Rethinking medication workflows for care managers

Rethinking medication workflows for care managers

Medication Hub is a module inside InCare, Innovaccer's care management platform. Care managers use it daily to review, add, and manage patient medications. The existing module was functional but frustrating - poor data defaults, blocking interactions, and no drug safety checks. I redesigned the experience end-to-end, from information architecture to interaction patterns to clinical safety features.

Medication Hub is a module inside InCare, Innovaccer's care management platform. Care managers use it daily to review, add, and manage patient medications. The existing module was functional but frustrating - poor data defaults, blocking interactions, and no drug safety checks. I redesigned the experience end-to-end, from information architecture to interaction patterns to clinical safety features.

Role

Solo Product Designer, User research, Interaction design, Design system contribution

Team

2 Engineers, 1 PM, 1 Designer, Clinical reviewers

Timeline

4 months

Confidentiality Notice


To respect confidentiality agreements, certain visuals have been blurred or omitted. If you'd like to learn more, I'd be happy to walk you through the original screens, design decisions, and outcomes during 1:1 conversation.

Problem

A daily clinical tool that created more work than it solved.

A daily clinical tool that created more work than it solved.

Every medication review required 5-6 clicks just to complete basic tasks. Multiply that by 30+ medications across 15+ patients a day. Care managers use InCare to verify every medicine a patient goes home with after hospital discharge. Get this wrong - the patient is back in hospital. The module they depended on for this was broken in ways that compounded daily.

Every medication review required 5-6 clicks just to complete basic tasks. Multiply that by 30+ medications across 15+ patients a day. Care managers use InCare to verify every medicine a patient goes home with after hospital discharge. Get this wrong - the patient is back in hospital. The module they depended on for this was broken in ways that compounded daily.

40%

40%

of medication errors during care transitions stem from poor reconciliation

of medication errors during care transitions stem from poor reconciliation

1 in 5

1 in 5

of those errors directly harm patients

of those errors directly harm patients

#1

#1

most-used module on the platform - every care manager, every day

most-used module on the platform - every care manager, every day

Before
Before
After
After
After
Before
Before
After

Context

Build for today. Designed for whats next

Build for today. Designed for whats next

Medication Review was already live - just not working well. Early in research, it became clear that Review and Reconciliation are fundamentally different workflows that would eventually both need to exist.


So we didn't just fix the module. We redesigned it so Reconciliation could plug in later without rebuilding from scratch.

Medication Review was already live - just not working well. Early in research, it became clear that Review and Reconciliation are fundamentally different workflows that would eventually both need to exist.


So we didn't just fix the module. We redesigned it so Reconciliation could plug in later without rebuilding from scratch.

Research

What care managers were actually struggling with- The module wasn't ugly. It was built around the wrong mental model.5 patterns that kept surfacing:

What care managers were actually struggling with- The module wasn't ugly. It was built around the wrong mental model.5 patterns that kept surfacing:

Every action blocked the table - blocking sidesheets cut care managers off from their data mid-task.

No structured view of changes - a flat list forced mental tracking of what was new, modified, or carried over.

The same drug appearing twice - different names from different sources, no detection, no merge.

Zero safety net - no drug interaction or allergy alerts of any kind.

Data that existed but wasn't surfaced - dose, route, frequency all typed manually despite living in the EMR.

What was shipped

Medication Review, rebuilt

Medication Review, rebuilt

A medication management system rebuilt from the ground up - with less clicking, more clarity, and clinical intelligence built in.
─────────────────────────────────────────────────────────────────────────

Care managers were spending more time managing the tool than managing their patients. Every action interrupted the workflow. Every field was manual. Every panel blocked the screen behind it.


The tool existed. It just made a hard job harder.

Self-serve medication review. Grouped, scannable, editable in place.
──────────────────────────────────────────────

Problem: The review table was a flat unsorted list. Care managers had to mentally track what changed between visits. Every field edit required leaving the table. Every panel that opened blocked it entirely.

Self-serve medication review. Grouped, scannable, editable in place.
──────────────────────

Problem: The review table was a flat unsorted list. Care managers had to mentally track what changed between visits. Every field edit required leaving the table. Every panel that opened blocked it entirely.

Key Interface Decision

Key Interface Decision

Grouped sections replaced a flat list.

Grouped sections replaced a flat list.

From last review / New medications / Modified medications / Discontinued - color-coded, always visible. Care managers see what changed the moment they open a review.

Inline editing, introduced to the design system.

Inline editing, introduced to the design system.

Our table component didn't support editable cells. I pushed for it. Fields now edit in place - no forms, no round trips, no lost context. This required working with the DS team to build the pattern from scratch.

Non-blocking panels replaced blocking sidesheets.

Non-blocking panels replaced blocking sidesheets.

Every action used to darken the table. Now the table stays live while a panel is open. Care managers can reference existing medications while adding a new one - the way they think during a patient call.

"Add Medication" was fixed to do what it said.

"Add Medication" was fixed to do what it said.

The button opened the history table. To actually add a medicine, you needed a second click inside the sidesheet. History now lives inline below the active table. The button opens the add form.

Auto-population with a safe fallback chain.

Auto-population with a safe fallback chain.

Dose, route, frequency now auto-fill: EMR table first → NDC lookup → clean blank. Never a zero, never a placeholder.

In clinical tools, a wrong default is more dangerous than an empty field. A care manager will fill in a blank. They might trust a pre-filled "0" and move on. This looks like a backend decision. Deciding what the user sees in every possible data state is a design decision.

Design System

Design System

A constraint in one module became infrastructure for the company.


The dense medication table exposed two gaps in our design system:

  • No small-size components - regular and large both failed at this data density

  • No outlined component variants - only filled styles were supported

Every data-heavy product at Innovaccer had the same problem. Nobody had pushed hard enough to fix it.

I drove it. Requirements, specs, production validation, usage guidelines.


Adopted across 8+ products company-wide. Now part of Innovaccer's Masala Design System.

Self-serve medication review. Grouped, scannable, editable in place.
──────────────────────────────────────────────

Problem: The review table was a flat unsorted list. Care managers had to mentally track what changed between visits. Every field edit required leaving the table. Every panel that opened blocked it entirely.

Self-serve medication review. Grouped, scannable, editable in place.
──────────────────────

Problem: The review table was a flat unsorted list. Care managers had to mentally track what changed between visits. Every field edit required leaving the table. Every panel that opened blocked it entirely.

Before
Before
After
After

Impact

Impact

2×

2×

customer adoption post-launch

8+

8+

products using the 12px component library

Top cited improvements

Top cited improvements

No more manual data entry. Working without constant context switching.

Reflection

Reflection

Structural decisions unlock everything else.

Structural decisions unlock everything else.

Understanding that Review and Reconciliation serve different purposes - and designing for both from day one - made every subsequent decision clearer. The architecture came before the screens.

Constraints are usually signals.

Constraints are usually signals.

When the table couldn't fit regular components, the right move wasn't a workaround. It was fixing it for everyone. What started as a personal problem became infrastructure for the whole company.

The hardest problems are the ones users stopped reporting.

The hardest problems are the ones users stopped reporting.

Duplicate medications, missing data, silent gaps - care managers had adapted to all of it. Finding what they'd normalised mattered more than fixing what they'd flagged.

Next Project

Next Project

Building the Workflow Engine

Built a no-code visual workflow builder and designed an AI copilot framework that reduced strategy setup from 3 weeks to 1 hour, solo end-to-end from research to coded prototype.

Building the Workflow Engine

Built a no-code visual workflow builder and designed an AI copilot framework that reduced strategy setup from 3 weeks to 1 hour, solo end-to-end from research to coded prototype.

Building the Workflow Engine

Built a no-code visual workflow builder and designed an AI copilot framework that reduced strategy setup from 3 weeks to 1 hour, solo end-to-end from research to coded prototype.