Feature Article

Case Report Forms - Critical to the Success of your Research Study
March 2008 Issue

Alice Bisbee, MPH, Assistant Director, Data Management,
BUSPH Data Coordinating Center

Author has nothing to disclose with regards to commercial support.

Educational Objectives:

At the end of this activity, participants should be able to:

• Recognize the role of Case Report Forms in clinical research.
• Coordinate the development of Case Report Forms.
• Identify the critical components of a Case Report Form.
• Identify design considerations in creating Case Report Forms.

PRINT | CLOSE WINDOW

• Introduction
• What makes a good Case Report Form (CRF)?
• An overview of the CRF development process
• Which data to include on the CRFs
• The anatomy of a CRF
• How to ask the questions on CRFs
• Examples
• Conclusion
• Quiz

Introduction

Case Report Form (CRF) design is an integral part of the study protocol preparation. The purpose of a CRF is to ensure consistency in data collection, with the goal of collecting analyzable data that will answer the study questions. A CRF is a “printed, optical, or electronic document designed to record all of the protocol-required information to be reported to the sponsor on each trial subject” (ICH GCP 1.11).

In most industry-sponsored trials, CRFs are designed and developed by the sponsor and supplied to the investigators. In investigator-initiated trials (where the investigator is also the sponsor) and non-FDA-regulated studies, the development of CRFs is usually the responsibility of the investigator. The aim of this article is to facilitate the development of these important documents.

What makes a good Case Report Form (CRF)?

A good CRF is user-friendly, uncluttered, and well-organized with clear instructions for completion. The terminology is familiar to the person filling out the form, and the reading level matches that of the study participants. The questions are asked clearly and usually only once. Data are collected as outlined in the protocol. Good CRFs also act as a valuable reference tool for the data analysts.

An overview of the CRF development process

Whether web- or paper-based, investigators must allow plenty of time to develop CRFs, and must coordinate the involvement of everyone who will collect or analyze the data. The entire study team (the clinical investigators, study coordinators, study nurses, research assistants, interviewers, clinical experts, etc.) will need to confirm whether the essential data are being captured to answer the study questions. The data management team (the database and web programmers and data managers) will make sure they have enough information to create a usable database. The analytic team (the data analysts and statistician) will determine whether the data can be appropriately quantified, and whether adequate data are being collected to answer the study questions.

After the forms are reviewed, pilot the forms by testing them with a sample of subjects from the intended study population (with IRB-approved consent forms) and with the clinicians who will be completing the forms. Look for common problems: Can clinicians use the forms to report study outcomes, or are they being asked for data not available to them? Are there insufficient place savers for the data being collected? For example, does the CRF inappropriately allow only two-digits for systolic blood pressure? Investigators must not get discouraged if they don’t get the CRF right the first time. Instead, allow time for revisions ... and more revisions!

Finally, allow time to pilot the data entry process. Do not start the study until the forms have been reviewed and tested and all corrections have been made. CRFs are often requested by IRB, as it is a way for the IRB to review the data elements being collected.

Now the forms are perfect … or are they? Re-evaluate the forms as the study progresses. Are there any items generating a high non-response rate? If so, perhaps the question should be reworded or dropped. Is useful information being obtained from open-ended questions? If so, consider creating a categorical variable from those responses, leaving a space for "other." If not, perhaps a new question is needed. Are the skip patterns being followed correctly? If not, perhaps some user training is needed, or maybe the forms need to be modified to clarify the skips.

Which data to include on the CRFs

Consider the administrative data that will assist in data analysis such as visits missed. In general, CRFs should only include data elements that will be analyzed. However, it is sometimes useful to add administrative information to a CRF that will not be part of the analysis, such as notes on the best time to reach a subject.

Usually, source data from source documents are transcribed onto the CRF. Examples of source documents are lab reports, progress notes, medical records, worksheets developed by the site team to record observations, x-ray readings, etc. It is also acceptable for CRFs or portions of CRFs to serve as source documents where data/observations are recorded directly onto the CRFs; however, the protocol should prospectively define which data may be treated in this way (ICH GCP 6.4.9). These data sections of the CRF should be signed or initialed and dated by the person collecting that information since source documentation should be attributable, per FDA guidance.

The anatomy of a CRF

Header: Study name. Form name. Site/location. Form date. Contact time point (e.g., baseline, 1 month visit, 13 month phone call). Space to identify the person entering the data (e.g., research assistant’s initials). Subject ID and subject initials if allowable (some studies of sensitive topics may not allow identifying information such as initials on forms). The subject ID must be unique for each subject and should appear on every page of every form, in case pages get separated. This ID will link the paper forms with a specific record in the database.

Footer: Full study name, grant/protocol number, CRF version number, and version date. Do not change anything on a CRF without changing the version number and date! Sometimes a check box is included to indicate when a form has been faxed or edited or data-entered.

Data Modules: This is where study data are collected. Some modules, such as demographics, may be duplicated between studies. Common modules are demographics, screening, adverse events, labs, medical history, medications, procedures, treatment discontinuation, study completion, and signatures.

How to ask the questions on CRFs

Study flow: Data collection should follow the flow of the study, and a CRF should only reflect data collected at one time point. For example, since height and weight will be measured before conducting exhausting neurological testing, those fields should appear earlier on the CRF. In most cases, a CRF should only contain data collected on one time point.

Visual clarity: The finished product should be easy on the eye and logical in its execution. Be consistent in the use of fonts and formatting. Check boxes should be aligned and in predictable locations. Illustrate skip patterns with arrows, if possible.

Inconsistencies: Avoid data inconsistencies by asking a question only one time and in only one place. The exception to this rule is when it is desirable to confirm a critical data point, such as the presence or absence of adverse events.

Open-ended questions: Open-ended questions are an invitation to analytical mayhem. “What is your date of birth?” may elicit a response such as,“My birthday was yesterday.” “Rate your asthma control in the past month” might return, “Pretty good.” Instead, provide place-savers for date of birth (__ __ / __ __ / __ __ __ __) and categorize response options (e.g., not controlled at all, poorly controlled, somewhat controlled, well controlled, completely controlled). Indicate whether only one category should be selected (e.g., primary diagnosis) or if “check all that apply” is applicable (e.g., race). Make sure that there really and truly will NEVER be more than one possible response before opting for “check one”. Changing the field mid-study is a major headache for the data entry person, the database manager who may have to split an existing field into several fields, and for the analyst who may have to make sense of data that were collected and categorized differently, depending on when a subject entered into the study.

“Other”: Always consider the “Other” category, as it might cut down on the number of questions left blank. “Other: specify: _____” should be positioned last in the list of possible responses to ensure that all anticipated responses are considered first. Continue to monitor the “Other” category for common responses which may indicate the need for a new category. Again, changing variables mid-study can be problematic. Any changes should be discussed with the data analyst and all changes must be approved by the IRB before implementing.

Units: Units of measurement should be clearly defined. Is temperature being measured in degrees Fahrenheit or Centigrade? Remember that the lab may report results using different units than those from other labs, so indicate all lab value units.

Missing data: There will always be missing data, but fewer opportunities for missing data on the CRFs means fewer reasons to question the accuracy of the data. Consider the possible levels of response: Is the information unknown, unavailable, not applicable, refused, or “no opinion”? Does it matter, or is it sufficient to know that the data will be labeled “missing” in the data analysis?

Examples

Examples of CRFs, good and bad, may be found in the Clinical Research Seminar presentation on this topic.

Conclusion

The importance of good CRF design is often not fully appreciated. Poorly designed or confusing CRFs may result in the collection of too much data, too little data, or the wrong data. The data collected from a good CRF will result in valid answers to study questions. CRF design and revisions will take some dedication and time, but as Voltaire once said, “perfection is attained by slow degrees; it requires the hand of time.”

Click here and close window if you are a BUMC researcher
and would like to take the quiz now.

Close Window