1 ADHD drugs and CV outcomes: Preliminary feasibility results and potential observational studies David J. Graham, MD, MPH on behalf of the FDA Epidemiology Contracts Study Team March 22, 2006
Jan 21, 2016
1
ADHD drugs and CV outcomes: Preliminary feasibility results
and potential observational studies
David J. Graham, MD, MPH
on behalf of the FDA
Epidemiology Contracts Study Team
March 22, 2006
2
FDA’s Epidemiology Contracts Program
• Replaces Cooperative Agreement Program
• Provides capability to study safety questions in a population setting
• 4 awardees Covered lives HMO Research Network 3.2 million Ingenix (i3 Drug Safety) 12 million Kaiser Permanente Research Institute 6.1 million TN and WA Medicaid 2.2 million
• Turnover: 1 yr: 8%-30%; 5 yrs: 25%-80%
• Funding 2005-06 $1.6 million 2006-07 $0.9 million
3
Feasibility study design
• Inception cohorts, all ages• Study period
Jan 1998-June 2005 (i3, KPRI, Medicaid) July 2000-June 2005 (HMO RN)
• Drugs of interest Amphetamine or dextroamphetamine Methylphenidate Atomoxetine
• Age-groups Children/adolescents (1-19 years) Adults (20-64 years)
• Outcomes of interest Sudden unexplained death Acute myocardial infarction Other ischemic heart disease Cerebrovascular accident Arrhythmia Hypertension Pulmonary hypertension
4
Details of ADHD drug exposure cohortsof children ≤ 19 years
Amphetamine Methylphenidate Atomoxetine Total
Base 7 M
No. 191 K 222 K 80 K 493 K
PYrs 160 K 200 K 40 K 399 K
% male 73 73 73 73
5
Persistency of ADHD drug use by drug in 0-19 year olds
0
50
100
150
200
250
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91
Months of use
Pat
ient
num
ber
(tho
usan
ds)
Methylphenidate
Amphetamine
Atomoxetine
6
Background rate per 105 per year
Sudden cardiac death 1-9
Acute myocardial infarction 1-20
Cerebrovascular accident 3
Background rates for cardiovascular events of interest in pediatric population age 1-19
7
1o All AMI 17 24 IHD 17 43 Cardiac arrest 14 55 CVA 49 90 Arrhythmia 245 445 HTN 66 353 Pulm HTN 10 22 Deaths† 241
†Incomplete
Counts of potential study outcomesin children age 1-19 years,
based on hospital discharge diagnoses
8
Deaths reported within pediatric ADHD-drug-exposed inception cohorts
• Deaths occurred at any time after cohort entry
• From any cause
• In-hospital only from 2 sites, none from 1 site
• Sudden out of hospital deaths included from 1 site 2 sites have death certificate linkage; SCD validated at 1 NDI search required with other 2 sites - turnover, time, $$
9
Power to identify a given risk ratio with abackground rate = 15 per 105 person-years
(AMI in children 1-19 years)P
ow
er (
1-b
eta)
Exposure cohort person-years (thousands)
40 80 120 160 200 240 280 320 360 400
0.00
0.20
0.40
0.60
0.80
1.00
RR=2
RR=3
RR=5
10
Power to identify a given risk ratio with abackground rate = 3 per 105 person-years
(CVA in children 1-19 years)
Po
wer
(1-
bet
a)
Exposure cohort person-years (thousands)
40 80 120 160 200 240 280 320 360 400
0.00
0.20
0.40
0.60
0.80
1.00
RR=2
RR=3
RR=5
RR=10
11
0102030405060708090
100
0 0.2 0.4 0.6 0.8 1
Power
Num
ber
of E
vent
s
RR=2
RR=3
RR=5
RR=10
Event number and statistical power requiredto confirm risk ratios from 2 to 10
12
Probability of excluding a given risk ratio, assuming true RR=1: background = 15 per 105 person-years
(AMI in children age 1-19)
0
0.2
0.4
0.6
0.8
1
0 80 160 240 320 400
Exposed cohort person-years (thousands)
Po
wer
RR=2
RR=3RR=5
13
Probability of excluding a given risk ratio, assumingtrue RR=1: background = 3 per 105 person-years
(CVA in children age 1-19)
0
0.2
0.4
0.6
0.8
1
0 80 160 240 320 400
Exposed cohort person-years (thousands)
Po
wer
RR=10
RR=5
RR=3
RR=2
14
Estimated risk ratio that can be detected with 80% probability by age- and drug-group
Sudden death
AMI CVA
Amphetamine 4-5 2-5 4-5
Methylphenidate 4-5 2-5 4-5
Atomoxetine 10 3-10 10
All combined 3 <2-3 3
15
Some additional power considerations
Age=1-19
AMI CVA
Background rate (x10-5) 15 3
Person-years (x10-5) 399 399
No. expected 4-60 12
No. reported 17 49
AMI: acute myocardial infarctionCVA: cerebrovascular accident (stroke)
16
Caveats regarding ADHD cohort study
• Preliminary data; relationship of drug exposures to outcomes not yet studied
• Crude definitions of exposure, outcome Hospital D/C diagnoses, not validated Outcome post 1st Rx; timing with current use not
known Out of hospital deaths (SCD) not captured at 2 sites
• Power calculations crude Uncertainty regarding background rates (i.e., AMI)
17
Proposal offered at February advisory meeting
• Observational echocardiographic study
• Within a large healthcare database:
Identify patients treated with ADHD drugs for varying durations of time
Select suitable untreated “controls” from same population
Perform echocardiography and assess
Left ventricular wall thickness
Contractility
18
Patient sampling for echocardiographic study
0
50
100
150
200
250
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91
Months of use
Pat
ient
num
ber
(tho
usan
ds)
Methylphenidate
Amphetamine
Atomoxetine
19
Summary
• Concern regarding potential for CV risk of ADHD drugs High prevalence of use in children; growing in adults
Sudden unexplained death of 1 interest; most difficult to study TN Medicaid and KPRI have death certificate linkage Other sites would require NDI search
Other CV outcomes
• Feasibility study Exposed person-time substantial for most ADHD drugs
CV outcomes require validation; timing with respect to exposure
Statistical power and uncertainty
Number of arrhythmia cases seems surprisingly high
Now in process of obtaining cost estimates for in-depth study
20
FDA Epidemiology Contracts ADHD Study Team (list by site)
• FDA, ODS Andrew Mosholder, MD, MPH Kate Gelperin, MD, MPH Judy A. Staffa, PhD David J. Graham, MD, MPH
• HMO Research Network Susan E. Andrade, PhD
• Ingenix (i3 Drug Safety) K. Arnold Chan, MD, ScD
• Kaiser Permanente Research Institute Joe Selby, MD, MPH
• Medicaid (TN and WA) William Cooper, MD, MPH