pregnancy registries: methodological points & real examples

Pregnancy Registries:

Methodological Points &

Real Examples

Sonia Hernández-Díaz, MD, DrPH

Harvard School of Public Health, Boston, USA

HARVARD SCHOOL OF PUBLIC HEALTH

Department of Epidemiology

2

Disclosure

Pregnancy Registries

The Pharmacoepidemiology Program at the HarvardSchool of Public Health is partially supported by traininggrants from Pfizer, Takeda, Bayer and Phrma.

SHD has consulted for AstraZeneca and GSK

3

Objective

Review methodological issues in Exposure


• Validity:

♦ Enrollment – missing data

♦ Selection of comparison groups

• Efficiency / Power

Real examples


4

Observational prospective cohort of women receiving a

biopharmaceutical product(s) of interest as part of their

routine clinical care who are enrolled voluntarily during

gestation, before outcome can be known. Participants are

followed until the end of pregnancy or longer to

systematically collect information on specific pregnancy

outcomes and evaluate their frequency relative to a

scientifically valid reference population(s)

Introduction: Pregnancy Registry


Introduction: FDA’s guidelines

Good drug candidates:

• Likely to be used in pregnancy, or

• Likely to be used by women of childbearing age, or

• Pose special risks, such as live vaccines

Recommended timing of enrollment:

• After exposure but before outcome of pregnancy is known

• However, it is expected that retrospective cases will arise

• Analyses should be conducted separately in these groups

Guidance for industry – Establishing pregnancy exposure registries – FDA 2002http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm071639.pdf

5Pregnancy Registries

Analysis: comparison groups

• Internal♦ Exposed to comparator drugs

♦ Unexposed

• External♦ surveillance systems

♦ background rates

♦ other pregnancy registries

Guidance for industry – Establishing pregnancy exposure registries – FDA 2002

http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm071639.pdf



In 2005, similar guidelines were published by the EMA.

In 2007, the Food and Drug Amendments Act (FDAAA)provided the authority under Title IX to require pregnancyregistries as a post marketing requirement (PMR)

FDAAA. US Food and Drug Administration Amendments Act (FDAAA) of 2007, Pub. L. No. 75-711, 52 Stat. 1040 (1938) as amended. . 2007.



8http://www.fda.gov/ScienceResearch/SpecialTopics/WomensHealthResearch/ucm134848.htm

9http://www.fda.gov/ScienceResearch/SpecialTopics/WomensHealthResearch/ucm134848.htm

10

Example:

North American Anti-Epileptic Drugs(AED) Pregnancy Registry


Methods - Study Design:

North American AED Pregnancy Registry

• Established in 1997 at Mass. General Hospital, Boston

• Enrolls pregnant women taking AEDs, and a reference group of friends

and family members not taking AED

• Women call

• Consent is obtained verbally for interview. Written consent obtained for

maternal and infant’s medical records

• Scientific Advisors supervise and approve release of findings

1-888-233-2334 (toll-free) http://www.AEDPregnancyRegistry.org


Methods - Data collection:

Three telephone interviews:• Enrollment (10-15 min.)

• 7 Months GA (5 min.)

• 2 Months postpartum (5 min.)

Ask about• AED use, demographic characteristics, epilepsy, vitamin use and the

presence of malformations

• For any AED data is collected on indication, dose, and start and stopdates


13

0

20

40

60

80

100

0 1 2 3 4 5 6 7 8 9 10

%

gestational months

Lunar months from last menstrual period to first interview

monthly

cumulative

Methods – Timing of first interview:


Methods – Timing of first interview:

Pure prospective: enrolled before having prenatal test

Traditional prospective: Participants who haveenrolled after having had an amniocentesis, chorionicvillus sampling test, nuchal translucency test, or anultrasound after 15 weeks gestation


Methods – Exposure definition:

To study teratogenicity: Use of specific AEDs anytime during the four lunar months after LMP

Classified according to number of AEDs

♦ Monotherapy

♦ Polytherapy


Methods – Outcome definition:

Major malformation diagnosed before 12 weeks• Overall

• Specific

Presence of malformations determined in interviews and

confirmed by medical records

Findings in infant reviewed by teratologist (L.Holmes)

blinded to exposure status


17

Objective



• Validity:

♦ Enrollment –Missing Data



Real examples


18

Timing of Enrollment

Depends on outcome(s) of interestConception (infertility) and contraception (OC failure).

Pregnancy loss (15% spontaneous abortions).

Health of the mother (preeclampsia).

Health of the offspring (preterm, birth defects). Preterm Delivery, 6-10%

Major Malformations, 2-4%

Long-term effects (neurodevelopment, late-diagnosis).


19


Because it depends on Relevant Timing ofExposure

Primary concern of drug use in pregnancy isteratogenesis.

Need to focus on etiologically relevant period: 1sttrimester.


21

Selection Bias

What would you expect to find if subjects are enrolled late in

pregnancy, after prenatal screening tests?

If diagnoses are included: potential overestimation

If diagnoses are excluded: potential underestimation



22

Selection Bias

What would you expect to find if exposed subjects are enrolled

during first trimester and unexposed are enrolled later in

pregnancy?



23

Selection Bias

Pregnancies

23

24

Pregnancies

Selection Bias

24

25

0

5

10

15

20

25

0 5 10

%

gestational month

Lunar month from last menstrual period to first interview

case

control

AED Registry – Timing of first interview:


26

Gestational age at enrollment

Margulis et al

Spontaneous abortion

27

All enrollees

RR

Unconditional logisticregression

5.1 (2.3;14.1)

Conditional on:

Trimester of enrollment 2.7 (1.2; 7.6)

Week 2.0 (0.9; 5.6)

Summary IRR 2.1 (0.9; 4.6)

SAB/N in AED users: 353/7,029nonusers: 6/581

Spontaneous abortion

28

All enrollees Pre-screening enrollees

RR RR


5.1 (2.3;14.1) 3.1 (1.4; 8.5)

Conditional on:

Trimester of enrollment 2.7 (1.2; 7.6) 2.5 (1.1; 7.1)

Week 2.0 (0.9; 5.6) 1.9 (0.8; 5.4)

Summary IRR 2.1 (0.9; 4.6) 2.0 (0.9; 4.4)

SAB/N in AED users: 353/7,029nonusers: 6/581

AED users: 349/4,304nonusers: 6/212

29

Gestational age at enrollment

30

All enrollees Pre-screening enrollees

RR RR


3.1 (1.4; 8.5) 2.1 (0.7; 10.2)

Conditional on:

Trimester of enrollment 3.2 (1.4; 9.0) 2.1 (0.7; 10.4)

Week 3.2 (1.4; 9.0) 2.0 (0.7; 10.1)

MCM/N in AED users: 210/6,676nonusers: 6/575

AED users: 117/3,955nonusers: 3/206

Major Malformations

Enrollment spanned through the entire pregnancy

• Different in antiepileptic drug users and nonusers

• Similar for users of different drugs

Analyses on SAB (early event) sensitive to gestational ageat enrollment

Analyses on MCM (often identified during prenatalscreening) sensitive to enrollment before/after prenatalscreening

31

Summary

Minimize left truncation (enroll early or detect earlypregnancies)

When the incidence rate of events changes over pregnancy(e.g., SAB), ensure that the distribution of time at risk for theoutcome is comparable between exposure groups

When an event may influence enrollment decisions (e.g.,prenatal screening), primary analyses must include onlysubjects who enrolled before this event

32

Conclusions

33

Selection Bias

Censoring of person-time “at risk”

• Start of follow up (left-censoring)

• End of follow up (right-censoring)

♦ Incidence versus prevalence at birth (birth defects)

♦ Intrauterine survival and elective abortions

• What would you expect to find in an analysis of teratogenicity ifexposure to drug XX causes early pregnancy losses?

34

Selection Bias

Pregnancies

34

35

Pregnancies

35

Selection Bias

36

Estimating the Proportion of all Observed Birth Defects Occurring inPregnancies Terminated by a Second-trimester AbortionElisabeth Svensson et al

■ 4% of all pregnancies ended in a second-trimester termination

■ The proportion of terminated pregnancies carrying birth defects (14%) is considerably greater than the corresponding proportion for pregnancies that endas live births or stillbirths (3%)

■ The proportion of birth defects unobserved at birth due to second-trimester terminations depends on type of defect and lethality:

Nervous system 48% Abdominal wall 39% Cardiac defects 5% Oral clefts 3% For many types of birth defects that proportion was less than 10%

37

Selection Bias Found RR=1

Assuming non-differential

miscarriage / termination for

social causes proportion of 20%

Correction for termination

among non-exposed from 10%

to 50%.

For differential termination

among exposed from 20% more

(selection -20%) to 20% less

(selection +20%)

38

Objective



• Validity:

♦ Enrollment



Real examples


39

Pregnancy registries often compare their estimates with

the background risk in “standard populations” :

• U.S. Centers for Disease Control and Prevention (CDC)

Metropolitan Atlanta Congenital Defects Program (MACDP):

frequency of malformations is 2.1% - 2.6% (first year of life)

• EUROCAT data in Europe

prevalence at birth is 2.0%

Reference: External vs. Internal


40

External comparison groups may be more efficient than

internal comparison groups:

• Large samples (stable estimates even for specific defects)

• Already collected

Are they valid enough?

Are risks comparable ?

Reference: External vs. internal


41

Example external comparison group:Prevalence of Birth Defects in Infants Born to Womentaking Antiretroviral Therapy in the first trimester


Birth Defect Prevalence after 1st Trimester ART Exposure

0.00%

1.00%

2.00%

3.00%

4.00%

5.00%

6.00%

7.00%

8.00%

9.00%

10.00%

APR(2

003)

NSHPC

APR(2

009)

NISDI

PACTG

316

Impaa

ct10

25

PACTG

219/2

19C

WIT

SECS

Cohort

Bir

thD

efe

ct

Pre

va

len

ce

(%)

1st tri Prevalence

LL

UL

42

Example internal comparison group:Prevalence of birth defects in ART exposed andunexposed infants by cohort


Birth Defects in ART Exposed vs. Unexposed Pregnancies

0.00%

1.00%

2.00%

3.00%

4.00%

5.00%

6.00%

7.00%

APR

(2003)

NSHPC APR

(2009)

NISDI PACTG

316

Impaact

1025

PACTG

219/219C

WITS ECS

Cohort

Bir

thD

efe

ct

Pre

vale

nce

(%)

Exposed

Unexposed

43

Why are these risks different?

Reasons for lack of comparability


44

Comparison Group

Comparable follow up time

Start of follow up

Last Menstrual Period (LMP) – Conception

After LMP (left-censoring) :

Intrauterine survival and elective abortions

Earlier follow up -> higher risks


45

Comparison Group

Comparable follow up time

End of follow up

Delivery or months after birth

Pregnancy losses without autopsy (right-censoring)

Complete and longer follow up -> higher risks


46

Comparison Group

Comparable outcome ascertainment

Sources of data: Maternal report, medical records,health care practitioner(s)

Diagnosis (ultrasounds and subclinical defects)


47

Comparison Group

Comparable outcome definition

Definition:

Period of diagnosis (prenatal, birth, life)

Exclusion / inclusion criteria

Classification:

Define etiologically relevant groups


48

Comparable populations (confounding)

SES, smoking, illegal drugs, BMI, genes…..

Indication (e.g. depression, asthma, epilepsy)

Reference with untreated condition (as severe as treated?)

Active comparison group treated with other drug:

Comparative effectiveness and safety research (CER)

Comparison Group


AED Registry – Reference groups:

1. Exposed (to other AED) internal reference group

• Lamotrigine, the most commonly reported AED.

2. Unexposed internal reference group

• Friends and family members not taking AED enrolled in the Registry

3. External reference group: Active Malformations SurveillanceProgram, BWH,1972 – 2000.

• 206,244 livebirths, stillbirths and elective terminations

• Malformations identified before 5 days of age

Nelson K, Holmes LB: N Engl J Med 320:19-23, 1989.Peller A et al: Obstet Gynecol 104 (5):957-964, 2004.


Risk of major malformations among infants exposed to aspecific AED in monotherapy during the first trimester

Pregnancy Registries 50

Unexposed lamotrigine carbamazepine valproate

n = 442** n = 1562 n = 1033 n = 323N (%) 5 (1.1%) 31 (2.0%) 31 (3.0%) 30 (9.3%)

(95%CI) (0.37 to 2.6%) (1.4 to 2.8%) (2.1 to 4.2%) (6.4 to 13.0%)

External reference

Relative risk 1.2 1.9 5.7

(95%CI) (0.9 to 1.7) (1.3 to 2.6) (4.1 to 8.1)

Major Congenital

Malformations*

* Diagnosed during pregnancy or before 12 weeks after birth. Confirmed by review of medical records.

**The unexposed internal comparison group were pregnant women, not taking an AED, who were

recruited from among the friends and family members of the enrolled women taking an AED.

*** Using the baseline prevalence rate of 1.62% in the unexposed comparison group. Active

Malformations Surveillance Program at Brigham and Women’s Hospital in Boston (n=69,277)

Hernández-Díaz S et al for the North American AED Pregnancy Registry.Neurology 2012;78:1692-1699

***

Risk of major malformations among infants exposed to aspecific AED in monotherapy during the first trimester


Unexposed lamotrigine carbamazepine valproate

n = 442** n = 1562 n = 1033 n = 323N (%) 5 (1.1%) 31 (2.0%) 31 (3.0%) 30 (9.3%)

(95%CI) (0.37 to 2.6%) (1.4 to 2.8%) (2.1 to 4.2%) (6.4 to 13.0%)

External reference

1.2 1.9 5.7

(0.9 to 1.7) (1.3 to 2.6) (4.1 to 8.1)

Unexposed internal reference

Relative risk reference 1.8 2.7 9.0

(95%CI) (0.7 to 4.6) (1.0 to 7.0) (3.4 to 23.3)

Exposed internal reference

Relative risk reference 1.5 5.1

(95%CI) (0.9 to 2.5) (3.0 to 8.5)

Major Congenital

Malformations*

* Diagnosed during pregnancy or before 12 weeks after birth. Confirmed by review of medical records.

**The unexposed internal comparison group were pregnant women, not taking an AED, who were

recruited from among the friends and family members of the enrolled women taking an AED.

*** Using the baseline prevalence rate of 1.62% in the unexposed comparison group. Active

Malformations Surveillance Program at Brigham and Women’s Hospital in Boston (n=69,277)

52

Enroll women as soon as possible after LMP, and before

outcome is known

Internal Reference Group with similar gestational age at

enrollment

• Active (exposed) reference group

External Comparison Groups if same follow up time can

be considered

• Internal followed 12 weeks, can be restricted to 5 days


AED- Comparable Follow up?

Exclusions AED Pregnancy Registry:

• Minor anomalies: preauricular sinus, simian crease• Birth marks: hemangioma, congenital mole• Prematurity related: undescended testes, PDA, PFO• Genetic disorders: Down syndrome, achondroplasia• Positional deformity: hip dysplasia with breech• Ultrasound only: unilateral renal agenesis• Technology induced: muscular VSDs, small ASD (<0.4 mm)

Approximately 17-27% of identified malformations excluded

Nelson K, Holmes LB: N Engl J Med 320:19-23, 1989

Ungar L et al: Birth Def Res (Part A): 76:373, 2006

Holmes LB et al: “Birth Def Res (Part A) 2011 (in press

21 October 2014 53Comparative Safety of AEDs

AED - Comparable Outcome ?

54

Internal reference group

Similar SES, smoking, illegal drugs, BMI, genes…..

Exposed internal reference♦ Similar indication

• Reduces confounding

• Clinically relevant comparison

Multi-drug registries (collaborations)

External reference if same baseline risk can be

assumed

AED - Comparable populations ?


55

Ideal reference group should have comparable:

• outcome definition (e.g., exclusion of minor anomalies)

• outcome assessment and timeframe for diagnosis (e.g.,

intensity of screening, frequency of terminations, inclusion

of prenatal diagnoses, gestational age at enrollment,

availability of diagnostic tests, start and stop of follow up)

• baseline risk for adverse pregnancy outcomes (e.g.

distribution of risk factors, including indication).

• methods for recruitment, enrollment, and data collection

Conclusion


56

Internal or

External Comparison Groups

• If same definition, ascertainment period, and

classification of malformations can be used

• To facilitate comparisons, studies could list criteria

and list all malformations

Conclusion


57

1. Is this drug another thalidomide or isotretinoin?

External reference• More efficient (cost and time)

• Selection, outcome misclassification and confoundingbiases probably not strong enough

2. Are there other, less frequent teratogenic risks?

Internal reference• More valid

• Limited power for specific malformations

What do we need to know for a given drug?


58

Objective



• Validity:

♦ Enrollment



Real examples


59

Power

Which are the outcomes of interest?

Teratogenic effects are often specific, i.e., a drugdoes not increase the risk of ALL malformations.

Need to consider specific birth defects. Major Malformations, 3 per 100

Specific Major Malformations, 1 per 1,000


60

Cohort StudiesFor a relatively common defect such as oral cleft

(baseline 1 per 1,000 births):

To identify arelative risk ofat least

Need to follow

Exposed Unexposed Total

20 150 300 450

12 300 600 900

5 1,500 3,000 4,500

Alpha=0.05 ; Beta=0.20; Reference group = 2x Exposed


61

Time trends AEDs in registry:


lamotrigine

carbamazepine

pheny

levetirac

valproic

topiramate

phenobarb0

10

20

30

40

50

60

1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

%

year

lamotrigine carbamazepine pheny levetirac valproic topiramate phenobarb

62

Sam

ple

size

and

conf

iden

cein

terv

als


0 200 400 600 800 1000 1200 1400

0 2 4 6 8 10 12 14

Hospital

Controls

lamotrigine

carbamazepine

levetiracetam

phenytoin

topiramate

valproate

phenobarbital

oxcarbazepine

gabapentin

zonisamide

clonazepam

Numberof Women Exposed

Prevalence MajorMalformations

63

Lamotrigine. Proportion of OralClefts and 95% confidenceinterval for every 100 infantsexposed.

Sample

size

Oral

celfts

Cumulative

risk (per 1000)

100 2 20.0

200 1 15.0

300 0 10.0400 1 10.0

500 0 8.0600 0 6.7700 1 7.1

800 0 6.3900 0 5.6

1000 0 5.01100 0 4.5

1200 0 4.21300 0 3.81400 1 4.3

1500 1 4.71600 0 4.4

1700 0 4.11800 0 3.9

63

Power considerations

64

Lamotrigine. Proportion of Oral Clefts and 95% confidenceinterval for every 100 infants exposed.


0

1

2

3

4

5

100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700

number of infants

Num

ber

of

malform

ations

0

1

2

3

4

5

Malf

orm

ati

on

s(%

)

Risk Low 95%CI upper 95%CI

64

65

When is it reasonable to conduct a registry ?


Conclusions

66

Considerations

Challenging recruitment

• Engage patient and provider populations

• Invest on recruitment and retention. Proactive enrollment strategies

Impossible recruitment

• Not enough patients to enroll♦ Rare diseases

♦ Contraindicated drug

• Alternatives♦ Global study

♦ Multidrug registries

♦ Other surveillance sources

67

Premarketing Approaches:

Pharmacologic or toxicologic studies - poor predictors

Animal studies - poor predictors

Clinical trials - exclude pregnant women

Alternative Sources


68

Postmarketing Approaches:

Case reports - clues; false alarms

Experimental studies - too small

Non-experimental (epidemiologic) studies

– Cohort

– Case-Control

Alternative Sources


69

Designed for Birth Defects Research

Wide range of exposuresCollaborative Perinatal Project

Specific exposures -- RegistriesTeratogen information services

Designed for Other Purposes

Automated claims databasesMedicaid Automated Database, HMOs

Computerized medical recordsGPRD in the UK

Cohort Studies


70

Prospective drug exposure information. No recall bias.

Concentrating on selected drugs can increase efficiency, particularlyto study uncommon drugs in relation to common events.

Longitudinal. Can estimate risk and risk differences.

Assess real use through maternal interview.

Can study multiple exposures and multiple outcomes


Advantages

71

Non-representative (self-referral bias, volunteers).

Often lack control group (non-comparable external comparison groups).

Inefficient (cost and time).

Short follow up (cannot assess late development).

Losses to follow-up (selective under-ascertainment).

Limited power, in particular for specific defects


Limitations

21 October 2014 72

To identify drugs with dramatic fetal risks (e.g., thalidomide,isotretinoins), pregnancy registries have appropriate efficiency andpower.

To identify drugs with intermediate fetal risks (e.g., carbamazepine,valproic acid), databases have appropriate efficiency and power ifuse is relative common.

To identify drugs with moderate fetal risks, case-control surveillancehas appropriate efficiency and power if use is relative common.

If outcome is rare, relative risk is modest, and prevalence of use islow… no currently known approach would have the power.

Conclusion

73

A Possible Approach (Allen Mitchell)

Cohortsn < 100

All new drugs

Yes

Possible

Phase I:Major teratogen?

Phase II:Moderate teratogen?

Case-control surveillancen = ongoing

No evidence of risk

Yes

Consider focusedcase-control study

Old drugs

Phase III:Lesser teratogen?

Cohortsn > 1000

No / not sure

Establish range of risks

No / not sure

Risk for specific defect(s)

Mitchell AA. Systematic Identification of Drugs That Cause Birth Defects-- A New Opportunity. New

England Journal of Medicine 2003;349:2556-2559.

pregnancy registries: methodological points & real examples

Documents