Phase 4 Trials (Studies) of Hormonal Contraception

Potential for Improving Information on Safety and

Effectivenessafter Marketing Approval

Diana Petitti, MDAdjunct Professor

Department of Preventive MedicineUniversity of Southern California

Keck School of Medicine

Advisory Committee for Reproductive HealthDrugs, FDA, January 23/24, 2007

Phase 4 The period after a drug is approved for marketing

Hormonal contraceptives are approved for marketing based on very small number of women from the point of view of assessing safety and relative safety

The populations included in studies done to obtain marketing approval are not representative of the women who will use the drugs in real life Age Smoking Medical history Motivation to contracept Education

Phase 4 The period after a drug is approved for marketing

Little is known about the effectiveness of use of hormonal contraceptives in real-life settings and in populations different from the ones that participate in clinical trials

Since the overwhelmingly most common reason for using contraceptives is to prevent pregnancy, this deficiency in post-marketing information is noteworthy

Phase 4 Specific issues in study of safety of hormonal contraceptives

We know a lot about what to expect from hormonal contraceptives based on the vast amount of research done over the last 5 decades Vascular events are the most important major

adverse event caused by combined estrogen/progestin contraceptives (ischemic stroke, venous thromboembolism, myocardial infarction)

Phase 4 Specific issues in study of safety of hormonal contraceptives (continued)

Vascular events These are rare (xx per 100,000) There are interactions (risk in women with

certain characteristics puts them at particularly high risk)Hypertension for strokeObesity for venous thromboembolismSmoking for myocardial infarction

Phase 4 Specific issues in study of safety of hormonal contraceptives (continued)

Vascular events We do not understand the pathophysiology of

vascular events caused by combined estrogen/progestin hormonal contraceptives and thus cannot predict whether or in what direction a change will affect these events

The inability to predict extends to changes in estrogen dose, estrogen type, progestin dose, progestin type, route of administration, cumulative dose, maximum dose, etc., etc., etc.

Phase 4 SurveillanceUnsystematic Activities

“Surveillance” assessing spontaneous reports of adverse effects to the FDA

Waiting for the “astute clinician” Waiting for an interested researcher

responding to report of an adverse effect from the FDA or from the astute clinician

Phase 4 SurveillanceUnsystematic Activities

Continuation of the unsystematic approach invites trouble false alarms based on faulty data

Waiting for “alarms” invites panic in response Alarms, whether ultimately false or true,

undermine the public’s confidence in the regulatory system and in the industry

Doing nothing in hopes there will be no alarms flies in the face of experience with hormonal contraception

Phase 4 Specific issues in study of the effectiveness of hormonal contraceptives

Old products are as poorly studied as new ones There may be population trends that would

affect use-effectiveness (obesity) both in old and in new products

Planned Phase 4 Studies

Phase 4 Studies

DESIGN Experimental Case-control Cohort Case-control nested within a cohort

Phase 4 RCTs

IDEAL IN THEORY BUT

Extremely costly if large enough to address vascular event differences

Difficult to choose appropriate comparator No such thing as a simple randomized trial

(???)

Phase 4 Studies

DESIGN Case-control Cohort Case-control nested within a cohort

Phase 4 Studies

DESIGN Case-control studies as a stand-alone design

are used most often to study exposures that occurred in the distant past for which exposure information cannot be reliably retrieved from records or computer-stored information sources

Phase 4 Studies

DESIGN The main disadvantage of the classical case-

control design that it is subject to recall bias—the tendency of the diseased to selectively over-report or under-report past exposure

Phase 4 Studies: Hybrid

DESIGN The increasing availability of computer

stored information on drug exposures makes it possible to combine the best features of cohort and case-control designs

Phase 4 StudiesCohort

DATA Prospective cohort with direct enrollment

and active follow-up Computer stored only Computer and physicians/records Computer and physicians/records plus

direct patient contact

Prospective Cohort with Direct Enrollmentand Active Follow-up(A New “Vessey” Study)

Advantages Comprehensive

information Good data on

confounders Ability to include

diverse populations

Disadvantages Costly Time to results is

long Power for rare

events is low

Phase 4 Cohort Studies

DATA Computer stored only Computer and physicians/records Computer and physicians/records plus

direct patient contact

Using Computer-Stored Data Only

Advantages Cheap Potential to yield

information quickly

Disadvantages Events cannot be

confirmed Patients are poorly

characterized Information on

important confounders is poor if it exists at all

Difficult to study effectiveness

Supplementing Computer Stored Data With Physicians/Records but Not Patient Contact

Advantages Events can be confirmed

or disconfirmed using standard criteria

Information is available on some confounders and effect-modifiers lacking in computer data

Disadvantages Lack (or inconsistent

availability) of information on some important confounders (family history; past medical history)

Uncertain accuracy of information on confounders

Difficult to study effectiveness

Supplementing Computer Stored Data and Physicians/Records With Direct Patient Contact

Advantages Potential to

characterize patients well

Good information on confounders and effect-modifiers

Accurate and complete information potentially available on effectiveness

Disadvantages Expensive Time to information

is long Subject to response

bias

Phase 4 Studies

ORGANIZATION Single source (e.g., on Health Plan) Multiple sources

Single Source of Data

Advantages Efficient Time to

information potentially short

Greater ability to trust study planners assertions

Disadvantages Non-

representative Can have

limitations due to formulary

Phase 4 Studies

COMPARATOR Historical Active

Comparison for New Contraceptive:Historical Compared with Active

Advantages Cheaper

Disadvantages Cannot be relied

on the yield the correct answer about anything

Recommended Design for Phase 4 Study of New Contraceptive Safety: Hybrid

Computer-based prospective cohort New product compared with old products

newly initiated Data from multiple sources

• Increases diversity of population of users• Assures mix of hormonal contraceptives

Confirmation using physician/hospital records and experts working based on specified criteria blinded to use

Recommended Design for Phase 4 Study of New Contraceptive

Collection of information on confounders by direct patient contact using a nested unmatched case-control design with oversampling of controls by a large fraction • Nesting decreases cost• Lack of matching and oversampling make it

possible to post-stratify

Case-Control Study with Patient Contact Nested in a Computer-assembled Cohort Derived from Multiple Sources

Advantages Cheaper than full

cohort contact Ability to focus on

maximizing response rates

Disadvantages Can’t be used to

study effectiveness

Some Final Comments

It is very difficult to mount a study that will definitively show the equivalence of a new product compared with something else because the events of greatest interest—vascular events--are rare and the power of feasible study designs is low except unless there are large differences

The ability of any feasible study with a cohort design and patient data collection to assess whether there are interactions (effect modification) between the new product and one or another of the factors known to increase the risk of vascular disease, such as obesity, smoking, family history) is very, very limited

Because the main reason for using these products in pregnancy prevention, more information about the comparative effects of the new products with old ones is very important

Some Final Comments

Collecting information about comparative effectiveness would require small numbers compared with collecting information about safety

Collecting information about comparative effectiveness could require a cohort design with regular direct patient contact since pregnancies are not reliably recorded in any computer-stored data source (no matter what they tell you)

High response rates would be essential to valid conclusions about comparative effectivenss