You are on page 1of 16

Pharmacoeconomics 2010; 28 (9): 765-780

ORIGINAL RESEARCH ARTICLE 1170-7690/10/0009-0765/$49.95/0

ª 2010 Adis Data Information BV. All rights reserved.

An Economic Evaluation of Colesevelam


when Added to Metformin-, Insulin-
or Sulfonylurea-Based Therapies in
Patients with Uncontrolled Type 2
Diabetes Mellitus
W. Robert Simons1,2 and Michael A. Hagan3
1 Global Health Economics & Outcomes Research, Inc., Summit, New Jersey, USA
2 Global Health Economics & Outcomes Research Pty Ltd, Sydney, New South Wales, Australia
3 Daiichi Sankyo, Inc., Parsippany, New Jersey, USA

Abstract Background: Several early studies demonstrated that bile acid sequestrants
were useful for lowering lipid levels in patients with hypercholesterolaemia
and may also be useful for lowering glucose levels in patients with type 2
diabetes mellitus (T2DM) uncontrolled on existing treatment (metformin-,
insulin- or sulfonylurea-based therapies).
Objective: This study modelled efficacy and safety data from the three clini-
cal trials to evaluate the cost effectiveness to US Managed Care Organiza-
tions of add-on treatment with colesevelam for reducing diabetes-related
complications.
Methods: Three randomized controlled trials in patients with T2DM and one
in hyperlipidaemia established that colesevelam lowered both glycaemic and
lipid parameters in adult patients participating in the studies. The validated
‘diabetic risk equation’ (DRE) and the ‘LIPID cardiovascular risk equation’
(LCRE) were used to translate the observed clinical benefits (surrogate mar-
kers related to T2DM [glycosylated haemoglobin {HbA1c} and fasting plas-
ma glucose] and cardiovascular disease [low-density lipoprotein cholesterol
{LDL-C}]). Performing an appropriate economic evaluation required the use
of both the DRE and the LCRE. These equations parameterize the clinical
efficacy measures as continuous, facilitating their application to clinical trial
results as well as the replication of other well established epidemiological data.
Tobit regressions were applied to a large commercially available managed
care administrative claims database (2000–6), Integrated Health Care Services
(IHCS), to evaluate the incremental costs associated with each type of diabetic
complication. Costs were inflated to 2010 values using the Healthcare Con-
sumer Price Index, while second- and third-year cost savings were discounted
766 Simons & Hagan

at 5% to the current year. Bootstrap sampling with 5000 samples of 100


patients per cohort was conducted, varying the number of events avoided as
well as their associated cost.
Results: With established metformin-, insulin- or sulfonylurea-based thera-
pies, the addition of colesevelam significantly reduced HbA1c by approxi-
mately 0.5% (p < 0.001) in all three studies. In addition, colesevelam reduced
placebo-adjusted LDL-C by 12.8–16.7% (p < 0.001). Using the DRE and
LCRE equations, the total savings from reductions in diabetes-related and
cardiovascular events were $US3543, $US4074 and $US3855 for colesevelam
added to metformin-, insulin- and sulfonylurea-based regimens in patients
with normal lipid levels. After subtracting the cost of colesevelam, first-year
savings were $US1326, $US1852 and $US1629 in the metformin, insulin and
sulfonylurea studies, respectively, for patients with raised lipid levels.
Conclusions: In adult patients with T2DM, the addition of colesevelam to
metformin-, insulin- or sulfonylurea-based therapies significantly improves
glycaemic control while also reducing LDL-C, and these improvements could
translate into substantial cost reductions due to reductions in the rates of
diabetes-related and cardiovascular complications.

Background macrovascular mortality due to progressive


atherosclerosis[8-10] and all-cause mortality.[11-14]
Studies in patients with diabetes mellitus (type 1 While the UKPDS 33[1] study compared dis-
and type 2 [T2DM]) have demonstrated a strong crete median thresholds of HbA1c 7.0% versus
associative relationship between hyperglycaemia 7.9% across a median of 10 years, other studies
and the incidence and progression of microvas- assessed high and low measures of HbA1c, typi-
cular (primarily diabetic retinopathy and nephro- cally on a single occasion. A companion study
pathy) and macrovascular (cardiovascular disease) (UKPDS 35[15]) sought to improve the evaluation
complications. The UKPDS (UK Prospective Dia- by granulating the HbA1c measures into categorical
betes Study)-33,[1] a controlled clinical trial with or polychotomous levels as well as integrating a
T2DM patients, compared the clinical benefits of temporal component (i.e. patient-years at HbA1c
tight glycaemic control versus conventional treat- of <6% [median 5.6%], 6 to <7% [median 6.5%],
ment. After a median of 10 years, the conventional 7 to <8% [median 7.5%], 8 to <9% [median 9.4%]
treatment group had a glycosylated haemoglobin and ‡10% [median 10.6%]). That is, UKPDS
(HbA1c) of 7.9%, while the intensive treatment 35[15] generated summary statistics for quintiles
group had an HbA1c of 7.0%, resulting in a 25% of HbA1c, across the 10 years of follow-up. While
risk reduction in microvascular endpoints. Other this statistical approach has been previously im-
studies have also evaluated thresholds of HbA1c plemented in the evaluation of all-cause mortality
and the risk of diabetes-related complications, and other cardiovascular endpoints[16] and uti-
but used a limited explanatory measure of HbA1c, lizes more of the distribution of the data (and to
either in precision or temporality or, more often, an extent its temporality), it also has its limita-
both. As noted by the UKPDS[1] group, a num- tions. Laboratory values, such as HbA1c, are cor-
ber of studies have assessed the relationship be- related over time. More importantly, the analysis
tween elevated HbA1c and an increased risk of loses considerable flexibility in its applicability to
microvascular complications,[2,3] sensory neuro- different clinical trials, especially if the objective
pathy,[3,4] myocardial infarction,[2,5,6] stroke,[7] is to translate clinical trial efficacy data measured

ª 2010 Adis Data Information BV. All rights reserved. Pharmacoeconomics 2010; 28 (9)
Economic Evaluation of Colesevelam for Diabetes Mellitus 767

in incremental units, not quintiles, into incremen- ing for ‘real-world’ serially correlated behaviour
tal changes in the number of events avoided as- with repeated HbA1c readings at variable frequen-
sociated with changes in HbA1c.[17] cies and durations between their measurement as
To address these shortcomings, we developed well as changes in treatment or lifestyle. See the
and validated three diabetic risk equations (DREs) technical appendix for the GEE specifications as
using generalized estimating equations (GEEs)[17] well as the Tobit regressions used for the cost of
that make use of all available HbA1c measure- diabetes-related complication evaluation described
ments and their variability across time. Those vali- below (Supplemental Digital Content 1, http://
dated equations replicated the coefficients within links.adisonline.com/PCZ/A91). The DRE mod-
1/100th in newly diagnosed patients with T2DM els HbA1c as an incremental continuous variable,
in the UK,[18] Germany[19] and the US[20] for the facilitating its adaptation or parameterization to
relationship between treatment and effects on incremental changes in HbA1c from clinical trials.
HbA1c and consequently the effects of HbA1c on Parameterization of HbA1c as a continuous vari-
incremental changes in diabetes-related compli- able also facilitates cross-validation with other
cations,[20] thus improving precision, specificity established epidemiological studies, such as the
and consistency. We used the US-validated DRE UKPDS, where HbA1c data were categorical; in-
to evaluate the number of diabetic complications deed, the equation can reproduce those data. The
avoided due to better glycaemic control in patients LCRE has the same advantage. LDL-C and HDL-C
treated with colesevelam who had previously not are parameterized as continuous variables instead
achieved control despite ongoing treatment with of dichotomous or categorical, facilitating their
metformin-, insulin- or sulfonylurea-based ther- application to incremental changes in LDL-C and
apies (monotherapy or combination therapy with HDL-C measured in clinical trials. That parame-
other oral antidiabetic [OAD] agents). Since most terization also facilitated cross-validation with
patients with T2DM also have hypercholesterol- the Prospective Trial Collaborative (PTA);[26]
aemia that is not at the recommended low-density that is, the LCRE reproduces the findings of the
lipoprotein cholesterol (LDL-C) goal and cole- well established epidemiological data of the PTA.
sevelam also improves LDL-C, non-high-density Again, the key is that the clinical efficacy vari-
lipoprotein cholesterol (non-HDL-C) and apoli- ables are continuous, facilitating their application
poprotein (apo)-B levels,[21-23] the DRE is coupled to clinical endpoints assessed in clinical trials.
with the ‘LIPID cardiovascular risk equation’ We employed regressions to evaluate the incre-
(LCRE), which translates improved lipid levels mental costs associated with each type of diabetic
with reductions in cardiovascular events.[24,25] complication annually for 3 years. Tobit regres-
Thus, this study evaluates the economic benefits sions were used due to the number of instances of
(cost savings) associated with the addition of co- zero or non-resource utilization; not everyone is
lesevelam to metformin-, insulin- and sulfonylurea- hospitalized every year or at any time during the
based diabetic treatment regimens in patients who duration of the study. The Tobit adjusts for this
had yet to achieve glycaemic control. The anal- distributional issue. The effects on HbA1c and lipid
ysis is conducted from the perspective of US levels measured in the clinical trials are plugged
managed care organizations. into the equations, yielding changes in the abso-
lute risk of diabetic complications and percent
changes in cardiovascular risk. These are multi-
Methods
plied by the cost of the events, while subtracting
Model Overview the cost of treatment.
Colesevelam hydrochloride (colesevelam;
The DRE and LCRE are two validated, dis- Welchol) is a bile acid sequestrant indicated (in
tinct models used to translate clinical effects into Europe and the US) as an adjunct to diet and ex-
effects on diabetes-related complications and car- ercise for the reduction of elevated LDL-C in adults
diovascular events. The DRE used GEEs, allow- with primary hyperlipidaemia, as monotherapy

ª 2010 Adis Data Information BV. All rights reserved. Pharmacoeconomics 2010; 28 (9)
768 Simons & Hagan

or in combination with an HMG-CoA reductase ablished metformin-, insulin- or sulfonylurea-based


inhibitor (statin), and for improvement of gly- therapies.[21-23] Inadequate glycaemic control suf-
caemic control in adults with T2DM. ficient for study inclusion was defined as an HbA1c
This study evaluated the cost effectiveness of between 7.5% and 9.5% in all three trials. The
colesevelam as an add-on therapy in adult pa- mean baseline HbA1c across all three studies was
tients with T2DM who have not achieved optimal approximately 8.0%. The duration of treatment
glycaemic control despite treatment with metformin-, was 26 weeks in the metformin[21] and sulfonyl-
insulin- or sulfonylurea-based therapies (mono- urea[23] trials and 16 weeks in the insulin trial.[22]
therapy or combination therapy with other OAD Because colesevelam also lowers LDL-C, a clin-
agents). While effectiveness is assessed in terms of ical trial evaluating the efficacy of colesevelam on
the clinical benefits, incremental cost differences lipid levels in adult patients with primary hyper-
are assessed by drug acquisition costs and savings cholesterolaemia was also integrated into the
due to reductions in diabetes-related complica- economic assessment.[27] This trial was a 24-week,
tions and cardiovascular disease. randomized, double-blind, multicenter, placebo-
controlled trial wherein four dosages of colese-
Study Patient Populations velam (2.3, 3.0, 3.75 or 4.5 g/day) were evaluated
to assess efficacy, safety and tolerability in 494 sub-
The study populations comprised adult patients
jects with mild-to-moderate primary hypercho-
with T2DM who had not achieved glycaemic
lesterolaemia (LDL-C level ‡130 and £220 mg/dL).
control despite current treatment. The presence
Data from the group receiving colesevelam
or absence of hyperlipidaemia was an additional
3.75 g/day were used in this economic evaluation,
confounder. While microvascular complications
as this is the dosage approved and recommended
are typically associated with hyperglycaemia,
by the US FDA.
macrovascular complications also affect patients
with T2DM, largely due to an accelerated rate of
atherosclerosis progression. Thus, the economic Models of Diabetes Mellitus-Related
model evaluates benefit of a reduction in both Complications and Glycosylated
microvascular and macrovascular complications, Haemoglobin
since colesevelam lowers both HbA1c and lipid
The economic evaluation of colesevelam trans-
levels in adult patients with T2DM.[21-23]
lates treatment changes in HbA1c (measured at
Data Sources baseline and end of study) into reductions in
diabetes-related complications that directly im-
The DRE and LCRE models were popu- pact on healthcare costs. The DRE was initially
lated with data from three randomized, placebo- estimated in two European longitudinal data-
controlled prospective clinical trials with similar bases with analogous characteristics (i.e. repeated
design wherein colesevelam or placebo were com- measures of HbA1c and diagnoses for complica-
pared as add-on therapy for adult patients receiving tions). It was then replicated, again, in a large US
established metformin-, insulin- or sulfonylurea- retrospective database (Integrated Health Care
based therapies, respectively.[21-23] In addition, a Services [IHCS]) consisting of 497 716 patients
randomized controlled trial (RCT) evaluating with T2DM, repeated HbA1c measures across time
colesevelam in adult patients with primary hyper- and documented occurrences of diabetes-related
cholesterolaemia was also used.[27] complications. The effects of pharmacological
The three randomized, placebo-controlled cli- treatment in patients with T2DM, estimated using
nical trials serve as the core set of pivotal trials the DRE developed from data in three different
and compared the efficacy and safety of addi- longitudinal databases, are consistently about
tional therapy with colesevelam (3.75 g/day) with a 1% decrease in HbA1c. This replication of the
placebo in adult patients with T2DM who had quantification strengthens the statistical associ-
not reached optimal glycaemic control with est- ation and the DRE. Moreover, the quantification

ª 2010 Adis Data Information BV. All rights reserved. Pharmacoeconomics 2010; 28 (9)
Economic Evaluation of Colesevelam for Diabetes Mellitus 769

is consistent with data from RCTs and, as such, economic model, these costs are accrued for the
certainly passes the litmus test for plausibility. duration of the economic evaluation, and years 2
More important for this economic evaluation and 3 are discounted at 5% per year to year 1. The
is the translation of the effects of blood glucose IHCS database, which was used to quantify the
control on the risk of diabetes complications. The relationship between diabetes-related complica-
DRE suggests that a 1% reduction in HbA1c trans- tions and incremental changes in HbA1c, was also
lates to about a 0.4% annual reduction in diabetic used to quantify the cost implications. Diabetic
complications. This, too, is consistently measured complications, identified by International Clas-
in the three different datasets, again establish- sification of Diseases, Ninth Revision (ICD-9)
ing consistency, specificity and plausibility impor- codes, were included in our analysis if both of the
tant to the interpretation and credibility of the following criteria were satisfied: (i) the encounter
results. in the healthcare system had a diagnosis asso-
ciated with one of the diabetes complications;
Low- and High-Density Lipoprotein and (ii) the encounter in the healthcare system
Cholesterol and Cardiovascular Risk Models also had a diagnosis associated with a particular
type of manifestation of the complication.
The economic model uses an updated, recently
Outpatient and inpatient costs were accrued
published meta-analysis by Brown et al.[28] of
for 1-, 2- and 3-year periods. Residual costs in the
23 randomized, placebo-controlled clinical trials.
second and third years that were associated with
In the meta-analysis, a 1% increase in HDL-C
an event that occurred in the first year were ac-
was associated with a 1.288% decrease in cardio-
crued to the first year event at a discounted rate
vascular events (p = 0.01), while a 1% decrease in
of 5% per year. Specifically, the cost models in-
LDL-C was associated with a 0.971% decrease in
cluded costs for outpatient and inpatient care
cardiovascular events (p = 0.003). As both HDL-C
accrued at 1, 2 and 3 years; these were the de-
and LDL-C are evaluated in the same regres-
pendent variables. Independent variables con-
sion model, these effects are independent. Similar
sisted of demographics and specific types of
to the results reported by Brown et al.,[28] the up-
microvascular and macrovascular complications
dated regression used in this analysis explained
(e.g. ketoacidosis, neurological disorders, diabe-
83% of the variation in percent changes in cardio-
tic coma, ophthalmic disorders, etc.), as well as
vascular events. The parameter estimates associated
other potential confounders. Incremental costs
with a 1% change in HDL-C and LDL-C are -1.51
are quantified by type of event by year.
(p = 0.02) and 1.02 (p = 0.02), respectively, or alter-
Because of the number of zero costs (i.e. not
natively, a 1% increase in HDL-C equates to a
everyone consumes resources such as hospitali-
1.5% reduction in cardiovascular events, while a
zation), Tobit regressions were used. A Tobit re-
1% reduction in LDL-C equates to a 1% reduc-
gression is a combination of a probit probability
tion in those events.
model and a linear regression (i.e. regression of
non-censored values is a truncated regression)
Costs
and has been used to resolve this disproportionality
Diabetes-Related Complications in resource utilization.[29]
and Cardiovascular Events
Costs (inflated to year 2010 values) were de- Colesevelam
rived from the 2000–6 IHCS database, a large Colesevelam is supplied as 625 mg tablets in
commercially available managed care adminis- bottles of 180 tablets for a cost of $US185.40
trative claims database that includes laboratory (National Drug Code [NDC] 65597-701-18).[30]
values such as HbA1c in addition to the variables The recommended dose is 3.75 g or six tablets of
common to administrative claims databases (e.g. 625 mg per day. The bottle of 180 tablets is suf-
demographics, inpatient and outpatient services ficient for a 30-day supply at a cost of $US6.18 per
with diagnoses, outpatient pharmacy, etc.). In the day or $US2255.70 annually. A new formulation of

ª 2010 Adis Data Information BV. All rights reserved. Pharmacoeconomics 2010; 28 (9)
770 Simons & Hagan

colesevelam was recently approved as a powder Results


for oral suspension, where one packet contains
3.75 g of colesevelam. Baseline Characteristics

Type 2 Diabetes Mellitus (T2DM)


Sensitivity Analysis Baseline demographic characteristics for the
Probablistic sensitivity analyses were conducted three clinical trials in adult patients with T2DM
on variants in the model that were variable, with colesevelam as an add-on to metformin-,
namely the number of events avoided (based on insulin- or sulfonylurea-based therapies, respect-
the three diabetes clinical trials) and the costs as- ively, are presented in table I. In all three trials,[21-23]
sociated with those avoided events (i.e. the 11 types demographic characteristics were comparable
of diabetes-related complications). Using the ex- across randomized treatment groups. Approxi-
pected costs associated with each type of event mately half of the population was male, and the
and their variances, 2000 costs were generated average age was 56 years.
using an algorithm written in SAS version 9.2,
producing data with identical statistical distribu- Hypercholesterolaemia
tions for each type of event, 1000 for each treat- The trial population characteristics presented
ment group; 2000 patients with treatment effects in table I for the evaluation of efficacy and safety
and variances as reported in the clinical trial re- of variable doses of colesevelam compared with
ports were also generated, again, using an algo- placebo in adult patients with hypercholesterol-
rithm written in SAS version 9.2; replication of aemia at baseline were generally comparable
the clinical trial efficacy results served as valida- across treatment arms. This 24-week study[27] had
tion. Non-parametric Bootstrap sampling of 5000 a population consisting of approximately 48%
sample means from 100 observations for incre- men, with an average age of 56 years. Approx-
mental effectiveness (i.e. change in the number of imately 90% of the patients were Caucasian.
events) and incremental costs randomly selected Enrolment included adult patients at varying de-
with replacement derived 5000 incremental cost- grees of risk for cardiovascular disease: 68% had
effectiveness ratios for each clinical trial. The 0–1 risk factors, 26% had ‡2 risk factors and 6%
same costs were used in populating diabetic or had pre-existing coronary heart disease (data not
cardiovascular events in each of the trials. shown).

Table I. Patient demographics in three studies in adult patients with type 2 diabetes mellitus (studies 1–3) and one study in patients with
primary hypercholesterolaemia (study 4)
Characteristic Study 1: metformin-based Study 2: insulin-based Study 3: sulfonylurea-based Study 4: primary
therapy[21] therapy[22] therapy[23] hypercholesterolaemia[27]
colesevelam placebo colesevelam placebo colesevelam placebo colesevelama placebo
(n = 159) (n = 157) (n = 147) (n = 140) (n = 230) (n = 231) (n = 95) (n = 88)
Sex [n (%)]
male 81 (50.9) 83 (52.9) 77 (52.4) 71 (50.7) 128 (55.7) 122 (52.8) 45 (47) 42 (48)
female 78 (49.1) 74 (47.1) 70 (47.6) 69 (49.3) 102 (44.3) 109 (47.2) 50 (53) 46 (52)
Race [n (%)]
Caucasian 89 (56.0) 94 (59.9) 94 (63.9) 89 (63.6) 135 (58.7) 128 (55.4) 85 (90) 82 (93)
Black 23 (14.5) 26 (16.6) 24 (16.3) 26 (18.6) 23 (10.0) 34 (14.7) 8 (8) 2 (2)
other 47 (29.6) 37 (23.5) 29 (19.7) 25 (17.9) 72 (31.3) 69 (29.9) 2 (2) 4 (5)
Mean age 55.7 (9.6) 56.9 (9.5) 56.9 (9.8) 56.3 (9.3) 56.6 (10.3) 57.0 (10.3) 56 (13) 55 (12)
[y (SD)]
a Includes only the cohort of adult patients receiving colesevelam 3.75 g/day.

ª 2010 Adis Data Information BV. All rights reserved. Pharmacoeconomics 2010; 28 (9)
Economic Evaluation of Colesevelam for Diabetes Mellitus 771
Table II. Glycaemic and lipid efficacy results from three studies in adult patients with type 2 diabetes mellitus (studies 1–3) and one study in adult patients with primary hyper-

Efficacy in Adult Patients with T2DM


change from

and Hypercholesterolaemia

2 [-1, 4]* a

0 [-2, 3]a
baseline

HbA1c = glycosylated haemoglobin; HDL-C = high-density lipoprotein cholesterol; LDL-C = low-density lipoprotein cholesterol; ND = not determined; * p < 0.001 vs baseline.
The mean HbA1c and lipid data at baseline
(22)*

(19)
-24
ND

ND

and at study endpoint for the three trials in pa-


0
hypercholesterolaemia[27]

tients with T2DM,[21-23] as well as for the trial in


patients with primary hypercholesterolaemia, are
47 [41, 59]a

50 [43, 59]a
presented in table II.[27] As indicated in table II,
Study 4: primary

baseline

despite ongoing treatment with metformin, insu-


(21)

(19)
159

155
ND

ND

lin or sulfonylurea, the average HbA1c levels at


baseline among patients receiving colesevelam were
ND

ND

95

88

95

88

8.1–8.3%. Treatment with colesevelam resulted


n

in placebo-adjusted reductions in HbA1c levels of


0.54%, 0.50% and 0.54% (all p < 0.001). LDL-C
change from
Study 3: sulfonylurea-based

and HDL-C levels were comparable between ran-


baseline

(0.86)*

(1.58)*
(1.02)

(1.53)
-17.7

domized groups and were within the normal range


(0.4)

(0.4)
-0.4

-1.1

-0.1

-0.0
0.2

at baseline and study completion. Treatment with


colesevelam resulted in placebo-adjusted reduc-
baseline

tions in LDL-C of 15.9%, 12.8% and 16.7%, respec-


(0.68)

(0.73)

(27.8)

(29.5)

(11.3)

(11.3)
104.3

105.8

44.5

44.8
therapy[23]

8.2

8.3

tively (all p < 0.001). For patients with primary


hypercholesterolaemia receiving colesevelam 3.75 g/
218

218

186

193

203

201

day, the mean LDL-C level at baseline was 159 –


n

21 mg/dL. At the end of the study, LDL-C levels


were 15% lower (p < 0.001).
change from
Study 2: insulin-based therapy[22]

baseline

(0.68)*

(24.9)*

Reductions in Diabetes-Related
(0.87)

(25.0)

(5.45)

(5.42)
-14.6
-0.4

-1.2

-0.6

-0.3
0.1

Complications and Cardiovascular Events

The clinical trial efficacy results were trans-


baseline

lated into changes in the occurrence of diabetes-


(0.61)

(0.62)

(27.7)

(29.1)

(11.0)
101.5

101.5

(9.5)
42.2

44.2

related complications using the DRE (changes in


8.3

8.2

HbA1c) and cardiovascular events using the LCRE


144

136

129

121

135

127

(changes in LDL-C). The measured relationships


n

of the impact of pharmacological interventions


on HbA1c, as well as the relationships for incre-
Study 1: metformin-based therapy[21]

change from

mental changes in HbA1c (i.e. a 1% change in


baseline

HbA1c) and the risk of diabetes-related compli-


(11.92)

(15.26)
(0.97)*

(22.7)*
(0.95)

(23.5)
-12.6
–0.4

cations, are presented in table III. For simplicity,


0.2

4.0

1.0

0.2

Data in square brackets are 95% CI.

table IV provides the number of events avoided


and percentages for the metformin trial only.
baseline

(0.65)

(0.61)

(33.8)

(29.0)

(11.6)

(10.1)
105.6

One thousand bootstrap samples, with replace-


99.0

44.9

44.6
8.1

8.1

ment of mean results from 100 observations from


[21-23]
cholesterolaemia (study 4)

[21-23]

each treatment arm and trial were generated, as


HDL-C [mg/dL (SD)]
LDL-C [mg/dL (SD)]

were their corresponding 95% CI.


Colesevelam 148

152

Colesevelam 125

126

Colesevelam 134

136
n

Table V presents the number of events avoid-


HbA1c [% (SD)]

ed yearly for each of the four trials. In the first


year, the addition of colesevelam to metformin
Placebo

Placebo

Placebo

therapy reduced the number of diabetes-related


and cardiovascular complications by 10 531 in
a

ª 2010 Adis Data Information BV. All rights reserved. Pharmacoeconomics 2010; 28 (9)
772 Simons & Hagan

Table III. Impact of pharmacological interventions on glycaemic control (glycosylated haemoglobin [HbA1c]) as well as the relationships for
incremental changes in HbA1c (i.e. a 1% change in HbA1c) and the risk of diabetes mellitus-related complications, predicted using the Diabetic
Risk Equations
Study country/primary endpoint UK study[18] German study[19] US database[20]
Population size 2137 3190 497 716
Effects of treatment on glycaemic control (%) -0.99 -0.92 -0.89
Effects of glycaemic control on the risk of diabetic complications (%) -0.388 -0.414 -0.436

100 000 patients with T2DM who lacked gly- cerebrovascular disease, stroke and myocardial
caemic control despite treatment. Of the specified infarction were reduced by 141 in 100 000 patients
complications, the complications with the great- in year 1. By year 2, a total of 18 702 diabetes-
est reductions were ophthalmic and neurological related and cardiovascular events were avoided
manifestations, which were each reduced by 14%, and, at the end of year 3, there was a reduction of
or 1474 events avoided per 100 000 patients. Peri- 27 691 events per 100 000 patients.
pheral circulatory disorder was second, with a re- The addition of colesevelam to insulin-based
duction of 7%, or 776 events avoided per 100 000 therapy resulted in a total of 12 112 events avoid-
patients. In addition to the reductions in diabetes- ed at the end of the first year. Reductions in
related complications associated with improved ophthalmic and neurological manifestations were
glycaemic control, cardiovascular events such as the greatest (of the specified complications), with

Table IV. Diabetes mellitus-related complications avoided by add-on treatment with colesevelam in adult patients (pts) with type 2 diabetes
inadequately controlled by metformin-based therapy (study 1[21])
Events avoided Year 1 Year 2 Year 3
% of all events % of all cumulative % of all cumulative
complications avoideda complications events avoideda complications events avoideda
Diabetic complications avoided due to decreases in HbA1c
ketoacidosis 3.73 388 3.72 690 3.72 1 022
hyperosmolarity 1.56 162 1.59 294 1.61 444
coma 0.50 52 0.55 102 0.59 161
renal manifestations 5.10 530 5.28 980 5.43 1 493
ophthalmic 14.19 1 474 14.62 2 710 14.96 4 116
manifestations
neurological 14.19 1 474 14.62 2 710 14.96 4 116
manifestations
peripheral circulatory 7.47 776 7.88 1 462 8.22 2 260
disorder
other specified 6.16 640 6.35 1 178 6.50 1 789
complication
unspecified 19.35 2 011 18.29 3 390 17.46 4 802
complication
congestive heart 10.02 1 002 15.73 1 573 20.85 2 084
failure
ischaemic heart 18.81 1 881 34.51 3 451 52.19 5 219
disease
Cardiovascular events avoided due to improvements in lipid levels (LDL-C and HDL-C)
cardiovascular events 1.34 141 0.86 161 0.67 185
Total events 10 531 18 702 27 691
a Per 100 000 pts.
HbA1c = glycosylated haemoglobin; HDL-C = high-density lipoprotein cholesterol; LDL-C = low-density lipoprotein cholesterol.

ª 2010 Adis Data Information BV. All rights reserved. Pharmacoeconomics 2010; 28 (9)
Economic Evaluation of Colesevelam for Diabetes Mellitus 773

Table V. 3-year cumulative diabetes mellitus-related and cardiovascular complications avoided (per 100 000 patients) by add-on treatment
with colesevelam in adult patients with type 2 diabetes (studies 1–3)a
Events avoided Study 1: metformin-based Study 2: insulin-based Study 3: sulfonylurea-based
therapy[21] therapy[22] therapy[23]
y1 y2 y3 y1 y2 y3 y1 y2 y3
Cumulative diabetic complications avoided due to decreases in HbA1cb
ketoacidosis 388 690 1 022 447 712 1 060 447 580 1 027
hyperosmolarity 162 294 444 186 304 460 186 345 446
coma 52 102 161 60 105 167 60 89 162
renal manifestations 530 980 1 493 610 1 011 1 548 610 1 181 1 500
ophthalmic manifestations 1 474 2 710 4 116 1 697 2 797 4 267 1 697 3 260 4 135
neurological manifestations 1 474 2 710 4 116 1 697 2 797 4 267 1 697 3 291 4 135
peripheral circulatory disorder 776 1 462 2 260 893 1 508 2 343 893 1 552 2 270
other specified complication 640 1 178 1 789 737 1 215 1 855 737 1 175 1 797
unspecified complication 2 011 3 390 4 802 2 315 3 499 4 978 2 315 3 035 4 824
ischaemic heart disease 1 881 3 451 5 219 2 166 3 561 5 411 2 045 3 396 5 243
congestive heart failure 1 002 1 573 2 085 1 154 1 624 2 161 1 089 1 548 2 095
Cardiovascular events avoided due to improvements in lipid levels (LDL-C and HDL-C)c,d
cardiovascular events 141 161 185 150 172 197 160 183 211
Total events 10 531 18 702 27 691 12 112 19 304 28 714 11 455 18 426 27 847
a Number of cardiovascular events avoided in adult patients with primary hypercholesterolaemia (study 4) were 337, 785, 1203 in years 1, 2
and 3, respectively, because patients had higher baseline LDL-C and therefore greater reduction in LDL-C and consequentially greater
benefit.[27]
b Reductions in diabetic events would correspond to those reported in studies 1–3.
c LCRE: 1.02 · %~LDL-C - (-1.51 · %~HDL-C) = %~CVD.
d Annual event rates based on data from the LIPID study[31] (table II) multiplied by the effect in footnote c.
CVD = cardiovascular disease; HbA1c = glycosylated haemoglobin; HDL-C = high-density lipoprotein cholesterol; LDL-C = low-density
lipoprotein cholesterol; LCRE = LIPID cardiovascular risk equation.

1697 events avoided in each category. In addition In adult patients with hypercholesterolaemia,
to the diabetes-related complications, 150 cardio- treatment with colesevelam resulted in 337 car-
vascular events were avoided. By year 2, a total of diovascular events avoided by the end of the first
19 304 diabetes-related and cardiovascular events year, 1785 events by year 2 and 1203 events per
were avoided, and by the end of the third year 100 000 patients by year 3.
there was a reduction of 28 714 events per 100 000
patients.
Cost Associated with Diabetes-Related
The addition of colesevelam to a sulfonylurea- Complications and Cardiovascular Disease
based therapy resulted in 11 455 events avoided at
the end of year 1. Following the previous trends, Incremental costs of diabetes-related and high
ophthalmic and neurological manifestations had LDL-C-related complications were obtained from
the greatest number of events avoided (of the the ICHS database. Using Tobit regressions, the
specified complications), with reductions of 1697 incremental costs (with 95% CI) were evaluated
in each category. In addition, 160 cardiovascular by type of diabetes-related complication as well
events were also avoided. By the end of the sec- as by setting (i.e. inpatient and outpatient) for the
ond year, a total of 18 426 diabetes-related and first year following the onset of the event and two
cardiovascular events were avoided and, at the subsequent years to measure residual costs (e.g.
end of year 3, there was a reduction of 27 847 rehabilitation in stroke patients, etc.). As indicated
events per 100 000 patients. in table IV, of specific disorders, manifestations

ª 2010 Adis Data Information BV. All rights reserved. Pharmacoeconomics 2010; 28 (9)
774 Simons & Hagan

Table VI. Incremental costs per event ($US, year 2010 values) of diabetes mellitus- and cardiovascular-related events by type in the first year
following the event as determined by Tobit analysis
Event Inpatient costs Outpatient costs Total costs (95% CI)
Cost of diabetes complications
Ketoacidosis 29 588 1031 30 619 (26 986, 34 254)
Hyperosmolarity 15 286 862 16 148 (9 400, 22 896)
Coma 24 995 977 25 972 (15 148, 36 797)
Renal manifestations 20 122 1267 21 389 (17 421, 25 356)
Ophthalmic manifestations 12 022 1040 13 062 (10 810, 15 314)
Neurological manifestations 16 669 1116 17 785 (15 680, 19 891)
Peripheral circulatory disorder 18 900 1081 20 081 (17 618, 22 323)
Other specified complication 30 037 1589 31 626 (29 369, 33 883)
Unspecified complication 18 228 1183 19 411 (17 654, 21 168)
Ischaemic heart disease 43 269 682 43 951 (41 691, 46 211)
Congestive heart failure 43 870 781 44 651 (42 531, 46 770)
Cost of cardiovascular complications
Cardiovascular disease 43 569 732 44 301 (42 111, 46 491)

of neurological disorders were the most frequent ketoacidosis cases is $US119 (95% CI 105, 133);
complication, at a total cost of $US17 785 (95% 0.003879 multiplied by $US30 619, the sum of
CI 15 680, 19 891) in the first year, increasing to a inpatient and outpatient costs. The event rates
total of $US22 368 by the end of year 2 and to are multiplied by their corresponding incremental
$US25 461 by the end of year 3 (data not shown). costs, yielding incremental costs avoided by event.
Ischaemic heart disease and congestive heart fail- These total cost reductions ($US2749 due to
ure associated with diabetes were among the fewer diabetes-related complications and $US62
most costly, at $US43 269 and $US43 870, respec- due to fewer lipoprotein-related events) sum to
tively, for inpatient costs in the first year. The first $US2812. When residual costs accruing in sub-
year inpatient and outpatient costs for cardiovas- sequent years are included, the cost reduction
cular complications were $US43 569 and $US732, totals $US3543.
respectively, for a total of $US44 301 (95% CI
42 111, 46 491). The inpatient and outpatient costs Cost Effectiveness
associated with diabetes-related ketoacidosis were
$US29 588 and $US1031, respectively, for a total Table VIII summarizes the assessment of the
first year cost of $US30 619 (95% CI 26 986, cost effectiveness of the use of colesevelam in the
34 254). Costs associated with specific diabetic or management of patients with T2DM who do not
cardiovascular events did not vary across the three experience optimal glycaemic control despite treat-
trials. Thus, reductions in costs associated with co- ment with metformin-, insulin- or sulfonylurea-
lesevelam are simply the multiplication of event based therapies (as monotherapy or combination
reductions (tables IV and V) by their corresponding therapy with OAD agents). The assessment was
costs (table VI) and divided by 100 000. Table VII performed for all patients in the three diabetes
presents these calculations. Not all data are shown. clinical trials in which colesevelam was added to
In table VII, the precise calculations are shown monotherapy or combination therapy with OAD
for study 1 (the addition of coleveselam to a agents. Because colesevelam also improves lipid
metformin-based regimen). Specifically, the event levels, the analyses were performed for those with
rate for each diabetes-related complication is mul- lipid levels as reported in the diabetes clinical trials
tiplied by the incremental cost associated with and as reported in the hypercholesterolaemia trial
that event. Thus, costs avoided due to fewer for adult patients receiving the recommended

ª 2010 Adis Data Information BV. All rights reserved. Pharmacoeconomics 2010; 28 (9)
Table VII. Costs (year 2010 values) associated with events avoided in year 1 per 100 000 adult patients (pts) previously treated with metformin: annual savings across 3 years
ª 2010 Adis Data Information BV. All rights reserved.

Economic Evaluation of Colesevelam for Diabetes Mellitus


Diabetes-related complication Number of Event rate Incremental cost Per pt cost
events inpatient outpatient total (95% CI) total (95% CI) total at y 2 total at y 3
avoided
Ketoacidosis 388 0.003879 29 588 1 031 30 619 (26 986, 34 254) 119 (105, 133) 135 152

Hyperosmolarity 162 0.001616 15 286 862 16 148 (9400, 22 896) 26 (15, 37) 31 27

Coma 52 0.000517 24 995 977 25 972 (15 148, 36 797) 13 (8, 19) 17 12

Renal manifestations 530 0.005302 20 122 1 267 21 389 (17 421, 25 356) 113 (92, 134) 151 169

Ophthalmic manifestations 1 474 0.014741 12 022 1 040 13 062 (10 810, 15 314) 193 (159, 226) 264 313

Neurological manifestations 1 474 0.014741 16 669 1 116 17 785 (15 680, 19 891) 262 (231, 293) 327 368

Peripheral circulatory disorder 776 0.007759 19 000 1 081 20 081 (17 618, 22 323) 156 (137, 173) 201 207

Other specified manifestations 640 0.006401 30 037 1 589 31 626 (29 369, 33 883) 202 (188, 217) 237 262

Unspecified complication 2 011 0.020108 18 228 1 183 19 411 (17 654, 21 168) 390 (355, 426) 492 515

Ischaemic heart disease 1 881 0.018814 43 269 682 43 951 (41 691, 46 211) 827 (784, 869) 938 949

Congestive heart failure 1 002 0.010021 43 870 781 44 651 (42 531, 46 770) 447 (426, 469) 514 499
Pharmacoeconomics 2010; 28 (9)

Subtotal 10 390 0.1039 273 086 11 609 284 695 (244 308, 324 865) 2749 (2501, 2996) 3307 3473

Lipoprotein-related complication

cardiovascular events 141 0.00141 43 569 732 44 301 (42 111, 46 491) 62 (59, 66) 71 71

775
Overall total 10 531 0.10531 316 655 12 341 328 996 (286 419, 371 355) 2812 (2560, 3062) 3379 3543
776 Simons & Hagan

dose of colesevelam (3.75 g/day). As indicated in primarily by lowering LDL-C. Cholesterol benefits
table VIII, treatment with colesevelam resulted accrue cost reductions totalling $US3543 for pa-
in cost savings when used as monotherapy or in tients with normal lipid levels. For those with either
combination with other OAD agents. That is, the elevated LDL-C or low HDL-C (<1.03 mmol/L),
cost associated with the diabetes-related compli- those savings increase to $US3582 (or $US1326
cations (even in the first year) is more than the after the cost of colesevelam has been deducted)
cost of colesevelam (i.e. $US2256 per year). per person in the first year. Savings associated
From table VII, the total incremental costs per with adding colesevelam to an insulin-based reg-
patient for year 1 are simply summed for a cost imen are $US1818 and $US1852 for those with
reduction due to improved HbA1c of $US2749 (95% normal and raised lipid levels, respectively. For
CI 2501, 2996) per patient in year 1 as shown in sulfonylurea-based treatment, the savings are
table VIII. Those events would continue to accrue $US1599 and $US1629, respectively, for those
costs of $US3307 and $US3473 into the second with normal and raised lipid levels (table VIII).
and third years, respectively, when discounted at The results from the sensitivity analyses (limited
5% per year to the present. Additionally, treatment to 500 samples for each of the clinical trials from
with colesevelam also improves cholesterol levels, the 5000) are presented in figure 1. The results are

Table VIII. Cost effectiveness (year 2010 values) of colesevelam in a cohort of 100 000 adult patients with type 2 diabetes mellitus with and
without elevated lipids; data based on the clinical trials
Aggregated y 1 Cost ($US, y 2009 values)
event rate avoided y 1 reduction y 2 accrued y 3 accrued savings
with colesevelam (95% CI) avoided cost avoided cost
(discounted) (discounted)
Study 1: metformin as comparator[21] and study 4[27]
All patients with lipid data from study 1 (insulin-based therapy)[21]
Improved HbA1c 0.10390 2749 (2501, 2996) 3307 3473
Improved lipids 0.00141 62 (59, 66) 71 71
Improved glycaemic and lipid control 0.10531 2812 (2560, 3062) 3379 3543 1287a
All patients with lipid data from study 4 (primary hypercholesterolaemia)[27]
Improved glycaemic and lipid control 0.10608 2846 (2593, 3098) 3418 3582 1326
Study 2: insulin as comparator[22] and study 4[27]
All patients with lipid data from study 2 (insulin-based therapy)[22]
Improved HbA1c 0.11963 3166 (2879, 3450) 3808 3999
Improved lipids 0.001499 66 (63, 70) 76 75
Improved glycaemic and lipid control 0.1211 3232 (2943, 3519) 3884 4074 1818b
[27]
All patients with lipid data from study 4 (primary hypercholesterolaemia)
Improved glycaemic and lipid control 0.10608 3262 (2971, 3551) 3919 4108 1852
Study 3: sulfonylureas as comparator[23] and study 4[27]
All patients with lipid data from study 3 (sulfonylurea-based therapy)[23]
Improved HbA1c 0.11295 2989 (2719, 3257) 3595 3775
Improved lipids 0.001499 71 (67, 74) 81 80
Improved glycaemic and lipid control 0.114449 3060 (2786, 3331) 3676 3855 1599c
[27]
All patients with lipid data from study 4 (primary hypercholesterolaemia)
Improved glycaemic and lipid control 0.115132 3086 (2811, 3359) 3706 3885 1629
a Cost of colesevelam: $US6.18 per day · 365 = $US2256 per year; $US3543 - $US2256 = $US1287.
b Cost of colesevelam: $US6.18 per day · 365 = $US2256 per year; $US4074 - $US2256 = $US1818.
c Cost of colesevelam: $US6.18 per day · 365 = $US2256 per year; $US3855 - $US2256 = $US1599.
HbA1c = glycosylated haemoglobin.

ª 2010 Adis Data Information BV. All rights reserved. Pharmacoeconomics 2010; 28 (9)
Economic Evaluation of Colesevelam for Diabetes Mellitus 777

a
500
0
−500
−1000
−1500
−2000
−2500
−3000
−3500
Incremental cost reductions ($US)

b
500
0
−500
−1000
−1500
−2000
−2500
−3000
−3500

c
500
0
−500 6 7 8 9 10 11
−1000
−1500
−2000
−2500
−3000
−3500
Incremental reduction in diabetic complications
per 100 000 patients (× 1000)

Fig. 1. Cost-effectiveness analysis showing incremental cost (year 2010 values) versus incremental number of diabetic complications for
(a) study 1 (metformin-based treatment),[21] (b) study 2 (insulin-based treatment)[22] and (c) study 3 (sulfonylurea-based treatment).[23] The
bootstrap model incorporates varying effectiveness as well as the cost associated with each type of complication and is based on 5000
samples of 100 with replacement from two treatment groups of 1000. Five hundred results for each clinical trial are shown graphically.

plotted in the south-east quadrant (quadrant II) of velam to OAD medications in adult patients with
the cost-effectiveness plane, as the results were cost T2DM is associated with significant reductions in
saving while more effective. These are plotted as a HbA1c, LDL-C, non-HDL-C, apo-B levels and
presentation of the uncertainty in the model in fig- with increases in triglycerides.[21-23,32] Glycaemic
ure 1, one for each of the three clinical trials (500 of control is extremely important, since post mor-
the 5000 samples are shown for each). The y-axis tem studies of patients with and without T2DM
reports the reduction in costs associated with num- have determined that there are a significantly greater
ber of diabetic events avoided less the cost of cole- number of deaths from cardiovascular causes,
sevelam. The x-axis reports incremental reductions including both ischaemic heart disease and cere-
in the number of events avoided. brovascular disease, in patients with T2DM than
in patients without T2DM. Death from renal
Discussion failure in the older population is also significantly
more common in patients with T2DM.[33]
The economic model described herein was used The significance of cholesterol control is evi-
to evaluate the cost effectiveness of colesevelam denced by the results of a landmark meta-analysis
in adult patients who have yet to achieve optimal of data from 61 prospective studies and almost
glycaemic control despite treatment with met- 900 000 adults to determine the effects of total
formin-, insulin- or sulfonylurea-based therapies. cholesterol and blood pressure on vascular mor-
Research has shown that the addition of colese- tality.[26] Results from this analysis demonstrated

ª 2010 Adis Data Information BV. All rights reserved. Pharmacoeconomics 2010; 28 (9)
778 Simons & Hagan

that total cholesterol was positively associated with by replicating the estimation of the relationship
ischaemic heart disease mortality in both middle between treatment and HbA1c, and HbA1c and
and old age and at all blood pressure levels. diabetes-related complications three times. Gen-
It was also determined that the total cholesterol/ eralizability was targeted by conducting the re-
HDL-C ratio was the strongest predictor of ischae- plication in three different countries as well as
mic heart disease mortality.[26] Furthermore, the two different continents and by replicating other
previously mentioned study by Brown et al.[28] well established epidemiological studies. Finally,
determined that patients with vascular disease treatment patterns are certainly different from
have a reduced cardiovascular risk when reduc- country to country, but the relationship between
tions in LDL-C are combined with elevations in treatment and HbA1c, and particularly HbA1c and
HDL-C. complications, are not country specific. Ethnicity
The current analysis specifically translated the is a prognostic factor in the aetiology of a number
clinical benefits of improved glycaemic and lipid of diseases, for example diabetes and hyperten-
control into reductions in diabetes-related com- sion, but there is no reason to believe that the
plications and cardiovascular events. While effi- relationship between HbA1c and complications
cacy was assessed in terms of the clinical benefits differ as a consequence of race. Thus, these results,
translated into effectiveness (outcomes), the events with their replication, validation and reproduc-
associated with improved lipid levels may seem tion of other well established epidemiological evi-
low. However, the events reported are based on dence are generalizable.
the first year of lipid modification therapy. In the These analyses are limited to benefits accrued
LIPID trial comparing pravastatin with placebo, in the first year of treatment. At present, no long-
the survival curves did not begin to diverge before term clinical trial data are available. Thus, this is
the second year.[31] In the first year, there were an important limitation of these analyses. How-
71 events in 4502 patients in the placebo group ever, the DRE is based on data with patients having
compared with 67 events in 4512 patients in the up to 10 years of follow-up. Thus, the parameteri-
pravastatin group for a difference of 0.00095, or zation of the relationships between drug treat-
95 events avoided in 100 000 patients, compared ment and HbA1c, and HbA1c and diabetes-related
with the 141 events avoided as reported in this complications, are supported with longer term
study. The 141 avoided events included other car- information.
diovascular events, while the 95 avoided events in
the LIPID trial represented cardiovascular deaths. Conclusions
In the second- to third-year interval, there were
85 deaths (0.0196) in the placebo group compared This cost-effectiveness analysis suggests that
with 67 deaths (0.0153) in the pravastatin group, using colesevelam for 1 year as an add-on treatment
for a difference of 0.00428. to other antidiabetic agents for adult patients
Validated DREs and LCREs were used to who have not yet experienced optimal glycaemic
translate the reductions in HbA1c and LDL-C. and lipid control can result in reduced healthcare
The US DRE was based on the analysis of nearly costs in the first year as well as a reduction in their
500 000 patients with T2DM in managed-care residual costs in years 2 and 3. Those savings ac-
settings in the US and was validated in two in- crue from microvascular and macrovascular events
dependent datasets (see table III). The LCRE was avoided due to improving HbA1c as well as LDL-C.
based on a meta-regression of all 25 available lipid
mortality/morbidity trials. In the future, these
equations should be reconciled with other epide- Acknowledgements
miological studies, such as the UKPDS.[1]
The results reported within are contingent on The authors affirm that the data analysis presented herein
is an accurate representation of the study results. Funding for
the validity and generalizability of the two equa- this study was provided by Daiichi Sankyo, Inc., marketer of
tions, the DRE in particular. Validity was targeted colesevelam HCI in the US. Michael Hagan is an employee of

ª 2010 Adis Data Information BV. All rights reserved. Pharmacoeconomics 2010; 28 (9)
Economic Evaluation of Colesevelam for Diabetes Mellitus 779

Daiichi Sankyo, Inc. The mathematical epidemiological models, proportional hazards regression model. Am J Epidemiol
DRE and LCRE, were developed by W. Robert Simons over 1992; 135 (6): 638-48
10 years, independent of Daiichi Sankyo, Inc.; thus, their 14. Sasaki A, Uehara M, Horiuchi N, et al. A long-term follow-
sponsorship had no influence on those models. Daiichi Sankyo, up study of diabetic patients in Osaka, Japan: mortality
Inc. funded W. Robert Simons for their application to their and causes of death. Tohoku J Exp Med 1983; 141 (Suppl.):
clinical trials. He collaborated on the design of the study and 639-44
conducted the statistical and econometric programming and 15. Stratton IM, Adler AI, Neil HA, et al. Association of gly-
analyses and participated in the writing of the manuscript. caemia with macrovascular and microvascular complications
The authors owe a debt of gratitude for the clinical expert- of type 2 diabetes (UKPDS 35): prospective observational
ise and business acumen contributed to this manuscript by study. BMJ 2000; 321 (7258): 405-12
Dr Soamnauth Misir, PharmD. 16. Diaz A, Bourassa MG, Guertin MC, et al. Long-term pro-
gnostic value of resting heart rate in patients with suspected
or proven coronary artery disease. Eur Heart J 2005; 26 (10):
967-74
References
1. UK Prospective Diabetes Study Group. Intensive blood- 17. Diggle P, Heagerty P, Liang K-Y, et al. Analysis of long-
glucose control with sulphonylureas or insulin compared itudinal data. 2nd ed. Oxford: Oxford University Press,
with conventional treatment and risk of complications in 2002: 126-67
patients with type 2 diabetes (UKPDS 33). Lancet 1998; 18. Simons WR, Bolinder B, Kemp R. A five year longitudinal
352 (9131): 837-53 analysis of the health benefits of transitioning toward in-
2. Klein R. Hyperglycemia and microvascular and macrovas- sulin therapy sooner in newly diagnosed type II diabetics.
cular disease in diabetes. Diabetes Care 1995; 18 (2): 258-68 ISPOR European Conference; 2000 Nov 5-7; Antwerp
3. Pirart J. Diabetes mellitus and its degenerative complications: 19. Simons WR, Vinod HD, Gerber RA, et al. Does rapid transi-
a prospective study of 4400 patients observed between 1947 tion to insulin therapy in subjects with newly diagnosed type 2
and 1973 (part 1). Diabetes Care 1978; 1: 168-88 diabetes mellitus benefit glycaemic control and diabetes-
4. Adler AI, Boyko EJ, Ahroni JH, et al. Risk factors for diabe- related complications? A German population-based study.
tic peripheral sensory neuropathy: results of the Seattle Exp Clin Endocrinol Diabetes 2006; 114 (9): 1-8
Prospective Diabetic Foot Study. Diabetes Care 1997; 20 (7): 20. Simons WR, Hagan MA. Replication and validation of the
1162-7 quantification of the relationship between pharmacologic
5. UK Prospective Diabetes Study Group. Risk factors for intervention, reductions in HbA1c and reductions in compli-
coronary artery disease in non-insulin dependent diabetes cations: its application in quantifying the benefits of adding
mellitus (UKPDS 23). BMJ 1998; 316 (7134): 823-8 colesevelam hydrochloride to metformin in type 2 diabetics
[abstract no. PDB16]. Value Health May 2009; 12 (3): A99
6. Kuusisto J, Mykkänen L, Pyörälä K, et al. NIDDM and its
metabolic control predict coronary heart disease in elderly 21. Bays HE, Goldberg RB, Truitt KE, et al. Colesevelam hydro-
subjects. Diabetes 1994; 43 (8): 960-7 chloride therapy in patients with type 2 diabetes mellitus
treated with metformin: glucose and lipid effects. Arch
7. Lehto S, Ronnemaa T, Pyörälä K, et al. Predictors of stroke
Intern Med 2008; 168 (18): 1975-83
in middle-aged patients with non-insulin-dependent dia-
betes. Stroke 1996; 27 (1): 63-8 22. Goldberg RB, Fonseca VA, Truitt KE, et al. Efficacy and
8. Standl E, Balletshofer B, Dahl B, et al. Predictors of 10-year safety of colesevelam in patients with type 2 diabetes mel-
macrovascular and overall mortality in patients with litus and inadequate glycemic control receiving insulin-
NIDDM: the Munich General Practitioner Project. Dia- based therapy. Arch Intern Med 2008; 168 (14): 1531-40
betologia 1996; 39 (12): 1540-5 23. Fonseca VA, Rosenstock J, Wang AC, et al. Colesevelam
9. Groeneveld Y, Petri H, Hermans J, et al. Relationship HCl improves glycemic control and reduces LDL choles-
between blood glucose level and mortality in type 2 diabetes terol in patients with inadequately controlled type 2 dia-
mellitus: a systematic review. Diabet Med 1999; 16 (1): 2-13 betes on sulfonylurea-based therapy. Diabetes Care 2008;
31 (8): 1479-84
10. Uusitupa MI, Niskanen LK, Siitonen O, et al. Ten-year
cardiovascular mortality in relation to risk factors and 24. Simons WR, Edwards DM. The relationship between high
abnormalities in lipoprotein composition in type 2 (non- density lipoprotein cholesterol (HDL-C), nicotinic acid
insulin-dependent) diabetic and non-diabetic subjects. and all-cause mortality [abstract no. PCV100]. Value
Diabetologia 1993; 36 (11): 1175-84 Health 2009; 12 (7): A331
11. Wei M, Gaskill SP, Haffner SM, et al. Effects of diabetes 25. Simons WR, Edwards DM. The quantification of the re-
and level of glycemia on all-cause and cardiovascular lationship between high density lipoprotein cholesterol
mortality: the San Antonio Heart Study. Diabetes Care (HDL-C) and all-cause mortality [abstract no. PCV47].
1998; 21 (7): 1167-72 Value Health 2009; 12 (7): A320-1
12. Hanefeld M, Fischer S, Julius U, et al. Risk factors for myo- 26. Prospective Studies Collaboration. Blood cholesterol and
cardial infarction and death in newly detected NIDDM: vascular mortality by age, sex, and blood pressure: a meta-
the Diabetes Intervention Study, 11-year follow-up. Dia- analysis of individual data from 61 prospective studies with
betologia 1996; 39 (12): 1577-83 55 000 vascular deaths. Lancet 2007; 370 (9602): 1829-39
13. Knuiman MW, Welborn TA, Whittall DE. An analysis of 27. Insull Jr W, Toth P, Mullican W, et al. Effectiveness of co-
excess mortality rates for persons with non-insulin-depen- lesevelam hydrochloride in decreasing LDL cholesterol
dent diabetes mellitus in Western Australia using the Cox in patients with primary hypercholesterolemia: a 24-week

ª 2010 Adis Data Information BV. All rights reserved. Pharmacoeconomics 2010; 28 (9)
780 Simons & Hagan

randomized controlled trial. Mayo Clin Proc 2001; 76 (10): low-density lipoprotein cholesterol and high-density lipopro-
971-82 tein cholesterol: results from the LIPID study. Eur Heart J
28. Brown BG, Stukovsky KH, Zhao XQ. Simultaneous low- 2004; 25 (9): 771-7
density lipoprotein-C lowering and high-density lipo- 32. Zieve FJ, Kalin MF, Schwartz SL, et al. Results of the glucose-
protein-C elevation for optimum cardiovascular disease lowering effect of WelChol study (GLOWS): a randomized,
prevention with various drug classes, and their combina- double-blind, placebo-controlled pilot study evaluating
tions: a meta-analysis of 23 randomized lipid trials. Curr the effect of colesevelam hydrochloride on glycemic control
Opin Lipidol 2006; 17 (6): 631-6 in subjects with type 2 diabetes. Clin Ther 2007; 29 (1):
29. Simons WR, Jones D, Buzdar A. Cost-effectiveness of ana- 74-83
strozole versus tamoxifen as first-line therapy for postmeno- 33. Roper NA, Bilous RW, Kelly WF, et al. Cause-specific
pausal women with advanced breast cancer. Clin Ther mortality in a population with diabetes: South Tees Dia-
2003; 25 (11): 2972-87 betes Mortality Study. Diabetes Care 2002; 25 (1): 43-8
30. US Department of Health and Human Services. US Food
and Drug Administration. National drug code directory
[online]. Available from URL: http://www.accessdata.fda. Correspondence: W. Robert Simons (Rob), Global Health
gov/scripts/cder/ndc/default.cfm [Accessed 2010 Aug 11] Economics & Outcomes Research Inc., 41 River Road,
31. Colquhoun D, Keech A, Hunt D, et al. Effects of pravastatin Summit, NJ 07901, USA.
on coronary events in 2073 patients with low levels of both E-mail: rsimons@globalhealtheconomics.com

ª 2010 Adis Data Information BV. All rights reserved. Pharmacoeconomics 2010; 28 (9)