Jefferson Maps Location Intelligence for Population Health Outreach

Operating Context

Short version: Jefferson Maps was a free, built-without-PHI location-intelligence layer for Jefferson Health Enterprise Population Health. I created it from practice-directory spreadsheets, public Jefferson location information, QGIS-created overlays, and Google Maps / Google My Maps functionality so outreach coordinators, navigators, practice coaches, and leaders could see the ambulatory network they were trying to activate.

Working definition: Population health location intelligence means making location, specialty, phone, hours, zip-code, transportation, landmark, and service-area information usable at the moment staff are trying to help a patient act.

This project came from a practical mismatch between the work and the tools.

Jefferson Health’s enterprise population health team was trying to close access and preventive-care gaps across a large ambulatory footprint. The team had excellent people and strong clinical leadership. Many colleagues had deep local Philadelphia knowledge. I was newer to Philadelphia, and that made a hidden workflow problem obvious: the staff member who knew the area best could move faster, but the system should not depend on who happened to know every practice, neighborhood, bus route, landmark, and specialty location from memory.

When a patient asked, “Do you have a location near me?” or “Can I get there without a car?” the answer required more than a practice name. It required proximity, specialty, phone number, hours, capacity awareness, transit access, landmarks, and sometimes a way to explain the route while the patient was still on the call.

The old pattern created four operating risks.

Operating risks

Local-memory routing

Staff could route patients to the practices they personally knew, not necessarily the best or closest available option.

Hidden availability

A practice with availability near the patient could be missed because it was not visible at call speed.

Navigation gaps

Patients who needed more navigation support could be told, implicitly or explicitly, to try again later.

Leadership blind spots

Leadership and practice coaches lacked a simple shared visual layer for zip codes, zones, practice distribution, and access gaps.

The project started as an outreach tool. It became broader population-health infrastructure.

What We Built

Jefferson Maps was built around a spreadsheet-backed practice directory and a Google Maps / Google My Maps-style operating layer. QGIS helped create geospatial layers and overlays. Google Maps provided the familiar staff-facing interface: pins, saved maps, route planning, transit directions, Street View, Earth view, phone/location lookups, and printable directions.

The core layer indexed practices by location and specialty. Over time, the maps could be duplicated, saved, and adapted for different operational questions.

Mapped use cases

Primary care access

Family medicine and internal medicine locations could be viewed as a practical access layer.

OB-GYN and Pap outreach

Staff could see relevant OB-GYN options instead of relying on the few locations they remembered.

Radiology and mammography

Screening-related locations became easier to find and explain during preventive-care outreach.

Ophthalmology

Diabetic eye-exam access could be treated as a location and navigation problem, not only a reminder problem.

GI and surgical centers

Colonoscopy pathways could include GI and outpatient surgical-center context.

Zones and overlays

Zip-code overlays, leadership zones, service areas, and practice-coaching territories could be saved as separate map views.

The work was not a custom app. That was the point. It used tools the team could actually open, understand, duplicate, and maintain.

Build Details

The initial data came from internal reference lists and public Jefferson directory information. Where the team had a partial practice list, I filled gaps from Jefferson’s own public site and organized the information into a spreadsheet. The data model was simple: practice and location information, not patient information.

Each location could carry the details staff actually needed during outreach.

Practice data fields

Practice name

The named location staff could reference during outreach.

Specialty or service line

The clinical category visible in the title, layer, color, or marker.

Address

The location anchor for proximity, driving, walking, and transit context.

Phone or extension

The call path staff needed without searching a separate directory.

Hours

Availability context when the information was available and useful.

Practical notes

Workflow notes that made the public directory facts usable during calls.

Pin and category

Map color and layer structure for quick scanning.

Visual marker

Specialty-specific marker cues so dense locations were easier to read.

The visual legend did real work. We used simple emoji-style marker conventions and color/category differences so staff could recognize specialties without reading every title. Multiple pins could exist at the same address when different specialties operated from the same location, which made dense locations more usable than a single generic marker.

The layer was maintained while I owned the project, then handed off before I left. I trained a colleague on the maintenance path so Jefferson Maps did not stay dependent on my personal knowledge.

Call Workflow

During outreach, the staff workflow could move from Epic queue to map context quickly.

The population-health work queues and supporting spreadsheets gave staff the patient outreach list. In later workflows, an address in Epic could open in the staff member’s browser map. If Google Maps was the default or selected mapping tool, the saved Jefferson Maps layer could sit on top of the patient’s location context. That meant staff could see nearby Jefferson practice options, specialty layers, and routing information without rebuilding the search from scratch.

That changed the call.

Before Jefferson Maps, a patient asking for a convenient practice might hear some version of, “I’m not familiar with that area; I can look it up and get back to you.” After Jefferson Maps, the staff member could say, “It looks like we have two locations within about 10 minutes of you, and one has Saturday availability. How does that sound?”

For patients without a car, the staff member could look at transit access, bus routes, estimated travel time, and walking context. For patients who navigated by landmarks, Street View and Earth view helped staff describe what the area looked like. Directions could be sent as a link, viewed in the patient’s own language through Google, or printed for patients who did not use a smartphone comfortably.

Why The Map Worked

Jefferson Maps solved three problems at once.

Operating value

Efficiency

Staff should not spend repeated minutes searching for the same practice addresses, phone numbers, hours, specialties, routes, and landmarks. Those minutes compound across thousands of outreach attempts.

Access

A patient who is willing to schedule may still fail if the offered location is too far away, unfamiliar, hard to reach, or not explained well enough.

Equity

Transportation, local knowledge, language, health literacy, fear, time, cost, and availability all shape whether a patient can act on outreach. A map layer cannot solve every access barrier, but it can stop the system from making those barriers invisible.

The project also protected staff from a bad operating assumption. The issue was not that outreach coordinators, navigators, or practice coaches needed to work harder. Jefferson Health had the strongest population health team I had seen: clinically led, analytically strong, and serious about access. The system around them needed to make the right next action easier.

We did not need better people. We needed better systems.

Standards, Governance, and Validation

Jefferson Maps was built around a conservative operating boundary: practice and location data, not patient data. The map layer could help staff understand the ambulatory footprint, routing context, specialty categories, and practical navigation options without turning the map itself into a patient-data system.

The governance challenge was maintenance. Location intelligence becomes unsafe if it drifts away from the practice directory, public location data, hours, phone numbers, or operational ownership. The map needed a maintainer, not only a creator, so the handoff and training were part of the actual work.

Results

The 2023 internal study context was focused on access and preventive-care outreach. I use rounded numbers here because the point is the operating pattern, not a false sense of precision.

The operating problem was large enough that small workflow improvements could change the program. The internal project material described roughly 28,000 Jefferson Health Plans patients in the Jefferson system, with roughly 6,200 overdue for either an annual primary-care visit or 3-year Pap measure. The target needed fewer than roughly 4,000 overdue patients.

Internal study frame

28k

patients in system

Roughly 28,000 Jefferson Health Plans patients in the Jefferson system.

6.2k

overdue patients

Roughly 6,200 overdue for annual primary-care visit or 3-year Pap outreach.

<4k

target threshold

The reimbursement target needed fewer than roughly 4,000 overdue patients.

The baseline funnel showed why small workflow improvements could change the result.

Baseline outreach funnel

47%

answered phone

Roughly 47 percent of patients answered the phone.

62%

agreed to schedule

Roughly 62 percent of those patients agreed to schedule.

80%

showed up

Roughly 80 percent showed up to the appointment.

23%

overall success

The overall success rate was roughly 23 percent.

When the reachable population is that constrained, the scheduling conversation has to be excellent. The staff member cannot waste the moment searching for a phone number, guessing which practice is close, or routing the patient only to the locations they personally know.

The broader 2023 process-improvement package that included Jefferson Maps was associated with a roughly 25 percent increase in call-to-schedule conversion. Nearly 90 percent of reached patients agreed to schedule in the post-improvement context. No-show complaints moved from roughly 2-3 per week to roughly 2-3 per month.

Those figures should be read as internal operational evidence, not as a randomized attribution claim for the map layer alone. The map was part of the system improvement.

Scale And Footprint

The documented Jefferson Maps layer covered at least 61 locations. My operating memory is that it expanded far beyond that as family medicine, internal medicine, affiliated practices, OB-GYN, radiology, ophthalmology, GI, and other outpatient locations were added.

Current public Jefferson Health location data gives a useful sense of the footprint. Jefferson’s public location finder now lists more than 450 location records, including more than 300 practices. That current count is not the exact 2023 project count, but it validates the broader point: a population health team working across this kind of ambulatory network needs a shared location layer, not personal memory.

How Different Teams Used It

Department use

Outreach coordinators

Used Jefferson Maps during calls to find nearby practice options, phone numbers, directions, and location context.

Navigators

Used the map to understand where patients could realistically go and how to explain the route.

Practice coaches

Used it to see the practices they supported and how those practices related to the broader geography.

Leaders

Used overlays to understand zones, zip codes, service areas, and project-specific operating footprints.

That portability was part of the value. Once the map existed, different teams could duplicate it, save a version, add an overlay, and answer their own question without waiting for a new software project.

The most important design choice was keeping the tool understandable. A spreadsheet, QGIS layer, and Google Maps workflow are not glamorous, but they gave the department a way to see the work.

Operating Method

The method I would reuse is straightforward.

Reusable method

01

Start with the staff decision

The decision was not 'where is a practice?' It was 'what nearby care option can I confidently offer this patient right now?'

02

Inventory the network

Capture location, specialty, phone, address, hours, and operational notes before designing anything.

03

Separate facts from workflow notes

Keep the source of truth visible enough that the map can be maintained.

04

Build around the user moment

The map has to be usable while a staff member is on the phone, not only useful in a planning meeting.

05

Use visual encoding sparingly

Color and emoji-style markers helped staff scan specialties quickly, but the title and notes still carried the official meaning.

06

Make overlays duplicable

Different leaders and projects need different zip-code, zone, and specialty views.

07

Train against local-memory bias

Even staff who know Philadelphia well should use the map to verify the closest and most appropriate option.

08

Assign maintenance

A map that is not maintained becomes a new source of misinformation.

My Operating View

This project is one of the clearest examples of how I think healthcare systems improve.

Technology does not have to be expensive to be useful. It has to fit the work. Jefferson Maps worked because it solved the problem at the level where the problem actually happened: a staff member, a patient, a care gap, a location question, and a short window where the patient was ready to act.

Healthcare often explains failure as individual failure. I almost never start there. In my experience, the useful answer is usually systems design: make the right action visible, fast, and reliable enough that excellent people can execute it consistently.

The newer academic population-health material sharpens that point. Value-based care and population health can sound abstract when they live in policy language: incentives, quality measures, prevention, interoperability, and care coordination. Jefferson Maps was the same idea at call speed. If the patient cannot understand where to go, how to get there, or which nearby option fits the care gap, the policy goal has not reached the workflow.

That was the point of Jefferson Maps: make the ambulatory network visible at the moment a patient was ready to act.