Self-Selection Studies for Nonprescription Drug Products

Administrative guidance, not enacted legislation. This captured publisher edition preserves the listed document and attachments; later developments are not independently consolidated.

Guidance for Industry
Self-Selection Studies for
Nonprescription Drug
Products
U.S. Department of Health and Human Services
Food and Drug Administration
Center for Drug Evaluation and Research (CDER)
April 2013
OTC

Guidance for Industry
Self-Selection Studies for
Nonprescription Drug
Products
Additional copies are available from:
Office of Communications, Division of Drug Information
Center for Drug Evaluation and Research
Food and Drug Administration
10903 New Hampshire Ave., Bldg. 51, rm. 2201
Silver Spring, MD 20993-0002
Tel: 301-796-3400; Fax: 301-847-8714; E-mail: druginfo@fda.hhs.gov
http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/default.htm
U.S. Department of Health and Human Services
Food and Drug Administration
Center for Drug Evaluation and Research (CDER)
April 2013
OTC

TABLE OF CONTENTS
I. 	INTRODUCTION............................................................................................................. 1
II. 	BACKGROUND ............................................................................................................... 2
III. 	STUDY DESIGN AND CONDUCT................................................................................ 2
A. 	Study Objectives ............................................................................................................................ 3
1. Primary Objective............................................................................................................................ 3
2. Secondary Objectives....................................................................................................................... 3
B. 	Study Population............................................................................................................................ 3
C. 	Statistical Considerations and Data Analysis.............................................................................. 4
1. Primary Endpoints, Success Criteria, and Mitigating Factors........................................................ 4
2. Sample Size Considerations............................................................................................................. 6
3. Data Analysis ................................................................................................................................... 7
D. 	Questionnaire Design..................................................................................................................... 7
1. Questions That Address the Study Objectives.................................................................................. 7
2. General Questionnaire Design Concepts ........................................................................................ 8
E. 	Study Conduct................................................................................................................................ 9
F. 	Validating the Self-Selection Decision.......................................................................................... 9
G. 	Data Collection and Recording..................................................................................................... 9
IV. 	FINAL REPORT ............................................................................................................ 10

Contains Nonbinding Recommendations
Guidance for Industry1
Self-Selection Studies for
Nonprescription Drug Products
This guidance represents the Food and Drug Administration’s (FDA’s) current thinking on this topic. It
does not create or confer any rights for or on any person and does not operate to bind FDA or the public.
You can use an alternative approach if the approach satisfies the requirements of the applicable statutes
and regulations. If you want to discuss an alternative approach, contact the FDA staff responsible for
implementing this guidance. If you cannot identify the appropriate FDA staff, call the appropriate
number listed on the title page of this guidance.
I. 	INTRODUCTION
This guidance is intended to provide recommendations to industry involved in developing and
conducting self-selection studies to support an application for nonprescription drug products. A
self-selection study assesses the ability of consumers to apply drug labeling information to their
personal health situation to make correct decisions about whether or not it is appropriate for them
to use a drug product.
This guidance covers general principles related to the conduct of self-selection studies, including
study design, methodology, and analyses, and should not be considered a substitute for an FDA
review of specific protocols. We recognize that self-selection data can be obtained in a variety
of different ways and that each development program raises distinct issues for which flexibility
and creativity is important. This guidance also incorporates advice obtained from the
Nonprescription Drugs Advisory Committee at a meeting on September 25, 2006, at which the
committee considered issues related to analysis and interpretation of consumer studies conducted
to support marketing of nonprescription drug products.2
FDA’s guidance documents, including this guidance, do not establish legally enforceable
responsibilities. Instead, guidances describe the Agency’s current thinking on a topic and should
be viewed only as recommendations, unless specific regulatory or statutory requirements are
cited. The use of the word should in Agency guidances means that something is suggested or
recommended, but not required.
1 This guidance has been prepared by the Division of Nonprescription Clinical Evaluation and the Office of
Biostatistics in the Center for Drug Evaluation and Research (CDER) at the Food and Drug Administration.
2 The transcript from the September 25, 2006, Nonprescription Drugs Advisory Committee meeting is available at
http://www.fda.gov/ohrms/dockets/ac/06/transcripts/2006-4230t.pdf.
1

Contains Nonbinding Recommendations
II. 	BACKGROUND
The development program for certain nonprescription drug products may include conducting
consumer studies, in addition to the required safety and efficacy studies. The consumer study
process may include label comprehension studies, which assess consumer understanding of
major communication elements,3 and self-selection studies, which test whether consumers can
apply the label information to their personal medical situations and make correct decisions about
whether it is appropriate for them to use or not use the drug product (self-selection decision).
We recommend conducting a label comprehension study to optimize the label before conducting
a self-selection study.
A self-selection study may be needed to test whether consumers can make appropriate self-
selection decisions based on the information contained in the label. Some of the circumstances
under which we might recommend a self-selection study include:
 	The drug product is for a new nonprescription indication
 	The drug product is for a new nonprescription target population
 	There are specific populations who should not use the proposed nonprescription drug
product (e.g., diabetics, transplant recipients)
 	A substantive labeling change has been proposed for an approved nonprescription drug
product that may affect the appropriate nonprescription population (e.g., a change in the
warnings, directions for use)
We encourage sponsors to seek FDA consultation and advice on any protocol for a self-selection
study.
A self-selection decision is not necessarily predictive of whether a consumer will actually use the
drug product or whether he or she will use it correctly. If information on consumer use is
needed, an actual use study should be considered. Actual use studies are outside the scope of
this guidance.
III. 	STUDY DESIGN AND CONDUCT
In general, self-selection studies can be open-label, uncontrolled studies. The following are
general recommendations for the design and conduct of a self-selection study. Subsequent
sections provide a more detailed discussion of each recommendation.
3 See the guidance for industry Label Comprehension Studies for Nonprescription Drug Products. We update
guidances periodically. To make sure you have the most recent version of a guidance, check the FDA Drugs
guidance Web page at
http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/default.htm.
2

Contains Nonbinding Recommendations
 	State the purpose and objectives of the study
 	Specify a study design that meets the study objectives and calculate the appropriate
sample size
 	Use labeling that is as similar as possible to the final drug product label
 	Construct a questionnaire that targets the study objectives
 	Minimize factors that may contribute to a biased study (e.g., sampling, recruitment
strategies, leading questions, interviews that bias the responses in a particular direction)
 	Enroll a population appropriate for the study objectives
 	Enroll subjects with low and normal literacy skills
 	When necessary, enrich the study with subjects who have relative or absolute
contraindications to use of the drug product
 	Consider pilot testing before conducting a large self-selection study
Self-selection can be studied in a variety of ways. Sometimes it can be evaluated as part of an
actual use study or sometimes in conjunction with a label comprehension study. A separate
study does not always need to be conducted to address self-selection.
A. 	Study Objectives
1. 	Primary Objective
In general, the primary objective(s) of a self-selection study should be to assess if subjects, after
reading the drug product label, can make a correct self-selection decision(s) based on their
medical condition(s).
2. 	Secondary Objectives
Secondary objectives may be included to assess self-selection accuracy in specific
subpopulations of interest and/or the reasons why subjects make incorrect self-selection
decisions. Thus, some of the data may come from questions asked to elicit reasons for incorrect
self-selection. It is important to determine why subjects make incorrect decisions to better
understand what label revisions may need to be made to improve correct self-selection.
B. 	Study Population
The study population should be defined based on the study objectives. Some studies may
include any subject who may have an interest in using the drug product, regardless of age, sex,
underlying medical conditions, and use of concomitant medications. Other studies may target a
3

Contains Nonbinding Recommendations
particular population of interest (e.g., subjects with a specific disease for whom use of the drug
product may be contraindicated). Because nonprescription drug products are available for
purchase without the input of a health care professional and no drug product is administered in
the study, exclusion criteria should be minimal (e.g., inability to speak, read, and understand
English).
Self-selection studies also should enroll an adequate number of subjects who have low literacy
skills to examine the ability of this subgroup to make a correct self-selection decision. Education
level is not a reliable substitute for literacy testing. At screening, the sponsor should assess
literacy levels of the study subjects by administering a validated instrument such as the Rapid
Estimate of Adult Literacy in Medicine (REALM) test,4 REALM-Teen for testing adolescents,5
or the Test of Functional Health Literacy in Adults (TOFHLA or S-TOFHLA).6,7,8 Investigators
should receive training to properly administer literacy tests.
C. 	Statistical Considerations and Data Analysis
1. 	Primary Endpoints, Success Criteria, and Mitigating Factors
The primary endpoint for a self-selection study should be the proportion of the study subjects
who make a correct self-selection decision. A correct self-selection decision is based on the
label element(s) that are needed to make an appropriate decision whether or not to use a drug
product (self-selection decision). A correct self-selection decision can be based on a single label
element or a composite of several label elements depending on the drug product.
For example, the drug product may have only one contraindication: diabetics should not use the
drug product. In this case, a correct self-selection decision for a diabetic (regardless of whether
he or she has the condition for which the drug product is indicated) would be that the drug
product is not appropriate for his or her use. In contrast, an incorrect self-selection decision
would be a diabetic who selects to use the drug product. An example of a composite of several
label elements would be a drug product that is only for women between 18 and 65 years of age
who do not have heart disease or hypertension. A correct self-selection decision would be a
woman between 18 and 65 years of age who has the labeled indication and does not have heart
4 REALM: Davis, TC et al., 1993, Rapid Estimate of Adult Literacy in Medicine: A Shortened Screening
Instrument, Family Medicine, 25:391-395.
5 REALM-Teen: Davis, TC et al., 2006, Development and Validation of the Rapid Estimate of Adolescent Literacy
in Medicine (REALM-Teen); A Tool to Screen Adolescents for Below-Grade Reading in Health Care Settings,
Pediatrics, 118(6):e1707-1714.
6 TOFHLA: Parker, RM et al., 1995, The Test of Functional Health Literacy in Adults: A New Instrument for
Measuring Patients’ Literacy Skills, Journal of General Internal Medicine, 10:537-541.
7 Baker, DW et al., 1999, Development of a Brief Test to Measure Functional Health Literacy, Patient Education
and Counseling, 38:33-42.
8 The REALM and TOFHLA were designed as rapid screening tools that were validated against the Wide Range
Achievement Test for literacy. Therefore, use of these instruments to screen literacy levels within the context of
health is appropriate.
4

Contains Nonbinding Recommendations
disease or hypertension who selects to use the drug product. An incorrect self-selection decision
would be a 55-year-old woman with hypertension who selects to use the drug product (whether
or not she has the indication).
Several analyses to illustrate self-selection decision making may be of interest; usually one or
two are chosen as primary endpoints, and the others may be used for exploratory analyses. Table
1 shows one way to consider choices for different endpoints, where A, B, C, and D represent the
number of subjects in the cells.
Table 1: Two-by-Two Table Illustrating Possible Self-Selection Choices
Appropriate to
select drug
Not appropriate to
select drug
Self-selected to take
drug
Correct
A
Incorrect
B 	A+B
Self-selected not to
take drug
Incorrect
C
Correct
D 	C+D
A+C 	B+D
The analysis chosen for the primary endpoint depends on the study design and the issue(s) of
greatest concern for nonprescription use. For example, the primary endpoint might be correct
self-selection for the entire population [(A+D)/(A+B+C+D)]. Alternatively, the primary
endpoint might be the proportion who select correctly among those who select to take the drug
([A/(A+B)]. As another example, the primary endpoint might be correct self-selection among
those who cannot take the drug [D/(B+D)]. The primary endpoint and how it will be calculated
should be defined in the protocol before the study starts, and discussed with the FDA.
Success criteria should be related to the predefined target level for correct self-selection. This
target level should be justified based upon a clinical rationale. Success criteria should be defined
using the confidence interval approach: the study can be claimed as a success only when the
lower limit of a predefined two-sided 95 percent (or one-sided 97.5 percent) confidence interval
for the correct self-selection rate is above the target level. We recommend using a two-sided 95
percent confidence interval to estimate the correct self-selection rate as well as to define the
success criteria. This approach allows consideration of variability within the study data and sets
the type I error rate for one-sided tests (2.5 percent) at half the conventional type I error rate (5
percent) used in two-sided tests.
Reasonable predefined mitigating factors may be acceptable in certain circumstances. Mitigating
factors are subject responses that would allow what appears to be an incorrect self-selection
decision to be considered a correct self-selection decision. Mitigations should be clinically
reasonable. The following is an example of a mitigating factor: subjects who make an incorrect
self-selection decision based on age, but who are within 1 year of the labeled age and verbalize
an understanding of the correct age. For example, a subject who is 54 makes an incorrect self-
selection decision based on age for a drug product that is labeled for adults 55 and older. After
being asked a nonleading, open-ended question to elicit more information about his or her
decision, the subject states that he or she knows the drug product is labeled for people aged 55 or
5

Contains Nonbinding Recommendations
older, but because he or she will be 55 within a month he or she feels it is okay to take the drug
product.
The definitions of a correct self-selection decision, mitigating factors, and the target success
criteria for correct self-selection should be determined in advance of study enrollment and
specified in the study protocol. Mitigation decisions should be auditable, and subject level data
should be presented so that the FDA can easily review and audit mitigating factors. If a sponsor
finds unexpected mitigating factors on post hoc analysis, the study report should explain why
and how these other factors were used to mitigate the endpoint(s).
With more complicated labels, such as those with multiple decision points, we recognize some
label elements have greater clinical significance than others if not heeded. Therefore, we
recommend that sponsors discuss with the FDA the label elements to be used to define correct
self-selection, the mitigating factors, and the predetermined success criteria before conducting
the study.
2. 	Sample Size Considerations
The number of subjects in a self-selection study should be large enough to provide a reliable
answer to the primary objective. Sizing of such a study should be based on the success criteria in
a hypothesis testing framework. This generally involves several elements: the predefined target
level for correct self-selection, the assumed percent of correct self-selection for the study
population, the type I error rate, and the desired study power for testing the predefined
hypothesis or hypotheses.
Each of the different elements in the sample size calculation should be determined as follows:
 	The predefined target level for correct self-selection should be based on a clinical
rationale based upon the medical significance of incorrect self-selection, as discussed in
the previous section.
 	The assumed percent of correct self-selection should be the best guess of what the true
self-selection rate is in the population of interest. Ideally, it should be based on results
from pilot studies on the same drug product or on self-selection studies for similar drug
products in a similar population.
 	The type I error rate should be set at 2.5 percent for a one-sided test corresponding to a
one-sided 97.5 percent confidence interval, or 5 percent for a two-sided test
corresponding to a two-sided 95 percent confidence interval.
 	The study power typically should be decided by sponsors based on their resources. We
recommend that the study power be 80 percent or above.
The number of subjects in a self-selection study should be large enough to evaluate the primary
endpoint for important subgroups of interest. Alternatively, a targeted self-selection study can be
conducted to assess a specific subgroup.
6

Contains Nonbinding Recommendations
3. 	Data Analysis
The principal features of the planned analysis should be defined in the protocol. The statistical
methods and the analysis of the primary and secondary endpoints also should be specified in the
protocol. Methods for constructing a two-sided 95 percent confidence interval to estimate and
define the success criteria for the correct self-selection rate should be described, including a
detailed description of the numerators and denominators. Methods for handling missing data
should be specified.
Typically, a comprehensive statistical analysis plan should be included in the protocol and
address all the details of the data analysis. In cases where the comprehensive statistical analysis
plan is prepared as a separate document, it should be prepared before the results of the study are
known and it should be submitted for FDA review.
D. 	Questionnaire Design
Typically, a questionnaire is used to collect data in a self-selection study. The questionnaire
design should: (1) reflect the study objectives; and (2) optimize the validity and interpretability
of the information collected. Wording, question structure, and question sequences can
significantly affect the validity and interpretability of the data collected. A detailed discussion of
questionnaire development is beyond the scope of this guidance. However, the following points
merit particular consideration.
1. 	Questions That Address the Study Objectives
The following are general recommendations about the types of questions that address the study
objectives.
 	The self-selection question: The first question asked should be an open-ended, self-
selection question (e.g., “Is it okay for you to use this medication?”) followed by an
open-ended, nonleading probing question. It is important to determine why subjects
make incorrect decisions to better understand what label revisions may be needed to
improve correct self-selection.
 	Open-ended probing questions: Open-ended questions that follow the self-selection
question should be asked because they may help provide additional data for the analysis.
Additional questions used to elicit more information should be nonleading and be used
sparingly so as not to prompt the subject to provide a desired response. An example of a
nonleading question for subjects who say that it is “okay for them to use the medication,”
would be to then ask, “Why do you say that?” This question enables multiple possible
responses including “after I talked to my doctor.” By contrast, an example of a leading
question, where the correct response is that a subject should talk to his or her doctor
before using the drug product, would be to ask, “Is there anything you would do before
you start using the medication?” This question may direct subjects toward a correct
7

Contains Nonbinding Recommendations
answer. Additionally, subjects who feel obligated to provide an answer may respond
with “talk to my doctor” as a logical default answer, which may introduce bias.
 	Medical history questions: Medical history questions should be asked after the self-
selection question to prevent bias that can occur from prompting the subject to focus on
specific label elements. The medical history questions should be asked in consumer-
friendly language and be based on obtaining health information related to the drug
product’s indications, warnings, and contraindications.
Sponsors sometimes choose to add a question asking subjects whether or not they would
purchase the drug product. We do not consider purchase decision data to be a reliable surrogate
for self-selection data because a purchase decision may be influenced by factors other than
personal health history, such as cost or the needs of someone else in the consumer’s home. If a
sponsor wishes to collect data on purchase decisions, these questions should be asked only
following the completion of the self-selection portion of the assessment.
2. 	General Questionnaire Design Concepts
The following are general recommendations about the design of questionnaires.
 	Simple vocabulary should be used, and questions should be pretested to ensure the
questions elicit the intended information.
 	Some questions should be direct, specific, and unambiguous, and should address a single
item or issue.
 	Some questions should involve a higher level of understanding where subjects may need
to put together several pieces of label information to arrive at the correct response.
 	Questions should be ordered so that information contained in a question does not bias a
subject’s ability to answer subsequent questions.
 	Response choices in multiple-choice questions should be independent and contain only
one correct answer.
 	When listing response categories for multiple-choice questions (e.g., to assess the
medical history), I don’t know should be included as one of the response categories to
give subjects permission to admit that they do not know so they avoid guessing. The
category other also should be included to allow subjects to add something that may not
be listed as a choice.
 	Questions intended to measure the behavioral intent of the subject should not be used.
Testing behavior is outside the scope of a self-selection study. If information about how
subjects would behave under real-world conditions is needed, an actual use study should
be conducted.
8

Contains Nonbinding Recommendations
The questionnaire can be pretested with a sample of respondents similar to the target population
to ascertain that the questionnaire is eliciting the intended information.
E. 	Study Conduct
Advertisements for the study should not contain any information about the proposed drug
product. If there is initial telephone contact, subjects should receive only information on how
and where to participate in the study. Occasionally, it may be necessary to use a clinical setting
for the study site so a specific population can be accessed. The study setting should be
comfortable and well lit for reading. Subjects should have adequate time to read the label and be
able to refer to it throughout the testing period. However, subjects should not be prompted to
refer to the label during testing.
Subjects should receive sufficient instruction on the format and conduct of the study and the
expected length of time it will take to participate. Two general approaches to administering the
questionnaire that can be considered include self-administration or asking the questions using a
trained interviewer. Using an interviewer may lessen the chance that low literate subjects will
respond incorrectly because they cannot comprehend the written question when, in fact, they
comprehend the label. Using an interviewer, however, may lead to interviewer bias, particularly
if the interviewer leads the subject to elicit a response. Therefore, interviewers should be
adequately trained and should adhere to established protocols and/or scripts. Inherent bias can
occur with any data collection method. Therefore, sponsors should provide a rationale for why a
particular method was chosen and should address any potential bias.
F. 	Validating the Self-Selection Decision
Answers to the self-selection question can be validated in a variety of ways. Sometimes this can
be accomplished by asking medical history questions, but other times collecting laboratory data
or conducting a physical examination also may be needed. Not all of these methods of validation
will be needed for every study.
G. 	Data Collection and Recording
Verbatim responses to all open-ended questions should be recorded. The procedure for coding,
categorizing, and analyzing verbatim responses to open-ended questions should be specified in
advance and described in the protocol. In addition, all correct and incorrect answers to closed-
ended questions should be prespecified. Any post-hoc coding for open-ended questions should
be documented.
Methods for verification of complete and accurate recording of study data should be described in
the protocol or the statistical analysis plan (i.e., subjects’ responses, data entry, missing data, and
data coding).
9

Contains Nonbinding Recommendations
IV. 	FINAL REPORT
The final report should summarize the study design, study conduct, and interpretation of the
study results. The report should assess whether the appropriate population has been enrolled to
adequately assess self-selection decision making. It also should describe the nature of the
recruitment effort and the response rate (i.e., the proportion of screened subjects who were
actually enrolled in the study). If possible, potential subjects who were excluded, or who chose
not to enroll in the study, should be characterized by demographic factors and the reasons for
nonparticipation. Enrolled subjects should be characterized as to relevant demographic factors
(including literacy) and whether or not they completed the entire study. Reasons why subjects
failed to complete the study should be provided in the final report.
The presentation of the study results should include both the overall correct self-selection rates
and correct self-selection rates in appropriate subsets (e.g., literacy level, sex, age, race, and
presence of high risk factors).
The acceptable success criteria for correct self-selection should be based on meeting the success
criteria that were established before the study began and were documented in the protocol and/or
the statistical analysis plan. The interpretation of the quantitative data should be augmented by
the verbatim responses collected from open-ended, nonleading questions used to assess the self-
selection. Thus, an analysis of both quantitative and qualitative data types should be provided to
support and interpret the study findings.
10