Skip to main content

Practice Operations

OB-GYN Work and Activity Letters, Pre-Filled From the Chart

Work and activity letters pile up in OB-GYN because every one is bespoke. How AI pre-fills the logistics from the episode record and leaves the clinical line.

7 min read

Work and activity letters are the quiet tax on an OB-GYN practice. A patient needs something in writing for an employer, a school, a gym, an airline, or a landlord, and each request arrives as a sentence rather than a form. There is no packet to fill out and no template that fits, so somebody opens a blank document, retypes the same header they typed yesterday, and walks it to a provider.

The volume is invisible because there is no queue for it. Requests come in through the portal, at the front desk, on a voicemail, and occasionally in a message a patient sends to a nurse about something else entirely. Nothing counts them, so nobody can say whether the practice writes fifteen of these a week or eighty.

The timing makes it worse. A pregnancy episode is a schedule of events, and the letter requests cluster around them: after the first visit, around the point where an employer asks for documentation, and again when duties or hours need to change. Those clusters are predictable in aggregate and unpredictable per patient, which is exactly the shape of work that ends up handled by whoever is standing closest.

Most of the letter is not the letter

Read ten of these side by side and the reusable portion is obvious. There is a letterhead, a date, a patient identity block, a statement of the care relationship with dates, a provider signature block with credentials, and one or two sentences of substance.

Everything except those one or two sentences is derivable. The episode record holds the dates that anchor the letter. The chart holds identity. The practice configuration holds the provider’s credential line and the location’s address, which matters more than it sounds for multi-site groups where staff routinely paste the wrong office’s footer.

The automation’s job is to produce a letter that is complete except for the substance, addressed correctly, with the right dates, on the right letterhead. The provider writes the substance. That is the entire division of labor and it holds for every letter type in this category.

The request is usually vaguer than the letter needs to be

Patients ask for a note. Employers want something specific. The gap between those two is where the second and third drafts come from.

An employer responding to an accommodation request under the Pregnant Workers Fairness Act is working from its own process, and the federal guidance is explicit that the law covers known limitations related to pregnancy, childbirth, or related medical conditions, with accommodations worked out between the worker and the employer. What the practice can usefully supply is narrow and factual. What it should not supply is a negotiation position.

So the highest-value automation here is not drafting. It is intake. When a request arrives, capture who the letter is for, what date range it needs to cover, whether the recipient wants a fax number or a portal copy, and whether the employer supplied its own form. Practices that collect those four items up front cut their rewrite rate sharply, because the second draft almost always exists to fix something nobody asked about the first time.

Episode-linked dates, and why the chart’s shortcuts fail

OB is a dependency graph rather than a series of independent visits. Intake, interval visits, and a shift in provider class later in the pregnancy all sit on a timeline, and a letter that references the wrong point on it is worse than no letter.

Pulling dates from the episode record rather than from the most recent appointment is what keeps this honest. The most recent appointment might be an unrelated visit, a rescheduled slot, or an entry created for a message. The episode carries the structure the letter is actually describing.

There is a second trap in multi-provider practices. Care is often shared across physicians, nurse-midwives, and nurse practitioners, and the person who signs the letter may not be the person on the last visit. The automation should propose a signer based on the episode’s care history and let staff override, rather than assuming the last encounter’s provider. Getting this wrong produces a letter a patient’s employer questions, which generates a call the practice then has to answer twice.

Delivery, tracking, and the copy nobody kept

A letter is not done when it is signed. It is done when the recipient has it and the practice can prove it.

That means the finished document goes back into the chart as a filed document, not just out as a fax or a portal attachment. Practices that skip the filing step discover the gap six months later when the patient asks for another copy and the only record is in a sent-fax log nobody can search.

It also means the outbound leg needs confirmation. Fax destinations for employers and third-party administrators fail more often than portal delivery, and a failed fax is silent. Close the tracked request only when the transmission is confirmed, and keep a short weekly list of letters signed but never confirmed delivered.

When the patient asks for the same letter again in three weeks with a different date range, the tracked history is what makes that a two-minute job instead of a fresh blank document.

The line the automation does not cross

The substantive sentence in a work or activity letter is a clinical statement about a specific patient. It is written by the clinician who knows the patient and it is signed by that clinician. Nothing in this workflow generates it, suggests it, or offers a menu of pre-written options for it.

That restraint is not a limitation to apologize for. It is what lets a practice put the rest of the workflow on autopilot without a compliance conversation. Staff can see exactly where the machine stops, so they stop checking the parts behind the line.

What the automation does own is everything a receptionist would recognize as clerical: capture the request, identify the patient and episode, assemble the document, address it, route it, deliver it, file it, and track it until the recipient has it.

Key Takeaways

  • Create a queue for letter requests. Most OB-GYN practices cannot state their weekly volume, which means they cannot staff for it or notice when it doubles.
  • Capture recipient, date range, delivery method, and whether the employer has its own form at intake. Missing those four is why most letters need a second draft.
  • Anchor dates to the episode record rather than the most recent appointment. Recent appointments are a noisy proxy for where a patient is in an episode.
  • Propose the signer from the episode’s care history and let staff override. In shared-care practices the last visit’s provider is often the wrong signer.
  • File the finished letter back into the chart as a document and confirm the outbound transmission before closing the request.
  • Keep the substantive clinical sentence entirely manual. The clarity of that boundary is what makes the automated remainder trustworthy to staff.

Nobody has ever hired a coordinator to write letters, and yet in most OB-GYN practices somebody spends a meaningful part of every week doing exactly that. The letter itself takes a clinician under a minute. Everything wrapped around it, the request, the header, the dates, the fax, the copy in the chart, is clerical work that can run on its own once somebody decides to treat it as a workflow instead of a favor.

Sources

Ready to See It in Action?

See how Pretty Good AI assembles letter drafts from the episode record and routes them for provider signature.

Schedule a Demo →

Written by Kevin Henrikson