Let’s continue exploring the art of eCRF design with a discussion of additional best practices! In Part 1 we discussed:
- Emulate paper forms.
- “Right-size” each form (not too little or too much data)
- Consider the site’s workflow
- Consider masking implications
- Group related information
Here are more tried and true best practice guidelines.
Clearly Label Fields and Use Relevant Database Names – When labeling data entry fields, consider the end user. Confusing labels lead to incorrect data entry. Let’s consider a field that is labeled “Amount”. Is this the number of pills? A liquid measure? A label such as “Number of Pills” is much clearer. When naming fields in the database, consider the needs of the statistician. A field name of “weight_in_kg” is clearer than a field name of “weight”.Here are some more best practice guidelines.
Incorporate Database Naming Conventions – Speaking of field database names… When standard abbreviations and format conventions are used to name database fields, it’s easier for both biostatisticians and regulators to understand the data in submissions. Fields holding dates are a good example. Let’s say there’s a visit date, a physical exam date, and a dosing date field used in a study. It’s better to use names such as “visit_date”, “phys_exam_date”, and “dosing_date” rather than “date_of_visit”, “exam_date”, and “when_dosed”.
Use “Skip Logic” (Controlled By) Where Appropriate – A good example of when this is appropriate is the case where there is a gender field and a pregnancy question on a form. If the gender is “male”, there’s no reason for the user to answer the pregnancy question. In this case, VISION’s “Controlled By” property can be used to control when the pregnancy field is displayed on a form. Incorporating this type of logic makes the form much more intuitive for the end user and reduces the likelihood of questions such as “is this a pregnant male?”.
If Same Data is Used on Multiple Forms, Copy it Between Forms – Let’s say there are multiple forms to fill out in a single visit day, and it would be useful to have the visit date displayed (and be resident in the database) for all those forms. Have the user enter the date on the first form, and then copy that entry to read only fields on the other forms using a calculation. This approach requires less data entry by the clinician and reduces errors.
Minimize Free Text Responses – Free text, while expressive, leaves something to be desired where data analysis is concerned and can also be error prone. Instead of asking a user to enter a number between 1 and 10, provide a radio with values from 1 to 10. Or create a drop down containing those numbers. Or use a Visual Analogue Scale. Such approaches introduce clarity in the data and are easier to quantify.
Avoid Excessive Error Checks – EDC is very powerful, and it’s easy to become overenthusiastic when implementing error checks. However! When error checks are too stringent edge cases become difficult for users to handle, resulting in additional work for monitors, project managers, and biostatisticians.
Use a Second Pair of Eyes – Sometimes, it’s surprising how another person interacts with a form the designer thinks is intuitive. Grab your favorite clinician and watch them (either in person or via screen share) fill it in. Note any questions or areas of confusion and use that information to make the form easier to use for others.
We hope you found these best practices practical and helpful! Read the previous article.