SA JOURNAL OF DIABETES & VASCULAR DISEASE
RESEARCH ARTICLE
VOLUME 7 NUMBER 1 • MARCH 2010
17
attending the paediatric diabetic clinics at Grey’s Hospital
and IALCH. At the time of the study there were 35 children
between the ages of six and 10 years with type 1 diabetes
from both sites. Criteria for inclusion in the study were the
presence of type 1 diabetes, age between six and 10 years,
duration of diagnosis of three months or more, and the
absence of mental retardation. Of the five subjects who were
excluded, one had mental retardation and four did not wish
to participate. Subjects who qualified for inclusion into the
study were identified on the day of their clinic visit and both
caregivers and subjects were given a brief verbal explanation
about the study while waiting for their consultation with the
doctor.
Diet-related knowledge
Diet-related knowledge was assessed using a multiple-choice
questionnaire, which consisted of 20 open-ended questions.
With each question there was a choice of three possible
answers with the fourth option being ‘I do not know’ to
minimise guessing. All questions had one correct answer
only, except for question 10 which had two correct answers
(this was indicated on the questionnaire). The questionnaire
was developed by a registered dietitian and consisted of
diet-related knowledge questions that a child with type 1
diabetes would be expected to answer on their own after
being diagnosed and given nutrition education for a period
of three months or more.
The questions were based on practical nutrition knowledge
relating to eating behaviour, insulin administration, control
of blood sugar levels, the understanding of the relationships
between carbohydrate intake, insulin and blood sugar levels
and food sources of carbohydrates, fat and fibre. The first half
of the questionnaire was constructed on the basis of essential
practical information to control blood sugar levels. The second
half tested practical knowledge of foods in relation to sugar,
carbohydrates, fat and fibre, which if answered correctly
and applied would be expected to lead to compliance with
the general dietary recommendations for children with type
1 diabetes as well as dietary recommendations from the
International Society for Pediatric and Adolescent Diabetes
(ISPAD) consensus guidelines (2002).
8
Content validity considers the subject matter of the
questionnaire and how representative and appropriate the
questions are in relation to the subject matter.
13,14
Face validity
considers how reasonable the individual questions and the
overall questionnaire are for the subjects from the point of
view of an expert group.
14
To ensure that the questionnaire
had both content and face validity, it was assessed by
two registered dietitians, experienced in the nutritional
management of children with type 1 diabetes. Changes
were made to the questionnaire to improve the wording and
overall understanding of the questions.
In order to test construct validity of the questionnaire,
ideally, two groups of children with different levels of
knowledge,
15
of a similar age to the study sample should
have been tested. It was not possible to find a sample of six-
to 10-year-old children expected to have a good knowledge,
as all such known children were part of the main study.
Construct validity of the questionnaire was therefore assessed
in a pilot study by comparing the scores of children with
type 1 diabetes older than 10 years (expected to have good
knowledge in this area) and scores from children between six
and 10 years without type 1 diabetes (not expected to have
good knowledge in this area), using the same procedure
as for the main study. The mean scores achieved were as
follows: children with type 1 diabetes older than 10 years (
n
=
9) 88% (range 76–100%) and children between six and 10
years without type 1 diabetes (
n
=
13) 19% (range 0–67%).
The significant difference between the scores (
p
=
0.000)
showed that the questionnaire had construct validity, based
on its ability to distinguish between groups with known
differences in knowledge levels on the subject matter.
15
The pilot test results of children with type 1 diabetes
older than 10 years were analysed for item difficulty, item
discrimination and internal consistency. Items should be
discarded if they are answered correctly by more than 80%
or less than 20% of the subjects.
16
A total of three items
were discarded as they were answered correctly by more
than 80% of the subjects. The score on each item was
correlated with the overall test score in order to assess item
discrimination. Items should be discarded if the item-to-total-
score correlation is below 0.2.
17
Two items was discarded as
the item-to-total-score correlation was below 0.2.
The questionnaire’s internal consistency reliability (ICR)
was measured using Cronbach’s
α
-value. Internal consistency
reliability refers to the extent to which individual questions
within a scale relate to the total score.
15
To demonstrate that
the questionnaire has ICR, the Cronbach’s
α
-value should be
above the minimum requirement of 0.7.
15
The Cronbach’s
α
-value for the pilot questionnaire was 0.48. There was an
improvement in the Cronbach’s
α
-value for the questionnaire
used in the main study, as it had increased to 0.74.
Because some of the subjects were Zulu speaking, the
questionnaire was translated from English into Zulu and from
Zulu back into English to check that the main themes were
not lost during the translation process. Only the researcher or
research assistant who also acted as the translator was present
in the room when the children completed the questionnaire.
Caregivers were not present while the children answered the
questionnaire to avoid influence on responses.
Children over the age of nine years were able to read and
answer the questionnaire on their own while those under
the age of nine were assisted by the researcher or research
assistant. In the case of children under the age of nine years,
the researcher read out the questions, gave the four possible
answers and waited for a response from the subject. In
some cases the questions had to be repeated. Zulu-speaking
subjects under the age of nine years were assisted by the
research assistant (translator) in the same way.
