Download S0735109711027628_mmc1

Online Appendix for the following JACC article
TITLE: Comparative Effectiveness and Cost-Effectiveness of Computed Tomography
Screening for Coronary Artery Calcium in Asymptomatic Individuals
AUTHORS: Bob J. H. van Kempen, BSc, Sandra Spronk, PhD, Michael T. Koller, MD,
Suzette E. Elias-Smale, MD, MSc, Kirsten E. Fleischmann, MD, MPH, M. Arfan Ikram,
MD, PhD, Gabriel P. Krestin, MD, PhD, Albert Hofman, MD, PhD, Jacqueline C. M.
Witteman, PhD, Myriam Hunink M.G., MD, PhD
1
APPENDIX
Technical Appendix.
Model.
Figure 1a to 1d show the cycle tree of the Markov model, with all states and possible transitions
between them. Within a cycle, the incidence of one type of event did not exclude the possibility of
another type.
CHD incidence.
The probability of incident CHD was determined by first converting observed 10 year probabilities
p to average rates r per time unit t (equation (1)).
r
 ln( 1  p )
t
(1)
An average rate was calculated which was assumed to be constant over time and extrapolated
beyond 10 years. This assumption was validated by observing the fairly constant rate over a 10year follow-up period (figures 2 and 3). Subsequently, the calculated rate was converted to an
annual probability, by:
p  1  e r
(2)
Adjusting for efficacy of Treatment.
Drug treatment efficacies, in terms of Relative Risks (RR) were obtained from meta analyses and
considered the relative risk in incidence of CHD (any of the following outcomes: non-fatal
myocardial infarction, CABG, PCI, and CHD mortality) or stroke (ischemic, hemorrhagic or
unspecified) compared to placebo. The relative risk of a certain treatment or intervention was
assumed to be constant over time.
2
Derived annual probabilities of CHD and stroke respectively, were multiplied by the RR of the
appropriate strategy.
pafter _ treatment  punadjusted  RR strategy
(3)
Adjusting the treatment efficacies for treatment adherence, baseline prevalence and
treatment goals.
The model incorporates 3 basic drug treatments: statins, anti-hypertensives and aspirin, each
with their own specific treatment goals. In order to estimate the combined effect, we made the
following assumptions:
1. An individual already on statins at baseline who had not reached the treatment goal, i.e.
>160 (4.92), >130 (3.37) and >90 (2.50) mg/dL (mmol/l) for the low, intermediate and
high risk category respectively, was assumed to switch to a higher dose or more potent
statin, and assigned half of the reduction in risk based on a full dose given to a non-user.
The same holds for an individual using anti-hypertensives at baseline and SBP >140,
>140, >130 mm Hg respectively.
2. When a combination of drugs was assigned, the net effect of the drugs together on risk
reduction was assumed to be the product of the individual RR’s, times a factor for
potential (dys)synergy, SF, where .90 < SF < 1.10. This range was chosen to make sure
that a combination of 2 or 3 drugs was at least as effective as the effect of a single drug.
For the base case analysis we used a synergy factor of 1. When 2 drugs were jointly
taken, the joint effect was corrected with SF. When 3 drugs were jointly taken, the joint
effect was corrected with SF2.
The RR of each strategy was corrected for adherence and baseline prevalence of statin, antihypertensives and aspirin use. To simplify, the adherence for aspirin, anti-hypertensives and
statin or any combination of them was considered to be equal.
3
For the individuals reclassified to the low risk group in the CT screening strategy, the RR’s for
both CHD and stroke (RR1) were determined by:
1


 f LDL , N  BP  , N  RR antiH  RR statin  SF  f LDL , N  BP  ,U   1  RR antiH   RR statin  SF  f LDL , N  BP   RR statin  
2




1
1
RR1  C  f LDL ,U  BP    1  RR statin   f LDL  BP  , N  RR antiH  f LDL  BP  ,U   1  RR antiH  

2
2


1
1
1


 f LDL ,U  BP  , N  RR antiH   1  RR statin   SF  f LDL ,U  BP  ,U   1  RR antiH    1  RR statin   SF

2
2
2


1  (1  f LDL  BP  )  C
(4)
Where C equals the percentage of therapy adherent individuals and f equals a fraction subindexed by:
LDL+ indicates LDL > 160 mg / dL (4.92 mmol/l); LDL- indicates LDL ≤ 160 mg / dL (4.92 mmol/l),
LDLU indicates statin use at baseline ; LDLN indicates no use at baseline,
BP+ indicates SBP > 140 mm Hg ; BP- indicates SBP < 140 mm Hg,
BPU indicates anti-hypertensives use at baseline ; BPU indicates no use at baseline,
RRStatin equals the relative risk of CHD (stroke) for someone taking statins versus placebo,
RRantiH equals the relative risk of CHD (stroke) for someone taking anti-hypertensives vs placebo,
SF equals the synergy factor discussed earlier.
The baseline fractions of f for strategy I are given in table 1.
For the current guidelines strategy, consisting of both statins and anti-hypertensives when
indicated, and the individuals reclassified to the intermediate risk group in the CT screening
strategy, the RR’s for both CHD and stroke (RR2) were determined by
4
1


 f LDL , N  BP  , N  RR antiH  RR statin  SF  f LDL , N  BP  ,U   1  RR antiH   RR statin  SF  f LDL , N  BP   RR statin  
2




1
1
RR 2  C  f LDL ,U  BP    1  RR statin   f LDL  BP  , N  RR antiH  f LDL  BP  ,U   1  RR antiH  

2
2


1
1
1


 f LDL ,U  BP  , N  RR antiH   1  RR statin   SF  f LDL ,U  BP  ,U   1  RR antiH    1  RR statin   SF

2
2
2


1  (1  f LDL  BP  )  C
(4)
Where C equals the percentage of therapy adherent individuals and f equals a fraction subindexed by:
LDL+ indicates LDL > 130 mg / dL (3.37 mmol/l); LDL- indicates LDL ≤ 130 mg / dL (3.37 mmol/l),
LDLU indicates statin use at baseline ; LDLN indicates no use at baseline,
BP+ indicates SBP > 140 mm Hg ; BP- indicates SBP < 140 mm Hg,
BPU indicates anti-hypertensives use at baseline ; BPU indicates no use at baseline,
RRStatin equals the relative risk of CHD (stroke) for someone taking statins versus placebo,
RRantiH equals the relative risk of CHD (stroke) for someone taking anti-hypertensives vs placebo,
SF equals the synergy factor discussed earlier.
The baseline fractions of f for strategy II are given in table 2.
For the individuals reclassified to the high risk group in the CT screening strategy, the RR’s were
determined by:
5


2
f


f
N
N
N  RR antiH  RR statin  RR asp  SF
N
N
U  RR antiH  RR statin  SF 
LDL  BP  A
 LDL  BP  A



1
1
2
 f LDLN  BP  ,U  A N  RR statin   1  RR antiH   RR asp  SF  f LDLN  BP  ,U  AU  RR statin   1  RR antiH   SF 
2
2


RR 3  C    f LDLN  BP  ,U  A N  RR statin  RR asp  SF  f LDLN  BP  ,U  AU  RR statin 



1
2
 f  ,U N N  1  1  RR
 1  RR statin   RR antiH  SF 
statin   RR antiH  RR asp  SF  f LDL ,U  BP N  AU 
 LDL  BP  A 2

2


 f  ,U  ,U N  1  1  RR statin   1  1  RR antiH   RR asp  SF 2

2
 LDL  BP  A 2

1
1
1


 f LDL ,U  BP  ,U  AU   1  RR statin    1  RR antiH   SF  f LDL ,U  BP  ,U  A N   1  RR statin   RR asp  SF  
2
2
2




1
C   f LDL ,U  BP  ,U  AU   1  RR statin   f LDL ,U  BP N  A N  RR antiH  RR asp  SF  f LDL ,U  BP N  AU  RR antiH 

2


1
1


 f LDL ,U  BP  ,U  A N   1  RR antiH   RR asp  SF  f LDL ,U  BP  ,U  AU   1  RR antiH   f LDL ,U  BP  ,U  A N  RR asp 
2
2


1  ( f LDL ,U  BP  ,U  AU )  C
(5)
All sub-indexes are similar to the ones used in (4), except for
RRasp which is the relative risk of incident CHD for someone taking aspirin vs placebo,
AU indicates aspirin use at baseline ; AN indicates no use,
LDL+ indicates LDL > 90 mg / dL (2.5 mmol/l); LDL- indicates LDL ≤ 90 mg / dL (2.5 mmol/l),
BP+ indicates SBP > 130 mm Hg ; BP- indicates SBP < 130 mm Hg
Note that aspirin is only prescribed in males. Fractions of f for strategy III are given in table 3.
Secondary prevention.
In high-risk individuals, the risk of a recurrent CHD event was assumed to be 1.5-fold higher than
their risk of a primary CHD event [22]. In low- and intermediate-risk individuals, the risk of a
recurrent CHD event was assumed to be similar to the risk of a primary CHD event in high-risk
6
individuals. Treatment for secondary prevention of CHD and stroke was assumed to be similar to
the medical treatment of high-risk individuals/secondary prevention of CVD.
Major bleeding rate.
We modeled the excess rate of major bleeding due to the use of aspirin in males. From the most
recent meta-analysis, the bleeding rate in the placebo group was converted to a yearly probability
PMajorBleeding. The annual probability of excess major bleeding PExcessMajorBleeding was calculated by:
PExcessMajorBleeding  ( RR  1)  PMajorBleeding
(6)
where RR equals the relative risk of major bleeding for males taking aspirin compared to placebo.
Death due to radiation.
A recent simulation study estimated the lifetime attributable risk (LAR) of cancer from a single CT
Coronary Calcium scan. We divided the LAR by the expected remaining lifetime of our cohort,
and used this as approximation for the annual cancer risk due to radiation. Making this simplifying
assumption led to slight overestimation of the radiation risk but since the risk is extremely low, it’s
influence will be negligible.
7
Figure 1a. Schematic presentation of the Markov simulation model. The cycle tree of each health state is presented.
8
Figure 1b. Second part of Markov model. The cycle tree of each health state is presented.
9
Figure 1c Third part of Markov model. The cycle tree of each health state is presented.
10
Figure 1d. fourth part of the schematic presentation of the Markov simulation model.
11
Figure 2: Probability of CHD in year t for men, conditional on survival up until beginning of year t. For
each risk category (low, intermediate and high), both the probability calculated assuming a constant
hazard rate and the Weibull distribution are drawn.
Conditional Probability
0.1
0.09
Weibull - low
0.08
0.07
Weibull - intermediate
0.06
Weibull - High
0.05
Constant hazard rate low
0.04
Constant hazard rate intermediate
0.03
Constant hazard rate - high
0.02
0.01
0
1
2
3
4
5
6
7
8
9
10
12
Figure 3: Probability of CHD in year t for women, conditional on survival up until beginning of year t.
For each risk category (low, intermediate and high), both the probability calculated assuming a
constant hazard rate and the Weibull distribution are drawn.
Conditional Probability
0.1
0.09
Weibull - low
0.08
0.07
Weibull - intermediate
0.06
Weibull - high
0.05
Constant hazard rate - low
0.04
Constant hazard rate intermediate
0.03
Constant hazard rate - high
0.02
0.01
0
1
2
3
4
5
6
7
8
9
10
13
Table 1: Fractions of different combinations of statin use, anti-hypertensives use, LDL-level and SBP
level at baseline, stratified by sex, used in the CT screening strategy for individuals reclassified to the
low risk category.
Fraction
men
women
LDL > 160 mg/dl
0.11
0.13
SBP > 140 mmHG
No Statin
No anti-hypertensives
LDL > 160 mg/dl
0.02
0.10
SBP > 140 mmHG
No Statin
anti-hypertensives use
LDL > 160 mg/dl
0.10
0.17
SBP < 140 mmHG
No Statin
-*
LDL > 160 mg/dl
0.00
0.01
SBP > 140 mmHG
Statin use
No anti-hypertensives
LDL > 160 mg/dl
0.00
0.01
SBP > 140 mmHG
Statin use
anti-hypertensives use
LDL > 160 mg/dl
0.00
0.01
SBP < 140 mmHG
Statin use
-*
LDL <160 mg/dl
0.29
0.19
SBP > 140 mmHG
-*
No anti-hypertensives
LDL <160 mg/dl
0.10
0.22
SBP > 140 mmHG
-*
anti-hypertensives use
LDL < 160 mg/dl
0.39
0.17
SBP <140 mmHG
-*
-*
*Irrespective of statin and anti-hypertensives use respectively
14
Table 2: Fractions of different combinations of statin use, anti-hypertensive use, aspirin use, LDL-level
and SBP level at baseline, stratified by sex, used in the ’current guidelines’ strategy, and for the
individuals reclassified to the intermediate risk category in the CT screening strategy.
Fraction
men
women
LDL > 130 mg/dl
0.24
0.23
SBP > 140 mmHG
No Statin
No anti-hypertensives
LDL > 130 mg/dl
0.09
0.22
SBP > 140 mmHG
No Statin
anti-hypertensives use
LDL > 130 mg/dl
0.30
0.26
SBP < 140 mmHG
No Statin
-*
LDL > 130 mg/dl
0.04
0.02
SBP > 140 mmHG
Statin use
No anti-hypertensives
LDL > 130 mg/dl
0.02
0.02
SBP > 140 mmHG
Statin use
anti-hypertensives use
LDL > 130 mg/dl
0.01
0.04
SBP < 140 mmHG
Statin use
-*
LDL <130 mg/dl
0.13
0.07
SBP > 140 mmHG
-*
No anti-hypertensives
LDL <130 mg/dl
0.06
0.08
SBP > 140 mmHG
-*
anti-hypertensives use
LDL < 130 mg/dl
0.12
0.04
SBP <140 mmHG
-*
-*
*Irrespective of statin and anti-hypertensives use respectively
15
Table 3: Fractions of different combinations of statin use, anti-hypertensive use, aspirin use, LDL-level
and SBP level at baseline, stratified by sex, used in the CT coronary calcium screening strategy for
individuals reclassified to the High risk category.
Fraction
-*
-*
No Statin
No anti-hypertensives
No Aspirin
-*
-*
No Statin
No anti-hypertensives
Aspirin use
-*
SBP > 130 mmHG
No Statin
Anti-hypertensives use
No Aspirin
-*
SBP > 130 mmHG
No Statin
Anti-hypertensives use
Aspirin use
-*
SBP < 130 mmHG
No Statin
Anti-hypertensives use
No Aspirin
-*
SBP < 130 mmHG
No Statin
Anti-hypertensives use
Aspirin use
LDL > 90 mg/dl
-*
Statin use
No Anti-hypertensives
No Aspirin
LDL > 90 mg/dl
-*
Statin use
No Anti-hypertensives
Aspirin use
LDL > 90 mg/dl
SBP > 130 mmHg
Statin use
Anti-hypertensives use
No Aspirin
LDL > 90 mg/dl
SBP > 130 mmHg
Statin use
men
0.50
women
0.33
0.14
0.07
0.04
0.19
0.04
0.04
0.05
0.19
0.01
0.04
0.09
0.02
0.04
0.02
0.06
0.04
0.01
0.04
16
Anti-hypertensives use
Aspirin use
LDL > 90 mg/dl
0.03
0.00
SBP <130 mmHg
Statin use
Anti-hypertensives use
No Aspirin
LDL > 90 mg/dl
0.00
0.04
SBP <130 mmHg
Statin use
Anti-hypertensives use
Aspirin use
LDL < 90 mg/dl
0.00
0.02
-*
Statin use
No Anti-hypertensives
No Aspirin
LDL < 90 mg/dl
0.00
0.00
-*
Statin use
No Anti-hypertensives
Aspirin use
LDL < 90 mg/dl
0.00
0.00
SBP > 130 mmHg
Statin use
No Anti-hypertensives
No Aspirin
LDL < 90 mg/dl
0.00
0.00
SBP > 130 mmHg
Statin use
No Anti-hypertensives
Aspirin use
LDL < 90 mg/dl
0.00
0.00
SBP < 130 mmHg
Statin use
Anti-hypertensives use
No Aspirin
LDL < 90 mg/dl
0.00
0.00
SBP < 130 mmHg
Statin use
Anti-hypertensives use
Aspirin use
*Irrespective of statin and anti-hypertensives use respectively
17
Table 4. Reclassification for <5%,5-20% and >20% for the low, intermediate and high risk category.
Parameters
Base-case value
men†
Base-case value
women†
Total: 329
Total: 247
37 (12%)
0.03
0.09
40 (16%)
0.03 (0.04, 0.16)
0.06 (0.04, 0.16)
212 (64%)
0.09
0.05
153 (62%)
0.14
0.14
80 (24%)
0.30
0.16
54 (22%)
0.23
0.14
Probabilities and Characteristics of
Reclassification groups
Individuals Reclassified to
alternative low risk group (<5%)
N (%)
Observed 10-year CHD Risk
Observed 10-year stroke Risk
Individuals reclassified to
alternative intermediate risk (5-20%)
group#
N(%)
Observed 10-year CHD Risk
Observed 10-year stroke Risk
Individuals Reclassified to
alternative high risk group (>20%)
N(%)
Observed 10-year CHD Risk
Observed 10-year stroke Risk
18
Table 5. Total number (baseline users plus new indicated users) of individuals on a certain
drug (statins, anti hypertensives or aspirin), per strategy in men.
statins
anti
hypertensives
aspirin
Men
Current practice
Number on drug (%)
39 (11.9%)
74 (22.5%)
67 (20.4%)
Number on drug (%)
Number (%) with
increased dosage/
more potent drug
247 (75.1%)
209 (63.5%)
67 (20.4%)
20 (6.1%)
53 (16.1%)
Number on drug (%)
Number (%) with
increased dosage/
more potent drug
228 (69.3%)
234 (71.1%)
29 (8.8%)
57 (17.3%)
# on drug (%)
329 (100%)
74 (22.5%)
Current guidelines
CT screening
128 (38.9%)
Statin therapy
67 (20.4%)
19
Table 6. Total number (baseline users plus newly users) of individuals on a certain drug
(statins, anti hypertensives or aspirin), per strategy in women.
statins
anti
hypertensives
aspirin
Women
Current practice
Number on drug (%)
37 (15.0%)
129 (52.2%)
44 (17.8%)
Number on drug (%)
Number (%) with
increased dosage/
more potent drug
215 (87.0%)
209 (84.6%)
44 (17.8%)
21 (8.5%)
81 (32.8%)
Number on drug (%)
Number (%) with
increased dosage/
more potent drug
183 (74.1%)
218 (88.3%)
22 (8.9%)
88 (35.6%)
Number on drug (%)
247 (100.0%)
129 (52.2%)
Current guidelines
CT screening
44 (17.8%)
Statin therapy
20
44 (17.8%)
Table 7. 10 year predicted risk, and a number of baseline risk factors of statin vs no statin
users and anti-hypertensive vs no anti-hypertensive users for the current practice, current
guidelines and CT calcium screening strategy in men.
statin user
no statin
user
antihypertensive
user
0.21
0.14
0.16
Current practice
Predicted 10 year risk of CHD
Age
Systolic blood pressure
(mmHg)
Total cholesterol (mg/dL)
HDL cholesterol (mg/dl)
Diabetes
Current smoking
ln(Calcium score)
69
69
67
154
210
48
8%
23%
6.08
142
222
48
6%
21%
4.75
149
222
52
5%
14%
5.18
Predicted 10 year risk of CHD
0.14
0.14
0.16
Current guidelines
Age
Systolic blood pressure
(mmHg)
Total cholesterol (mg/dL)
HDL cholesterol (mg/dl)
Diabetes
Current smoking
ln(Calcium score)
69
71
69
144
235
48
5%
21%
4.87
144
186
48
9%
23%
5.01
154
222
50
8%
18%
5
Predicted 10 year risk of CHD
0.18
0.08
0.17
69
70
69
145
233
48
6%
21%
5.59
142
201
48
6%
22%
3.354
151
222
49
7%
18%
5.22
CT screening
Age
Systolic blood pressure
(mmHg)
Total cholesterol (mg/dL)
HDL cholesterol (mg/dl)
Diabetes
Current smoking
ln(Calcium score)
21
Table 8. 10 year predicted risk, and a number of baseline risk factors of statin vs no statin
users and anti-hypertensive vs no anti-hypertensive users for the current practice, current
guidelines and CT calcium screening strategy in women.
statin user
no statin
user
antihypertensive
user
0.14
0.14
0.14
Current practice
Predicted 10 year risk of CHD
Age
Systolic blood pressure
(mmHg)
Total cholesterol (mg/dL)
HDL cholesterol (mg/dl)
Diabetes
Current smoking
ln(Calcium score)
73
75
73
148
226
48
19%
14%
4.58
149
245
50
17%
13%
4.29
147
238
51
19%
7%
4.4
Predicted 10 year risk of CHD
0.14
0.12
0.14
Current guidelines
Age
Systolic blood pressure
(mmHg)
Total cholesterol (mg/dL)
HDL cholesterol (mg/dl)
Diabetes
Current smoking
ln(Calcium score)
74
76
74
147
250
50
15%
15%
4.37
157
199
50
28%
3%
4.08
152
238
50
19%
11%
4.33
Predicted 10 year risk of CHD
0.16
0.07
0.14
73
75
74
147
254
49
15%
15%
4.92
154
215
51
22%
8%
2.66
151
238
50
18%
12%
4.44
CT screening
Age
Systolic blood pressure
(mmHg)
Total cholesterol (mg/dL)
HDL cholesterol (mg/dl)
Diabetes
Current smoking
ln(Calcium score)
22