¼ 7.4).
Item impact study
Of the 14 items selected for the final questionnaire,
five had prevalences of 50–75%, six had prevalences
of 20–49% and only three had prevalences of less
than 20%. The majority had mean importance rat-
ings of 1.8 or more. The impact scores of these items
ranged from 23.1 to 141.7. The seven items not
included and their prevalences and impact scores
were: having to cancel or change plans (13.0%; 16.3);
avoided or been unable to go out (8.6%; 14.7); child
refused to take part in family activities (11.0%; 9.9);
limitations in the type of activities undertaken by the
family (9.7%; 9.7); family’s ability to pick up and go
at a moment’s notice limited (4.3%; 5.4); family foods
changed (7.4%; 1.9) and unable to get a baby sitter
(0.0%; 0.0).
Factor analysis and identification of subscales
Data from the 266 subjects taking part in the relia-
bility and validity study were used for the factor
analysis. Thirteen of the 14 items were included in
this analysis. The remaining item, which addressed
financial difficulties, was excluded, since it
addressed economic rather than psychosocial or
behavioural impacts. This item was analysed sepa-
rately when subscale analyses were being under-
taken. Three factors were identified which explained
54.7% of the variance. Based on their item content,
they were labelled: (i) parental and family activities
(five items); (ii) parental emotions (four items) and
(iii) family conflict (four items). The loadings of the
items on these factors are shown in Table 3.
Table 3. Results of the factor analysis: item loadings on the three factors
Item
Parental/family activity
Have you or the other parent taken time off work?
Has your child required more attention from you or the other parent?
Have you or the other parent had less time for yourselves or other family members?
Has your sleep or that of the other parent been disrupted?
Have family activities been interrupted?
Parental emotions
Have you or the other parent been upset?
Have you or the other parent felt guilty?
Have you or the other parent worried that your child will have fewer life opportunities?
Have you felt uncomfortable in public places?
Family conflict
Has your child argued with you or the other parent?
Has your child been jealous of you or other family members?
Has your child’s condition caused disagreement or conflict in the family?
Has your child blamed you or the other parent?
Child oral and oro-facial conditions
Nature and extent of family impact
Scores for the overall scale ranged from 0 to 33, with
a mean of 8.28 and a standard deviation of 6.83. This
indicates that adequate variability is detected by the
measure to locate subjects on a continuum of family
impact (22). For the overall scale, floor effects were
minimal, with only 10.2% of subjects having a score
of zero. There were no ceiling effects, that is, subjects
with maximum scores.
Table 4 shows the distribution of responses to the
14 family impact items by the 266 parent–caregiver
informants included in the validity study. Using the
per cent responding ‘sometimes’ and ‘often/every-
day’ as an indicator, the most commonly reported
issues were the child requiring more attention,
financial difficulties, taking time off work, parents
feeling guilty, worried and upset about the child’s
condition and the child being argumentative. Each
of these seven items was reported by more than one-
fifth of the informants. Overall, 72.9% of the infor-
mants reported some family impact ‘sometimes’ or
‘often/everyday’ from the child’s oral or oro-facial
condition during the previous 3 months. Impact on
parental or family activities of this frequency was
reported by 53.0%, impact on parental emotions by
44.0%, conflict in the family by 31.6% and financial
difficulties by 31.2%.
Discriminant and construct validity
Using the Kruskal–Wallis one-way analysis of var-
iance, significant differences across the three clinical
groups were observed for 10 of the 14 items. For six
of these items, the oro-facial group had the highest
scores. Two of these (time off work, P < 0.01; child
required more attention, P < 0.05) came from the
Factor 1 Factor 2 Factor 3
0.640
0.450
0.763
0.652
0.700
0.622
0.786
0.784
0.476
0.545
0.495
0.604
0.803
443
Locker et al.

Table 4. Distribution of responses to the family impact items (n ¼ 266)

During the past 3 months, how often . . . Never Once or twice Sometimes Often/everyday

Parental/family activity
Have you or the other parent taken time off work? 44.0 30.8 19.5 5.6
Has your child required more attention from you or 45.5 21.8 26.3 6.4
the other parent?
Have you or the other parent had less time for 64.7 19.9 13.9 1.5
yourselves or other family members?
Has your sleep or that of the other parent been 71.1 18.0 10.2 0.8
disrupted?
Have family activities been interrupted? 72.6 13.9 12.0 1.5
Parental emotions
Have you or the other parent been upset? 50.4 26.3 20.3 3.0
Have you or the other parent felt guilty? 58.3 16.9 18.0 6.8
Have you or the other parent worried that your child 63.5 13.9 18.4 4.1
will have fewer life opportunities?
Have you felt uncomfortable in public places? 85.3 6.8 6.8 1.1
Family conflict
Has your child argued with you or the other parent? 50.4 26.3 18.4 4.9
Has your child been jealous of you or other family 70.3 18.0 9.4 2.3
members?
Has your child’s condition caused disagreement or 74.1 16.9 6.4 2.6
conflict in the family?
Has your child blamed you or the other parent? 89.1 5.3 5.3 0.4
Financial burden
Has your child’s condition caused financial 55.6 13.2 17.3 13.9
difficulties for your family?

parental/family activity subscale, three from the
parental emotions subscale (upset, P < 0.05; worried
about child’s opportunities, P < 0.001; felt guilty,
P < 0.01) and one from the family conflict subscale
(child jealous of others, P < 0.001). The paedodontic
group had the highest score for three items, two for
parental/family activity (family activities inter-
rupted, P < 0.05; parents’ sleep disrupted, P < 0.05)
and one for family conflict (disagreement or conflict
in the family, P < 0.05). The orthodontic group had
the highest score for the item referring to financial
difficulties only (P < 0.0001). These data suggest that
the nature of the family impact varied across the
clinical groups but was more pervasive for the oro-
facial group. They also suggest that most of the
individual items discriminated between the clinical
groups.
Overall Family Impact Scale scores for the three
clinical groups were highest for the oro-facial group
(mean ¼ 9.40) and lowest for the paedodontic group
(mean ¼ 7.33), with the orthodontic group falling
between the two (median ¼ 8.25). These differences
were not statistically significant. A similar pattern
was observed with respect to the proportions in each
clinical group reporting one or more items ‘some-
times’ or ‘often/everyday’ (Table 5). When the three
multi-item subscales were examined, the groups
were significantly different with respect to the par-
ental emotions subscale. Differences in mean scores
on the parent/family activity subscale almost
reached significance with a P-value of 0.07. The
proportions reporting financial difficulties as a
result of the child’s oral condition were 19.7% for
the paedodontic group, 49.1% for the orthodontic
group and 16.3% for the oro-facial group (P < 0.0001;
chi-square test).
Within-group analysis indicated that the correla-
tion between Family Impact Scale scores and the
number of decayed tooth surfaces for the paedodon-
tic group with a diagnosis of caries was 0.30
(P < 0.05). This analysis could not be undertaken
with the orthodontic group, since the diagnostic
classification was nominal rather than ordinal. In
the oro-facial group, the number of children with
each diagnosis was too small to allow for statistical
testing.
With respect to construct validity, there were sig-
nificant associations between the Family Impact
Scale score and parent–caregiver global ratings of
the child’s oral health (P < 0.001) and overall well-
being (P < 0.0001) in the expected direction. For
example, the mean score of those reporting that
the child’s overall well-being was ‘not at all’ affected
by their condition was 4.28, while the mean score for

444
 Child oral and oro-facial conditions

Table 5. Mean scale and subscales statistics by clinical group

Paedodontic (n ¼ 67) Orthodontic (n ¼ 117) Oro-facial (n ¼ 82) P

Family Impact Scale

Mean score 7.33 8.25 9.40 NS
Per cent 68.2 73.3 77.5 NS
Subscales
Parent/family activity
Mean score 3.36 2.78 3.73 0.07
Per cent 48.5 49.1 62.5 NS
Parental emotions
Mean score 1.86 2.21 3.21 <0.01
Per cent 34.8 41.4 57.5 <0.05
Family conflict
Mean score 1.53 1.84 1.94 NS
Per cent 25.8 34.5 33.8 NS

P-values obtained from one-way analysis of variance or chi-square tests.

With one or more items with response of sometimes, often or everyday.

those reporting that it was affected ‘very much’ was
14.67. As expected, the association was stronger for
global ratings of well-being (r ¼ 0.47; P < 0.001) than
for global ratings of oral health (r ¼ 0.24; P < 0.001).
In addition, the global rating of overall well-being
was associated with scores on all three subscales as
well as the financial difficulty item (P < 0.0001 for all
analyses), while the global rating of oral health was
associated with the parental emotions subscale score
(P < 0.0001) and the single item concerning financial
difficulties (P < 0.0001).
The correlation between the family impact score
and the score derived from the 31 items addressing
parental–caregiver perceptions of the child’s physi-
cal, social and emotional functioning was 0.82
(P < 0.001). There was no association between the
family impact score or the subscale scores and the
child’s gender or age or the gender of the parent–
caregiver informant.
Reliability
Cronbach’s alpha for the 14-item scale was 0.83,
indicating good internal consistency reliability. This
was not improved by the deletion of any item. All
item–total score correlations were significant at the
P < 0.001 level. Alphas for the three multi-item sub-
scales were: parental and family activity, 0.72;
parental emotions, 0.70 and family conflict, 0.64.
The test–retest reliability assessment was based on
data from 56 parents–caregivers whose children had
orthodontic or oro-facial conditions. The remaining
24 cases taking part in the test–retest study were not
included because the informant indicated that the
child’s condition had changed between the two
administrations of the questionnaire, or because a
different parent–caregiver completed the second
questionnaire. The intraclass correlation coefficient
was 0.80, indicating substantial to perfect agreement
(20). Since internal consistency estimates approxi-
mate test–retest estimates (8), Cronbach’s alpha was
calculated separately for the 66 parents–caregivers
whose children were in the paedodontic clinical
group and who were not included in the test–retest
study because of the high probability of short-term
change. The alpha for these cases was 0.83.
Discussion
Although the issue of the family impact of oral and
oro-facial conditions was first raised by Sheiham &
Croog (23) in 1981, to date, no measure of this
dimension of oral health-related quality of life has
been reported. In order to begin to fill this gap in the
literature, this paper describes the development and
evaluation of a Family Impact Scale that can be used
in oral health research. The aim was to produce a
generic measure, applicable to a wide range of oral
and oro-facial conditions, that would serve as an
outcome measure in clinical trials or evaluation
research. The measure we produced consisted of
14 items that attempted to capture the effect of a
child’s oral or oro-facial condition on parental and
family activities, parental emotions, family conflict
and family finances.
Given our measurement goals, we used the devel-
opment process described by Guyatt et al. (10) and
Juniper et al. (11). This has been used in developing

445
Locker et al.
numerous measures of disease-specific health-
related quality of life questionnaires including a
measure of the quality of life of the parents of
children with asthma (9). The defining characteristic
of this approach is that it uses the item-impact
method, rather than statistical procedures such as
factor analysis, to select items for the questionnaire.
In practical terms, this means that the questionnaire
consists of items describing frequent problems or
problems that cause the maximum bother to those
affected by the conditions in question. In theoretical
terms, it transforms a health status measure into a
measure of health-related quality of life by ensuring
that its items address issues of importance to the
patient population being studied (14). In addition,
by maximizing the number of high prevalence
items in the measure, it also reduces floor effects
and promotes sensitivity to change (15). Here, this
approach was used to select 14 items from a list of
21 describing various aspects of family impact.
Using these 14 items, Family Impact Scale scores
ranged from 0 to 33, indicating that the measure was
sensitive to variations in family impact among the
sample included in the study. Floor and ceiling
effects were minimal, suggesting that the measure
is likely to perform well in studies intended to
measure change. Since the study described here
was cross-sectional, the ability of the measure to
detect small but important changes and the long-
itudinal construct validity of change scores derived
from repeat administrations of the measure could
not be assessed. Since our intention was to produce
an evaluative measure, properties such as respon-
siveness and minimal important differences need
to be determined in longitudinal studies which
involve changes in subjects’ health status. This is
the next step in evaluating our Family Impact
Scale.
In terms of the discriminant validity of the mea-
sure, significant differences were observed across
the three clinical groups included in the study for 10
of the scale’s 14 items. However, the pattern of the
association differed across items. The paedodontic
clinical group had the highest item scores for items
such as sleep disturbance and interruption in family
activities; the orthodontic group had the highest
score for the financial difficulties item and the
oro-facial group had the highest scores for six items
including time off work, parents feeling guilty and
worried about their child’s future, and the child
being jealous of other family members. Neverthe-
less, there is a certain logic to these associations. The
items most frequently reported by parents–care-
446
givers of the paedodontic group, almost all of whom
had high levels of dental decay, were those most
likely to be affected by pain. The fact that the ortho-
dontic group reported more financial difficulties
reflects the pattern of public funding for oral and
oro-facial conditions in the city of Toronto. The more
extreme scores of the oro-facial group on two items
from each of the three multi-item subscales reflects
the relative severity of these conditions.
This inconsistent pattern of associations probably
explains why overall scale scores did not discrimi-
nate between the three clinical groups, even though
they differed in terms of clinical severity. The fact
that the overall scale scores of the oro-facial group
were not as extreme as expected may also reflect the
fact that these patients have been the recipients of
high-quality specialist clinical and psychosocial care
since birth which helps families adjust by mitigating
many of the effects of the condition itself. By con-
trast, paedodontic conditions such as caries and
orthodontic problems emerge later in life and, in
the case of the former, may develop over the short-
term. Also, the paedodontic and orthodontic sam-
ples were recruited prior to the start of treatment or
immediately after treatment had been initiated, so
little had been done to alleviate the problems they
experienced. It is also possible that the exclusion of
low-prevalence items compromised the ability of the
measure to discriminate between groups (15). In
studies that aim to discriminate rather than evaluate,
the performance of the measure may be improved
by the addition of some of the seven low-prevalence
items that were dropped at the item impact stage. It
should be noted, however, that the parental emo-
tions subscale did discriminate between groups,
with means and proportions being the highest for
the oro-facial group and lowest for the paedodontic
group as predicted.
All hypotheses pertaining to construct validity
were confirmed. Overall scale scores showed sig-
nificant associations with parent–caregiver global
ratings of their child’s oral health and overall
well-being in the direction expected. There was also
a strong and significant correlation between family
impact scores and a score derived from those items
which measured parental–caregiver perceptions of
the child’s oral health-related quality of life. How-
ever, when evaluating these associations, the possi-
bility of caregiver burden bias needs to be kept in
mind. As mentioned in the introduction, proxy
reports of a child’s distress may reflect parent–care-
giver distress rather than that of the child (3). Con-
sequently, a better test of construct validity would be

the association between parental–caregiver report of
family impact and the child’s self-report of oral
health-related quality of life. Data are currently
being collected from parent/caregiver–child pairs
using both the parent–caregiver and child question-
naires so that these associations can be examined.
The discriminant validity of the scale also needs to
be tested by further studies examining variations
within clinical groups according to the severity of
the child’s condition. Although we collected clinical
data during this study, most of the clinical ratings
took the form of nominal rather than ordinal diag-
nostic classifications or the numbers of children with
specific diagnoses were small, thereby precluding
testing of variations in family impact within groups.
However, within the paedodontic group, there was
a significant moderate correlation between family
impact scores and the number of decayed tooth
surfaces, indicating more family impact with more
extensive untreated decay.
The reliability of the scale was clearly demons-
trated. Internal consistency reliability and test–retest
reliability statistics were both excellent, with a
Cronbach’s alpha and intraclass correlation coeffi-
cient of 0.83 and 0.80, respectively. The intraclass
correlation coefficient demonstrated that the scale
was reproducible in groups who are stable over
time.
The study also provided some data concerning the
nature and extent of family impact resulting from
oral and oro-facial conditions in children. These data
demonstrate the pervasive effects these conditions
can have on the functioning of parents–caregivers
and the family as a whole. Almost three-quarters of
the parents–caregivers reported frequent family
impact in the previous 3 months. The most common
were the child requiring more attention, financial
difficulties, taking time off work, parents feeling
guilty, worried and upset about the child’s condition
and the child being argumentative. These data indi-
cated that oral and oro-facial conditions affect parent
and family activities, impact on parental emotions
and can result in conflict in the family. While these
effects may not be as prominent as those observed
with chronic conditions such as asthma and cancer,
they are an important source of family distress and
should be taken into account when measuring child
oral health. However, since the study was based on
convenience samples, the prevalence estimates and
scores apply only to those who took part in the
study. Further data are required from different
locations and different patient populations to verify
the technical properties of the measure we have
Child oral and oro-facial conditions
developed and to confirm and extend the findings
on family impact reported here.
Acknowledgements
The study on which this paper was based was supported
by the Hospital for Sick Children Foundation, Grant XG99-
085.
References
1. Cohen L, Jago J. Toward the formulation of socio-
dental indicators. Int J Health Serv 1976;6:681–7.
2. Slade GD, editor. Measuring oral health and quality
of life. Chapel Hill, University of North Carolina:
Department of Dental Ecology; 1997.
3. Osman L, Silverman M. Measuring quality of life for
young children with asthma and their families. Eur
Respir J 1996;9(suppl.):35s–41s.
4. Jenney M, Kane R, Lurie N. Developing a measure of
health outcomes in survivors of childhood cancer: a
review of the issues. Med Pediatr Oncol 1995;24:
145–53.
5. Rothman ML, Hendrick SC, Blucroft KA, Hickman
DH, Rubenstein LZ. The validity of proxy generated
scores as measures of patient health status. Med Care
1991;29:115–24.
6. Fink R. Issues and problems in measuring children’s
health status in community health research. Soc Sci
Med 1989;29:715–9.
7. Landgraf JM, Abetz L, Ware JE Jr. Child Health
Questionnaire: a users’ manual. Boston, MA: The
Health Institute, New England Medical Center; 1996.
8. Stein R, Riessman C. The development of the Impact-
on-Family Scale: preliminary findings. Med Care
1980;18:465–72.
9. Juniper EF, Guyatt GH, Feeny D, Ferrie P, Griggith L,
Townsend M. Measuring quality of life in parents of
children with asthma. Qual Life Res 1996;5:27–36.
10. Guyatt GH, Berman LB, Townsend M, Pugsley SO,
Chambers LW. A measure of quality of life for
clinical trials in chronic lung disease. Thorax 1987;42:
1394–7.
11. Juniper EF, Guyatt GH, Jaeschcke R. How to develop
and validate a new health-related quality of life
instrument. In: Spiker, B, editor. Quality of life and
pharmacoeconomics in clinical trials, 2nd edn. Phila-
delphia: Raven Publishers; 1996.
12. DeVellis R. Scale development: theory and applica-
tions. Newbury Park: Sage Publications; 1991.
13. Guyatt G, Bombardier C, Tugwell P. Measuring
disease-specific quality of life in clinical trials. CMAJ
1986;134:889–95.
14. Guyatt GH, Cook DJ. Health status, quality of life and
the individual. JAMA 1994;272:630–1.
15. Locker D, Allen PF. Developing short form measures
of oral health-related quality of life. J Public Health
Dent 2002;62:13–20
16. Juniper EF, Guyatt GH, Streiner D, King D. Clinical
impact versus factor analysis for quality of life
questionnaire construction. J Clin Epidemiol
1997;50:233–8.
447
Locker et al.
17. Shrout PE, Fleiss JL. Intraclass correlation coefficients:
uses in assessing rater reliability. Psych Bull 1970;86:
420–8.
18. Kraemer C, Thiemann S. How many subjects? Statis-
tical power analyses in research. Beverly Hills: Sage;
1987.
19. Brent E, Mirelli J, Thompson A. EX-SAMPLE: an
expert system for determining sample size, Version 3.
Columbia, MO: Idea Works Inc.; 1993.
20. Donner A, Eliasziw M. Sample size requirements for
reliability studies. Stats Med 1987;6:441–8.
448
21. Landis J, Koch G. The measurement of observer
agreement for categorical data. Biometrics 1977;33:
159–74.
22. Varni J, Rode C, Seoid M, Katz E, Friedman-Bender
A, Quiggins D. The Pediatric Cancer Quality of
Life Inventory-32 (PCQL-32). Part II. Feasibility
and range of measurement. J Behav Med 1999;22:
397–406.
23. Sheiham A, Croog SH. The psychosocial impact of
dental diseases on individuals and communities. J
Behav Med 1981;4:257–72.