Safe & accessible methadone dosing

.

How we made Sonara’s virtual dosing experience faster, easier, and safer for patients on methadone.
01
A primer

Sonara for methadone treatment

.

Sonara is a virtual medication adherence platform that allows methadone clinics to offer patients a safe alternative to the daily clinic visits stipulated by state and federal regulations.
Rather than going to the clinic everyday, patients can record themselves taking their medication from the comfort of their home using safeguards including a patented, tamper-aware QR code label placed on each, individual liquid dose. Once a patient submits their video it’s uploaded to a clinic-facing web app to be reviewed.
Directly observed, in-person dosing

vs

.

Remotely observed, at-home dosing
02
Background

Project background

.

In 2022, I led the redesign of Sonara’s remote medication dosing experience, a core feature of Sonara’s mobile-first web app for patients on methadone. Over the prior year, our user base had grown and our analytics and user research infrastructure had matured, revealing a series of pain points in the dosing flow that were hurting the overall patient experience.
Our primary goal was to minimize user frustrations around the biggest pain point - the frequent failure of the QR scanning feature on many android devices. We also wanted to address usability issues users had reported in the broader dosing flow to create a faster and more seamless experience.
Company:
Sonara Health
Timeline:
1 month
Role:
Lead Product Designer
Team:
Head of product, Lead Software Engineer, Senior Software Engineer

The impact

.

80
%
Faster verification
For patients who couldn’t scan their QR code successfully
0
%
Drop off
Between starting recording and completing  bottle verification
Lower
Anxiety
Fewer messages from patients about QR scanning issues
03
The challenge

Designing around limitations

.

Sonara's tamper aware QR code labels were designed to make it easy for clinicians to detect possible diversion when reviewing dose videos in Sonara. But many patient devices were unable to scan the QR code reliably, resulting in a longer, more frustrating dosing experience for some patients.
We already knew the QR scanning issue was a limitation of our tech stack, but we couldn't remove it altogether - as a proprietary security feature it was critical to our sales process. We would have to design around it. On the bright side it created an opportunity to address other patient-reported issues in the dosing experience that had been waiting in the backlog.
"We knew the QR scanning issue was a limitation of our tech stack...we would have to design around it."
Intact Sonara label

vs

.

Destroyed Sonara label

Where we started

.

01
Low QR scan rate
30% of our users were unable to scan their QR code 50% of the time. 8% were never able to scan successfully.
02
Added time to skip scan
Patients who ultimately skipped the QR scan spent over 1 minutes longer on this step than patients who had no issues.
03
High ticket volume for QR issues
We received dozens of patient messages per month expressing anxiety about being unable to scan the QR code.
04
High-friction flow
Patients reported that the dosing experience felt cumbersome and long.
04
Discovery

The percieved
importance of QR scans

.

We couldn't improve the QR scan rate through technical means, so we framed our solution space around improving the dosing experience for patients with QR scanning issues, without compromising on patient safety or. To inform our explorations, we interviewed patients and clinicians with 3 high-level research questions to guide us:
  • Do clinicians have other ways to verify the integrity of bottles when patients skip the QR scan?
  • How much importance do users give to a successful QR scan?  
  • Why do patients spend so long trying to scan their QR code before skipping the scan?

Research insights

.

01
Bottles can be verified visually
From clinicians we learned as long as they could get a good look at the front and back of the bottle in a dose recording, they could visually determine whether it had been tampered with.
02
Anxiety about skipping the scan
Patients spent so much time trying to scan their QR code because they were anxious they would be penalized for skipping the scan. The flow reinforced anxiety by encouraging multiple scanning attempts before skipping.
03
Too many taps, too much reading
Prior user feedback and onsite testing told us that patients were frustrated by the double tap to move between steps in the dosing flow. We also observed that patients with lower literacy levels struggled to read the steps.
Our team at a clinic implementation where I observed patients dosing with Sonara.
05
Our users

Designing for equity

.

Our patients represent a spectrum of characteristics that can influence their experience using Sonara. Consistent with our mission of expanding access to treatment, we design for the breadth of our user population’s cognitive ability, tech-literacy and reading comprehension, among other factors.
06
User Flows

Skipping the scan

Manual verification

.

Our research sparked the idea for ‘manual verification’ - patients with scanning issues would be asked to show the front and back of their bottle before opening it. This would allow clinicians to visually detect tampering and give patients an alternative way to verify their bottle, ameliorating anxiety about scanning.
From our data, we knew that most successful scans happened within 15 seconds of starting the recording. By moving patients automatically to manual verification after 15 seconds, we removed the burden of the decision to skip the scan and in doing so, hoped to reinforce that both ways of verifying were equally valid.
" [Manual verification]  would allow clinicians to visually detect tampering and give patients an alternative way to verify their bottle, ameliorating anxiety about scanning."
Intact Sonara label

vs

.

Destroyed Sonara label
07
Early Design Decisions

Failure-proof verification

.

In transitioning users from a failed QR scan to manual verification, our goal was to avoid sharing information that would create unnecessary anxiety for patients without causing confusion.

Our team agreed it didn't benefit users to know their QR scan had failed. Rather than treating the QR scan and manual verification as two separate steps, we decided to treat them as two possible exit points from the same step. Users could receive success feedback at either exit. The idea was to help users feel like they had successfully verified their bottle regardless of whether their QR has scanned.
"The idea was to help users feel like they had successfully verified their bottle regardless of whether their QR has scanned."
08
Explorations

Showing, not telling

.

I designed the manual verification screen in parallel with the redesign of the rest of the dosing steps, which allowed us to repurpose familiar elements across the entire flow.

My explorations of the new design for the post-verification steps focused on:
  • Reducing cognitive load by representing key information visually
  • Creating a faster flow feeling flow through efficient interactions
  • Building user confidence by reinforcing successful actions
Mapping key info to visual cues
Before exploring design concepts, I identified the key pieces of information that users would need on each step and how they could be translated to visual cues.
Indexing towards familiar
In my early ideation, I tried some experimental ways of representing progress and strep instructions. Ultimately, I chose to move forward with one of the simpler concepts that used familiar components and patterns
Indexing towards familiar
In my early ideation, I experimented with ways of representing progress and strep instructions.  I chose to move forward with one of the simpler concepts that used familiar components and patterns.
Efficient and encouraging interactions
I realized reinforcing a successfully completed step in context, during the transition between steps would feel most natural and space efficient. I tried a number of ways to accomplish this using a micro interaction on our primary CTA button.

The resulting step design

09
More explorAtions

The manual verification screen

.

The biggest risk I saw with the manual verification screen was that it was only presented to some users some of the time. If users weren't paying attention (and we didn't expect them to be), they might not realize what step they were on.

I explored different ways to distinguish manual verification from other steps but, after sharing they with the team, we agreed to test the version that followed the same pattern as the other steps in the flow. This decision was partially motivated by concerns around performance issues and the ergonomics of a smaller video frame. In taking this route, I advocated strongly for A/B testing.

Manual verification concepts

09
Interaction design

Animation for communication

.

Animations within and between the dosing steps were key to communicating with users throughout the flow - I felt it was important to test the clarity of these cues.

I used Protopie to build a high fidelity prototype of the dosing flow, including manual verification, to test with users. Later, this also came in handy for communicating the CSS properties and values of the animations to engineering.
10
testing & releasing

Testing and de-risking

.

In consideration of our timeline and resource constraints, we chose to conduct usability testing remotely with current Sonara patients. We saw positive outcomes across the board:
  • 100% of users understood understood and completed the manual verification step successfully.
  • 100% of users said the dosing experience felt faster and easier.
Despite good feedback, I was still concerned about the modularity of the flow tripping some users up. Since there was no way to test this in a controlled environment, we agreed to A/B test the new design with a small subset of users and monitor for issues. Once we felt confident in the new flow, we released it to all patients.
"Since there was no way to test [the risk of the modular flow] in a controlled environment, we decided to A/B test the new design...and monitor for issues"
11
Results

Faster, safer, better

.

The redesign of the dosing dramatically reduced the amount of time patients spent on bottle verification without increasing drop off which was already close to 0%.
We didn't have a standardized metric for patient satisfaction, like NPS, so we used the volume of support tickets related to QR scanning issues as a proxy. While the volume decreased significantly, it wasn't entirely eliminated. We suspected that the way we solicited patient feedback at the end of the dosing flow might be responsible for this.
  • 100% of bottles verified, either visually or via QR code scan
  • 80% less time spent on verification by patients who couldn't scan their QR code and shorter recording times for patients without issues
  • 0% drop off during verification (no increase)
  • Fewer support tickets about QR scanning issues
12
Lessons Learned

Lessons from low-end devices

.

After launching the feature, we discovered a technical issue in the performance of animations on Medicaid phones. We realized that most of the phones distributed to patients through Medicaid were obscure brands that weren't available to through device farms, so we hadn't testing them.

We were able to roll out a quick fix to affected users, but going forward we obtained several "Obama phones" made a policy of testing the performance of animations on these devices before releasing them.
Some of the Medicaid devices we now use for design QA.
Home
Next project