motivational enhancement therapy with and without cognitive behaviour therapy for type 1 diabetes:...
TRANSCRIPT
Article: Treatment
Motivational enhancement therapy with and without
cognitive behaviour therapy for Type 1 diabetes:
economic evaluation from a randomized controlled trial
A. Patel, E. Maissi, H.-C. Chang, I. Rodrigues*, M. Smith, S. Thomas†, T. Chalder, U. Schmidt,J. Treasure and K. Ismail
Institute of Psychiatry, King’s College London, *Guy’s, King’s and St Thomas’ School of Medicine and †Diabetes Centre, St Thomas’ Hospital, London, UK
Accepted 10 November 2010
Abstract
Aims To assess the cost-effectiveness of motivational enhancement therapy and cognitive behaviour therapy for poorly
controlled Type 1 diabetes.
Methods Within-trial prospective economic evaluation from (i) health and social care and (ii) societal perspectives. Three
hundred and forty-four adults with Type 1 diabetes for at least 2 years and persistent, suboptimal glycaemic control were
recruited to a three-arm multi-centre randomized controlled trial in London and Manchester, UK. They were randomized to (i)
usual care plus four sessions of motivational enhancement therapy (ii) usual care plus four sessions of motivational enhancement
therapy and eight sessions of cognitive behaviour therapy or (iii) usual care alone. Outcomes were (i) costs, (ii) Quality-Adjusted
Life Year gains measured by the EuroQol 5-dimensional health state index and the 36-item Short Form and (iii) diabetes control
measured by change in HbA1c level at 1 year.
Results Both intervention groups had significantly higher mean health and social care costs (+ £535 for motivational
enhancement therapy and + £790 for combined motivational enhancement and cognitive behavioural therapy ), but not societal
costs compared with the usual-care group. There were no differences in Quality Adjusted Life Years. There was a significantly
greater HbA1c improvement in the combined motivational enhancement and cognitive behavioural therapy group (+ 0.45%;
incremental cost-effectiveness ratio = £1756), but the not in the motivational enhancement therapy group. Cost-effectiveness
acceptability curves suggested that both interventions had low probabilities of cost-effectiveness based on Quality Adjusted Life
Years (but high based on HbA1c improvements). Imputing missing costs and outcomes confirmed these findings.
Conclusions Neither therapy was undisputedly cost-effective compared with usual care alone, but conclusions vary depending
on the relative importance of clinical and quality-of-life outcomes.
Diabet. Med. 28, 470–479 (2011)
Keywords cost, cost-effectiveness, diabetes, psychological therapy
Abbreviations CBT, cognitive behaviour therapy; EQ5D, EuroQol 5-dimensional health state index; MET,
motivational enhancement therapy; QALY, quality-adjusted life year
Introduction
Diabetes and its complications incur substantial costs for health
services [1,2], patients and wider society as a result of typically
early onset (Type 1 diabetes), increasing prevalence (Type 2),
chronicity and premature mortality. Providing cost-effective
treatment and long-term management approaches is particularly
important for those at high risk of complications and those
with suboptimal glycaemic control and difficulties in adhering
to their self-care regimen. Self-care difficulties are often asso-
ciated with psychological problems, which may be improved
by psychological treatments. Although the effectiveness of
psychological therapies to improve glycaemic control in such
individuals has shown promise in Type 2 diabetes [3] and
adolescents with Type 1 diabetes [4], effectiveness and
Correspondence to: Dr Anita Patel, Health Service and Population Research
Department, Institute of Psychiatry, King’s College London, De Crespigny
Park, London SE5 8AF, UK. E-mail: [email protected]
(Clinical Trials Registry No. ISRCTN 77044517)
DIABETICMedicine
DOI:10.1111/j.1464-5491.2010.03198.x
ª 2011 The Authors.470 Diabetic Medicine ª 2011 Diabetes UK
cost-effectiveness in adults with Type 1 diabetes has not been
established. We recently reported a three-arm, multi-centre,
randomized controlled trial which compared motivational
enhancement therapy (MET) with and without cognitive
behaviour therapy (CBT) for adults with Type 1 diabetes
against usual care [5]. We found that motivational enhancement
therapy with cognitive behaviour therapy resulted in significant
improvement in HbA1c levels over 1 year, but motivational
enhancement therapy alone did not. We now report a
comprehensive economic evaluation from that trial.
Patients and methods
The economic evaluation was undertaken from health and social
care, and societal, perspectives over 1 year. Individual-level
economic data were collected within a three-arm, multi-centre
randomized controlled trial, full details of which are described by
Ismail et al. [5]. To summarize, 344 adults (18–65 years) with
Type 1 diabetes for at least 2 years and persistent, suboptimal
glycaemic control were recruited from five hospitals in south-east
London and three hospitals in Greater Manchester, England.
Ethics review and approval was obtained from all relevant
committees.
Interventions
Participants were randomized to receive either (i) four sessions
of a diabetes-specific manualized motivational enhancement
therapy over 2 months (MET) in addition to their usual diabetes
care, (ii) four sessions of motivational enhancement therapy
(over 2 months) followed by eight sessions of manualized
cognitive behaviour therapy (MET + CBT) (over an additional
4 months) in addition to their usual diabetes care or (iii) usual
diabetes care alone.
The first motivational enhancement therapy session was a
standardized computerized self-assessment of diabetes-relevant
behaviours, followed by feedbackandan assessment of the rating
of the level of importance, confidence and readiness to change.
Remaining sessions covered obstacles to, and options for,
changing behaviours, emphasizing participants’ own problem-
solving skills andcollaborative formulationofa changeplan.The
therapy was supplemented with homework writing tasks to
increase motivation.
The cognitive behaviour therapy manual covered developing a
collaborative individualized programme, with each session
structured around agenda setting, homework planning and
feedback. Strategies included: normalizing dietary, exercise,
diabetes-related self-monitoring and lifestyle-related behaviours;
anxiety, worry and stress management; challenging diabetes-
specific negative automatic thoughts; improving impulse control;
behavioural experiments; activity scheduling; strategies for
eliciting social support; and assertiveness training.
Both interventions were delivered in diabetes clinics separate
from usual clinic visits by six trained nurses assessed for
competency.
Data collection
Individual-level economic data related to impacts attributable to
diabetes or related illness were collected using an adapted Client
ServiceReceipt Inventory (CSRI) [6]. It includedquestionsabout:
socio-demographic and socio-economic profile; time off work;
use of health and social care resources; informal care and any
time carers tookoff work to provide such care; and out-of-pocket
expenses. It was administered retrospectively at three assessment
points: by face-to-face interview at baseline (covering the
previous 3-month period), and by telephone interview at
6 months and 12 months after randomization (covering the
previous 6-month periods).
Additionally, economic impacts of attending a typical
intervention session were measured using a self-complete
questionnaire at participants’ last therapy session (or posted for
self-completion if they did not attend all sessions) to avoid
revealing randomization allocation during follow-up interviews.
This covered: time taken to attend a session; whether they took
time off work to do so and, if so, in which way (annual leave, sick
leave, unpaid leave, made up the time or other arrangement); lost
pay; and travel costs.
Health-related quality life was measured using the 36-item
Short Form [7] for the purpose of estimating quality-adjusted life
years (QALYs). As general population utility weights were not
yet available for this measure when this study was designed, we
additionally included the EuroQol 5-dimensional health state
index (EQ5D) [8]. Both were administered at baseline and
12 months.
Costs
Each participant’s resource-use data were multiplied by
appropriate unit costs (see also Supporting Information,
Appendix S1) to calculate costs for each assessment period.
Follow-up costs were summed to calculate total 1-year costs and
are reported in pounds sterling (£) at 2005 ⁄ 2006 prices.
Discounting was unnecessary as all costs were related to a
1-year period.
Total health and social care costs consisted of: hospital
inpatient and outpatient services, primary care services, other
community-based services, social services, medications, insulin-
related equipment, other equipment and adaptations and
intervention costs. Total societal costs consisted of the same
plus: informal care; participants’ and families’ out-of-pocket
expenses (including travel to intervention sessions); lost
productivity because of work absence; and lost productivity,
lost leisure time and lost pay because of attending intervention
sessions.
Motivational enhancement therapy and cognitive behaviourtherapy unit costs
After identifying time and material inputs, including training and
supervision, directly associated with an average session of each
DIABETICMedicineOriginal article
ª 2011 The Authors.Diabetic Medicine ª 2011 Diabetes UK 471
therapy, we estimated motivational enhancement therapy and
cognitive behaviour therapy unit costs (from the healthcare
perspective) as an average cost per session ⁄ per person for each
intervention, assuming resource inputs for one session of each
intervention did not significantly vary between sessions or
persons (Table 1).
One 50-min motivational enhancement therapy session cost
£49 and £48 per session, including and excluding training,
respectively. Despite having equivalent patient contact time,
cognitive behaviour therapy involved more training and
supervision inputs than motivational enhancement therapy, so
respective costs for cognitive behaviour therapy were £81 and
£73.
Individual-level intervention costs were calculated by
multiplying these unit costs by the number of each type of
therapy session attended.
Outcome measures
We linked costs with the trial’s primary outcome measure,
diabetes control as measured by change in HbA1c level between
baseline and 12 months, and QALYs. Although relevant36-item
Short Form utility weights were available by the end of the study,
we additionally calculated EQ5D-based QALYs to maximize
potential for comparisonswithother evidence.Utilityweights for
eachmeasure [9,10]wereattached tohealth states atbaselineand
12 months to calculate QALYs using the total area under the
curve approach with linear interpolation between assessment
points (and baseline adjustment for comparisons) [11].
Analyses
We analysed data using SPSS for Windows release 12.0.1 (SPSS
Inc., Chicago, IL, USA;1989–2001), Stata 8.2 for Windows
(StataCorp., College Station, TX, USA; 1985–2004) and Stata
for Windows 10.1 (StataCorp; 1985–2008).
Costs and outcomes are presented as mean values with
standard deviations. Mean differences and 95% CIs were
obtained by non-parametric bootstrap regressions (1000 repeti-
tions) to account for the non-normal distribution commonly
found in economic data. Although the three groups were
expected to be balanced at baseline, baseline costs and
outcomes could be predictors of follow-up costs. To provide
more relevant treatment-effect estimates [12], the regression on
costs included covariates for the baseline value for the same cost
category and baseline HbA1c. Similarly, regressions on outcomes
included covariates for the baseline value of the same outcome.
All groups were compared against each other in turn, keeping
participants in the randomization group to which they were
assigned regardless of the number ⁄ type of intervention sessions
they attended.
Collecting Client Service Receipt Inventory data by interview,
rather than self-complete, resulted inminimal itemnon-response.
In the few instances of missing items, values were imputed to
enable estimation of cost subtotals ⁄ totals. Where there was
indication of resource use (e.g. duration of contact was provided,
but number of contacts was missing), the mean value for other
users of that resource in the same randomization group at the
same assessment point was assumed. Where there was no such
indication of use, it was assumed not and a zero cost was
allocated for that resource. Where medication name was missing,
but other information (e.g. dose) indicated use, an average
national prescription cost was assumed. Where medication name
was provided without usage quantity, an average national
prescription cost for that particular medication was assumed.
Client Service Receipt Inventory non-responders at 6-month
and ⁄ or 12-month follow-up were excluded from analyses
because both assessments were needed to compute 1-year costs.
Table 1 Motivational enhancement therapy (MET) and cognitive behaviour therapy (CBT) resource inputs and unit costs (£, 2005 ⁄ 2006 prices)
Resources MET unit cost CBT unit cost
Delivery to patient Therapist* contact and non-contact time 24 26
Therapist supervision Therapist and supervisor� contact and non-contact time 22 46
Therapist training Therapist and trainer� contact and non-contact time 1 8
Materials Manual, information sheets, Accu-Test CD-ROM, tape
recorder� and tapes�1 1
Other inputs Therapist time to chase non-attenders 1 < 1
Total cost per 50-min session
(cost assuming 20% higher attendance rate)
49 (46) 81 (73)
Total cost per 50-min session, excluding
training costs (cost assuming 20% higher
attendance rate)
48 (45) 73 (66)
*Therapist salary and on-costs [10] based on a nurse on the mid-point of National Health Service Agenda for Change (NHS AfC) Band 6
(£0.39 per min).
�MET supervisor ⁄ trainer costs [10] were based on a clinical psychiatrist on the mid-point of NHS AfC Band 8A (£0.75 per min); CBT
supervisor ⁄ trainer salary and on-costs [10] were based on a senior CBT therapist on the mid-point of NHS AfC Band 8A (£0.75 per min) and
a junior CBT therapist on the low-point of NHS AfC Band 8A (£0.55 per min).
�Office Depot Business Solutions Catalogue, 2007.
DIABETICMedicine Cost-effectiveness of psychological therapies for Type 1 diabetes • A. Patel et al.
ª 2011 The Authors.472 Diabetic Medicine ª 2011 Diabetes UK
Similarly, outcome calculations required both baseline and
12 months’ values, so participants with either of these missing
were also excluded. Examinations of cost data on their own are
based on available cases regardless of availability of outcomes
data, and vice versa for examinations of outcomes data.
However, cost-effectiveness and cost-utility analyses only
include participants with both cost and relevant outcome data.
Sensitivity analyses
To explore the potential impact of excluding cases with missing
data, we examined characteristics of included and excluded
cases. We also imputed missing 1-year costs and outcomes using
the multiple imputation procedure in Stata for Windows 10.1.
Cost imputations were based on variables expected to predict
follow-up costs: randomization group, baseline age, sex, baseline
HbA1c, baseline value for the same cost category and the number
of therapy sessions attended. Predictor variables for outcome
imputations were the same, except that they included the
baseline value of the same outcome rather than cost. We then
compared mean cost and outcome differences for this imputed
full sample.
Cost-effectiveness and cost utility
With two cost perspectives, three outcomes and three arms,
there were 18 cost–outcome combinations to examine. We
calculated incremental cost-effectiveness ratios (ICERs; mean
cost difference divided by mean outcome difference) for any
combination which showed both significantly higher costs and
outcomes in one group compared with another.
Given difficulties around estimating confidence intervals for
incremental cost-effectiveness ratios, we explored uncertainty
using cost-effectiveness acceptability curves (CEACs) based on
the net-benefit approach [13]. These represent the probability
that one intervention is cost-effective compared with another,
accounting for hypothetical monetary values that policymakers
may place on point improvements in each outcome.
Cost-effectiveness acceptability curves were constructed by first
calculating a series of net benefits for each individual, using the
following formula, wherek represents the monetary value placed
on one additional unit of outcome: net benefit = (k · outcome)
– cost.
We examined k values ranging £0 to £45 000 (in £2500
increments) to cover the £20 000–£30 000 per QALY gain
threshold range currently specified for National Institute for
Health and Clinical Excellence (NICE) decision making in
England and Wales [14]. We then calculated between-group
differences in mean net benefits (for each value of k) using series
of bootstrapped linear regressions (1000 repetitions), which
included covariates for the baseline value of the same cost
category and outcome. We then counted the proportion of times
one group had a greater net benefit than another (for each value
of k) and plotted these proportions (or probabilities) for all 18
cost–outcome combinations.
Results
Participant characteristics and response rates
Of participants, 216 ⁄ 344 (62.8%) had both 6- and 12-month
cost data, i.e. data required for the calculation of 1-year costs.
Those with cost data were on average 2 years older and had
better mean HbA1c levels (9.13% ⁄ 76 mmol ⁄ mol vs. 9.32% ⁄78 mmol ⁄ mol) at 12 months compared with the full study
sample, although differences were not explored statistically
(Table 2).The samplewas further reduced forcost-effectiveness ⁄cost-utility analyses. Of particular note, EQ5D-based analyses
included only 50.4% of the MET group and 36-item Short
Form-based analyses included only 48.8% of the usual-care
group.
Resource use
Resource use was broadly comparable between groups at all
assessments (not tested statistically to avoid multiple testing).
Table 2 Sample characteristics
Full sample
(n = 344)
Subsample with
1-year cost data
(n = 216)
Subsample with
1-year cost data
and HbA1c data
(n = 207)
Subsample with
1-year cost data
and SF-36-based
QALY gain data
(n = 189)
Subsample with
1-year cost data
and EQ5D-based
QALY gain data
(n = 185)
Valid no. Valid no. Valid no. Valid no. Valid no.
Mean age (years) 344 36 216 38 207 38 189 38 185 38
Number male (%) 344 136 (39.5) 216 87 (40.3) 207 83 (40.1) 189 74 (39.2) 185 72 (38.9)
Number female (%) 344 208 (60.5) 216 129 (59.7) 207 124 (59.9) 189 115 (60.8) 185 113 (61.1)
Mean HbA1c at baseline,
% (sd; mmol ⁄ mol)
344 9.63 (1.15; 81) 216 9.60 (1.17; 81) 207 9.58 (1.17; 81) 189 9.52 (1.09; 80) 185 9.52 (1.08; 80)
Mean HbA1c at 12 months,
% (sd; mmol ⁄ mol)
305 9.32 (1.52; 78) 207 9.13 (1.35; 76) 207 9.13 (1.35; 76) 186 9.03 (1.29; 75) 182 9.03 (1.32; 75)
EQ5D, EuroQol 5-dimensional health state index; QALY, quality-adjusted life year; SF-36, 36-item Short Form.
DIABETICMedicineOriginal article
ª 2011 The Authors.Diabetic Medicine ª 2011 Diabetes UK 473
Participants were high users of hospital-based specialist diabetes
services, general practitioner surgery services, chiropody and
opticians (Table 3) and used few other community-based
services.
Costs
Mean intervention costs were £195 for the MET group and £660
for the MET + CBT group (Table 4). While specific follow-ups
showed no cost differences, total 1-year health and social care
costs were higher for both intervention groups compared with
the usual-care group (but did not differ between the two
intervention groups). There were no differences in costs unre-
lated to the interventions, suggesting that the additional costs of
the interventions were neither fully offset by savings elsewhere
nor led to additional costs. Patient ⁄ family and lost productivity
costs, which were small compared with health and social care
costs, did not differ between the groups. Total societal costs
showed no differences, although confidence intervals do suggest
a tendency towards higher costs for both intervention arms
compared with usual care.
Outcomes
There was no significant difference in HbA1c improvement
between the MET and usual-care groups or between the MET
and MET + CBT groups, but the MET + CBT group showed a
significantly greater improvement of 0.45% compared with the
usual-care group (Table 5).
Neither the EQ5D nor the 36-item Short Form suggested
differences in mean QALYs. Despite some quantitative dif-
ferences in results derived from the two measures (with the
EQ5D indicating greater mean total QALYs per group and thus
slightly greater mean differences), both suggested the same
direction of difference.
Sensitivity analyses
The available case analyses are likely to be representative of the
full study sample because imputing missing costs and outcomes
produced similar findings (Tables 4 and 5). The only notable
difference following imputation was a very small QALY
disadvantage (–0.0001 QALYs), rather than advantage (0.003
QALYs) for the MET + CBT group compared with usual care.
This would alter cost-effectiveness conclusions based on that
particular comparison, but the meaningfulness of this is unclear
given the small size and lack of statistical significance of these
differences.
Cost utility and cost-effectiveness
Only one cost–outcome combination showed significant dif-
ferences in both costs and outcomes: the MET + CBT group had
higher health and social care costs (+ £790) and greater HbA1c
improvement (+ 0.45 points) compared with the usual-care Tab
le3
Most
com
monly
use
dhea
lth
and
soci
alca
rese
rvic
esat
6an
d12
month
s(d
uri
ng
pre
vious
6m
onth
s)
Unit
ME
TM
ET
+C
BT
Usu
al
care
6m
onth
s(n
=84)
12
month
s(n
=96)
6m
onth
s(n
=82)
12
month
s(n
=88)
6m
onth
s(n
=77)
12
month
s(n
=102)
Use
rs(n
)M
ean
*sd
Use
rs(n
)M
ean
*sd
Use
rs(n
)M
ean
*sd
Use
rs(n
)M
ean
*sd
Use
rs(n
)M
ean
*sd
Use
rs(n
)M
ean
*sd
Sec
ondary
care
Inpati
ent
ward
adm
issi
on
Nig
hts
812
99
83
67
27
52
311
39
62
Dia
bet
iccl
inic
Vis
its
56
21
72
22
59
21
64
22
52
11
72
23
Dia
bet
esfo
ot
clin
icV
isit
s6
610
——
—9
34
——
—5
55
——
—
Dia
bet
esey
ecl
inic
Vis
its
27
11
23
11
31
11
21
10
24
11
26
10
Ophth
alm
olo
gy
Vis
its
71
113
10
81
09
11
51
-14
11
Phle
boto
my
Vis
its
22
1—
——
82
1—
——
32
1—
——
Pri
mary
care
and
com
munit
y-b
ase
dse
rvic
es
Gen
eral
pra
ctit
ioner
Vis
its
27
21
40
21
26
21
36
21
30
33
48
47
Dia
bet
essp
ecia
list
nurs
eV
isit
s6
22
15
21
82
112
21
71
113
11
Pra
ctic
enurs
eV
isit
s9
10
14
23
14
11
82
28
22
17
21
Chir
opodis
tV
isit
s7
22
12
48
91
19
11
82
19
21
Opti
cian
Vis
its
——
—16
10
——
—16
10
——
—15
10
*M
ean
for
use
rsonly
.
CB
T,
cognit
ive
beh
avio
ur
ther
apy;
ME
T,
moti
vati
onal
enhance
men
tth
erapy.
DIABETICMedicine Cost-effectiveness of psychological therapies for Type 1 diabetes • A. Patel et al.
ª 2011 The Authors.474 Diabetic Medicine ª 2011 Diabetes UK
Tab
le4
Mea
nco
sts
and
mea
nco
stdif
fere
nce
sat
bas
elin
e(f
or
pre
vious
3m
onth
s)an
dove
r1
year
(£,2005
⁄2006
pri
ces)
Cost
cate
gory
ME
T
n=
73
ME
T+
CB
T
n=
73
Usu
al
care
n=
70
ME
Tvs.
usu
al
care
ME
T+
CB
Tvs.
usu
al
care
ME
Tvs.
ME
T+
CB
T
Mea
nsd
Mea
nsd
Mea
nsd
Adju
sted
mea
n
dif
fere
nce
*95%
CI
Adju
sted
mea
n
dif
fere
nce
*95%
CI
Adju
sted
mea
n
dif
fere
nce
*95%
CI
ME
T⁄M
ET
+C
BT
inte
rven
tions
195
57
660
301
0—
195
183,
208
660
590,
727
465
390,
536
Base
line
Hea
lth
⁄soci
al
care
�540
605
564
561
497
301
43
–102,
213
67
–60,
228
24
–166,
215
Pati
ent
⁄fam
ily�
66
247
97
414
49
246
17
–66,
94
49
–53,
174
32
–64,
157
Lost
pro
duct
ivit
y§
129
531
22
54
65
382
63
–87,
212
–43
–160,
17
–106
–243,
–2
Soci
etal–
734
1024
684
831
611
602
124
–146,
403
73
–141,
329
–50
–354,
237
1-y
ear
foll
ow
-up
Hea
lth
⁄soci
al
care
�,ex
cludin
gM
ET
⁄CB
T1811
2032
1574
1305
1356
851
340
–23,
646
127
–155,
404
–287
–690,
146
Hea
lth
⁄soci
al
care
,in
cludin
gM
ET
⁄CB
T2006
2034
2234
1326
1356
851
535
171,
857
790
507,
1072
178
–229,
619
Pati
ent
⁄fam
ily�
766
1982
605
1474
592
3195
30
–638,
825
–229
–919,
432
–231
–751,
226
Lost
pro
duct
ivit
y§
149
749
150
757
57
126
48
–56,
156
88
–32,
303
100
–28,
305
Soci
etal–
,ex
cludin
gM
ET
⁄CB
T2725
3180
2329
2405
2005
3504
449
–599,
1357
149
–848,
914
–323
–1044,
416
Soci
etal,
incl
udin
gM
ET
⁄CB
T2920
3179
2989
2442
2005
3504
643
–414,
1549
814
–176,
1586
144
–581,
894
Sen
siti
vit
yanaly
sis—
impute
dfu
llsa
mple
Hea
lth
⁄soci
al
care
,ex
cludin
gM
ET
⁄CB
T1809
1748
1694
1303
1511
1115
258
39,
456
144
–88,
364
–141
–397,
128
Hea
lth
⁄soci
al
care
,in
cludin
gM
ET
⁄CB
T1999
1744
2250
1261
1501
1112
458
242,
653
707
474,
926
226
–17,
492
Soci
etal,
excl
udin
gM
ET
⁄CB
T2607
2620
2488
2371
2248
2998
350
–340,
895
154
–474,
636
–186
–666,
289
Soci
etal,
incl
udin
gM
ET
⁄CB
T2797
2617
3044
2363
2239
2995
550
–142,
1104
718
90,
1197
182
–309,
656
*C
om
pari
sons
of
6-m
onth
,12-m
onth
and
1-y
ear
cost
sin
clude
covari
ate
sfo
rth
ebase
line
valu
eof
the
sam
eco
stca
tegory
and
base
line
HbA
1c.
�Hea
lth
⁄soci
al
care
cost
(excl
udin
gin
terv
enti
on
cost
)in
cludes
cost
sof
seco
ndary
care
,pri
mary
⁄com
munit
y-b
ase
dca
re,
med
icati
ons
and
equip
men
tas
are
sult
of
dia
bet
esand
rela
ted
illn
esse
s.
�Pati
ent
⁄fam
ily
cost
sin
clude
cost
sof
info
rmal
care
and
out-
of-
pock
etex
pen
ses
as
are
sult
of
dia
bet
esor
rela
ted
illn
esse
sand
lost
pay,
lost
leis
ure
tim
eand
travel
cost
sto
att
end
inte
rven
tion
sess
ions.
§L
ost
pro
duct
ivit
yco
sts
incl
ude
the
cost
sof
part
icip
ants
’ti
me
off
work
and
thei
rfa
mil
y⁄f
rien
ds’
tim
eoff
work
topro
vid
eca
refo
rth
emas
are
sult
of
dia
bet
esor
rela
ted
illn
esse
sand
tim
eoff
work
toin
terv
enti
on
sess
ions.
–Soci
etal
cost
isth
eto
tal
of
hea
lth
⁄soci
al
care
,pati
ent
⁄fam
ily
and
lost
pro
duct
ivit
yco
sts.
CB
T,
cognit
ive
beh
avio
ur
ther
apy;
ME
T,
moti
vati
onal
enhance
men
tth
erapy.
DIABETICMedicineOriginal article
ª 2011 The Authors.Diabetic Medicine ª 2011 Diabetes UK 475
group, translating into an incremental cost-effectiveness ratio of
£1756. Further incremental cost-effectiveness ratio calculations
were unnecessary as all other cost–outcome combinations
suggested either ‘neutrality’ (no significant differences in costs
and outcomes) or ‘dominance’ (one treatment option intuitively
preferable because of lower costs and better outcomes). Informal
incremental cost-effectiveness ratio calculations for the ‘neutral’
scenarios showed that, compared with usual care, motivational
enhancement therapy plus cognitive behaviour therapy gener-
ally had lower incremental cost-effectiveness ratios than
motivational enhancement therapy alone based on HbA1c
improvements, but higher incremental cost-effectiveness ratios
based on QALYs.
Ata thresholdof£25 000peradditionalQALY(themid-point
of the range specified by the National Institute for Health
and Clinical Excellence), probabilities of cost-effectiveness
for motivational enhancement therapy and motivational
enhancement therapy plus motivational enhancement therapy
compared with usual care and compared with each other were
low (0.33 maximum) from both perspectives (Fig. 1). Cost-
effectiveness acceptability curves based on the 36-item Short
Form and EQ5D were similar.
Probabilities of cost-effectiveness based on HbA1c improve-
ments were high for both interventions, but the implications of
this are unclear given that there is currently no policy-based cost-
effectiveness threshold based on HbA1c.
Discussion
Among several good-quality economic evaluations of various
approaches to management of diabetes and its complications
[15,16], there are none of psychological treatments for
improving glycaemic control in Type 1 diabetes. This
comprehensive, randomized controlled trial-based economic
evaluation suggests that neither motivational enhancement
therapy plus cognitive behaviour therapy nor motivational
enhancement therapy alone is an undisputedly cost-effectiveTab
le5
Mea
noutc
om
esan
dm
ean
outc
om
edif
fere
nce
sove
r1
year
Outc
om
e
ME
TM
ET
+C
BT
Usu
al
care
ME
Tvs.
usu
al
care
ME
T+
CB
Tvs.
usu
al
care
ME
Tvs.
ME
T+
CB
T
nM
ean
sd
nM
ean
sd
nM
ean
sd
Mea
n
dif
fere
nce
*95%
CI
Mea
n
dif
fere
nce
*95%
CI
Mea
n
dif
fere
nce
*95%
CI
Base
line
uti
lity
base
don
EQ
5D
114
0.7
60.2
5106
0.7
80.2
7119
0.7
90.2
5–0.0
3–0.1
0,
0.0
3–0.0
06
–0.0
8,
0.0
60.0
2–0.0
4,
0.0
9
12-m
onth
uti
lity
base
don
EQ
5D
90
0.7
90.2
683
0.7
80.2
894
0.7
90.2
7–0.0
01
–0.0
7,
0.0
8–0.0
1–0.0
9,
0.0
7–0.0
1–0.0
8,
0.0
7
Base
line
uti
lity
base
don
SF-3
6117
0.7
00.1
2106
0.7
00.1
4121
0.7
10.1
2–0.0
2–0.0
5,
0.0
1–0.0
1–0.0
4,
0.0
20.0
1–0.0
3,
0.0
4
12-m
onth
uti
lity
base
don
SF-3
696
0.7
40.1
385
0.7
30.1
489
0.7
40.1
1–0.0
01
–0.0
4,
0.0
4–0.0
1–0.0
4,
0.0
3–0.0
1–0.0
5,
0.0
3
QA
LY
sbase
don
EQ
5D
90
0.7
80
0.2
283
0.7
80
0.2
592
0.8
04
0.2
20.0
11
–0.0
2,
0.0
40.0
03
–0.0
3,
0.0
3–0.0
08
–0.0
4,
0.0
2
QA
LY
sbase
don
SF-3
696
0.7
21
0.1
185
0.7
19
0.1
389
0.7
29
0.1
00.0
04
–0.0
1,
0.0
20.0
002
–0.0
1,
0.0
1–0.0
04
–0.0
2,
0.0
1
HB
A1c
impro
vem
ent
105
0.2
41.4
695
0.5
91.3
8105
0.1
21.1
70.1
4–0.2
2,
0.4
80.4
50.1
1,
0.8
00.2
8–0.0
5,
0.6
6
Sen
siti
vit
yanaly
sis—
impute
dfu
llsa
mple
QA
LY
sbase
don
EQ
5D
117
0.7
70
0.2
3106
0.7
82
0.2
5121
0.7
89
0.2
30.0
07
–0.0
1,
0.0
3–0.0
001
–0.0
2,
0.0
2–0.0
07
–0.0
3,
0.0
2
QA
LY
sbase
don
SF-3
6117
0.7
14
0.1
1106
0.7
16
0.1
3121
0.7
26
0.1
10.0
02
–0.0
1,
0.0
10.0
002
–0.0
1,
0.0
1–0.0
02
–0.0
2,
0.0
1
HB
A1c
impro
vem
ent
117
0.2
51.3
9106
0.5
51.3
2121
0.1
21.0
90.1
6–0.1
7,
0.4
70.4
50.1
5,
0.7
50.2
8–0.0
4,
0.6
4
*O
utc
om
eco
mpari
sons
incl
uded
covari
ate
sfo
rth
ebase
line
valu
eof
the
sam
eoutc
om
e.
CB
T,
cognit
ive
beh
avio
ur
ther
apy;
EQ
5D
,E
uro
Qol
5-d
imen
sional
hea
lth
state
index
;M
ET
,m
oti
vati
onal
enhance
men
tth
erapy;
QA
LY
,quali
ty-a
dju
sted
life
yea
r;SF-3
6,
36-i
tem
Short
Form
.
0
0.2
0.4
0.6
0.8
1
25000
5000
7500
10 000
12 500
15 000
17 500
20 000
22 500
25 000
27 500
30 000
32 500
35 000
37 500
40 000
42 500
45 000
Willingness to pay (£) for each outcome
Prob
abilit
y of c
ost-e
ffecti
vene
ss
HbA1c point improvement, health and social care perspectiveHbA1c point improvement, societal perspectiveQALY gain (SF-36), health and social care perspectiveQALY gain (SF-36), societal perspectiveQALY gain (EQ5D), health and social care perspectiveQALY gain (EQ5D), societal perspective
FIGURE 1 Cost-effectiveness acceptability curves: MET + CBT vs. usual
care. CBT, cognitive behaviour therapy; EQ5D, EuroQol 5-dimensional
health state index; MET, motivational enhancement therapy; QALY,
quality-adjusted life year; SF-36, 36-item Short Form.
DIABETICMedicine Cost-effectiveness of psychological therapies for Type 1 diabetes • A. Patel et al.
ª 2011 The Authors.476 Diabetic Medicine ª 2011 Diabetes UK
treatment approach compared with usual care alone over 1 year.
Both cost perspectives suggested similar conclusions, thus
avoiding potential dilemmas of trade-offs between different
stakeholders.
Conclusions vary between clinical and quality-of-life
outcomes. There is evidence to suggest that, if small HbA1c
improvements are sustained for a reasonable duration, they can
reduce future complications [17,18] and confer significant
healthcare cost savings within a relatively short time [19].
Quality-of-life measures can detect differences among diabetes
subgroups (e.g. UK Prospective Diabetes Study [20]), so we
cannot rule out the possibility that our measures were too
insensitive. However, reaching similar conclusions using two
quality-of-life measures (or with the Diabetes Quality-of-Life
measure [5]) supports the robustness of our conclusions. It also
suggests that providing psychological therapies to this group had
no negative effect on the measured quality-of-life domains. It is
likely that quality-of-life improvements occur in the longer term
and our time horizon was too short to detect these. For example,
a Type 2 diabetes study suggested that a 1% reduction in HbA1c
is equivalent to a 0.4 QALY gain (and £108 cost saving) over
20 years [21]. Another study examining the longer term cost-
effectiveness of professional-directed and patient-centred
diabetes guideline implementation strategies suggested that a
0.2% and 0.3% reduction in HbA1c at 1 year for each group,
respectively, translated into lifetime QALY gains of 0.29 and
0.59, respectively [22]. Different quantifications of QALYs
between measures (which is not unusual—see Grieve et al. [23]
and McCrone et al. [24]) raises important issues about choice of
health-related quality-of-life measures in economic evaluations
and the comparability of variably produced QALYs for
policymaking.
Both interventions increased health and social care costs with
no apparent cost offsets. They paradoxically aimed to increase
diabetes resource use for a difficult-to-treat group, in which
conventional intensive multidisciplinary diabetes care had
already failed to improve glycaemic control; for example, both
therapies included topics aboutmakingbestuseofdiabetes teams
and assertively seeking help to optimize self care. So, while
increased costs elsewhere may have reflected improved self care,
use of needs-appropriate services and investments in long-term
societal gains, such effects were either absent or too subtle to
affect overall care costs.
This study had some limitations. Firstly, while any data
collection biases are likely to be equal across the groups, we
cannot discount the possibility of double-counting if
intervention group participants mistakenly reported therapy
sessions as part of their usual diabetes care, in which case
healthcare costs for those two groups may be overstated and
any potential savings concealed. More generally, the reliability
of self-reported resource use over a 6-month period is unclear.
However, the multi-site nature of this study, broad perspective
(attributable to the breadth of diabetes impacts) and lack of
comprehensive computerized medical records necessitated this
approach. Using shorter recall periods would have required
more frequent assessments (which were infeasible because of
resource constraints as we used interviews to maximize quality)
or assessment gaps. The latter risked finding artificial cost
differences simply because of timing variations in routine
annual patient care reviews—temporary ‘flurries’ of healthcare
activity can occur around these reviews and thus lead to cost
variations depending on inclusion ⁄ exclusion of such periods.
Secondly, longer time horizons may be necessary to identify
relevant longer-term outcomes; for example, quality-of-life
gains may be more evident alongside reductions in future
complications. Finally, the sample size for the economic
evaluation was disappointing despite all reasonable attempts
at follow-up. The imputation method used in the sensitivity
analyses assumed that data were missing at random but, at this
level of missing data, we cannot rule out the possibility of non-
random causes for dropout (e.g. less improvement in outcomes,
negative views of therapy), which may have differently
impacted on both costs and outcomes. The sensitivity
analyses may thus present more optimistic scenarios than
ones which would be presented under more informed cost and
outcome imputations.
At a minimum of £48 636 per QALY gain, neither
intervention met the current cost-effectiveness threshold used
in England and Wales [14]. Longer-term cost-effectiveness,
taking account of possible reductions in complications and
future quality-of-life gains, needs to be evaluated. Other
interventions for Type 1 diabetes which have shown
significant HbA1c improvements in the short to medium term
have generally suggested good value for money in models of
long-term cost-effectiveness. For example: implementing
intensive rather than conventional diabetes management is
suggested to cost an additional $28 661 per life year gained
(lifetime model) [25]; treatment with continuous subcutaneous
insulin infusion compared with multiple daily injections is
estimated to cost £25 648 per QALY gained (lifetime model)
[26]; and a structured treatment and teaching programme
compared with standard practice is estimated to produce a 0.12
QALY gain and £2200 cost saving per patient (10-year model)
[27].
Motivational enhancement therapy plus cognitive behaviour
therapy has potential to provide additional HbA1c improvements
at a lower total additional cost and a greater probability of cost-
effectiveness, despite its greater cost. As our conclusions depend
on the outcome measure chosen, decisions regarding the
provision of nurse-delivered motivational enhancement therapy
and cognitive behaviour therapy for adults with poorly
controlled Type 1 diabetes depend on the relative importance
of these outcomes and of broader aims, such as increasing
diabetes nurse specialists’ skills (in the context of limited supply
of ⁄ access to psychological treatments) and targeting difficult-
to-treat patients.
Competing interests
Nothing to declare.
DIABETICMedicineOriginal article
ª 2011 The Authors.Diabetic Medicine ª 2011 Diabetes UK 477
Acknowledgements
This study was presented at the 45th EASD Annual Meeting in
Vienna, Austria, September–October 2009. The UK Department
of Health’s Health Technology Assessment Programme funded
the study (http://www.nchta.gov.uk project no 01 ⁄ 17 ⁄ 05). We
would like to thank and are indebted to the following: the
participants for their time and commitment in participating in
this study; the diabetes physicians (Professor Stephanie Amiel,
King’s College Hospital; Dr Jake Powrie, Guy’s Hospital; Mrs
Judy Adcock, Lewisham Hospital; Dr Richard Savine, Mayday
Hospital; Dr Robert Davies, Manchester Royal Infirmary;
Professor Phil Wiles, North Manchester General Hospital; and
Dr Ngai Kong, Stepping Hill Hospital, Stockport) who gave
permissionandenduring support; their clinic and laboratory staff
who assisted in recruiting and following up participants; general
practitioners who assisted with data collection; Dr Kirsty
Winkley; Ms Judy Jackson (research psychologist) who did the
recruitment, follow-up and data collection in the Manchester
sites; the Trial Steering Committee members (Professor Glyn
Lewis, Dr Dennis Barnes and Dr Bianca de Stavola) and the
Data Ethics and Monitoring Committee (Professor Graham
Dunn, Professor Robert Peveler and Dr Peter Watkins) for
their intellectual guidance; to Ms Suzanne Roche (cognitive
behaviour therapy therapist, Maudsley Hospital for contributing
to the cognitive behaviour therapy training and manual
development); the nurse therapists who delivered the
treatments; the Clinical Trials Unit, Institute of Psychiatry,
King’s College London for randomization and allocation
concealment; Dr Keith Wiener (North Manchester General
Hospital) for his personal communication about glycated
haemoglobin measures; and Barbara Barrett for invaluable
assistance with the economic analyses. Finally, we would
like to especially acknowledge the contribution of Mr Phil
Dickinson (dietician, Manchester Royal Infirmary), who died
recently; he managed the diabetes database and contributed
enthusiastically to this project and will be greatly missed by his
colleagues.
References
1 Gray A, Fenn P, McGuire A. The cost of insulin-dependent diabetes
mellitus (IDDM) in England and Wales. Diabet Med 1995; 12:
1068–1076.
2 Dixon S, Currie CJ, Peters JR. The cost of diabetes: time for a
different approach? Diabet Med 2000; 17: 820–822.
3 Ismail K, Winkley K, Rabe-Hesketh S. Systematic review and meta-
analysis of randomised controlled trials of psychological interven-
tions to improve glycaemic control in patients with type 2 diabetes.
Lancet 2004; 363: 1589–1597.
4 Hampson SE, Skinner TC, Hart J, Storey L, Gage H, Foxcroft D
et al. Behavioural interventions for adolescents with type 1
diabetes: how effective are they? Diabetes Care 2000; 23: 1416–
1422.
5 Ismail K, Thomas S, Maissi E, Chalder T, Schmidt U, Bartlett J et al.
Motivational enhancement therapy with and without cognitive
behaviour therapy to treat type 1 diabetes. Ann Intern Med 2008;
149: 708–719.
6 Beecham J, Knapp M. Costing psychiatric interventions. In:
Thornicroft G, Brewin C, Wing J, eds. Measuring Mental Health
Needs. Oxford: Oxford University Press, 1992: 200–224.
7 Ware J, Sherbourn C. The MOS, 36 item Short-Form Health Survey
(SF-36). I. Conceptual framework and item selection. Med Care
1992; 30: 473–483.
8 The EuroQol Group. EuroQol: a facility for the measurement
of health-related quality of life. Health Policy 1990; 16: 199–
208.
9 Brazier J, Roberts J, Deverill M. The estimation of a preference-
based measure of health from the SF-36. J Health Econ 2002; 21:
271–292.
10 Dolan P, Gudex C, Kind P, Williams A. A Social Tariff for Euro-
Qol: Results from a UK Population Survey. York: University of
York, 1995.
11 Manca A, Hawkins N, Sculpher MJ. Estimating mean QALYs in
trial-based cost-effectiveness analysis: the importance of controlling
for baseline utility. Health Econ 2005; 14: 487–496.
12 Assmann SF, Pocock SJ, Enos LE, Kasten LE. Subgroup analysis and
other (mis)uses of baseline data in clinical trials. Lancet 2000; 355:
1064–1069.
13 Briggs AH. A Bayesian approach to stochastic cost-effectiveness
analysis. Health Econ 1999; 8: 257–261.
14 NICE. Guide to the Methods of Technology Appraisal. London:
National Institute for Health and Clinical Excellence, 2008.
15 Klonoff DC, Schwartz DM. An economic analysis of interventions
for diabetes. Diabetes Care 2000; 23: 390–404.
16 Gray A, Raikou M, McGuire A, Fenn P, Stevens R, Cull C et al.
on behalf of the UK Prospective Diabetes Study Group. Cost-
effectiveness of an intensive blood glucose control policy in
patients with type 2 diabetes: economic analysis alongside rando-
mised controlled trial (UKPDS 41). Br Med J 2000; 320: 1373–
1378.
17 The Diabetes Control and Complications Trial ⁄ Epidemiology of
Diabetes Research Group. Retinopathy and nephropathy in patients
with Type 1 diabetes 4 years after a trial of intensive therapy. N
Engl J Med 2000; 342: 381–389.
18 The Diabetes Control and Complications Trial ⁄ Epidemiology of
Diabetes Interventions and Complications Study Research
Group. Intensive diabetes treatment and cardiovascular disease in
patients with Type 1 diabetes. N Engl J Med 2005; 353: 2643–
2653.
19 Wagner EH, Sandhu N, Newton KM, McCulloch DK, SD
Ramsey, Grothaus LC. Effect of improved glycemic control on
health care costs and utilization. J Am Med Assoc 2001; 285:
182–189.
20 UK Prospective Diabetes Study Group. Quality of life in type 2
diabetic patients is affected by complications but not by intensive
policies to improve blood glucose or blood pressure control
(UKPDS 37). Diabetes Care 1999; 22: 1125–1136.
21 McEwan P, Peters JR, Bergenheim K, Currie CJ. Evaluation of the
costs and outcomes from changes in risk factors in type 2 diabetes
using the Cardiff stochastic simulation cost-utility model (Diab-
Forecaster). Curr Med Res Opin 2006; 22: 121–129.
22 Dijkstra RF, Niessen LW, Braspenning JC, Adang E, Grol RT.
Patient-centred and professional-directed implementation strategies
for diabetes guidelines: a cluster randomized trial-based cost-effec-
tiveness analysis. Diabet Med 2006; 23: 164–170.
23 Grieve R, Grischchenko M, Cairns J. SF-6D versus EQ-5D: reasons
for differences in utility scores and impact on reported cost-utility.
Eur J Health Econ 2009; 10: 15–23.
24 McCrone P, Patel A, Knapp M, Schene A, Koeter M, Amaddeo F
et al. A comparison of SF-6D and EQ-5D utility scores in a study of
patients with schizophrenia. J Ment Health Policy Econ 2009; 12:
27–31.
DIABETICMedicine Cost-effectiveness of psychological therapies for Type 1 diabetes • A. Patel et al.
ª 2011 The Authors.478 Diabetic Medicine ª 2011 Diabetes UK
25 Diabetes Control and Complications Trial Research Group. Life-
time benefits and costs of intensive therapy as practiced in the
diabetes and complications trial. J Am Med Assoc 1996; 276: 1409–
1415.
26 Roze S, Valentine WJ, Zakrzewska KE, Palmer AJ. Health-eco-
nomic comparison of continuous subcutaneous insulin infusion
with multiple daily injection for the treatment of Type 1 diabetes in
the UK. Diabet Med 2005; 22: 1239–1245.
27 Shearer A, Bagust A, Sanderson D, Heller S, Roberts S. Cost-
effectiveness of flexible intensive insulin management to enable
dietary freedom in people with Type 1 diabetes in the UK. Diabet
Med 2004; 21: 460–467.
Supporting Information
Additional Supporting Information may be found in the online
version of this article:
Appendix S1. Unit cost summary (£, 2005 ⁄ 2006 prices).
Please note: Wiley-Blackwell are not responsible for the content
or functionality of any supporting materials supplied by the
authors. Any queries (other than for missing material) should be
directed to the corresponding author for the article.
DIABETICMedicineOriginal article
ª 2011 The Authors.Diabetic Medicine ª 2011 Diabetes UK 479