48826dft.docx
04/30/19
Postapproval
Pregnancy Safety
Studies
Guidance for Industry
DRAFT GUIDANCE
This guidance document is being distributed for comment purposes only.
Comments and suggestions regarding this draft document should be submitted within 60 days of
publication in the Federal Register of the notice announcing the availability of the draft
guidance. Submit electronic comments to https://www.regulations.gov. Submit written
comments to the Dockets Management Staff (HFA-305), Food and Drug Administration, 5630
Fishers Lane, Rm. 1061, Rockville, MD 20852. All comments should be identified with the
docket number listed in the notice of availability that publishes in the Federal Register.
For questions regarding this draft document, contact (CDER) Denise Johnson-Lyles at 301-796-
6169 or (CBER) the Office of Communication, Outreach, and Development at 800-835-4709 or
240-402-8010.
U.S. Department of Health and Human Services
Food and Drug Administration
Center for Drug Evaluation and Research (CDER)
Center for Biologics Evaluation and Research (CBER)
May 2019
Clinical/Medical
Postapproval
Pregnancy Safety
Studies
Guidance for Industry
Additional copies are available from:
Office of Communications, Division of Drug Information
Center for Drug Evaluation and Research
Food and Drug Administration
10001 New Hampshire Ave., Hillandale Bldg., 4th Floor
Silver Spring, MD 20993-0002
Phone: 855-543-3784 or 301-796-3400; Fax: 301-431-6353; Email: druginfo@fda.hhs.gov
https://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/default.htm
and/or
Office of Communication, Outreach, and Development
Center for Biologics Evaluation and Research
Food and Drug Administration
10903 New Hampshire Ave., Bldg. 71, rm. 3128
Silver Spring, MD 20993-0002
Phone: 800-835-4709 or 240-402-8010; Email: o[email protected].gov
https://www.fda.gov/BiologicsBloodVaccines/GuidanceComplianceRegulatoryInformation/Guidances/default.htm
U.S. Department of Health and Human Services
Food and Drug Administration
Center for Drug Evaluation and Research (CDER)
Center for Biologics Evaluation and Research (CBER)
May 2019
Clinical/Medical
TABLE OF CONTENTS
I. INTRODUCTION............................................................................................................. 1
II. BACKGROUND ............................................................................................................... 2
III. PHARMACOVIGILANCE — CASE REPORTS AND CASE SERIES .................... 4
IV. PREGNANCY REGISTRIES.......................................................................................... 5
A. Overview ......................................................................................................................................... 5
B. Registry Design Considerations .................................................................................................... 6
1. Objectives ......................................................................................................................................... 7
2. Study Population for Inclusion ........................................................................................................ 7
3. Outcome Definition(s) and Ascertainment ....................................................................................... 7
4. Sample Size and Statistical Power ................................................................................................... 8
5. Safety Evaluation When a Pregnancy Registry Is Not Feasible ...................................................... 8
6. Comparator Selection — Reference Group(s) ................................................................................. 9
7. Exposure Definition and Ascertainment .......................................................................................... 9
8. Covariates — Potential Confounders .............................................................................................. 9
9. Data Collection .............................................................................................................................. 10
10. Data Analysis and Presentation ................................................................................................ 10
11. Privacy and Human Subject Protection Issues ......................................................................... 11
12. Independent Data Monitoring Committee/Scientific Advisory Board ....................................... 11
13. Recruitment and Retention Plans .............................................................................................. 11
a. Recruitment ............................................................................................................................. 12
b. Retention ................................................................................................................................. 13
14. Multiproduct Pregnancy Registries ........................................................................................... 14
15. Pregnancy Registry Discontinuation ......................................................................................... 14
16. Lactation Study Added on to a Pregnancy Registry .................................................................. 14
V. COMPLEMENTARY STUDIES .................................................................................. 15
A. Electronic Data Sources .............................................................................................................. 15
1. Methods to Identify Pregnancies ................................................................................................... 16
2. Estimates of Conception and Gestational Age ............................................................................... 16
3. Linkages to Offspring ..................................................................................................................... 17
4. Study Outcome Ascertainment and Validation .............................................................................. 17
B. Population-Based Surveillance and National Registries or Registers ..................................... 18
C. Population-Based Case Control Studies .................................................................................... 19
1. Selection of Pregnancy-Related Cases and Controls .................................................................... 19
2. Exposure Assessment ..................................................................................................................... 20
3. Examples of Pregnancy Case-Control Studies in the United States .............................................. 20
REFERENCES ............................................................................................................................ 22
APPENDIX A: LIST OF DATA COLLECTION ELEMENTS ........................................... 26
Contains Nonbinding Recommendations
Draft — Not for Implementation
1
Postapproval Pregnancy Safety Studies 1
Guidance for Industry
1
2
3
4
5
6
This draft guidance, when finalized, will represent the current thinking of the Food and Drug 7
Administration (FDA or Agency) on this topic. It does not establish any rights for any person 8
and is not binding on FDA or the public. You can use an alternative approach if it satisfies the 9
requirements of the applicable statutes and regulations. To discuss an alternative approach, 10
contact the FDA staff responsible for this guidance as listed on the title page. 11
12
13
14
15
I. INTRODUCTION 16
17
The purpose of this guidance is to provide sponsors
2
and investigators with recommendations on 18
how to design investigations to assess the outcomes of pregnancies in women exposed to drugs 19
and biological products regulated by FDA (i.e., pregnancy safety studies). The goal of 20
postapproval pregnancy safety studies is to provide clinically relevant human safety data that can 21
inform health care providers treating or counseling patients who are pregnant or anticipating 22
pregnancy about the safety of drugs and biological products through inclusion of the information 23
in a product’s labeling. 24
25
In the years since FDA issued guidance on this topic, pregnancy safety studies required by FDA 26
have expanded beyond those using data from pregnancy exposure registries (pregnancy 27
registries)
3
to also include other types of epidemiologic studies and pregnancy surveillance 28
programs. This guidance should be used in conjunction with other epidemiological literature on 29
the design, conduct, and interpretation of observational studies. The development of pregnancy 30
safety studies requires specialized knowledge in a variety of areas, including expertise in the 31
fields of epidemiology, clinical teratology, obstetrics, pediatrics, clinical genetics, and statistics 32
when designing a study.
4
33
34
1
This guidance has been prepared by the Postapproval Pregnancy Safety Studies working group in the Center for
Drug Evaluation and Research in cooperation with the Center for Biologics Evaluation and Research and the Office
of Women’s Health in the Office of the Commissioner at the Food and Drug Administration.
2
For the purposes of this guidance, sponsors refer to persons or entities that conduct or fund studies for approved
products.
3
A pregnancy registry collects data that are then analyzed to address a safety question. For the purposes of this
guidance, pregnancy registry refers to both the data collection and the study that uses the data.
4
The previous guidance for industry Establishing Pregnancy Exposure Registries published August 23, 2002, has
been withdrawn.
Contains Nonbinding Recommendations
Draft — Not for Implementation
2
In general, FDA’s guidance documents do not establish legally enforceable responsibilities. 35
Instead, guidances describe the Agency’s current thinking on a topic and should be viewed only 36
as recommendations, unless specific regulatory or statutory requirements are cited. The use of 37
the word should in Agency guidances means that something is suggested or recommended, but 38
not required. 39
40
41
II. BACKGROUND 42
43
Pregnant women represent an important segment of the population, with over 6 million 44
pregnancies occurring per year, based on national vital statistics (Curtin et al. 2015). Pregnant 45
women may have chronic conditions, such as diabetes, seizure disorders, or asthma, that need to 46
be treated during pregnancy, or pregnant women may develop acute or serious medical 47
conditions during pregnancy that require treatment. In addition, nearly half of all pregnancies in 48
the United States may be unintended, which could result in potential inadvertent exposure to 49
drugs and biological products in pregnancy if a woman is exposed to a drug when she is not 50
aware she is pregnant (Finer 2016). Therefore, there is an important need for safety information 51
on product exposure during pregnancy. 52
53
During clinical development of most drugs and biological products, pregnant women are actively 54
excluded from trials, and if pregnancy does occur during a trial, the usual procedure is to 55
discontinue treatment and monitor the women to assess pregnancy outcomes. Consequently, at 56
the time of a drug or biological product’s initial marketing, except for drugs and biological 57
products developed to treat conditions unique to pregnancy, there are no or limited human data to 58
inform the safety of a drug or biological product taken during pregnancy. 59
60
Section 505(o)(3) of the Federal Food, Drug, and Cosmetic Act (FD&C Act) (21 U.S.C. 61
355(o)(3)), added by section 901 of the Food and Drug Administration Amendments Act of 2007 62
authorizes FDA to require certain postmarketing studies or clinical trials for prescription drugs 63
approved under section 505(b) of the FD&C Act and biological products approved under section 64
351 of the Public Health Service Act (42 U.S.C. 262). Under section 505(o)(3), FDA can require 65
such studies or trials at the time of approval to assess a known serious risk related to the use of 66
the drug, to assess a signal of serious risk related to the use of the drug, or to identify an 67
unexpected serious risk when available data indicates the potential for a serious risk. Under 68
section 505(o)(3), FDA can also require such studies or trials after approval if FDA becomes 69
aware of new safety information.
5
Postapproval studies using data collected in pregnancy 70
registries may be required to assess potential serious risks to the pregnancy that may affect the 71
health of the fetus or the woman due to drug or biological product use during pregnancy.
6
72
However, gaps in safety data in pregnant women still exist. 73
5
Defined at section 505-1(b)(3) of the FD&C Act. Also see the guidance for industry Postmarketing Studies and
Clinical Trials — Implementation of Section 505(o)(3) of the Federal Food, Drug, and Cosmetic Act (April 2011).
We update guidances periodically. For the most recent version of a guidance, check the FDA guidance web page at
https://www.fda.gov/RegulatoryInformation/Guidances/default.htm.
6
See the guidance for industry Postmarketing Studies and Clinical Trials — Implementation of Section 505(o)(3) of
the Federal Food, Drug, and Cosmetic Act.
Contains Nonbinding Recommendations
Draft — Not for Implementation
3
74
FDA held a 2-day public meeting in 2014 where stakeholders, including birth defect experts 75
from academia, industry, professional organizations, and patient groups, discussed the conduct of 76
pregnancy registries and epidemiologic studies using different study designs.
7
In addition, FDA 77
conducted reviews of pregnancy registries listed on the FDA’s Office of Women’s Health web 78
page (Gelperin et al. 2017). Based on FDA reviews and the 2014 public meeting, FDA 79
understands that pregnancy registry data have contributed to labeling changes and clinical 80
guidelines, but their potential has not been fully realized, often because of feasibility issues. 81
82
Pregnancy registries remain an important tool for safety data collection in the postmarketing 83
setting because of the prospective design and the ability to collect detailed patient level data. 84
However, because of the recurring challenges of achieving sufficient enrollment, pregnancy 85
registries generally are not sufficient by themselves to assess the safety of products during 86
pregnancy; therefore, other study methods capable of appropriately assessing the occurrence of 87
specific major congenital malformations (MCMs) (e.g., birth defects and congenital anomalies)
8
88
and other pregnancy outcomes are needed. In addition, use of complementary approaches may 89
help address the limitations inherent to a specific study design and provide greater confidence in 90
the conclusions. Input received from the 2014 public meeting and findings from FDA reviews 91
were used to develop this guidance. 92
93
The following sections describe three general approaches (pharmacovigilance, pregnancy 94
registries, and complementary data sources) that can be used in the postmarket setting to evaluate 95
drug or biological product safety during pregnancy. These approaches are not intended to imply 96
a hierarchy of evidence from the different study methods. Rather, each approach may uniquely 97
contribute to the overall safety assessment of a product during pregnancy. When considering 98
postmarketing approaches, the selection of any one or combination of these assessments and 99
timing of initiation may vary by drug or biological product. Consideration can be given to 100
experience with similar drugs and biological products, knowledge of the underlying disease and 101
its risks (maternal and fetal), potential use of the drug or biological product in females of 102
reproductive potential and pregnant women, existing knowledge of a safety concern, and the 103
potential for capturing the same pregnancy in two different assessments (double counting). 104
Moreover, evaluation of the strengths and limitations inherent to each type of assessment allows 105
FDA to recommend or require the appropriate method of postapproval risk assessment.
9
106
107
7
See transcripts from the FDA public meeting “Study Approaches and Methods to Evaluate the Safety of Drugs and
Biological Products During Pregnancy in the Post-Approval Setting,” May 28-29, 2014, at
https://www.fda.gov/Drugs/NewsEvents/ucm386560.htm.
8
For the purposes of this guidance, the following terms are used interchangeably: congenital malformations,
congenital anomalies, and birth defects, and are referred to as MCM throughout this guidance.
9
The authority to require a responsible person to conduct a postapproval study or studies or clinical trial(s) of a drug
under section 505(o)(3) includes the authority for FDA to set parameters for the study or trial to be conducted,
including how the study or trial is to be done and the population and indication. In other words, under section
505(o)(3), we can require a study or clinical trial that is well designed and adequate to address the serious safety
concern. Our current thinking on this and other matters is set forth in the guidance for industry Postmarketing
Studies and Clinical Trials — Implementation of Section 505(o)(3) of the Federal Food, Drug, and Cosmetic Act.
Contains Nonbinding Recommendations
Draft — Not for Implementation
4
108
III. PHARMACOVIGILANCE — CASE REPORTS AND CASE SERIES 109
110
Good pharmacovigilance practice involves the collection of comprehensive data on adverse 111
pregnancy outcomes to detect safety signals and develop a case series for analysis. Sources can 112
include spontaneous reports submitted to the sponsor and FDA, as well as case reports from the 113
medical literature or clinical studies. Well-documented and informative case reports can be used 114
to identify a signal, particularly if the pregnancy outcome is rare in the absence of drug exposure. 115
Safety signals generally indicate the need for further investigation, which may or may not lead to 116
the conclusion that the product caused the outcome or increased the risk of the outcome. The 117
importance of astute clinicians and clinical judgment in identifying a distinctive and unique 118
pattern of congenital malformations associated with a particular pregnancy exposure has been 119
critical in identifying teratogens (Shepard 1994; Obican and Scialli 2011). The quality of the 120
reports is critical for appropriate evaluation of the potential relationship between the product and 121
adverse outcomes. FDA recommends that sponsors make a reasonable effort to obtain complete 122
information for case assessment during initial contacts and subsequent follow-up. 123
124
Case reports are the most common source of reports of adverse pregnancy outcomes but can 125
often be challenging to interpret because information is often incomplete or there are additional 126
risk factors for the adverse pregnancy outcome, which case reports may not address. In addition, 127
one needs to consider the background rates of adverse pregnancy outcomes. Good case reports 128
include numerous important elements for conducting adequate pharmacovigilance. Specific 129
critical factors in evaluating the effects of product exposure in human pregnancies may include, 130
but are not limited to, the following:
10
131
132
• A detailed description of the adverse pregnancy outcome 133
134
• A detailed description of the exposure including the specific medication, the dose, 135
frequency, route of administration, and duration 136
137
• The timing of the exposure in relation to the gestational age 138
139
• The maternal age, medical and pregnancy history, and use of concomitant medications, 140
supplements, and other substances 141
142
• Exposures to known or suspected environmental teratogens 143
144
FDA has occasionally considered case reports and case series to be adequate data sources for 145
establishing a causal association for a human teratogenic exposure, such as with isotretinoin 146
(Centers for Disease Control and Prevention (CDC) 1984; Rosa 1983), or a serious adverse 147
event, such as oligohydramnios with trastuzumab (Zagouri et al. 2013). In general, such 148
evidence has been evaluated on a case-by-case basis. Case reports have been most useful and 149
influential in situations where the adverse pregnancy outcome rarely occurs as a background 150
10
See the guidance for industry Good Pharmacovigilance Practices and Pharmacoepidemiologic Assessment
(March 2005).
Contains Nonbinding Recommendations
Draft — Not for Implementation
5
event, and the adverse outcome is well-documented. A suspected safety signal arising from case 151
reports and case series that is not initially confirmed should be viewed as the start of an iterative 152
process, and not necessarily conclusive evidence of absence of risk. 153
154
Known limitations of spontaneous postmarketing reports (such as under-reporting, lack of a 155
denominator, and incomplete information) pose considerable challenges in analyzing cases and 156
determining whether a causal relationship exists between a product exposure and an adverse 157
pregnancy outcome.
11
Thus, routine pharmacovigilance usually will be insufficient for a 158
conclusive assessment regarding the potential risk of an exposure during pregnancy because of 159
the inability to quantify risk. Observational studies such as pregnancy registries and other 160
pharmacoepidemiological studies usually are needed to provide additional information including 161
a control group to derive and compare rates on safety outcomes of drugs and biological products 162
used during pregnancy. A sponsor should have a structured approach for pregnancy surveillance 163
with targeted questionnaires to obtain follow-up information on all potentially exposed 164
pregnancies of which the sponsor becomes aware, regardless of whether the pregnant woman 165
chooses to enroll in a registry. Pregnant women should be able to decline participation or 166
additional follow-up at any time at their discretion. 167
168
169
IV. PREGNANCY REGISTRIES 170
171
A. Overview 172
173
A pregnancy registry actively collects information on drug or biological product exposure during 174
pregnancy and associated pregnancy outcomes, which can be used to conduct a prospective 175
observational study (women are enrolled before the pregnancy outcome). Pregnancy registries 176
depend on the voluntary participation of women who have been exposed to a specific drug or 177
biological product during pregnancy and unexposed women who enroll into the comparator 178
cohort. Pregnancy registry data are prospectively collected by maternal interview and medical 179
record documentation and may include results of the clinical examination of the newborn. 180
Because of the prospective design of pregnancy registries, they may support assessment of 181
multiple maternal, obstetrical, fetal, and infant outcomes, including pregnancies that do not result 182
in a live birth. 183
184
A pregnancy registry may be U.S.-based or international in its scope. When submitting interim 185
and final pregnancy registry study reports, sponsors should include cumulative analyses of 186
worldwide pregnancy surveillance data to provide perspective on registry feasibility and updates 187
on available safety data in pregnant women that may not be included in the registry. 188
189
Pregnancy registries have the following strengths: 190
191
• By enrolling women exposed to the product of interest, pregnancy registries can be an 192
efficient way to collect data on the effects of rare exposures during pregnancy. 193
194
11
Ibid.
Contains Nonbinding Recommendations
Draft — Not for Implementation
6
• A pregnancy registry can be initiated and start to accrue real-time data as soon as a 195
product becomes commercially available, in contrast to the use of claims data and 196
electronic health records where there will be a lag time in data availability. 197
198
• Prospective enrollment facilitates ascertainment of an exposure of interest close to the 199
time it occurs and before information about the pregnancy outcome is known. 200
201
• Pregnancy registries have the potential to obtain accurate information about whether 202
exposure occurred and the timing of the exposure in relation to gestational age, dose, 203
frequency, and duration of the exposure, as well as covariates, and may therefore reduce 204
exposure misclassification, recall bias, and confounding. 205
206
• A pregnancy registry can potentially collect data on a variety of pregnancy and infant 207
outcomes, including postnatal outcomes. 208
209
• A pregnancy registry can be designed to include data from physical examination of the 210
newborn, and periodic clinical assessment of the offspring of exposed mothers, enabling 211
access to detailed clinical information about outcomes of interest, without relying on 212
International Classification of Diseases (ICD) codes. 213
214
Pregnancy registries have the following limitations: 215
216
• Analyses of collected data may have minimal statistical power to detect associations for 217
rare pregnancy outcomes. 218
219
• Most pregnancy registries are designed primarily to collect data used to assess the overall 220
risk of MCMs. Effects on less common, specific MCMs may be missed for all but the 221
most potent teratogens. 222
223
• Patient recruitment and retention are often challenging, and identification of an 224
appropriate comparator group may not always be feasible. 225
226
• Data from pregnancy registries generally are not sufficient by themselves to assess the 227
safety of products during pregnancy, and other study methods such as retrospective 228
cohort studies or case control studies may be needed to corroborate registry findings. 229
230
The ability of a pregnancy registry to provide safety data that can be used to inform product 231
labeling depends on factors such as the availability and quality of key clinical data and the 232
number of patients enrolled into the registry. Sponsors should address registry design 233
considerations (discussed below) in a written protocol and statistical analysis plan that include 234
considerations of study feasibility. 235
236
B. Registry Design Considerations 237
238
A well-written protocol for a pregnancy registry should describe its objectives, which may range 239
from open-ended safety surveillance to testing a specific hypothesis. The following issues 240
Contains Nonbinding Recommendations
Draft — Not for Implementation
7
should be addressed in the protocol to ensure consistency of data collection and analysis that will 241
provide scientifically valid results. 242
243
1. Objectives 244
245
The protocol should state the objectives of the registry for all study outcomes. An effective 246
pregnancy registry has the potential to serve as an early warning system to identify a previously 247
unrecognized major teratogen soon after market introduction by identifying MCMs in infants of 248
exposed mothers. For less potent teratogens or for drugs and biological products that cause other 249
adverse pregnancy outcomes, a pregnancy registry can function as a signal detection study and 250
generate hypotheses that can be tested using other methods that may be better powered to assess 251
specific birth defects or other abnormalities. 252
253
2. Study Population for Inclusion 254
255
Ideally, women in the exposed and unexposed cohort should be enrolled in a pregnancy registry 256
prospectively (i.e., before the conduct of any prenatal tests that could provide knowledge of the 257
outcome of pregnancy). If the condition of the fetus has already been assessed through prenatal 258
testing (e.g., targeted ultrasound, amniocentesis), such reports traditionally have been considered 259
retrospective. However, because it may be difficult to obtain enrollment before prenatal testing 260
on a consistent basis, the study population should include all women, including those who have 261
had early prenatal testing, and the protocol should address how pregnancies with prenatal testing 262
before enrollment will be evaluated in statistical analyses to avoid potential bias. 263
264
3. Outcome Definition(s) and Ascertainment 265
266
A pregnancy can result in live birth, miscarriage (loss before 20 weeks), elective termination, or 267
fetal death/stillbirth (loss after 20 weeks). Within each of these categories the fetus or infant can 268
be evaluated for the presence or absence of the primary outcome. As part of the study design, 269
the protocol should state a priori criteria for defining study outcomes. Criteria for defining birth 270
defects as major should be clearly stated. For example, MCMs might be defined as 271
“abnormalities in structural development that are medically or cosmetically significant, are 272
present at birth, and persist in postnatal life unless or until repaired.” Similarly, criteria should 273
be established for abnormalities that will be excluded from the definition of outcome (e.g., those 274
that are minor, transient, chromosomal abnormalities, genetic syndromes, positional defects, 275
prematurity related) (Holmes and Westgate 2011). A standardized classification system should 276
be used, as appropriate. An expert clinical geneticist or dysmorphologist should review and 277
classify medical records and reports of all MCMs. The clinical expert reviewer and method of 278
assessment should be the same for both the exposed and comparator group(s) and the reviewer 279
should be blinded to the exposure status. 280
281
Some examples of other outcomes that may be primary or secondary on a case-by-case basis 282
include: 283
284
• Measures of fetal growth deficiency (small for gestational age) 285
• Preterm delivery 286
Contains Nonbinding Recommendations
Draft — Not for Implementation
8
• Other pregnancy complications 287
• Developmental milestones or neurologic abnormalities in offspring of exposed mothers 288
• Abnormalities of immune system development in offspring of exposed mothers 289
290
4. Sample Size and Statistical Power 291
292
A written protocol for a pregnancy registry should include a statistical analysis plan with a 293
description of target sample size based on power calculations, and assessment of feasibility of 294
the study in the patient population of interest. When estimating the target sample size, it is 295
important to take into consideration the expected background rate of pregnancy loss, and cases 296
that may be lost to follow-up or otherwise unevaluable. Estimated rates based on the general 297
population may not apply to specific disease groups (e.g., diabetes). 298
299
Determination of an adequate sample size depends on the objective(s) and design of the registry 300
and the background rate of the outcome in the study population. If more than one pregnancy 301
outcome is considered, sample size determination should be based on the outcome with the 302
lowest background rate (e.g., MCMs). Consideration should be given to the prevalence of the 303
disease in females of reproductive potential and pregnant women and anticipated frequency of 304
product exposure in pregnant women. 305
306
No known teratogen increases the risk of all MCMs. Typically, a specific defect or pattern of 307
defects is associated with a specific teratogenic exposure during a critical period. Specific 308
MCMs occur rarely in the general population (i.e., fewer than 1 in 1,000 live births). 309
Historically, pregnancy registries have not had sufficient sample size or power to evaluate 310
increased risks for specific MCMs unless the relative risks are large (Gelperin 2017; Bird et al. 311
2018). Therefore, most registries compare the overall proportion of the total combined number 312
of various MCMs observed in the exposed group to the overall proportion in the comparator 313
group(s). Sponsors should include justification for the choice of expected background rates for 314
outcomes of interest in their proposed sample size and power calculations. 315
316
5. Safety Evaluation When a Pregnancy Registry Is Not Feasible 317
318
In some situations, a pregnancy registry may never have adequate power to allow statistical 319
inference. Achievement of an adequate sample size may not occur when the likelihood of 320
exposure in pregnancy is low, or use of a product is not recommended during pregnancy. 321
Anticipated issues with registry study feasibility should be stated in the protocol and 322
appropriately addressed, for example by expanding the inclusion criteria to include all reports of 323
exposed pregnancies (both prospective and retrospective). For products that are anticipated to be 324
used rarely during pregnancy (e.g., treatment of advanced cancer), sponsors can consider a 325
pregnancy surveillance program (a structured approach for data collection with targeted 326
questionnaires to obtain follow-up information on all exposed pregnancies of which sponsors 327
become aware). This type of case series of exposed pregnancies can inform clinical and 328
regulatory decision-making. Worldwide safety data collection is usually needed to identify a 329
sufficient number of exposed pregnancies for clinical safety assessment. 330
331
Contains Nonbinding Recommendations
Draft — Not for Implementation
9
6. Comparator Selection — Reference Group(s) 332
333
The strategy for selection of an appropriate comparison group(s) should be made when designing 334
the pregnancy registry and should be included in the protocol. Ideally, the registry should enroll 335
a concurrent internal comparison group of pregnant women unexposed to the evaluated 336
treatment. In addition, patients with the same disease (and disease severity, if feasible) should be 337
compared, because confounding due to the underlying condition may arise. Cohorts exposed to 338
different treatment regimens, when available, can serve as additional internal comparator groups 339
when evaluating a specific drug or biological product used as one treatment in a multiproduct or 340
disease-based registry study (for example, autoimmune diseases). 341
342
A background rate or the prevalence of congenital anomalies in a population-based surveillance 343
system (e.g., Metropolitan Atlanta Congenital Defects Program (MACDP))
12
or from another 344
pregnancy registry may be the only available comparator in certain situations. However, if 345
background rates or information from the external population-based surveillance system are 346
chosen as a comparison group, it is important to be aware of the limitations of whatever existing 347
system is used so that appropriate analyses can be designed, and results interpreted correctly. 348
For example, while MACDP prevalence data are well-documented and stable over time, they 349
have several characteristics that limit their validity as a comparator group for a pregnancy 350
registry. Limitations include the small geographic region from which the data are drawn 351
(metropolitan Atlanta); inclusion and exclusion criteria for outcomes of interest that differ from 352
the registries (particularly with regard to chromosome abnormalities); and the duration of 353
postnatal follow-up. Importantly, because external comparators typically estimate risk in the 354
general, mostly healthy, population, they may not be helpful to discern effects of the exposure of 355
interest and the underlying disease of the pregnant woman undergoing treatment, such as 356
diabetes or asthma. 357
358
When available and feasible, sponsors can consider use of external databases with data on 359
background rates in the disease population of interest to ensure comparability of groups. 360
Selection of an appropriate comparator is important because comparing dissimilar populations 361
could bias the study results, indicate a risk when none exists, or mask an increased risk that 362
exists. When feasible, selection of multiple comparator groups may be informative. 363
364
7. Exposure Definition and Ascertainment 365
366
Sponsors should collect detailed information on start and stop dates for all products taken during 367
pregnancy, as well as dose, frequency, duration, and indication. Exposure information in the 368
time period just before pregnancy is also often important, especially for products with a long 369
half-life. Accurate information about specific gestational timing of exposure(s) can help identify 370
critical exposure periods during gestation and biological plausibility for specific effects. 371
372
8. Covariates — Potential Confounders 373
374
Sponsors should consider the potential for confounding by indication, which makes it difficult to 375
determine whether any observed effects are caused by the drug or biological product or the 376
12
https://www.cdc.gov/ncbddd/birthdefects/macdp.html
Contains Nonbinding Recommendations
Draft — Not for Implementation
10
underlying disease. Data should be collected on the pregnant woman’s pertinent medical history, 377
current disease status, and overall management. Other potential confounders for which data 378
should be collected include, for example: socioeconomic status, maternal age, tobacco and 379
alcohol use, illegal drug use, maternal body mass index, folic acid and vitamin use during the 380
pregnancy, obstetrical history, medical history, family history of adverse pregnancy outcomes 381
including MCMs, and other relevant confounders (Caton 2012). 382
383
9. Data Collection 384
385
The value of the pregnancy registry depends on the accuracy and comprehensiveness of its data. 386
All data collection efforts should be identical among exposed and comparator study groups to 387
minimize bias. 388
389
The objective(s) of the registry should determine the type, extent, and length of patient follow-390
up. The feasibility of obtaining reliable pregnancy and infant outcome information is a critical 391
consideration in pregnancy registry design. Although prenatal health care providers are a good 392
source of information on outcomes, such as miscarriage, elective terminations, live births, and 393
pregnancy complications, they are not a good resource for information on infant conditions not 394
readily diagnosed at or soon after birth. The infant’s health care provider is the best resource for 395
full information on the health status of the infant after birth. The protocol should also specify 396
inclusion of pertinent findings from postmortem examination of pregnancies with nonlive birth 397
outcomes to avoid bias due to under-ascertainment of major malformations (Holmes and 398
Westgate 2011). 399
400
The protocol should include a plan and rationale for follow-up contacts during and/or after 401
pregnancy. The follow-up contact should obtain details on the pregnancy course, outcome, 402
status of the infant, and any evidence of abnormalities. 403
404
See Appendix A for a list of recommended data elements to include when designing a pregnancy 405
registry. 406
407
10. Data Analysis and Presentation 408
409
Validation of cases should be performed through medical record review and adjudication of 410
outcomes by a clinical dysmorphologist or appropriate specialist for both the exposed and 411
comparator group(s). 412
413
Inferential statistics should be applied to test prespecified hypotheses regarding the potential 414
association between the exposure and the outcome(s) of interest. 415
416
Potential biases should be discussed, as well as possible methods for mitigation, if applicable. 417
Descriptive statistics are the primary approach for summarizing patient characteristics and 418
additional data from a pregnancy registry. Given the heterogeneous nature of data obtained in 419
pregnancy registries, there is no one format for data presentation that is applicable for all studies. 420
The choice of a final format depends on outcomes identified in the registry protocol, 421
unanticipated findings, and expert advice. We encourage sponsors to develop forms of data 422
Contains Nonbinding Recommendations
Draft — Not for Implementation
11
presentation and analysis that fully capture outcomes of concern within their particular registry. 423
Separate analyses should be performed for each pregnancy outcome (miscarriage, elective 424
termination, fetal death/stillbirth, live birth) and stratified by gestational timing of exposure (with 425
a separate analysis of first trimester exposures for MCMs). Additional analytical approaches 426
should be used to assess covariates and factors that may affect the study findings, such as 427
gestational timing of enrollment (Margulis et al. 2015). 428
429
11. Privacy and Human Subject Protection Issues 430
431
Sponsors should consider privacy (including data protection) and human subject protection 432
(including obtaining informed consent and institutional review board (IRB) oversight) when 433
designing a pregnancy registry and developing protocols for the subsequent use of the data from 434
the registry. FDA recommends that an IRB be consulted when developing a pregnancy registry 435
to ensure that the collection of data and all other procedures associated with the registry will 436
withstand scientific and ethical scrutiny. 437
438
Because pregnancy registries typically do not involve the administration of an investigational 439
product, there is not likely to be any foreseeable risk or harm to the pregnant woman, fetus, or 440
resulting child from participating in the registry other than risk associated with inappropriate 441
disclosure of identifiable private information. The patient should be requested to sign medical 442
record release forms to allow collection of the records from the health care provider(s) of the 443
mother and infant. Investigators are responsible for ensuring that any data releases are compliant 444
with the Health Insurance Portability and Accountability Act and that all research performed 445
complies with standards of privacy of individually identifiable health information. 446
447
If the registry involves the collection of information on the child after birth, either through a 448
physical examination or specimen collection, considerations should be given to 21 CFR part 50, 449
subpart D, Additional Safeguards for Children in Clinical Investigations (for FDA-regulated 450
human subjects research). 451
452
12. Independent Data Monitoring Committee/Scientific Advisory Board 453
454
To ensure scientific integrity and appropriate patient protection, we encourage each registry to 455
have an independent data monitoring committee (or scientific advisory board) similar to those 456
used for clinical studies. Members of the committee could include experts in obstetrics, 457
embryology, teratology, pharmacology, epidemiology, pediatrics, clinical genetics, and any 458
relevant therapeutic areas. The committee could assist in the review of data, classification of 459
specific pregnancy outcomes including MCMs when relevant, and the dissemination of 460
information to ensure that results are interpreted and reported accurately. We recommend that 461
the role and duties of the committee or scientific advisory board be specified in the protocol. 462
463
13. Recruitment and Retention Plans 464
465
Successful recruitment and retention strategies are critical to the success of pregnancy studies 466
such as registries or other studies requiring enrollment of study subjects. We recommend a 467
robust recruitment and retention plan that includes a multipronged approach to ensure 468
Contains Nonbinding Recommendations
Draft — Not for Implementation
12
widespread coverage of the eligible population. Early enrollment also may improve detection of 469
pregnancy outcomes such as miscarriage. These plans should be flexible and continuously 470
reassessed throughout the study to ensure the registry maintains an adequate number of eligible 471
pregnant women in both the exposure and comparator groups. 472
473
a. Recruitment 474
475
Engaging health care providers and patients before the initiation of recruitment increases 476
awareness of the study and provides an opportunity to seek feedback from these stakeholders 477
regarding the study plan. We encourage sponsors to collaborate with entities such as existing 478
registries, patient advocacy groups, medical societies, and other relevant organizations to engage 479
in awareness activities. 480
481
Under the Pregnancy and Lactation Labeling Rule requirements, if there is a pregnancy registry 482
for the product, relevant contact information must be included in product labeling under the 483
subheading Pregnancy Exposure Registry.
13
Suggested modes of contact information include a 484
toll-free telephone number or a website’s uniform resource locator (URL). 485
486
The FDA’s Office of Women’s Health (OWH) maintains an online list of pregnancy registries 487
that are actively enrolling women to raise awareness about pregnancy registries and connect 488
consumers and health professionals to registries. The registries are posted to the FDA’s OWH 489
web page based on a sponsor’s or investigator’s request to list its registry. FDA encourages 490
sponsors and investigators to submit a pregnancy registry listing to OWH at 491
[email protected]. FDA does not endorse any registry and is not responsible for the 492
content of registries listed on the web page.
14
493
494
Recruitment strategies can be described as facility-based, health care provider-initiated, or 495
patient-initiated. 496
497
• Facility-based recruitment can occur at the level of a practice or health system. 498
Electronic health records can be used to identify drug or biological product users to 499
facilitate the enrollment process for providers. For example, an automated alert of a 500
pregnancy registry can be generated in response to positive pregnancy test results and/or 501
specific drug or biological product prescriptions. 502
503
• Health care provider-initiated recruitment of patients is an important deciding factor for 504
many pregnant women. Provider recruitment approaches include: 505
506
‒ Announcement of the registry study and contact information in the product labeling 507
‒ Promotional materials and product Internet pages 508
13
21 CFR 201.57(c)(9)(i)(A).
14
The Pregnancy Registries web page is located at
https://www.fda.gov/ScienceResearch/SpecialTopics/WomensHealthResearch/ucm251314.htm. The OWH mailbox
address and the web page URL may change. See the FDA website for the most recent information
(https://www.fda.gov/).
Contains Nonbinding Recommendations
Draft — Not for Implementation
13
‒ Announcements in professional journals and newsletters 509
‒ Personal mailings to specialists 510
‒ Presentations and exhibits at professional meetings 511
512
• Patient-initiated recruitment efforts rely on patients to contact the registry study staff and 513
self-enroll. Because pregnancy is often recognized by the patient first, registries that 514
enroll patients directly can allow for recruitment of patients earlier in pregnancy. Useful 515
avenues to notify pregnant women of pregnancy registries include: 516
517
‒ Print media including publications, press releases, and articles in newspapers and 518
magazines with pregnant women among their readership 519
520
‒ Distribution of flyers and posters in locations such as hospitals, ultrasound clinics, 521
laboratories, prenatal classes, community centers, stores, and coffee shops (Webster 522
et al. 2012) 523
524
‒ Social media 525
526
‒ Downloadable applications for mobile devices or personal computers could enable 527
broader participation through ease of providing information
15
528
529
Successful strategies to encourage the participation of pregnant women in medical research that 530
may be applicable to postapproval safety studies include: 531
532
• Incentives that facilitate study participation (Webb et al. 2010) 533
534
• Employing empathetic, culturally sensitive, and personable study staff (El-Khorazaty et 535
al. 2007). 536
537
b. Retention 538
539
Even though recruitment materials may yield strong initial recruitment results, we recommend 540
implementing a robust retention plan to ensure that an adequate number of pregnant women 541
remain in the registry. The retention plan should address specifics of patient retention strategies, 542
contingency plans to obtain follow-up information, methods to track follow-up rates over time, 543
and implementation steps to improve follow-up if expected follow-up rates are not met. 544
545
FDA also recommends that retention efforts focus on participating health care providers to 546
improve retention rates and reduce the burden of data collection (e.g., implementing streamlined 547
processes and succinct forms). Access to pregnancy registry results provides a strong incentive 548
for the participation of health care providers, particularly obstetric care providers, and the 549
provision of interim data reports to participating health care providers may bolster retention. 550
Additionally, high levels of retention have been achieved by pregnancy registries that 551
communicate directly with patients. Emphasizing the mission of the pregnancy registry may 552
15
See the FDA’s MyStudies Application (App) web page at
https://www.fda.gov/Drugs/ScienceResearch/ucm624785.htm.
Contains Nonbinding Recommendations
Draft — Not for Implementation
14
reinforce participants’ motivation to remain in the study. Sharing study results through a 553
newsletter or website has been found to be effective in reinforcing patients’ altruistic reasons for 554
participation. Establishing and maintaining a longitudinal relationship between participant and 555
interviewer can reduce loss to follow-up. As with other longitudinal studies, collecting contact 556
information of family members or friends in case the patient cannot be reached can aid in 557
retention. Recruitment and retention of pregnant women may be aided by a flexible follow-up 558
schedule (e.g., conducting follow-up interviews by telephone, during evening and weekend hours 559
or over a secure online platform), because participants may be balancing work and childcare 560
responsibilities. 561
562
14. Multiproduct Pregnancy Registries 563
564
To prevent overburdening patients, physicians, and health delivery systems with multiple 565
requests to participate in individual studies, we encourage sponsors to work together directly or 566
through consortiums to develop or support multiproduct registries. A multiproduct pregnancy 567
registry actively collects information on exposure to various product therapies in specific 568
diseases, such as human immunodeficiency virus or epilepsy (Hernández-Díaz et al. 2012). In 569
some cases, a general multiproduct registry, such as that conducted by a teratogen information 570
service, collects information on products for unrelated indications.
16
Multiproduct registries 571
have advantages over single-product registries with respect to efficiency and economy. They 572
also have the advantage of having comparison groups of pregnant women unexposed to the drug 573
or biological product of interest readily available (see section IV.B.6., Comparator Selection — 574
Reference Group(s)). 575
576
15. Pregnancy Registry Discontinuation 577
578
We recommend that a pregnancy registry be continued until one or more of the following occurs: 579
580
• Sufficient information has accumulated to meet the scientific objectives of the registry 581
582
• The feasibility of collecting sufficient information diminishes to unacceptable levels 583
because of low exposure rates, poor enrollment, or loss to follow-up 584
585
• Other methods of gathering appropriate information become achievable or are deemed 586
preferable 587
588
16. Lactation Study Added on to a Pregnancy Registry 589
590
There is also often a need to collect lactation data to provide information on the safety of drugs 591
and biological products during breast-feeding. Pregnancy registries can be used to recruit and 592
enroll breast-feeding women in lactation studies. Some women enrolled in a pregnancy registry 593
are already taking a drug or biological product during pregnancy, and because they may be likely 594
to continue treatment after delivery, these women are an ideal population in which to study 595
16
See the MotherToBaby pregnancy studies conducted by the Organization of Teratology Information Specialists
available at https://mothertobaby.org/pregnancy-studies/.
Contains Nonbinding Recommendations
Draft — Not for Implementation
15
product levels in milk. For information on how to conduct a lactation study, see the draft 596
guidance for industry Clinical Lactation Studies: Considerations for Study Design.
17
597
598
599
V. COMPLEMENTARY STUDIES 600
601
Use of complementary studies with different study designs may help address the limitations 602
inherent to a pregnancy registry. Additionally, as more postmarketing safety information 603
becomes available from interim registry reports, spontaneous reports, or case series, a more 604
specific safety signal may become apparent. Thus, additional studies that complement data 605
obtained from pregnancy registries and other sources, referred to as complementary studies in 606
this guidance, can be implemented as the need arises to better understand the specific effects of 607
using a drug or biological product during pregnancy, and to more precisely quantify the 608
magnitude of an association between a pregnancy exposure and a specific outcome. 609
610
Complementary studies can be retrospective in design, using secondary data (i.e., data collected 611
for purposes other than to assess the safety of one specific drug or biological product).
18
612
Common retrospective data sources and study designs used for complementary studies for 613
purposes of pregnancy-related research can include the following: 614
615
• Electronic data sources (e.g., insurance claims and electronic health record databases) 616
• Population-based surveillance and national registries or registers 617
• Population-based case control studies 618
619
These data sources and designs are discussed in the following subsections.
19
620
621
A. Electronic Data Sources 622
623
Electronic data sources often contain a large number of records available for research. At the 624
time of publication of this guidance, electronic data sources readily available for pregnancy 625
research include electronic administrative claims databases and/or electronic health record 626
(EHR) databases, referred to collectively as electronic health care data (EHD) in this guidance. 627
Best practices for studies using these data sources have been described in guidance
20
and also 628
apply to pregnancy studies using EHDs. 629
630
17
When final, this guidance will represent the FDA’s current thinking on this topic. For the most recent version of a
guidance, check the FDA guidance web page at https://www.fda.gov/RegulatoryInformation/Guidances/default.htm.
18
As the need arises, secondary data can be supplemented with additional data collection (e.g., maternal interview).
19
Methods used to identify and evaluate pregnancy outcomes in a pregnancy registry study described in section IV.,
Pregnancy Registries (e.g., study objective(s), outcome(s), comparators, exposure, confounders, statistical analysis
plan) also apply when considering complementary studies and will not be repeated in this section. This section
addresses concerns specific to the data sources selected for complementary studies.
20
See the guidance for industry and FDA staff Best Practices for Conducting and Reporting
Pharmacoepidemiologic Safety Studies Using Electronic Healthcare Data (May 2013).
Contains Nonbinding Recommendations
Draft — Not for Implementation
16
Regardless of the specific type of electronic data sources and study design used, investigators 631
should fully understand and describe the strengths and limitations of the data source proposed 632
(including the population(s) covered, data elements captured and their validity, system(s) of care, 633
and system-specific clinical and pharmacy data) to evaluate whether the data source is 634
appropriate to address specific pregnancy-related hypotheses. 635
636
Pregnancy and/or live birth data from EHD sources have been developed and used in a variety of 637
ways to evaluate product exposure and/or safety during pregnancy (Devine et al. 2010; Andrade 638
et al. 2012; Taylor et al. 2015; Huybrechts et al. 2014). Despite its successful and growing use, 639
selection of an EHD source to evaluate drug or biological product safety in pregnancy should 640
reflect consideration of methods used to identify pregnancies, estimates of conception and 641
gestational age, linkage to offspring records, and ascertainment and validation of pregnancy and 642
birth outcomes. Each of these considerations is discussed below. 643
644
1. Methods to Identify Pregnancies 645
646
The ability to identify clinically recognized pregnancies and births using EHD is central to use of 647
any database capable of assessing product safety during pregnancy. Identifying live births in an 648
EHD is relatively straightforward because delivery codes are available and relatively reliable. 649
650
Sponsors should consider the implications of limiting a study population to that of only live 651
births, because birth defects likely to result in non-live birth outcomes would not be captured. 652
Failure to include non-live births in a study population primarily affects study generalizability; 653
however, it also may result in a biased relative risk estimate if the rate of pregnancy loss or 654
termination caused by the defect is higher in one group than the other. 655
656
Use of EHD to identify non-live birth pregnancy outcomes for assessment of safety signals is 657
challenging. Non-live birth outcomes may be identified in EHD by the presence of diagnostic 658
and/or procedure codes specific to the outcome. However, gestational age at the time of the 659
outcome may be difficult to estimate if gestational age-specific codes accompanying the outcome 660
codes are unavailable or unreliable. Without a reasonable estimate of gestational age, a reliable 661
assessment of pregnancy exposure is difficult unless the investigator has access to ultrasound or 662
laboratory data. 663
664
2. Estimates of Conception and Gestational Age 665
666
A valid estimate of gestational age, from which a conception date may be estimated, is critical 667
for determining the timing of an exposure during pregnancy. For studies assessing pregnancy 668
outcomes among live births only, several methods exist for identifying gestational age. These 669
include: 670
671
• U.S. birth certificates (when available) 672
• Diagnostic ICD codes found in EHD databases
21
and algorithms using these codes 673
• EHR or ultrasound report 674
21
Given the potential variability in code validity by data source and outcome type (e.g., live birth versus stillbirth),
codes to identify gestational age should be validated in each database.
Contains Nonbinding Recommendations
Draft — Not for Implementation
17
675
3. Linkages to Offspring 676
677
Common methods for mother-infant linkages in the United States include linkages using birth 678
certificates and linkages using unique data elements within the same EHD source (Andrade et al. 679
2012). Linkages of pregnancies identified in EHD to offspring using birth or fetal death 680
certificates or other sources (e.g., medical records, national or state birth defect surveillance 681
registries) can provide the investigator access to several important variables that are not captured, 682
poorly captured, or captured with inadequate detail in EHD sources (e.g., maternal/paternal 683
race/ethnicity, maternal smoking status, parity, birth defects, some drug exposure, and precise 684
estimates of gestational age and birthweight of the newborn). 685
686
Even when only EHD sources are available, study data can be enhanced by linking those from 687
the mother to the offspring. Many EHD sources contain unique identifiers assigned to both the 688
mother and infant that may reflect the relationship to the primary health insurance policyholder. 689
Matching this number, as well as the mother’s delivery date, to the newborn’s date of birth often 690
successfully links the mother’s pregnancy to the infant’s health records. However, if the 691
newborn is covered under a different insurance policy than the mother, the linkage may be 692
impossible or at least limited to the clinical information available on the birth certificate or other 693
data sources. 694
695
In the United States, linkages of non-live birth outcomes identified in EHD sources to other data 696
sources are limited. Some states require reporting of fetal deaths (after 20 weeks), and this 697
information may be available to investigators on a case-by-case basis via the state’s vital records 698
department. Information collected by the state is often similar to that collected on a birth 699
certificate, but specific data elements vary by state. 700
701
4. Study Outcome Ascertainment and Validation 702
703
Diagnostic and procedure codes contained in EHD sources can be used to identify and study 704
product-associated MCMs. However, the presence of any single diagnostic code does not 705
necessarily imply a correct diagnosis. Diagnostic codes may reflect coding errors, rule-out 706
diagnoses, actual diagnoses, or the presence of an abnormality that has not yet been validated or 707
characterized. The validity of diagnostic codes for specific birth defects varies greatly by 708
specific defect and data source (Cooper et al. 2008; Palmsten et al. 2014). Outcome validation is 709
still needed for all outcomes unless a high-performing algorithm has been previously validated 710
for the specific outcome in the same (or similar) database under consideration. Some outcomes 711
can be ascertained in multiple ways. For instance, preterm birth and “small for gestational age” 712
may be identified through the presence of diagnostic codes or may be calculated using 713
gestational age and birth weight data found on the birth certificate and/or medical record. 714
Investigators should validate these outcomes in the specific database of interest if considering 715
their use as endpoints in EHD studies. 716
717
For all birth outcomes identified using EHD, sponsors should use a gold standard method of 718
validation such as a medical chart for the development of a testable algorithm. For MCMs, 719
sponsors should use reviews by clinical experts (geneticists or dysmorphologists) and/or linkage 720
Contains Nonbinding Recommendations
Draft — Not for Implementation
18
to birth defect registries and/or birth certificate data. The use of only EHD without access to 721
such gold standard sources or, at a minimum, a high-performing validated algorithm measuring 722
the same outcome in the specific database being considered may result in inaccuracy. 723
724
B. Population-Based Surveillance and National Registries or Registers
22
725
726
Population-based birth defect data sources are part of surveillance networks that extend to an 727
entire group of people having similar demographics (e.g., the entire nation in some European 728
countries), or to similar groups of people (e.g., state or regional births in the United States). The 729
advantage of using birth defect surveillance registries for MCM identification or validation is 730
that the identified MCM cases have already been adjudicated. Many of these registries capture 731
and adjudicate birth defect information for live births, stillbirths/fetal deaths, and elective 732
terminations. Some international birth defect registries follow guidelines developed by the 733
World Health Organization, in collaboration with the CDC and the International Clearinghouse 734
for Birth Defects Surveillance and Research (ICBDSR). Birth defect definitions in these 735
registries include MCMs associated with chromosomal abnormalities, which may not be 736
applicable to outcomes associated with drug or biological product exposures. 737
738
If maternal exposure information is collected, much of it is obtained from obstetrical records. If 739
sponsors consider population-based birth defect registries for exposure-based complementary 740
studies, they may need to supplement the registries with drug or biological product exposure 741
information from targeted maternal interviews and/or link to prescription information when 742
personal interviews are not possible.
23
743
744
Population-based birth defect registries have the substantial advantage of having large sample 745
sizes that allow the study of relatively rare MCMs. 746
747
Examples of population-based birth defect surveillance networks include: 748
749
• State-based Surveillance (United States) 750
751
• Vaccine and Medications in Pregnancy Surveillance System (VAMPSS) (United 752
States)
24
753
754
• The ICBDSR
25
755
756
22
For the purposes of this section, the term registry is used interchangeably with register (a term more commonly
used in Europe).
23
International population-based birth defect registries, usually European, can link to other databases to obtain drug
or biological product exposure and outcome information.
24
http://www.bu.edu/slone/research/studies/vampss/
25
http://www.icbdsr.org/resources/annual-report/
Contains Nonbinding Recommendations
Draft — Not for Implementation
19
• The European Registration of Congenital Anomalies and Twins, registries
26
757
758
Those that capture MCMs as a result of mandatory reporting allow for an accurate estimate of 759
incident birth defects in the network, especially when the numerator can be easily linked to the 760
number of pregnant women in the country or the region as the denominator during the study 761
period. 762
763
Regardless of the type of surveillance or registry selected for analysis, limiting observation only 764
to MCMs increases the risk of missing important toxic product effects that may be incompatible 765
with life or that may occur at different times during the pregnancy. Some registries, however, do 766
include stillbirths and elective terminations. Therefore, it is important to thoroughly understand 767
and describe what information is and is not available in the population-based registries 768
considered for a study, including what information is available on maternal drug or biological 769
product exposures. 770
771
C. Population-Based Case Control Studies 772
773
Case-control study designs (including nested designs) are frequently considered when there is a 774
need to collect additional information from the mothers through personal interviews, to obtain 775
additional information on infants, to request permission to review medical records, or to perform 776
long-term follow-up of the offspring. Case-control studies also may be needed if the registry is 777
unable to collect sufficient data to assess a safety signal previously identified from another data 778
source. 779
780
1. Selection of Pregnancy-Related Cases and Controls 781
782
Cases with pregnancy or infant outcomes of interest can be identified from EHD, or regional, 783
national, or international birth defect registries. The same concerns identified earlier in this 784
guidance for selection of controls or comparators for pregnancy registry studies (internal or 785
external controls) also apply to selection of controls or comparators for complementary case-786
control studies (see section IV.B.6., Comparator Selection — Reference Group(s)). For any 787
study, it is most important to ensure that comparators or controls are selected from the same 788
disease population (internal controls) when possible. Controls can be identified from the same 789
EHD or vital statistics departments or from general (state, regional, or national) birth records 790
giving rise to the cases; alternatively, birth outcomes (cases and controls) can be identified from 791
exposure- or disease-based registries. 792
793
When a case-control design is considered to evaluate a pregnancy outcome, regardless of the 794
source from which cases and controls were identified, sponsors should validate case or control 795
status using medical records or other reliable sources such as birth defect registries or review by 796
clinical experts. Documentation of validation should be provided when selecting cases from 797
these data sources. Case status identified from national or international networks are usually 798
already validated. 799
800
26
http://www.eurocat-network.eu/aboutus/datacollection/guidelinesforregistration/malformationcodingguides
Contains Nonbinding Recommendations
Draft — Not for Implementation
20
2. Exposure Assessment 801
802
The advantages of obtaining additional information by interviewing the mother as part of a case-803
control study include the ability to collect data on all types of drug or biological product 804
exposures, including those not covered by insurance (e.g., over-the-counter, supplements). An 805
additional strength is the ability to extend or adapt the interview to capture information not 806
available from other databases: personal or family history; race and other demographics; dose, 807
timing, and duration of product use; history of maternal disease or indication for medication; 808
comorbidities; and potential confounders such as body mass index, tobacco and alcohol use, 809
reproductive history, occupation (maternal and paternal), and the occurrence of breast-feeding. 810
At the interview, investigators can obtain informed consent to review medical records to confirm 811
diagnoses or to identify brand or lot, among others. If relevant, investigators can request 812
biological specimens (e.g., breast milk samples, buccal swabs for DNA testing) to test for 813
product penetrance or assess hereditary effects. Direct access to the mothers allows specialized 814
physical examinations and developmental follow-up of the offspring. 815
816
Exposure recall bias is always a concern for information obtained from maternal interviews, 817
because such self-reported data are collected after the pregnancy outcome (i.e., case status) is 818
known. Recall bias could be introduced if the accuracy of reported exposure is different between 819
cases and controls, for example mothers of birth defect cases may more accurately recall 820
exposures during pregnancy versus mothers of unaffected infants. Attempts to minimize this 821
bias could include selecting as controls mothers with other adverse pregnancy outcomes (e.g., 822
malformed infants with chromosomal defects or with malformations other than the one(s) of 823
interest) or other serious medical problems. Another approach to minimize recall bias is the use 824
of pharmacy records among cases and controls to confirm reported drug or biological product 825
exposures, when available, although pharmacy data only provide information on prescription 826
fills and not necessarily on quantity consumed and may not include over-the-counter products. 827
828
3. Examples of Pregnancy Case-Control Studies in the United States 829
830
Examples of case-control studies are listed below and can be used as a starting point for 831
designing a study. Note, however, that data from these studies, although population-based, are 832
only specific to the populations studied and may not be relevant to the study population under 833
consideration. If comparisons are to be made to these studies, every effort should be made to 834
understand and explain the similarities and differences and to identify resulting confounding and 835
biases. 836
837
• The National Birth Defects Prevention Study
27
838
839
• Birth Defects Study to Evaluate Pregnancy exposures
28
840
841
27
http://nbdps.org/
28
http://www.bdsteps.org/
Contains Nonbinding Recommendations
Draft — Not for Implementation
21
• Pregnancy Health Interview Study (Birth Defects Study), a multicenter case-control study 842
based at the Slone Epidemiology Center at Boston University, a collaborator of the 843
VAMPSS
29
844
845
29
http://www.bu.edu/slone/research/studies/phis/
Contains Nonbinding Recommendations
Draft — Not for Implementation
22
REFERENCES 846
847
Literature 848
849
Andrade, SE, Davis RL, Cheetham TC, Cooper WO, Li DK, Amini T, Beaton SJ, Dublin S, 850
Hammad TA, Pawloski PA, Raebel MA, Smith DH, Staffa JA, Toh S, Dashevsky I, Haffenreffer 851
K, Lane K, Platt R, and Scott PE, 2012, Medication Exposure in Pregnancy Risk Evaluation 852
Program, Matern Child Health J, 16(7):1349–1354. 853
854
Andrade, SE, Scott PE, Davis RL, Li DK, Getahun D, Cheetham TC, Raebel MA, Toh S, Dublin 855
S, Pawloski PA, Hammad TA, Beaton SJ, Smith DH, Dashevsky I, Haffenreffer K, and Cooper 856
WO, 2013, Validity of Health Plan and Birth Certificate Data for Pregnancy Research, 857
Pharmacoepidemiol Drug Saf, 22(1):7–15. 858
859
Bird, S, Gelperin K, Taylor L, Sahin L, Hammad H, Andrade SE, Mohamoud MA, Toh S, and 860
Hampp C, 2018, Enrollment and Retention in 34 United States Pregnancy Registries Contrasted 861
With the Manufacturer’s Capture of Spontaneous Reports for Exposed Pregnancies, Drug Safety, 862
41:87–94. 863
864
Caton, AR, 2012, Exploring the Seasonality of Birth Defects in the New York State Congenital 865
Malformations Registry, Birth Defects Research (Part A), 94(6):424–437. 866
867
CDC, 1984, Isotretinoin — A Newly Recognized Human Teratogen, Morb Mortal Wkly Rep., 868
Apr 6, 33(13):171–173. 869
870
Cooper, WO, Hernández-Díaz S, Gideon P, Dyer SM, Hall K, Dudley J, Cevasco M, Thompson 871
AB, and Ray WA, 2008, Positive Predictive Value of Computerized Records for Major 872
Congenital Malformations, Pharmacoepidemiol Drug Saf, 17(5):455–460. 873
874
Council for International Organizations of Medical Sciences, 2010, Practical Aspects of Signal 875
Detection in Pharmacovigilance: Report of CIOMS Working Group VIII. 876
877
Curtin, SC, Abma JC, and Kost K, 2015, 2010 Pregnancy Rates Among U.S. Women, NCHS 878
health e-stat. 879
880
Devine, S, West S, Andrews E, Tennis P, Hammad TA, Eaton S, Thorp J, and Olshan A, 2010, 881
The Identification of Pregnancies Within the General Practice Research Database, 882
Pharmacoepidemiol Drug Saf, 19(1):45–50. 883
884
El-Khorazaty, MN, Johnson AA, Kiely M, El-Mohandes AA, Subramanian S, Laryea HA, 885
Murray KB, Thornberry JS, and Joseph JG, 2007, Recruitment and Retention of Low-Income 886
Minority Women in a Behavioral Intervention to Reduce Smoking, Depression, and Intimate 887
Partner Violence During Pregnancy, BMC Public Health, 7:233. 888
889
Finer, Zolna, 2016, Declines in Unintended Pregnancy in the United States, 2008-2011, N Engl J 890
Med, 374(9):843–852. 891
Contains Nonbinding Recommendations
Draft — Not for Implementation
23
892
Frew, PM, Saint-Victor DS, Brewinski Isaacs M, Kim S, Swamy GK, Sheffield JS, Edwards 893
KM, Villafana T, Kamagate O, and Ault K, 2014, Recruitment and Retention of Pregnant 894
Women Into Clinical Research Trials: An Overview of Challenges, Facilitators, and Best 895
Practices, Clin Infect Dis, 59 Suppl 7:S400–S407. 896
897
Gelperin, K, Hammad H, Leishear K, Bird ST, Taylor L, Hampp C, and Sahin L, 2017, A 898
Systematic Review of Pregnancy Exposure Registries: Examination of Protocol-Specified 899
Pregnancy Outcomes, Target Sample Size, and Comparator Selection, Pharmacoepidemiol Drug 900
Saf, 26:208–214. 901
902
Gliklich, R, Dreyer N, and Leavy M, editors, 2014, Registries for Evaluating Patient Outcomes: 903
A User’s Guide, third edition, two volumes. (Prepared by the Outcome DEcIDE Center 904
(Outcome Sciences, Inc., a Quintiles company) under Contract No. 290 2005 00351 TO7.) 905
AHRQ Publication No. 13(14)-EHC111, Rockville, MD: Agency for Healthcare Research and 906
Quality, April 2014. 907
908
Hernández-Díaz, S, Smith CR, Shen A, Mittendorf R, Hauser WA, Yerby M, and Holmes LB, 909
2012, Comparative Safety of Antiepileptic Drugs During Pregnancy, Neurology, May 22, 910
78(21):1692–1699. 911
912
Holmes, LB, Mittendorf R, Shen A, Smith CR, and Hernández-Díaz S, 2011, Fetal Effects of 913
Anticonvulsant Polytherapies: Different Risks From Different Drug Combinations, Arch Neurol, 914
68(10):1275–1281. 915
916
Holmes, LB and Westgate MN, 2011, Inclusion and Exclusion Criteria for Malformations in 917
Newborn Infants Exposed to Potential Teratogens, Birth Defects Research (Part A), 91:807–812. 918
919
Huybrechts, K, Bateman BT, Palmsten K, Desai RJ, Patorno E, Gopalakrishnan C, Levin R, 920
Mogun H, and Hernández-Díaz S, 2015, Antidepressant Use Late in Pregnancy and Risk of 921
Persistent Pulmonary Hypertension of the Newborn, JAMA, Jun 2, 313(21):2142–2151. 922
923
Huybrechts, KF, Palmsten K, Avorn J, Cohen LS, Holmes LB, Franklin JM, Mogun H, Levin R, 924
Kowal M, Setoguchi S, and Hernández-Díaz S, 2014, Antidepressant Use in Pregnancy and the 925
Risk of Cardiac Defects, N Engl J Med, 370(25):2397–2407. 926
927
Margulis, AV, Mittleman MA, Glynn RJ, Holmes LB, and Hernández-Díaz S, 2015, Effects of 928
Gestational Age at Enrollment in Pregnancy Exposure Registries, Pharmacoepidemiol Drug Saf, 929
Apr, 24(4):343–352. 930
931
National Birth Defects Prevention Network, 2004, Guidelines for Conducting Birth Defects 932
Surveillance, Sever LE editor. 933
934
Obican, S and Scialli AR, 2011, Teratogenic Exposures, American Journal of Medical Genetics 935
Part C (Seminars in Medical Genetics), 157:15–169. 936
937
Contains Nonbinding Recommendations
Draft — Not for Implementation
24
Palmsten, K, Huybrechts KF, Kowal MK, Mogun H, and Hernández-Díaz S, 2014, Validity of 938
Maternal and Infant Outcomes Within Nationwide Medicaid Data, Pharmacoepidemiol Drug Saf, 939
23(6):646–655. 940
941
Rosa, FW, 1983, Teratogenicity of Isotretinoin, Lancet, 2(8348):513. 942
943
Shephard, T, 1994, Proof of Teratogenicity, Teratology, 50:97–98. 944
945
Taylor, LG, Thelus Jean R, Gordon G, Fram D, and Coster T, 2015, Development of a Mother-946
Child Database for Drug Exposure and Adverse Event Detection in the Military Health System, 947
Pharmacoepidemiol Drug Saf, 24(5):510–517. 948
949
Webb, DA, Coyne JC, Goldenberg RL, Hogan VK, Elo IT, Bloch JR, Mathew L, Bennett IM, 950
Dennis EF, and Culhane JF, 2010, Recruitment and Retention of Women in a Large Randomized 951
Control Trial to Reduce Repeat Preterm Births: The Philadelphia Collaborative Preterm 952
Prevention Project, BMC Med Res Methodol, 10:88. 953
954
Webster, GM, Teschke K, and Janssen PA, 2012, Recruitment of Healthy First-Trimester 955
Pregnant Women: Lessons From the Chemicals, Health & Pregnancy Study (CHirP), Matern 956
Child Health J, 16(2):430–438. 957
958
WHO, CDC, and ICBDSR, 2014, Birth Defects Surveillance: A Manual for Programme 959
Managers, Geneva, World Health Organization. 960
961
Zagouri, F, Sergentanis TN, Chrysikos D, Papadimitriou CA, Dimopoulos MA, and Bartsch R, 962
2013, Trastuzumab Administration During Pregnancy: A Systematic Review and Meta-963
Analysis, Breast Cancer Res Treat, 137(2):349–357. 964
965
Guidances
1
966
967
Draft guidance for industry Clinical Lactation Studies: Considerations for Study Design
2
968
969
Draft guidance for industry Postmarketing Safety Reporting for Human Drug and Biological 970
Products Including Vaccines
3
971
972
Guidance for industry Good Pharmacovigilance Practices and Pharmacoepidemiologic 973
Assessment 974
975
Guidance for industry Postmarketing Studies and Clinical Trials — Implementation of Section 976
505(o)(3) of the Federal Food, Drug, and Cosmetic Act 977
1
We update guidances periodically. To make sure you have the most recent version of a guidance, check the FDA
guidance web page at https://www.fda.gov/RegulatoryInformation/Guidances/default.htm.
2
When final, this guidance will represent the FDA’s current thinking on this topic. For the most recent version of a
guidance, check the FDA guidance web page at https://www.fda.gov/RegulatoryInformation/Guidances/default.htm.
3
When final, this guidance will represent the FDA’s current thinking on this topic.
Contains Nonbinding Recommendations
Draft — Not for Implementation
25
978
Guidance for industry and FDA staff Best Practices for Conducting and Reporting 979
Pharmacoepidemiologic Safety Studies Using Electronic Healthcare Data 980
981
Contains Nonbinding Recommendations
Draft — Not for Implementation
26
APPENDIX A: 982
LIST OF DATA COLLECTION ELEMENTS 983
984
The following data elements should be included when designing a pregnancy registry. 985
986
General 987
988
Patient identifier 989
Name of reporter at initial contact with the registry 990
Date of initial contact with the registry 991
Dates of any follow-up contacts 992
Telephone number and email address of reporter 993
Additional contact names, telephone numbers, and email addresses (if reporter is the patient) 994
995
Maternal Information 996
997
Source of information (e.g., obstetrician, pregnant woman) 998
Birth date 999
Race 1000
Occupation 1001
Height, weight, body mass index 1002
Maternal medical history (e.g., hypertension, diabetes, seizure disorder, autoimmune 1003
disease, known risk factors for adverse pregnancy outcomes including environmental or 1004
occupational exposures) 1005
Obstetrical history: 1006
Number of pregnancies and outcome of each (live birth, miscarriage, pregnancy 1007
termination (elective or therapeutic), ectopic pregnancy) 1008
Previous maternal pregnancy complications 1009
Previous fetal/neonatal abnormalities and type 1010
Current pregnancy: 1011
Date of last menstrual period 1012
Ultrasound results for gestational dating 1013
Prenatal test results (including dates) 1014
Pregnancy weight gain of mother 1015
Obstetric complications (e.g., preeclampsia, premature delivery) 1016
Complications during pregnancy (including any adverse product reactions) and dates 1017
Number of fetuses 1018
Disease course(s) during pregnancy and any complications 1019
Drug or biological product exposures (prescription drugs, over-the-counter products, and 1020
dietary supplements): 1021
Name 1022
Dosage and route 1023
Date of first use and duration 1024
Indication 1025
Recreational drug use (e.g., tobacco, alcohol, illicit drugs) and amount 1026
Contains Nonbinding Recommendations
Draft — Not for Implementation
27
Family history (specify type, maternal or paternal, among others): 1027
Malformations 1028
Genetic disorders 1029
Multiple fetuses/births 1030
1031
Neonatal Information 1032
1033
Initial: 1034
Source of information (e.g., obstetrician, pediatrician, mother) 1035
Date of receipt of information 1036
Date of birth or termination 1037
Gestational age at birth or termination 1038
Gestational outcome (live born, fetal death/stillborn, miscarriage, elective termination, and 1039
termination for a fetal anomaly) 1040
Sex 1041
Obstetric complications (e.g., preeclampsia, premature delivery) 1042
Pregnancy order (singleton, twin, triplet) 1043
Results of neonatal physical examination including 1044
Anomalies diagnosed at birth or termination (including autopsy results) 1045
Anomalies diagnosed after birth 1046
Weight at birth indicating whether small, appropriate, or large for gestational age 1047
Length at birth 1048
Head circumference at birth indicating whether small, appropriate, or large for gestational 1049
age 1050
Condition at birth (including, when available, Apgar scores at 1 and 5 minutes, umbilical 1051
cord vessels and gases, need for resuscitation, admission to intensive care nursery) 1052
Neonatal illnesses, hospitalizations, drug therapies 1053
1054
Follow-up: 1055
Source of information (e.g., pediatrician, mother) 1056
Date of receipt of information 1057
Anomalies diagnosed since initial report 1058
Developmental assessment 1059
Infant illnesses, hospitalizations, drug therapies 1060
1061
