Designing forms for error prevention
PLIÉ is a clinical research study through UCSF Weill Institute of Neurosciences at the San Francisco VA Medical Center. PLIÉ aims to help Veterans with dementia with their cognitive and physical function through a group movement class.
While I was working as a Clinical Research Coordinator, I discovered patterns of mistakes on paper forms. These errors came from all sources: Veteran participants, research assistants, or even myself. As I kept a log of these errors, I noticed the same errors emerging and began to examine why the same errors kept occuring.
01 / Timeline
I worked on this project while I was a Clinical Research Coordinator for PLIÉ from 2016-2019.
02 / The challenge
Paper forms are used to collect primary outcome data. This data is used for a clinical research study, so accuracy is a very high priority. The goal was to minimize human error.
03 / Research methods
I discovered instances of participant confusion during neurocognitive assessments, and took notes on Veteran’s confusion. I discovered errors in data entry on paper forms, and in the outcomes database, and kept a log of errors.
04 / Deliverables
The deliverables for this project:
1. Updated paper forms
2. Implementation of changes in our outcomes database
05 / Tools
I used Microsoft word to edit the forms, and the changes were updated in Access database & Redcap database.
Form redesign #1 [Affect-Mobility-Social Feelings Scale]
Participant facing self-report scale
This scale was given to veterans during their assessment. The document was originally pieced together from several different scales. After administering this measure to several veterans with cognitive impairment and observing their confusion, I redesigned it to make it more accessible.
Version 0
The original version was confusing to veterans because the gray columns were disorienting.
The numbers going down the columns were intended to be helpful, but for someone with cognitive impairment, it was difficult to follow along.
The font was also very small, making it difficult for someone with low vision to read.
Version 0, original scale [click to view]
Version 1
The gray column and rows were reformatted to highlight only the meaningful information, because the name of the scale is not important for the participants to read.
The font was made bigger, but this caused the questions to not fit on one page.
When folks with cognitive impairment are given a set of questions, they very often do not look at the back of a page, so it is common practice to keep forms single sided.
Version 1, some edits made [click to view]
Version 2
The most important change on this version is that the numbers were removed and replaced with the word choices.
This lets veterans stay on the question, without needing to refer back to the top of the page and potentially lose their place.
The margins were also made narrower, so all of the questions could fit on one page.
The titles of the scale were also moved to the left, because it is not important for the participant to see.
Version 2, the final version [click to view]
Form redesign #2 [Assessment Completion List]
Research assistant facing checklist
Before
The PIDN (participant ID number) was located at the bottom, making it difficult to keep track of documents.
The title of this form is long and did not distinguish it from other study-related documents.
Items in this document were not categorized by topic or chronologically, their order was random.
Before redesign [click to view]
After
The space to write in the PIDN was moved to the top. A checkbox was added so a research participant could check off “Visit 1” or “Visit 2” rather than write it out.
The first hierarchy of grouping for items is relevancy, and the second ordering is chronological.
The title of this document was changed from “Primary Participant Visit Completion Checklist” to “Assessment Completion List”
A label was placed by each scale, noting if the measure should be administered by a research assistant or the participant.
After redesign [click to view]
Learnings from this project: the lessons I will be taking with me
Lesson #1
Is it human error or is it a problem with the interface?
If the same mistake continues to emerge, that is a great starting place for research and design. In the workplace it can be tempting to place blame on a user for mistakes, but when mistakes are continuously repeated, it is worth investigating to see if there is a solution.
Lesson #2
Consult cross-functionally and examine constraints.
On this project, I consulted with the constraints of the database we used to insure that the new format could be implemented. In research, databases can come with heavy restrictions, and something as simple as adding a question can throw off the entire numbering system. Additionally, form changes need approval from Institutional Review Boards (IRB) to ensure all personal information is protected. I consulted with the Principal Investigator, the database manager, research participants, and other research staff as I was updating these forms.