What Pre-Visit Chart Review Actually Looks Like in an FQHC

Pre-visit chart review is the single highest-value prep work a clinic can do. It is also the task most FQHCs skip or do poorly because it takes too much time.

The math is simple. A thorough chart review takes 5 to 15 minutes per patient. For a provider seeing 71 patients in a week, that is 6 to 18 hours of manual work. Most clinics cannot spare that time, so one of three things happens: the nurse does a rushed version that misses critical gaps, the provider wings it during the visit and loses time digging through old notes, or nobody does it and care gaps pile up.

This is not a staffing problem. This is a workflow problem. Chart review is high-value work that should run automatically. When it does, providers walk into clinic with a clear picture of what every patient needs before the visit starts.

Here is what that looks like.

What Pre-Visit Chart Review Actually Covers

Most clinics think chart review means "scan the last note and check for refills." That is not chart review. That is triage.

Real chart review covers six areas:

Problem list reconciliation. Active problems that should be marked resolved. Resolved problems that flared up again. Conditions documented in notes but never added to the problem list. A patient with uncontrolled diabetes in every progress note but no diabetes diagnosis on the problem list is a documentation gap and a coding gap.

Medication reconciliation. What is prescribed in the EHR versus what the pharmacy dispensed versus what the patient reports taking. If the chart says the patient is on metformin 1000mg twice daily but the last fill was six months ago, that matters. If the patient is on three different blood pressure meds but their BP is still 160/95, that matters.

Lab trends. A1c trending up over three visits. eGFR declining steadily. A lipid panel that is overdue by 18 months. TSH that was high six months ago with no follow-up order. These are not one-off data points. These are patterns that only surface when you look at the chart longitudinally.

Care gaps. Overdue screenings (mammogram, colonoscopy, diabetic eye exam). Missing immunizations (flu, pneumococcal, Tdap). Gaps in chronic disease management (annual foot exam for diabetics, urine microalbumin for CKD). Every FQHC has a list of HEDIS or UDS measures they track. Chart review is where you catch them before the visit, not after.

HCC gaps. Conditions the chart supports but are not coded properly. A patient with documented diabetic neuropathy but no ICD-10 code for it. A patient on diuretics for heart failure who only has "hypertension" on the problem list. Under CMS HCC V28, specificity matters. Diabetic nephropathy, diabetic retinopathy, and morbid obesity all have HCC implications that affect risk adjustment. If the documentation supports it but it is not coded, you lose capture.

Open referrals and stalled orders. A cardiology referral placed three months ago with no follow-up. A colonoscopy order that was never scheduled. A specialist report that came back but was never reviewed. These are the things that fall through the cracks when nobody is watching the full chart.

This is what a complete chart review looks like. And this is why most clinics do not do it. It takes too long.

What It Looks Like When It Runs Automatically

When MedScrub connects to eClinicalWorks and syncs overnight, chart review stops being a manual task.

Here is the workflow. The clinic schedules patients for the week. MedScrub pulls the appointment list from eCW overnight via FHIR. For each scheduled patient, it pulls their full clinical record: demographics, encounters, problem list, medications, labs, immunizations, referrals.

Each chart runs through the review workflow. Problem list reconciliation happens first. The system checks for conditions documented in notes but missing from the problem list. It flags active problems that have not been addressed in six months. It surfaces resolved conditions that were reactivated in recent visits.

Medication reconciliation happens next. Prescribed meds are cross-referenced with pharmacy dispense data. Gaps are flagged. If a patient is supposed to be on a statin but has not filled it in nine months, that shows up. If a med was discontinued in the last visit but is still on the active med list, that shows up.

Lab trends get analyzed. A1c values over the past year. eGFR trajectory for CKD patients. Lipid panels, thyroid panels, vitamin D levels. The system does not just pull the most recent value. It shows the trend and flags abnormals that need follow-up.

Care gaps get checked against the clinic's quality measures. Overdue screenings, missing immunizations, chronic disease management gaps. These are not generic reminders. These are specific to the patient's age, sex, problem list, and previous care history.

HCC gaps get surfaced based on V28 hierarchies. If the chart documents diabetic nephropathy but only codes diabetes, the system flags it. If the patient is on heart failure meds but the problem list only says "hypertension," the system suggests heart failure as a diagnosis to review.

Open referrals and pending orders get pulled from eCW's referral and order modules. Any referral older than 30 days with no documented follow-up gets flagged. Any pending lab or imaging order that was never completed gets flagged.

By the time the provider opens MedScrub before clinic, all 71 charts are reviewed. The provider sees a dashboard with every patient's summary: key problems, current meds, recent labs, care gaps, HCC opportunities, open referrals. No digging through eCW. No pulling up old notes. No manual checklist.

The provider clicks into a patient chart and sees exactly what needs attention. A diabetic patient with an A1c of 8.1 who is overdue for a foot exam and an eye exam. A hypertensive patient whose BP has been above goal for three visits and who has not had labs in 14 months. A patient on metformin whose last pharmacy fill was six months ago.

This is not a summary. This is a pre-visit action plan. The provider walks into the visit knowing exactly what to address.

How the eClinicalWorks Connection Works

MedScrub connects to eClinicalWorks via FHIR R4. This is not screen scraping. This is not a manual export. This is a registered eCW FHIR app that syncs data nightly using eCW's FHIR API.

eClinicalWorks exposes patient data through its FHIR server. The FHIR resources MedScrub pulls include:

• Patient: demographics, contact info, preferred language
• Encounter: visit dates, visit types, chief complaint
• Condition: problem list with onset dates, clinical status (active, resolved, recurrence)
• MedicationRequest: prescribed meds, dosage, frequency, start/stop dates
• Observation: labs, vitals, social history
• Immunization: vaccines, dates, lot numbers
• ServiceRequest: referrals, pending orders

The sync happens overnight. MedScrub pulls the appointment schedule for the next day or week, identifies the patients, and fetches their clinical records. The data gets stored in MedScrub's local database, de-identified, and processed through the chart review workflow.

eCW's FHIR surface is not perfect. Some data is harder to pull than others. Referral status tracking is inconsistent. Medication dispense data depends on whether the clinic has integrated with SureScripts or a pharmacy benefits manager. Clinical notes are available as DocumentReference resources, but the text is often unstructured and needs NLP to extract meaningful information.

That said, eCW's FHIR implementation is solid for the core clinical data. Problem lists, meds, labs, and immunizations come through clean. If you want deeper technical details on the eCW FHIR connection, see our eClinicalWorks integration guide.

Where the AI Fits

MedScrub uses AI to analyze chart data, but it does not send raw patient records to OpenAI or Anthropic. Every record gets de-identified before it touches an AI model.

For structured data, de-identification is straightforward. Patient names, MRNs, dates of birth, addresses, and phone numbers get stripped out. What the AI sees is a field map: [PATIENT_1] with problem list entries, med list entries, lab values, and care gap flags. Dates get shifted consistently so temporal relationships stay intact, but the actual dates do not.

For unstructured data like clinical notes, de-identification is harder. A progress note might say "Jane Doe presents today with worsening shortness of breath. She was last seen on 10/02 and her A1c was 8.1." That note contains a name, a date, and clinical information.

MedScrub runs five layers of NLP de-identification:

1. Regex patterns for common identifiers (phone numbers, SSNs, MRNs, zip codes)
2. scispaCy for biomedical entity recognition (patient names, provider names, locations)
3. Stanza for linguistic parsing and named entity recognition
4. BioBERT for contextual understanding of medical terms versus identifiers
5. MedScrub's custom models trained on FQHC note patterns

After all five layers run, the note becomes: [PATIENT_1] presents today with worsening shortness of breath. [PRONOUN] was last seen on [DATE_1] and [PRONOUN] A1c was 8.1.

The AI processes the de-identified text and returns structured findings: problems to add to the problem list, meds to reconcile, labs to follow up, care gaps to address. Those findings get re-identified and linked back to the patient record before the provider sees them.

The clinic controls which AI model runs. MedScrub supports OpenAI (GPT-4o), Anthropic (Claude 3.7), Google (Gemini 1.5), and local models via Ollama. If a clinic wants to keep everything on-premise, they can run Llama 3.3 locally and never send data to a third party. If they want the best accuracy, they can use Claude or GPT-4o with full de-identification.

The AI is not making clinical decisions. The AI is surfacing patterns the provider needs to see. The provider reviews the findings and decides what to act on.

What Changes for the Clinic

The immediate change is time. Chart review that used to take 6 to 18 hours per week now takes zero manual effort. The nurse or MA who was spending an hour before each clinic session prepping charts can focus on patient calls, prior authorizations, or care coordination.

The provider spends less time hunting through eCW during the visit. They open MedScrub, see the pre-visit summary, and walk into the room knowing what the patient needs. The visit becomes more efficient. There is less "let me pull up your labs real quick" and more "your A1c came back at 8.1, let's talk about adjusting your metformin dose."

HCC capture improves. MedScrub flags conditions that meet V28 criteria but are not coded. A patient with documented diabetic retinopathy who only has "diabetes" on the problem list gets flagged. The provider reviews the documentation, confirms the diagnosis, and adds it to the problem list. That captures the HCC. Over a year, that adds up to better risk adjustment and better revenue.

Care gaps close faster. Overdue mammograms, missed flu shots, and pending colonoscopies get surfaced before the visit, not six months later during a quality review. The provider can address the gap during the visit or delegate it to the care team.

Medication adherence improves. When the system flags that a patient has not filled their metformin in six months, the provider can ask about it during the visit. Maybe the patient stopped taking it because of side effects. Maybe they could not afford it. Maybe they just forgot. Either way, the gap gets addressed instead of ignored.

Referrals stop falling through the cracks. If a cardiology referral has been open for three months with no follow-up, MedScrub flags it. The provider can check in with the patient during the visit and make sure the appointment got scheduled.

The clinic's quality scores improve. UDS measures improve. HEDIS scores improve. Patient outcomes improve because gaps get caught and closed before they become problems.

This is not a hypothetical. This is what happens when chart review runs automatically.

How to Get Started

If your clinic runs eClinicalWorks, MedScrub connects in minutes. The FHIR integration is already built. You point MedScrub at your eCW FHIR server, provide the necessary credentials, and the sync starts.

We offer a free 2-week evaluation on your own data. No mock patients. No demo environment. Your actual scheduled patients, your actual charts, your actual workflows. You see exactly what MedScrub surfaces for every patient before you commit.

Start your evaluation or talk to us if you have questions about the integration, the AI models, or how chart review fits into your clinic's workflow.

Pre-visit chart review is too valuable to skip and too time-consuming to do by hand. When it runs automatically, providers walk into every visit prepared. That is what good clinical workflow looks like.