The shared core plus specialty extension model
Specialty-specific EHR templates work best when they are not actually separate templates. The durable pattern is a shared clinical core that every note in the hospital contains — identifiers, presenting complaint, allergies, current medications, vitals, assessment, and plan — extended by a specialty block that adds only what that specialty genuinely needs.
This matters for two reasons. Clinically, a physician covering a ward at night needs to find the same information in the same place regardless of which specialty admitted the patient. Structurally, a shared core means the fields that feed decision support, reporting, and the discharge summary exist identically everywhere, so downstream logic does not need a variant per specialty.
The failure pattern is the opposite: each department designs its own note from scratch, the core fields end up named differently and positioned differently in each, and within three years the hospital has a hundred templates that share nothing and can be reported on only through text search.
Defining the core, and defending it
The core should be small enough that no specialty finds it burdensome and complete enough that it can stand alone as a usable note. Getting agreement on it is a governance exercise, not a design one — the content is rarely controversial, but the willingness of each department to accept a common structure often is.
Anchor the discussion in downstream dependencies rather than preference. A field belongs in the core if something else consumes it: a decision-support rule, the discharge summary, a coding workflow, a quality indicator, a statutory return. A field that feeds nothing has no claim on universal placement, whatever anyone's habit.
Once agreed, the core needs protection. Specialties may add to their block but may not remove, rename, or reposition core elements. Enforce this in the configuration tooling rather than by policy alone, because a rule that depends on everyone remembering it will not survive a year of change requests.
Typical contents of a defensible shared core
- Patient identifiers, encounter context, and treating unit
- Presenting complaint and relevant history
- Allergies and current medications, drawn from the coded record
- Vital signs and any early warning score
- Assessment, coded diagnosis, and plan
What each specialty actually needs to add
Specialty blocks should be built from the clinical decisions that specialty makes repeatedly, not from a wish list. Cardiology needs functional class, structured ECG and echocardiographic findings, and risk assessment elements that feed treatment decisions. Orthopaedics needs laterality captured unambiguously, mechanism of injury, range of motion, neurovascular status, and implant details that must be traceable years later.
Paediatrics needs weight-band and age-appropriate fields, growth and development parameters, immunisation status, and feeding history — and, critically, dose calculation that depends on a current weight being captured as a structured value rather than written into narrative. Obstetrics needs gestational age, gravidity and parity, antenatal risk factors, and a structure that spans a pregnancy rather than a single encounter.
The common thread is that each specialty block is justified by an action. Laterality in orthopaedics prevents wrong-site procedures. Weight in paediatrics drives dosing. Gestational age in obstetrics drives every clinical threshold. Fields that do not drive an action belong in narrative, where they cost nothing to maintain.
The test for a proposed specialty field
- Does a clinical decision, calculation, or protocol depend on it?
- Does a report, indicator, or statutory return consume it?
- Would its absence be noticed within a week?
- Is it captured reliably by the person the form asks?
- If none of the above apply, does it belong in narrative instead?
Design rules that keep templates usable
The most common template defect is length. A form that requires scrolling past fields irrelevant to today's encounter trains clinicians to skip, and skipping is indiscriminate. Show the specialty block conditionally where the system allows it, keep the default view to what is needed for a typical encounter, and put the uncommon fields behind an expansion.
Be sparing with mandatory fields. Each one is a promise that the information is genuinely required and will genuinely be used, and a form with many of them produces placeholder entries rather than data. Reserve mandatory status for the small set where a blank is unsafe, and use soft prompts elsewhere.
Default values deserve specific scrutiny because they silently become the recorded data. A pre-ticked normal examination is a documentation integrity problem waiting to appear in a case review. Defaults are appropriate for administrative fields and dangerous for clinical findings.
Governance: who may create a template
Template sprawl is a governance failure with a predictable mechanism. A consultant asks for a small change, the request is handled as a support ticket, a copy of an existing template is made and edited, and nobody ever retires the original. Repeat this a hundred times and the catalogue becomes unmanageable.
The fix is a single owning group with a defined process: requests are submitted with a stated clinical purpose, changes are made to the existing template where possible, a new template requires explicit justification for why an extension will not serve, and every template has a named clinical owner and a review date. Copying a template must not be a self-service action.
Maintain a visible catalogue: every template, its owner, its purpose, its usage volume, and when it was last reviewed. HealUDoc template configuration can be versioned and audited so a change is a controlled release with a rollback path rather than an edit made directly in the live environment.
“We found eleven versions of the same orthopaedic admission note. Nine had been created by copying, none had an owner, and three were still in daily use by one consultant each.”
Controlling sprawl once it has already happened
Most hospitals reading this already have the problem, and the remedy is a rationalisation exercise. Start with usage data: pull every template with its usage count over the last year. There will be a long tail used a handful of times, and that tail can usually be retired with limited consultation.
Then cluster the remainder by what they actually capture. Near-duplicates that differ in field order and wording rather than in content can be merged into one template with the union of genuinely distinct fields. This is where clinical involvement is essential, because the differences that look cosmetic occasionally are not.
Retire rather than delete. Mark a template inactive so no new note can be created from it, while existing notes render exactly as they were authored. Rewriting historical notes to fit a new template is never acceptable — the record must continue to show what was documented at the time.
Measuring whether the templates are healthy
Template quality is measurable without subjective review. Look at completion rates for optional fields, because a field completed almost never is either badly placed or unnecessary. Look at fields where one value dominates overwhelmingly, which usually indicates a default nobody changes. Look at the ratio of structured entry to free text within the same note, which shows where clinicians are routing around the form.
Add a time measure: median time from encounter end to note completion, by specialty and template. A template that is materially slower than its peers is either capturing more, or capturing worse. HealUDoc dashboards can report field completion and note timing per template, which turns template review from an opinion exercise into an evidence-based one. Comparing a slow template against what its extra fields actually feed usually resolves which of the two explanations applies.
Review a small number of templates each month rather than attempting a full audit annually. A standing rhythm keeps owners engaged, catches drift early, and — most usefully — makes the retirement of a field an ordinary event rather than a negotiation.
