Evaluation of an informed choice invitation for type 2 diabetes screening.

vi
im
a,
it,
Hea
tabo
s C
for
Patient Education and Counseling1. Introduction
The importance of informed consent is enshrined within UK
health policy [1]. There is a strong emphasis on the need for an
individual to be provided with sufficient information regarding
possible risks, and the lack of certainty with respect to benefits,
prior to agreeing to treatment. The General Medical Council
(GMC) makes similar recommendations [2], advocating
provision of the following information relevant to screening
programmes: details of possible diagnosis; uncertainty around
diagnosis; treatment options; the purpose of the investigation,
and details of the likely experience including side effects;
details of benefits and harms, including potential lifestyle
changes; the nature of ongoing assessment; and an unambig-
uous statement regarding freedom of choice to proceed or not.
These recommendations have yet to be implemented in
practice. Invitations to attend for screening traditionally
provide information about its population benefits, aimed at
achieving high rates of uptake [3]. Indeed, a recent Cochrane
Review of interventions that communicated individuals’ risk of
disease in screening programmes found that they generally
increase screening uptake [4,5]. However, the review noted that
insufficient data were available to examine effects on informed
choice and related outcomes, mirroring the typical lack of
mention of possible adverse effects in screening programmes.
This approach reflects a greater concern with potential public
health benefits than with individual autonomy. This may reflect
a reluctance of those organising screening programmes to
* Corresponding author. Tel.: +44 1223 330456; fax: +44 1223 762515.
E-mail address: ik261@medschl.cam.ac.uk (I. Kellar).
0738-3991/$ – see front matter # 2008 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.pec.2008.04.005Objective: To evaluate an innovative invitation designed to facilitate informed choices for undergoing screening for type 2 diabetes.
Methods: Four hundred and seventeen people aged 40–69 years (sex: F 53%/M 47%), without known diabetes, recruited from street locations.
Participants were randomised to receive one of two hypothetical invitations for screening for type 2 diabetes; one based on General Medical
Council guidelines and combined with a decisional balance sheet, the other a brief traditional invitation. Informed choice was assessed
immediately after the invitation and 3 weeks later using measures of knowledge, attitudes and intentions.
Results: Twoweeks after receipt of the invitation, the proportion of informed choices was significantly higher among participants who received the
informed choice invitation compared with those who received the traditional invitation (42.9% versus 11.2%; difference = 31.7%, 95% CI: 22.5–
40.5%; p < 0.001). Mean knowledge scores were significantly higher after the receipt of the invitation designed to facilitate informed choices than
after the traditional invitation (5.49 versus 3.90; t(405) = 10.106, p < 0.001). Intentions to participate in screening were unaffected by receipt of the
informed choice invitation.
Conclusion: Compared with a traditional invitation, receipt of the invitation designed to facilitate informed choices increased the proportion of
informed choices about type 2 diabetes screening attendance.
Practice implications: : Although the new invitation was associated with better knowledge of screening it had no differential effect on intention
and its effect on attendance still requires evaluation.
# 2008 Elsevier Ireland Ltd. All rights reserved.
Keywords: Informed choice; Decision-making; Screening; Diabetes; KnowledgeAbstractEvaluation of an informed choice in
Ian Kellar a,*, Stephen Sutton a, S
Ann Louise Kinmonth
a General Practice & Primary Care Research Un
University of Cambridge, Institute of Public
b MRC Epidemiology Unit, Institute of Me
c Health Psychology Section, King
Received 3 May 2007; received in revisedtation for type 2 diabetes screening
on Griffin b, A. Toby Prevost a,
Theresa M. Marteau c
Department of Public Health & Primary Care,
lth, Cambridge CB2 2SR, United Kingdom
lic Science, Cambridge, United Kingdom
ollege London, United Kingdom
m 14 April 2008; accepted 14 April 2008
www.elsevier.com/locate/pateducou
72 (2008) 232–238

is: p
and Counseling 72 (2008) 232–238 233implement a policy change that is unevaluated in practice and is
feared may privilege concern for informed choice to the neglect
of achieving the public health benefits of screening [3]. In
relation to screening for diabetes, there is increasing evidence
that psychological consequences are limited in terms of anxiety
in the short-term [6], and in the long term with respect to health-
related quality of life [7], perceived health status and well-being
[8]. However, whilst the psychological effects of screening do
appear to be minimal, there are burdens associated with
diagnosis following screen detection [9,10]. It is possible that
information about the type and likelihood of the burdens
associated with the screening procedure as well as the possible
treatment, when presented alongside the type and likelihood of
the benefits, could deter some people from participating in
screening programmes [11–14].
In preparation for a trial to test the impact of an informed
choice invitation for screening for type 2 diabetes on attendance
and subsequent motivation to adopt recommended preventive
actions (the DICISION trial), we have developed an invitation
designed to foster informed choices. This paper reports an
evaluation of the effect of this invitation on the proportion of
participants who make an informed choice compared with a
traditional invitation in an experimental analogue study in
which participants were asked to imagine that they had received
an invitation for screening for type 2 diabetes.
Type 2 diabetes is a chronic condition that meets many of the
Fig. 1. Classification of choices, based on the three dimensions of attitudes (x-ax
low, high). Adapted from [17].
I. Kellar et al. / Patient Educationcriteria for screening [15]. Undiagnosed diabetes is common
and typically asymptomatic, but is associated with significant
consequent morbidity and premature mortality. Between one-
third and one-half of people with type 2 diabetes are
undiagnosed, and at diagnosis up to 30% exhibit evidence of
diabetic complications [16]. However, there is continuing
uncertainty over the extent to which the benefits of early
detection outweigh the harms and whether screening is a cost-
effective use of limited health service resources [15].
1.1. Conceptualising informed choices
An informed choice can be defined as a behaviour with two
core characteristics: first, it is based upon knowledge of
relevant information; and second, it reflects the decision-
maker’s values [17–20]. This conceptualisation of informedchoice is consistent with a multidimensional measure of
informed choice [17,21], which categorises behavioural
choices as either uninformed or informed based on measures
of knowledge and attitudes (Fig. 1). In this operationalisation,
uninformed choices are those in which knowledge is poor, or in
which knowledge is good but the decision-maker’s behaviour
does not reflect their attitudes. Intention is specified as themost
proximal determinant of behaviour by several theories,
including the theory of planned behaviour, and protection
motivation theory [22–24], and is used as a proxy for behaviour
in this study.
Based on this model of informed choice, we developed an
invitation for screening for type 2 diabetes that was designed to
provide good quality information to increase knowledge of the
relevant aspects of the screening test and that included a
decision aid to increase participants’ awareness of their
attitudes and thereby to increase attitude–intention congruence.
1.2. Aim and hypotheses
This study aimed to evaluate an innovative invitation to
increase informed choice in relation to screening for type 2
diabetes. The following hypotheses were tested:
H1. The proportion of informed choices will be higher follow-
ing an informed choice invitation compared with a traditional
ositive, negative), intentions (y-axis: positive, negative), and knowledge (z-axis:appointment invitation.
H2. The proportion of choices in which attitudes and inten-
tions are congruent will be higher following an informed choice
invitation compared with a traditional appointment invitation.
H3. Knowledge about screening will be higher following an
informed choice invitation compared with a traditional appoint-
ment invitation.
To reflect the likely gap between receipt of an invitation and
attending for screening, and therefore the need for choices to be
sustained over this period, the study assessed choice 2 weeks
later.1
1 Identical data were collected immediately following presentation of the
invitations, and a similar pattern to the results presented here was observed.

2. Methods This text was followed by two open text boxes, entitled ‘‘Good
things from screening for diabetes’’ and ‘‘Bad things from
screening for diabetes’’.3
2.3. Measures
I. Kellar et al. / Patient Education and Counseling 72 (2008) 232–2382342.1. Participants
Eligible participants were volunteer members of the
public aged between 40 and 69 years, with no previous
diagnosis of diabetes, who agreed to provide demographic
details and accept a follow-up visit at their place of residence
2 weeks later. Additionally, a quota was set of 50% of
participants having finished full-time education at 16 or
before.
2.2. Study materials
2.2.1. Traditional invitation
The traditional invitation was based on previous invitations
to attend for screening tests for diabetes and coronary heart
disease [25,26]. It contained information (200 words) regarding
the seriousness of diabetes, the nature of the screening test, and
details of how the appointment would be arranged. Readability
scores indicated that it could be comprehended by 11-year-olds
(Flesch–Kincaid Grade level = 6.34; Flesch Reading Ease
score = 71.512).
2.2.2. Informed choice invitation
The information contained in both informed choice
invitation was based upon the GMC guidelines [2] described
above. The draft invitation text was developed iteratively using
‘‘think-aloud’’ methods with volunteers [27]. The method
requires respondents to verbalise their thoughts whilst reading
the invitation and has typically been to examine decision-
making [28,29]. In this case the verbal reports were used to
ascertain where information was not being understood as
intended, and readability tools [30] were then used to refine the
text. The final invitation comprised 857 words and retained an
equivalent reading comprehension score to the traditional
invitation (Flesch–Kincaid Grade level = 5.76; Flesch Reading
Ease score = 72.88). Additionally, two pie charts were used to
represent data described in the text regarding the likelihood of a
positive screening result and the likelihood of health benefits
from early detection.
A decision balance sheet decision aid followed the informed
choice text. This required participants to write down the
anticipated consequences for themselves of undergoing
screening in terms of pros and cons, prompted by the
instruction:
When making your decision, it might help you to think
about the good and bad things that could happen if
you are screened for diabetes. Spend a few minutes
thinking about having screening for diabetes. Please list
all the good and bad things that go through your mind.
Please underline the things that are most important to
you.2 Approximates to a reading age of around 11 years.Knowledge about screening for type 2 diabetes was
measured using eight multiple-choice items derived from the
GMC guidelines, with between three and six response options
(e.g. ‘‘the screening test for diabetes is. . . [a finger prick test; a
test where blood is taken from a vein in the arm; a urine test; a
fitness test; I don’t know]’’). The number of correct items was
used as the knowledge score, with a potential range of 0–8.
Attitude regarding undergoing screening for type 2 diabetes
were measured by six items (e.g. ‘‘For me, having the screening
test for diabetes would be. . . [beneficial–not beneficial], each
scored on 7-point scales scored from 1 to 7. The mean of the six
items was used as the score. Cronbach’s alpha was 0.89 at time
1 and 0.92 at time 2. Intention regarding undergoing screening
for type 2 diabetes was measured by a five-item scale (Would
you have the diabetes screening test? [definitely yes; probably
yes; do not know; probably not; definitely not]) and a 7-point
scale (‘‘Would you have the diabetes screening test?’’
[extremely likely–extremely unlikely]). Both scores were
converted to z-scores. The mean of the two items was used
as the score. Cronbach’s alpha was 0.72 at time 1 and 0.84 at
time 2.
2.4. Procedure
Data were collected by a market research company
(Cambridge Market Research) from members of the public
across the UK between February and April 2006. Participants
were recruited from street locations by trained interviewers.
Participants were invited to immediately complete a brief
questionnaire containing demographic measures that were used
to ensure that the inclusion criteria (see Section 2.1) were
satisfied. Those who met the criteria were then asked to read
either a traditional invitation, an instrumental-informed choice
invitation or an affective-informed choice invitation from a pile
of invitations that were randomly ordered using a random
number generator, and were told ‘‘[it is] an invitation to attend a
diabetes screening appointment. The appointment will not take
place, but please VIVIDLY IMAGINE that you have received
this from your GP regarding a REAL appointment’’. Twoweeks
later, participants were visited at their homes by interviewers,
and completed a questionnaire. Participants received £5
immediately on completion of the follow-up questionnaire.
The Cambridge University Psychology Research Ethics
Committee granted ethical approval for the research.
3 We developed 2 versions of the invitation; identical except that in the first,
the decision aid referred to ‘‘good things’’ and ‘‘bad things’’ about screening for
diabetes, whereas the other referred to ‘‘good feelings’’ and ‘‘bad feelings’’.
However, this instrumental/affective manipulation had no significant effects on
any outcome. Subsequently, we have treated the two invitations as a single
group.

Table 1
Demographic characteristics of study participants (%(n))
Variable Total sample % (n = 417) Traditional % (n = 139) Informed % (n = 278)
Social grade C2, D and E 32.4 (135) 37.4 (52) 29.9 (83)
Home owner 86.1 (359) 79.9 (111) 89.2 (248)
Sex (male) 47.0 (196) 50.4 (70) 45.3 (126)
Over 50 years 50.6 (211) 52.5 (73) 49.6 (138)
Left school at or before 16 years 61.6 (257) 66.2 (92) 59.4 (165)
Highest formal qualification at or below NQFa level 2 69.7 (232) 74.3 (84) 67.3 (148)
Not working 26.6 (111) 23.7 (33) 28.1 (78)
. QC
I. Kellar et al. / Patient Education and Counseling 72 (2008) 232–238 2352.5. Analyses
The study was designed with 140 participants per invitation,
providing 90% power to detect an absolute increase of 20% in
the percentage of participants making an informed choice when
comparing the invitations using a chi-squared test at the 5%
level of significance.
2.5.1. Defining criteria for informed choices
Drawing upon the multidimensional measure of informed
choice [17], participants were categorised as having made an
informed choice using the knowledge, attitude, and intention
measures. Participants were assigned as having made an
informed choice if they scored above the median on
knowledge and were jointly above or below the median on
attitude and intention (see Fig. 1: cells 1 and 4, respectively).
Participants were assigned as having made a congruent
choice if they scored jointly above or below the median on
attitude and intention (see Fig. 1: cells 1 & 5 and 4 & 8,
respectively).
Chi-square tests were used to compare groups on categorical
variables (e.g. whether or not the choice was congruent) and t-
tests for variables that could be treated as continuous (e.g.
knowledge score); for the latter, the standardised mean
difference (d) was used as the measure of effect size. To
maximise precision of results, rather than deleting a case that
had any missing data, we retained such cases for analysis if they
had available data. Analyses were carried out using SPSS 12 for
Windows.
3. Results
a Education was represented by National Qualifications Framework level 35Four hundred and seventeen people completed demographic
measures at time 1 and, of these, 407 (97.6%) completed the
Table 2
Rates of choice (%(n)) categorised by MMIC by invitation type
Category Invitation
Traditional
Informed choice (cells 1 and 4) 11.2 (15)
High knowledge, incongruent choice (cells 2 and 3) 0.7 (1)
Low knowledge, congruent choice (cells 5 and 8) 64.9 (87)
Low knowledge, incongruent choice (cells 6 and 7) 23.1 (31)
Congruent choice (cells 1, 4, 5 and 8) 76.1 (102)
***p < 0.001.questionnaire 2 weeks later, of whom 53% were female and
47% were male. There were no missing data for demographic
items at time 1, and of the responses to the questionnaire 2
weeks later, 2.4% (range 1.9–3.4%) were missing.
3.1. Randomisation checks
Table 1 shows demographic characteristics of study
participants, which were not significantly different between
study groups.
3.2. Informed and congruent choices
The proportions of informed and congruent choice and chi-
square statistics for tests of difference between participants
assigned the traditional and informed invitations are given in
Table 2. Two weeks after presentation of the invitation, the
informed choice invitation resulted in a significantly higher
proportion of informed choices compared with the traditional
invitation (42.9% and 11.2%, respectively; difference = 31.7%,
95% CI: 22.5–40.5%; x2 = 41.1 (1), p < 0.001). Thus, the
hypothesis (H1) that the proportion of informed choices is
higher following an informed choice invitation was supported.
The proportion of attitude–intention congruent choices was
uniformly high, with no differences between the traditional and
the informed choice invitations (76.1% and 74.0%, respec-
tively). Thus, the hypothesis that the proportion of choices in
which attitudes and intentions are congruent is higher following
an informed choice invitation (H2) was not supported. To
further investigate the lack of effect on congruence, post hoc
tests were performed on the attitude and intention measures
(see Table 3). There were no significant differences between the
A. The structure of the Nqf: QCA, 2007.informed and traditional invitation recipients on attitudes or
intentions.
x2 (d.f.)
(n = 134) Informed (n = 273)
42.9 (117) 41.1 (1)***
15.4 (42) 20.4 (1)***
31.1 (85) 42.1 (1)***
10.6 (29) 11.2 (1)***
74.0 (202) 0.6 (1)

3.3. Knowledge
4. Discussion and conclusion
4.1. Discussion
Compared with the traditional invitation, the informed
choice invitation resulted in a significantly higher proportion of
informed choices as defined by our conceptualisation. This
increase reflected increased type 2 diabetes screening-related
knowledge but not increased attitude–intention congruence.
Inspection of the knowledge items revealed that whilst the
written materials effectively conveyed the consequences of
diabetes, the likely results of screening, the meaning of a
negative screening result, and the likely harms of screening,
Table 3
Knowledge, attitudes and intentions by invitation type
Variables Invitation t-Test
Traditional
(n = 134)
Informed
(n = 273)
Knowledge 3.90 (1.47) 5.49 (1.53) 10.1 (405)***
Attitude 6.15 (0.98) 6.25 (0.89) 1.0 (405)
Intention (z-score) �0.11 (0.99) 0.06 (0.88) 1.7 (404)
***p < 0.001. The intention scale z-scores rescaled to a 7-point scale are 6.0
and 6.2 for the traditional and informed groups, respectively.
pe
I. Kellar et al. / Patient Education and Counseling 72 (2008) 232–238236Table 3 presents the group means and tests of between group
differences on the component constructs of the informed choice
measure. There was a significant difference in knowledge when
comparing the informed with the traditional invitation
recipients (informed M = 5.49, traditional M = 3.90;
t(405) = 10.1, p < 0.001; d = 0.95). Thus, the hypothesis that
knowledge about screening is higher following an informed
choice invitation (H3) was supported.
To further investigate the effect of the informed choice
invitation on knowledge, post hoc tests were performed upon
the responses to the eight multiple-choice knowledge items (see
Table 4). Where significant differences were found, these were
in the expected direction, with proportions of correct answers
uniformly associated with receipt of the informed choice
invitation. High proportions of correct answers in both groups
on items relating to the seriousness of diabetes and the nature of
the screening test indicate either that this knowledge is
available in both the traditional and informed choice invitations
or that the majority of participants already knew this
information. Low proportions of correct answers on the
meaning of a positive test result in both groups indicate that
this information was not adequately conveyed. Additionally,
despite higher proportions of correct scores on the item relating
to the efficacy of early treatment to reduce the consequences of
diabetes being significantly associated with receipt of the
informed choice invitation, only 19.6% were correct on this
item. This proportion of correct answers was markedly lower
than where other significant associations occurred, with
proportions higher than 75% on those items.
Table 4
Correct responses (%(n)) to each of the eight knowledge items by invitation tyQuestion Invita
Tradi
Seriousness of diabetes 89.7 (
Common long-term problems 42.1 (
Nature of screening test 86.8 (
Most likely test result 52.9 (
Meaning of negative screening result 40.4 (
Meaning of positive screening result 27.9 (
Efficacy of early treatment to reduce consequences of diabetes 3.7 (
Possible harms of screening 47.1 (
*p < 0.05.
***p < 0.001.only around one-third of recipients understood that a positive
type 2 diabetes screening test result was not diagnostic or that
following early treatment following the diagnosis of diabetes
from screening tests, a minority will benefit but there will be
costs for all.
Compared with the information that was successfully
conveyed, explaining the meaning of a positive screening
result and the effectiveness of early treatment involved the
presentation of more complicated numerically based informa-
tion. We drew upon review findings suggesting the effective-
ness of using graphs to present complex outcomes [31], and
used pie charts as well as text that had been subjected to ‘‘think-
aloud’’ testing. However, the information conveyed required
comparison of the probabilities of more than a single potential
outcome. In the case of those receiving positive screening test
results, subsequent diagnostic testing would either confirm a
diagnosis of diabetes or not. Concerning the effectiveness of
early treatment following the diagnosis of diabetes from
screening tests, a minority will benefit but there will be costs for
all. The failure to benefit the majority may be due to treatment
failure, or to treatment being unnecessary (i.e. the individual
would not have developed long-term complications of
diabetes). A recent study comparing comprehension of
different types of risk information indicates that comprehension
may be lower when individuals are asked to compare multiple
outcomes [32]. We elected to convey the benefits and harms of a
procedure within the same graphical and textual elements of the
information provided so as to reduce the amount of text and
number of graphical representations presented. Further
tion Test
tional (n = 134) Informed (n = 273)
122) 93.7 (254) 2.1 (0.149)
56) 81.5 (220) 64.0 (0.000)***
118) 93.0 (252) 4.2 (0.039)*
72) 84.1 (228) 45.5 (0.000)***
55) 78.2 (212) 57.3 (0.000)***
38) 27.1 (74) 0.0 (0.858)
5) 19.6 (53) 18.7 (0.000)***
64) 76.1 (207) 34.3 (0.000)***