www.thelancet.com Vol 371 January 5, 2008 41
Effi cacy of a theory-based behavioural intervention to
increase physical activity in an at-risk group in primary care
(ProActive UK): a randomised trial
Ann-Louise Kinmonth, Nicholas J Wareham, Wendy Hardeman, Stephen Sutton, A Toby Prevost, Tom Fanshawe, Kate M Williams, Ulf Ekelund,
David Spiegelhalter, Simon J Griffi n
Background Declining physical activity is associated with a rising burden of global disease. Eff orts to reverse this
trend have not been successful. We aimed to assess the effi cacy of a facilitated behavioural intervention to increase the
physical activity of sedentary individuals at familial risk of diabetes.
Methods We enrolled 365 sedentary adults who had a parental history of type 2 diabetes. They were recruited from
either diabetes or family history registers at 20 general practice clinics in the UK. Eligible participants were randomly
assigned to one of two intervention groups, or to a comparison group. All participants were posted a brief advice
leafl et. One intervention group was off ered a 1-year behaviour-change programme, to be delivered by trained
facilitators in participants’ homes, and the other the same programme by telephone. The programme was designed to
alter behavioural determinants, as defi ned by the theory of planned behaviour, and to teach behaviour-change
strategies. The principal outcome at 1 year was daytime physical activity, which was objectively measured as a ratio to
resting energy expenditure. Analysis was by intention to treat. This study is registered as ISRCTN61323766.
Findings Of 365 patients, we analysed primary endpoints for 321 (88%) for whom we had data after 1 year of follow-up.
At 1 year, the physical-activity ratio of participants who received the intervention, by either delivery route, did not
diff er from the ratio in those who were given a brief advice leafl et. The mean diff erence in daytime physical-activity
ratio, adjusted for baseline, was −0·04 (95% CI −0·16 to 0·08). The physical-activity ratio did not diff er between
participants who were delivered the intervention face-to-face or by telephone (mean diff erence −0·05; 95% CI
−0·19 to 0·10).
Interpretation A facilitated theory-based behavioural intervention was no more eff ective than an advice leafl et for
promotion of physical activity in an at-risk group; therefore health-care providers should remain cautious about
commissioning behavioural programmes into individual preventive health-care services.
Declining physical activity and the associated rising
burden of disease is a major public-health problem.
Physical inactivity is thought to account for at least 11·7%
of all deaths in developed countries. A third of
premature deaths from coronary artery disease, colon
cancer, and diabetes in Canada and the US can be
attributed to inactivity.1 The challenge for public-health
policymakers is to identify eff ective strategies to reverse
this behavioural trend in populations and defi ned
The diffi culty of this challenge has been shown by
reviews of the eff ectiveness of interventions that are
targeted at the population level or at high-risk individuals
and community groups via health-care settings.3–8 In
general, trials have been characterised by small eff ect
sizes and by important limitations such as short
follow-up and high attrition rates. Many studies have
used imprecise self-reported measures of activity as
their main outcomes; therefore, reporting biases could
have infl ated diff erences between study groups. Such
measures also focus on the most readily quantifi ed
aspects of physical activity and cannot easily capture
changes in overall daytime energy expenditure related
to physical activity.6,9 Target groups and the interventions
themselves have often been poorly described, which
complicates attempts to identify the eff ective or
ineff ective elements of an intervention.
More eff ective strategies might target everyday activities,
and might better specify behavioural determinants and the
techniques to alter them.8,10,11 Theory-based interventions
have been associated with larger and longer term eff ects
than those without an explicit basis in theory,5,8,12 and the
theory of planned behaviour has shown good predictive
power for self-reported physical activity in short-term
We aimed to assess whether a 1-year intervention based
on theory and evidence would increase physical activity
in adults who were at risk of diabetes because of a
parental history of type 2 diabetes and a self-reported
sedentary lifestyle.15 We also aimed to assess whether
diff erent methods of delivery of this targeted behavioural
intervention (ie, in person or by telephone) aff ected the
acceptability of the intervention or its effi cacy for
changing behaviour (and its clinical and psychosocial
Lancet 2008; 371: 41–48
See Comment page 5
General Practice and Primary
Care Research Unit,
Department of Public Health
and Primary Care, University of
(Prof A-L Kinmonth MD,
W Hardeman MSc,
Prof S Sutton PhD,
A T Prevost PhD, T Fanshawe MA,
K M Williams PhD); Medical
Research Council Epidemiology
Unit, Cambridge, UK
(Prof N J Wareham PhD,
U Ekelund PhD, S J Griffi n DM);
and Medical Research Council
Biostatistics Unit, Cambridge,
UK (Prof D Spiegelhalter FRS)
Prof Ann-Louise Kinmonth,
General Practice and Primary
Care Research Unit, Department
of Public Health and Primary
Care, University of Cambridge,
CB2 0SR, UK
www.thelancet.com Vol 371 January 5, 2008
Study design and participants
The trial methods have been described elsewhere.16
Between March, 2001, and October, 2003, we recruited
the children of patients with type 2 diabetes, either
because their parents were identifi ed from diabetes
registers at 20 general practice clinics in the UK, or
because they were identifi ed from family history records
at seven of those clinics. Ethical approval was obtained
from the East of England
MREC 02/5/53). All participants gave written informed
We identifi ed 1521 potentially eligible people who were
aged 30–50 years, and had a parental history of diabetes
without known diabetes.16 Of these, 1123 completed and
returned a questionnaire: 286 of them were unwilling to
participate and 343 were excluded because they were
highly active (as defi ned by the questionnaire).17,18 29 were
excluded because either they had been prescribed β
blockers that aff ected heart-rate variability; they were
unable to walk briskly across fl at terrain for 15 min; they
lived further than 30 min by car from the study
coordination centre; or they had illness or social
obligations that would prevent participation. We screened
the remaining 465 potentially eligible participants by
telephone to check eligibility, and confi rm willingness to
participate. At this stage, 66 were excluded: 31 because
they did not meet inclusion criteria, and 35 who refused to
participate. We took baseline measurements for
399 people. 24 of these people partici pated in a pilot study
to assess the acceptability, logistics, and delivery of the
measures and the intervention. Before randomisation, we
excluded 10 more people (seven did not meet inclusion
criteria, two were unwilling to participate, and one agreed
to participate after recruitment had closed).
Participants were randomised from a central site.
Trial staff who did measurement and data entry were
unaware of the groups to which participants had been
assigned. We used a partial minimisation procedure
that dynamically adjusted randomisation probabilities
to balance six baseline covariates: physical activity,
body-mass index, age, sex, living with children, and
intention to increase physical activity. Siblings were
assigned to the same group.
Each participant was randomly assigned to one of three
groups: one to take part in a behavioural change
programme delivered by a facilitator over the telephone
(distance); a second group to take part in the same
programme, but delivered in the home (face to face); and
a comparison group (advice). Participants in all groups
were sent a leafl et by post with brief motivational advice
on the benefi ts of increased activity. Those in the
comparison group received only the advice leafl et.
The protocol-driven programme that was off ered to
the other two groups was developed by experts, piloted
with volunteers, and delivered by trained facilitators
from a range of health professions. Training for
facilitators consisted of a detailed manual and a 5-day
course on established behaviour-change techniques and
their practical application.16,19 The same group of
facilitators delivered interventions by both methods.
Participants in the two intervention groups were taught
to maximise personal advantages and opportunities,
and to minimise disadvantages and obstacles to
becoming more physically active. The intervention
focused on eight self-regulatory strategies for behavioural
change, including goal-setting, action-planning, self-
monitoring, using rewards, goal-review, using prompts,
107 analysed for primary outcome
5 excluded because of missing data
111 analysed for primary outcome
2 excluded because of missing data
121 allocated to comparison group
103 analysed for primary outcome
4 excluded because of missing data
124 allocated to intervention by telephone120 allocated to intervention in person
365 randomly assigned
12 lost to follow-up after 1 year
1 moved away
5 unable to contact
1 unable to attend
5 withdrew consent
8 lost to follow-up after 1 year
1 moved away
3 unable to contact
2 unable to attend
1 withdrew consent
13 lost to follow-up after 1 year
4 moved away
2 unable to contact
3 unable to attend
4 withdrew consent
15 did not receive intervention as per protocol
2 discontinued intervention
21 did not receive intervention as per protocol
4 discontinued intervention
Figure 1: Trial profi le15
www.thelancet.com Vol 371 January 5, 2008 43
building support from family and friends, and
prevention of relapses.16,19
The intervention programme lasted 1 year. Both
methods of delivery were introduced with a session in
the home of the participant. The telephone-based
programme included four 45-min calls and two 15-min
support calls during the 5-months’ intensive phase,
followed by monthly postal contact for the remaining
7 months. The home-based programme included four
1-hour home visits and two 15-min telephone calls
during the 5-month intensive phase, and monthly 30-
min follow-up phonecalls for the rest of the year. The
introductory meeting and the four 45-min telephone
calls and four 1-hour home visits were regarded as core
sessions for the purpose of the per-protocol analysis.
Facilitators followed protocols for all sessions, and
taped some contacts with participants for quality
assurance. A lead facilitator supervised facilitators by
monthly review of tapes and discussion of best practice
in following the protocols. Facilitators also met for
fortnightly team discussions.
The primary outcome was energy expenditure on
daytime physical activity, expressed as a ratio to measured
resting energy expenditure (to account for diff erences in
body size between individuals). We measured energy
expenditure at baseline and after 1 year by monitoring
participants’ heart rates over 3 consecutive days, as they
lived their normal lives.16,20 We asked each participant to
come to a study centre to do a submaximal exercise test on
a treadmill to establish the association between individual
heart rate and oxygen uptake (as measured by indirect
Secondary outcome measures were measured at
baseline and at 1 year. Maximal cardiorespiratory fi tness
(VO2max) was estimated as the oxygen uptake at maximal
heart rate (220 beats per min minus age), and self-reported
physical activity was assessed by valid ated questionnaire.21
Weight was measured with standard scales (SECA;
London, UK), which were calibrated every 3 months;
body-fat percentage by bioelectrical impedance (Bodystat,
Isle of Man, UK); and blood pressure with an automatic
sphygmomanometer (Accutorr, Datascope, Mahwah NJ,
USA). Other clinical measures—glyco sylated haemo-
globin, fasting plasma glucose, lipids, and insulin—were
measured in one laboratory with established methods
and quality assurance systems.16
We used self-reported measures for psychosocial
outcomes. We assessed wellbeing with the SF-36 survey,22
a well-validated instrument that scored eight variables
from 0 to 100, such that high scores would denote excellent
physical and emotional wellbeing, with no limitations in
work or social activity due to health problems. Changes of
the order of 3–5 scale points might be clinically relevant.23
The SF-36 also includes a measure of change in health
over the past year across fi ve categories from much worse
to much better, such that a score of 50 would equate to no
change. We also assessed anxiety, with a six-item test that
scored feelings on a four-point Likert scale, such that non-
anxious states would score about 37 and highly anxious
states would score about 48.24 To measure worry about
diabetes and perceived comparative risk of diabetes, we
used instruments adapted from previous studies,25,26 such
that participants who only rarely worried about diabetes
would score 12 or lower, and those who thought that their
chance of getting diabetes were the same or less than for
others of their age would score 3 or below.
Intention to be more physically active over the next year
was measured on two Likert scales developed from
recommendations by Ajzen,14 such that 4 would represent
a moderately strong intention on a scale of 1 to 5.
We also collected all self-reported questionnaires,
except the SF-36, by post at 6 months after baseline to
analyse psychosocial aspects. At 6 months and 1 year, we
asked participants about the acceptability of the
programme and its delivery (specifi cally their enjoyment
and satisfaction, and the clarity of the advice, measured
on three to four point Likert scales developed for the
trial), and about whether or not they had used the eight
behaviour change strategies. We used contact registers
and trial-specifi c questionnaires at 1 year to assess the
frequency and severity of any injuries related to physical
activity. Independent assessments of intervention fi delity
will be reported elsewhere.
We tested the principal hypothesis after 1 year by
comparison of outcome data, by use of linear regression,
and adjusting for baseline outcome data and randomisation
stratifi ers. We fi rst compared the two intervention groups
with the group who were sent leafl ets only, and then
compared the three groups with each other. We used a
separate dummy variable to include participants for whom
we did not have baseline data, according to the missing
indicator method.27 In the analysis of each outcome we
excluded extreme outliers—individuals with a value at
least four standard deviations from the mean. We used the
same methods to analyse secondary outcomes.
We undertook other preplanned analyses. A per-protocol
analysis included participants who had attended all core
sessions. We assessed whether the identity of the facilitator
in the intervention groups aff ected the outcome by use of
a linear mixed-eff ects model with facilitator as a random
eff ect. We did sensitivity analyses to investigate the eff ect
of having excluded participants with missing data. We
used multiple imputation, and assumed fi rst a scenario in
which people with missing data had the same results as
others in their group, and second a scenario in which the
missing data were the same as for the comparison group.
We used a linear mixed-eff ects model to assess the eff ects
of clustering of participants within families. We used
S-plus software (version 2000) for clustered and missing
data and SPSS software (version 12·5) for other analyses.
Analysis of self-reported measures of physical activity
and psychosocial variables at 6 months has been reported
www.thelancet.com Vol 371 January 5, 2008
to aid interpretation of intervention eff ects on psychosocial
variables, by use of standardised eff ect sizes: 0·2 for
small, 0·5 for moderate, and 0·8 for large eff ects.28
We calculated sample size with data from an observa-
tional cohort study that included a repeated measures sub-
study in which the residual standard deviation of 1-year
change in physical-activity ratio, adjusted for baseline, was
0·53.29 We calculated that a sample size of 100 individuals
in each group would provide 80% power to detect, with
95% confi dence, a diff erence in change in mean
physical-activity ratio of 0·18 between the combined inter-
vention groups and the comparison group. This diff erence
is roughly equivalent to 30 min of brisk walking on fl at
terrain every day. Based on this sample size, we calculated
that the observed diff erence in mean physical-activity ratio
between any two study groups would have a 95% CI that
ranged 0·15 either side of the diff erence, which would be
equivalent to a diff erence of 25 min of brisk walking.16,30
This study is registered as ISRCTN61323766.
Role of the funding source
The sponsor of the study had no role in study design, data
collection, data analysis, data interpretation, or writing of
the report. The corresponding author had full access to all
the data in the study and had fi nal responsibility for the
decision to submit for publication.
Figure 1 shows the trial profi le. We randomised 365 people.
They included 32 sibling-pairs and two sibling-triplets
(70/365 parti ci pants), who were cluster-randomised.
332 participants attended a 1-year follow-up, at which we
obtained a valid measure of primary outcome for 321 of
the 365 randomised patients.
We analysed the primary endpoint for these 321 (88%)
patients for whom we had data. We excluded 44 patients
from this analysis: we did not have complete data for 11;
ten could not be contacted; ten withdrew consent; six
moved away; six were unable to attend follow-up; and
one died. We did analyse six patients who discontinued
the intervention and 36 who did not receive the
intervention as per protocol. The proportions analysed in
diff erent trial groups did not diff er (p=0·29).
Baseline characteristics, including stratifi ers, in the
three trial groups were similar (table 2). These
321 patients were generally sedentary, overweight, and
white, and their mental and physical health and anxiety
levels were similar to population norms.22,24 As a group,
they were not worried about diabetes, and perceived
their risks of developing it as only just greater than
others of their age. They were not socially disadvantaged;
98% had cars available for their use and 89% owned
their homes. The mean age at which they had fi nished
full-time education was 18 (SD 3) years, and 55% were
in managerial or professional jobs. The mean age of
participants at randomisation was 40·6 (SD 6·0) years.
62% were female; 75% lived with children; and
52% reported a strong intention to increase their
Table 1 and fi gure 2 show that the combined
intervention groups did not have a higher daytime
energy expenditure, as measured by heart-rate
monitoring and expressed as a ratio to resting energy
expenditure, than did those in the comparison group.
Face-to-face delivery showed no advantage over telephone
delivery. Over 12 months, the physical-activity ratio
increased in all participants by an average of 0·11
(95% CI 0·05–0·18), which is equivalent to an additional
20 min of brisk walking per day.20
The intervention did not aff ect cardiorespiratory fi tness
or total physical activity at 1 year, as measured by
questionnaire. Self-reported physical activity at 6 months
was similar to that at 1 year. Table 2 shows that, compared
with those in the comparison group, participants assigned
Comparison (advice)Intervention by telephone Intervention in personAdjusted inter vention eff ect*
Baseline6 months 1 yearBaseline6 months 1 yearBaseline6 months1 year
Cardiorespiratory fi tness
(predicted VO2max in L/min)†
Total reported activity
(total metabolic equivalent
in h per week)‡
–0·04 (95% CI –0·16 to 0·08)
0·10 (95% CI –0·001 to 0·21)
84·4 (55·7)99·1 (56·9) 101·4 (58·6)89·3 (52·1) 104·2 (51·9)105·2 (51·8)87·4 (47·2) 97·8 (47·9)97·1 (45·7) –0·23 (95% CI –9·68 to 9·23)
Data are mean (SD) unless otherwise specifi ed. *Adjusted mean intervention eff ect’ refers to mean diff erence in 12-month measurements between combined intervention groups and comparison group,
adjusted for baseline measurement and minimisation stratifi ers. †321 people for whom we had data at 12 months. ‡324 participants at baseline and 12 months, and 301 at 6 months.
Table 1: Adjusted mean intervention eff ect on physical activity variables at 6 and 12 months for combined intervention groups versus comparison group
Decrease in activity
Increase in activity
Combined intervention groups
Intervention by telephone
Intervention in person
Adjusted mean difference
Combined interventions vs comparison
–0·04 (95%CI –0·16 to 0·08)
–0·05 (95%CI –0·19 to 0·10)
Figure 2: Adjusted mean diff erence in physical-activity ratio from baseline to 12 months
Adjusted mean diff erence refers to mean diff erence in 12-month measurements between groups, adjusted for
baseline measurement and minimisation stratifi ers.
www.thelancet.com Vol 371 January 5, 2008 45
to the inter vention did not have improved weight,
body-mass index, waist circumference, fat per centage,
blood pressure, glycosylated haemoglobin, or bio-
Figure 3 shows that participants randomised to the
intervention scored better on self-reported health status
on six of the eight SF-36 scales, on change in health over
the year, and on anxiety than those in the comparison
group. Eff ect sizes were small for physical function
(0·20), general health (0·21), and anxiety (0·29); small-
to-moderate for social function (0·41), energy levels
(0·38), and change in health (0·43); and moderate for
aspects of mental health (0·51) and their impact on daily
activities (0·48). The intervention did not aff ect worry
about diabetes or perceived risk of diabetes. Diff erent
delivery methods for the intervention did not aff ect
psychosocial or self-reported health. 6-month data were
similar to those at 1 year.
Intention to be more physically active was not stronger
in the treatment group than in the comparison group at
1 year. However, at 6 months, intention was stronger in
both intervention groups than in the comparison group,
with a moderate standardised intervention eff ect (0·52).
88% (109/124) of participants for whom delivery was
by telephone and 83% (99/120) of those who had
face-to-face delivery completed all core meetings. We
treated this group as the per-protocol group.
Questionnaires on acceptability were returned by 87%
(108/124) and 91% (109/120) of participants, respectively,
at 1 year. All respondents reported that they enjoyed the
programme and thought that the advice about physic-
ally activity was clear. 97% (210/217) of respondents
reported that they were fairly or very satisfi ed with the
Most participants in the combined intervention groups
reported at both 6 and 12 months that they had used the
eight behaviour change strategies. More than 75%
reported that they had set goals and used action plans
and self-monitoring activities, and more than 60%
reported they had used family support and techniques
for dealing with setbacks.
32 participants (11 with intervention by telephone,
10 with intervention in person, and 11 in the comparison
group) reported that within a year of randomisation they
had visited either a family doctor or an emergency
department or hospital outpatients’ department for pain
or injury to muscles, joints, or bones during or after
physical activity. One participant who had the intervention
by face-to-face delivery had an unrelated hospital
admission. One person in the comparison group died
from unrelated causes.
The results of the per-protocol analysis did not diff er
from the analysis reported here. Neither primary nor
secondary outcomes diff ered between participants who
received the intervention from diff erent facilitators. A
sensitivity analysis showed that results were unaff ected
by allowing for clustering or by including the
44 participants for whom we did not have data for the
Sedentary middle-aged women and men with a parental
history of diabetes who participated in a facilitated
theory-based behavioural intervention did not do more
physical activity than those who were given a brief
motivational advice leafl et. The method for measurement
of physical activity, whether objectively, with heart-rate
monitoring, or subjectively, with a questionnaire, did
Comparison (advice)Intervention by telephone Intervention in personn Adjusted intervention eff ect*
Baseline 1 yearBaseline 1 yearBaseline 1 year
Body-mass index (kg/m²)
Waist circum ference (cm)
Fat percentage (%)
Systolic blood pressure (mm Hg)
Diastolic blood pressure (mm Hg)
Plasma glucose (mmol/L)
Total cholesterol (mmol/L)
HDL cholesterol (mmol/L)
LDL cholesterol (mmol/L)
0·001 (95% CI -0·87 to 0·88)
–0·04 (95% CI –0·35 to 0·27)
–0·10 (95% CI –1·20 to 1·00)
0·05 (95% CI –0·67 to 0·77)
–0·30 (95% CI –2·33 to 1·74)
0·98 (95% CI –0·56 to 2·53)
0·06 (95% CI 0·003 to 0·12)
0·06 (95% CI –0·03 to 0·15)
0·09 (95% CI –0·05 to 0·24)
–0·04 (95% CI –0·09 to 0·02)
0·10 (95% CI –0·03 to 0·23)
0·14 (95% CI –0·01 to 0·29)
0·09 (95% CI –0·03 to 0·21)
–1·77 (95% CI –6·94 to 3·39)
Data are mean (SD), unless otherwise specifi ed. *Adjusted mean intervention eff ect refers to mean diff erence in 12-month measurements between combined intervention groups and comparison group,
adjusted for baseline measurement and minimisation stratifi ers.
Table 2: Adjusted mean intervention eff ect on secondary outcomes for combined intervention groups versus comparison group
www.thelancet.com Vol 371 January 5, 2008
not aff ect these results. Similarly, we recorded no
diff erence between the groups in clinical and
biochemical measures at 1 year, or in self-reported
physical activity at six months.
The absence of an intervention eff ect on physical
activity was not due to failure to deliver the programme.
Programme attendance and acceptability were high.
Change in physical activity did not vary between
participants according to the facilitator who delivered
the inter vention. The degree of adherence to the protocol
did not aff ect its eff ectiveness, and neither did intensity
of intervention (ie, delivery by telephone or in the home).
Most intervention participants reported at both 6 months
and 1 year that they had used key behavioural change
strategies. Although the quality of their use was not
measured, the intervention was associated with a
moderate eff ect on intention to be more physically active
at 6 months. The theory of planned behaviour has not
been proven to predict changes in objectively assessed
physical activity over time, and further analyses of the
cognitive determinants of behavioural change in the
cohort are planned.
We did record some small-to-moderate eff ects of the
intervention for improvement of self-reported health
status and reduction of anxiety. These positive eff ects
occured independently of any charge in physical activity,
which suggested a mechanism that was independent of
Other trials have also shown that interventions have
aff ected wellbeing but not behaviour or clinical
endpoints.31 This eff ect might have been mediated by a
sense of false reassurance from participation in the
intervention.32 However, measures of self-reported health
have been independently associated with mortality risk,
and thus the improvement in self-rated health status
might warrant further study.33
Whereas between-group diff erences in physical activity
were small, the cohort analysis showed that physical
activity increased, across all groups combined, by an
amount that was equivalent to about 20 min of brisk
walking every day. However, despite the overall increase
in activity, participants were still quite sedentary at
follow-up, which suggests that an eff ective intervention
could increase physical activity more.
Other trials of interventions to promote physical activity
have also reported that physical activity increased in all
groups, and that counselling had little advantage over brief
advice delivered in a consultation.34,35 This might be due, at
least in part, to an intervention eff ect of the measurement
of physical activity and its determinants,36 the magnitude
Favours comparison group
Favours combined interventions
Baseline 6 month 1 year
Baseline 6 month 1 year
Combined interventions vs comparison group
Adjusted intervention effect (95% CI)
3·01 (0·72 to 5·30)
2·74 (–1·93 to 7·41)
13·0 (7·09 to 19·0)
6·31 (3·19 to 9·43)
Mental health 75·2
7·71 (4·98 to 10·4)
6·81 (3·66 to 9·97)
1·66 (–1·73 to 5·06)
3·75 (0·86 to 6·64)
Change in health
6·02 (2·44 to 9·61)
–3·61 (–5·98 to –1·24)
(2·99) (2·70) (2·47)
9·70 9·49 9·38
0·35 (–0·15 to 0·85)
risk of diabetes
–0·07 (–0·22 to 0·08)
Baseline 6 month 1 year
Intention to be
0·10 (–0·04 to 0·24)
Figure 3: Adjusted mean intervention eff ect on 6-month and 12-month psychosocial measures
Data are mean (SD), unless otherwise specifi ed. Adjusted mean intervention eff ect refers to mean diff erence in 12-month measurements between combined intervention groups and comparison group,
adjusted for baseline measurement and minimisation stratifi ers. *Scale 0–100. †Scale 20–80. ‡Scale 6–27, converted to percentage. §Scale 1–5. Normative mean from reference 22.
www.thelancet.com Vol 371 January 5, 2008 47
of which might vary between baseline and follow-up. We
used questionnaires that might have drawn attention to
the sources of everyday activities, and to the psychological
mediators of these behaviours. Similarly, our use of an
objective measure of physical activity, with detailed
physiological measurement, might also have increased
the salience of activity to participants. We recruited
participants throughout the year, so seasonal variation was
unlikely to have caused the eff ect. However, we did select
participants who were motivated to change: at trial entry
they viewed their risk of diabetes as slightly above average,
and more than half had a strong intention to do more
activity, which suggest that the group had pre-existing
motivation to change their behaviour.
The complexity of the intervention could have confused
rather than strengthened the messages in the advice
leafl et given to all participants, and interacted to reduce
rather than increase its eff ectiveness. Our results suggest
that less intensive support might be as eff ective as more
Reviews of trials of a range of community-based
approaches have reported some eff ects on physical
activity, although activity did not increase to current
public health recommendations.8 However, these reviews
included short-term studies that assessed self-reported
physical activity.4–8 A recent synthesis of trials showed
that a range of approaches to increasing physical activity
in community samples had a moderate positive eff ect on
self-reported physical activity (pooled standardised mean
diff erence 0·31; 95% CI 0·12–0·50) and on measures of
cardiorespiratory fi tness.5 None of these studies captured
overall energy expenditure directly.
A recent synthesis of reviews underlined the potential
of brief advice supported by written materials to increase
physical activity, and identifi ed common attributes of
eff ective interventions.8 These included provision of
individually tailored advice, goal-setting, self-monitoring,
ongoing support, promotion of moderate-intensity
activity, and exploring beliefs about the costs and benefi ts
of physical activity.8 Our trial intervention combined all
of these attributes.
We recruited an at-risk population through primary
care; for example, at diagnosis of type 2 diabetes in a
family member. Therefore our results are more relevant to
health services than previous studies which have recruited
Our study investigated a homogeneous social and ethnic
group. Most participants were sedentary, overweight,
female, and living with children. The link between these
characteristics and responsiveness to physical-activity
interventions in other studies is inconsistent and might
be dependent on the intervention.5
Other trials in groups at risk of diabetes through
obesity, family history, life stage, or ethnic group have
focused on weight loss or prevention of weight gain, and
have shown that promotion of physical activity can cause
small eff ects, usually in combination with dietary
control.37,38 Data from observational studies suggest that
sustained prevention of weight gain might require much
greater increases in energy expenditure than those that
are usually targeted or attained in trials.39
We aimed to address study design issues that have
hindered interpretation of previous studies; however, in
doing so we created a situation in which study participation
seemed to have a greater eff ect on physical activity than
did the intervention. Although objective assessment of
usual physical activity has advantages over questionnaire
assessment, estimation of this complex variable remains
a challenge.9 Similarly, the advantages of measuring
cognitive processes along the hypothesised causal path to
behavioural change to investigate mechanisms of eff ect
must be balanced by the imprecision of measurement.
Moreover, the questionnaires themselves could have an
intervention eff ect. Interpretation of our fi ndings would
have been assisted by inclusion of a control group who
received no leafl et. Future studies should assess the
intervention eff ects of measurement in pilot work or
through trial designs that include measurement control,
and eff orts should be made to minimise the overall
burden of trial participation.
Participants who did the programme reported that the
intervention was acceptable and that they felt physically
and emotionally better than did those in the comparison
group. The cohort analysis suggests that increases in
physical activity are possible in this group and points to
new avenues for individual interventions based on, for
example, self-monitoring of behaviour after brief advice.
Such interventions might be cheaper than more intensive
behavioural interventions that require investment in
Our fi ndings suggest that approaches based on personal
education and individual behaviour change alone are
unlikely to increase physical activity in an environment
where there are plentiful inducements to keep still.
Preventive health-care systems that aim to reverse the
trend towards sedentary living in populations and defi ned
at-risk groups should consider the need for wider
public-health strategies and sociocultural change.2
A-LK, NW, WH, SS, ATP, DS, and SG participated in the design of the trial
and intervention. A-LK and WH directed the intervention team; NW
directed the clinical measurement team; SS directed psychological
measurements; and SG directed the trial coordination team. All authors
participated in the acquisition and analysis of data, and in critical revision
of the manuscript, and have seen and approved the fi nal version.
Confl ict of interest statement
We declare that we have no confl ict of interest.
We thank study participants and practice teams; the trial coordination
team (led by KW and Julie Grant); measurement team (led by UE and
Emanuella De Lucia-Rolfe); intervention facilitators (Rosamund Cole,
Philippa Gash, Kathryn Julian, Nan Millette, Nicky McLean, Chris Swain,
Sue Boase, and Judith Argles); Marie Johnston, Susan Michie, Shirley
Pearce, Imogen Hobbis, Tony White, and Will Hollingworth for
contributions to to the development and delivery of the intervention; Larry
Pulley for design of educational materials; the chair of the Trial Steering
Articles Download full-text
www.thelancet.com Vol 371 January 5, 2008
Committee (Janet Darbyshire), and its external members (Dick Eiser and
Martin White); Esther van Sluijs and David Mant for helpful comments on
the manuscript; and the Medical Research Council, National Health
Service R&D, the Royal College of General Practitioners’ Scientifi c
Foundation, and Diabetes UK. The General Practice and Primary Care
Research Unit is part of the National Institute of Health Research School
for Primary Care Research.
1 Murray CJL, Lopez AD. The global burden of disease. Harvard, MA,
USA: Harvard School of Public Health ,1996.
2 World Health Organization. The World Health Report 2002.
Reducing Risks, Promoting Healthy Life. Geneva, Switzerland:
3 US Preventative Services Task Force. Behavioral counseling in
primary care to promote physical activity: recommendation and
rationale. Ann Intern Med 2002; 137: 205–07.
4 Eden KB, Orleans CT, Mulrow CD, Pender NJ, Teutsch SM. Does
counseling by clinicians improve physical activity? A summary of the
evidence for the U.S. Preventive Services Task Force. Ann Intern Med
2002; 137: 208–15.
5 Hillsdon M, Foster C, Thorogood M. Interventions for promoting
physical activity. Cochrane Database Syst Rev 2005; 1: CD003180.
6 Simons-Morton DG, Calfas KJ, Oldenburg B, Burton NW. Eff ects of
interventions in health care settings on physical activity or
cardiorespiratory fi tness. Am J Prev Med 1998; 15: 413–30.
7 NICE. NICE Public Health intervention guidance: Four commonly
used methods to increase physical activity: brief interventions in
primary care, exercise referral schemes, pedometers and
community-based exercise programmes for walking and cycling.
London, UK: National Institute for Health and Clinical
Excellence, 2006. http://www.nice.org.uk/guidance/index.jsp?action
=byID&o=11373#documents (accessed on Dec 10, 2007).
8 Hillsdon M, Foster C, Cavill N, Crombie H, Naidoo B. The
eff ectiveness of public health interventions for increasing physical
activity among adults: a review of reviews. London, UK: Health
Development Agency, 2005.
9 Wareham NJ, Rennie KL. The assessment of physical activity in
individuals and populations: why try to be more precise about how
physical activity is assessed? Int J Obes Relat Metab Disord 1998;
22 (Suppl 2): S30–38.
10 Michie S, Abraham C. Interventions to change health behaviours:
evidence-based or evidence-inspired? Psychol Health 2004; 19: 29–49.
11 Dishman RK, Buckworth J. Increasing physical activity: a quantitative
synthesis. Med Sci Sports Exerc 1996; 28: 706–19.
12 Griffi n SJ, Kinmonth AL, Skinner C, Kelly J. Educational and
psychosocial interventions for adults with diabetes. London, UK:
British Diabetic Association, 1999.
13 Armitage CJ, Conner M. Effi cacy of the Theory of Planned Behaviour:
a meta-analytic review. Br J Clin Psychol 2001; 40: 471–99.
14 Ajzen I. The Theory of Planned Behavior. Organ Behav Decis Process
1991; 50: 179–211.
15 Sargeant LA, Wareham NJ, Khaw KT. Family history of diabetes
identifi es a group at increased risk for the metabolic consequences of
obesity and physical inactivity in EPIC-Norfolk: a population-based
study. The European Prospective Investigation into Cancer.
Int J Obes Relat Metab Disord 2000; 24: 1333–39.
16 Williams K, Prevost AT, Griffi n S, et al. The ProActive trial protocol
—a randomised controlled trial of the effi cacy of a family-based,
domiciliary intervention programme to increase physical activity
among individuals at high risk of diabetes [ISRCTN61323766].
BMC Public Health 2004; 4: 48.
17 Godin G, Shephard RJ. A simple method to assess exercise behavior
in the community. Can J Appl Sport Sci 1985; 10: 141–46.
18 Wareham NJ, Jakes RW, Rennie KL, et al. Validity and repeatability of
a simple index derived from the short physical activity questionnaire
used in the European Prospective Investigation into Cancer and
Nutrition (EPIC) study. Public Health Nutr 2003, 6: 407–13.
19 Hardeman W, Sutton S, Griffi n S, et al. A causal modelling approach
to the development of theory-based behaviour change programmes
for trial evaluation. Health Educ Res 2005; 20: 676–87.
20 Wareham NJ, Hennings SJ, Prentice AM, Day NE. Feasibility of
heart-rate monitoring to estimate total level and pattern of energy
expenditure in a population-based epidemiological study: the Ely
Young Cohort Feasibility Study 1994–5. Br J Nutr 1997; 78: 889–900.
21 Wareham NJ, Jakes RW, Rennie KL, Mitchell J, Hennings S, Day NE.
Validity and repeatability of the EPIC-Norfolk Physical Activity
Questionnaire. Int J Epidemiol 2002; 31: 68–74.
22 Jenkinson C, Layte R, Coulter A, Wright L. The UKSF-36 An analysis
and interpretation manual. Oxford, UK: Health Services Research
Unit, University of Oxford; 1996.
23 Samsa G, Edelman D, Rothman ML, Williams GR, Lipscomb J,
Matchar D. Determining clinically important diff erences in health
status measures: a general approach with illustration to the Health
Utilities Index Mark II. PharmacoEconomics 1999; 15: 141–55.
24 Marteau TM, Bekker H. The development of a six-item short-form of
the state scale of the Spielberger State-Trait Anxiety Inventory (STAI).
Br J Clin Psychol 1992; 31: 301–06.
25 Lerman C, Trock B, Rimer BK, Jepson C, Brody D, Boyce A.
Psychological side eff ects of breast cancer screening. Health Psychol
1991; 10: 259–67.
26 Weinstein ND. Perceptions of personal susceptibility to harm. In:
Mays VM, Albee GW, Schneider SF, eds. Primary prevention of
AIDS: Psychological approaches. Newbury Park, CA, USA: Sage,
27 White IR, Thompson SG. Adjusting for partially missing baseline
measurements in randomized trials. Stat Med 2005; 24: 993–1007.
28 Cohen J. Statistical power analysis for the behavioural sciences. 2nd
ed. Hillsdale, NJ, USA: Lawrence Erlbaum Assoc, 1988.
29 Wareham NJ, Wong MY, Day NE. Glucose intolerance and physical
inactivity: the relative importance of low habitual energy expenditure
and cardiorespiratory fi tness. Am J Epidemiol 2000; 152: 132–39.
30 Ainsworth BE, Haskell WL, Whitt MC, et al. Compendium of physical
activities: an update of activity codes and MET intensities.
Med Sci Sports Exerc 2000; 32 (Suppl): 498–504.
31 Kinmonth AL, Woodcock A, Griffi n S, Spiegal N, Campbell MJ.
Randomised controlled trial of patient centred care of diabetes in
general practice: impact on current wellbeing and future disease risk.
The Diabetes Care From Diagnosis Research Team. BMJ 1998;
32 Marteau TM, Kinmonth AL, Thompson S, Pyke S. The psychological
impact of cardiovascular screening and intervention in primary care:
a problem of false reassurance? Br J Gen Pract 1996; 46: 577–82.
33 Idler EL, Benyamini Y. Self-rated health and mortality: a review of
twenty-seven community studies. J Health Soc Behav 1997; 38: 21–37.
34 The Writing Group for the Activity Counseling Trial Research Group.
Eff ects of physical activity counseling in primary care: The Activity
Counseling Trial: a randomized controlled trial. JAMA 2001; 286: 677–87.
35 van Sluijs EM, van Poppel MN, Twisk JW, Chin APMJ, Calfas KJ, van
Mechelen W. Eff ect of a tailored physical activity intervention
delivered in general practice settings: results of a randomized
controlled trial. Am J Public Health 2005; 95: 1825–31.
36 van Sluijs EM, van Poppel MN, Twisk JW, van Mechelen W. Physical
activity measurements aff ected participants’ behavior in a randomized
controlled trial. J Clin Epidemiol 2006; 59: 404–11.
37 Hamman RF. Prevention of Type 2 Diabetes. In: Williams R,
Herman W, Kinmonth AL, Wareham N, (eds) Evidence Base for
Diabetes Care. Chichester: Wiley, 2002: 75–176.
38 Wing RR, Venditti E, Jakicic JM, Polley BA, Lang W. Lifestyle
intervention in overweight individuals with a family history of
diabetes. Diabetes Care 1998; 21: 350–59.
39 Erlichman J, Kerbey AL, James WP. Physical activity and its impact
on health outcomes. Paper 2: Prevention of unhealthy weight gain
and obesity by physical activity: an analysis of the evidence. Obes Rev
2002; 3: 273–87.