Ask a radiologist about structured reporting and you get a reasonable objection before you finish the sentence.

The RealityA paragraph a person reads, or data a system can use

⚠️

The Problem

Every measurement locked in prose.

A downstream system that needs one value has to parse a sentence, or a human has to retype it.

🔧

The Solution

Structure the fields that get reused.

Discrete elements where something downstream consumes them, narrative where only a person reads it.

✅

The Resolution

A report that serves both readers.

Still readable by a clinician, now queryable by a registry, a recall list or a comparison.

The written report requirement, which is the floor

Before the structure question, there is a baseline obligation worth stating precisely.

The Medicare Claims Processing Manual, Chapter 13, covering radiology services and other diagnostic procedures, states the rule for the professional component in one sentence: the interpretation of a diagnostic procedure includes a written report.

That is the floor, and it is about existence rather than format. Nothing in it prescribes structure. What it does establish is that the report is not an optional artifact of the interpretation; it is part of it. Everything in this piece is about making that required document do more work than it currently does.

What the standard defines for measurement exchange

The mechanism for moving structured findings between systems is older and more specific than most people assume.

DICOM describes itself as the standard for the communication and management of medical imaging information and related data, addressing the exchange of digital information between medical imaging equipment and other systems. One point of attribution is worth getting right: DICOM is developed by the DICOM Standards Committee, an independent international standards development organisation. The National Electrical Manufacturers Association and its Medical Imaging and Technology Alliance division serve as the Secretariat. Saying DICOM is maintained by NEMA is a common error.

The standard goes considerably further than images. Its content mapping resource defines report templates for specific clinical domains, including a set for obstetric and gynaecological ultrasound covering patient characteristics, procedure summary, fetal biometry, fetal biometry ratios, long bones, cranium, biophysical profile, amniotic sac, early gestation, ovaries, follicles, pelvis and uterus, vascular measurement groups and the fetal anatomy survey. That is fifteen named templates in one domain alone.

Report templates and where adoption actually stalls

The failure mode is consistent enough to describe precisely.

A practice adopts a template set built by somebody else, for a different case mix, with every possible field present because the author could not know which would be needed. The radiologist meets a form with forty fields for a study that needs nine, works slower, produces a report padded with normal statements nobody asked for, and concludes reasonably that structured reporting is a step backwards.

What works instead is narrower: structure only the elements something downstream consumes, and leave the rest as narrative. That set is growing, because the certification criteria at 45 CFR 170.215 now oblige certified systems to expose patient data through a standardised API, which means more downstream consumers exist than did five years ago. If no system, registry, recall process or comparison ever reads a field, that field has no reason to be discrete. The test is not “could this be structured” but “does anything read it”. Getting that judgement right is the whole job, and it is why EHR template development is an engagement rather than a download.

Coverage determination and what has to be documented

One more documentation obligation sits underneath the report, and it is about necessity rather than format.

CMS’s national coverage determination on ultrasound diagnostic procedures states plainly that failure to provide documentation of the medical necessity of tests may result in denial of claims, that tests for screening purposes performed in the absence of signs, symptoms, complaints or personal history of disease or injury are not covered except as explicitly authorised by statute, and that tests not ordered by a treating physician or other qualified treating non-physician practitioner acting within the scope of their licence will be denied as not reasonable and necessary.

That determination dates from 2007 and still references ICD-9-CM, which is worth knowing when you read it. It remains current policy. The practical consequence is that the order and its indication matter as much as the report, and a structured report that captures the indication as a field rather than as an opening sentence makes the pair far easier to audit.

Where to start if you are starting

A short and unglamorous sequence.

Find the three or four fields something downstream already needs and currently gets by parsing text or by a person retyping. Structure those and nothing else. Measure report turnaround before and after, and be prepared to revert a field that costs more than it returns. Expand only where a new consumer appears. A template set that grows because a system needs a value will be used; one that grows because a field existed will not.

• Structure what something reads. Not what could be structured.
• Measure turnaround, all of it. Including the fields you regret.
• Keep the narrative. Both readers matter.
• Capture the indication as a field. The order is audited alongside the report.

Qventive works as a structured radiology reporting consultant for New Jersey practices, normally within IT support for radiology practices rather than as a standalone project.

Where reporting depends on how systems exchange data, that work runs through radiology EHR and IT support and our interface engine, and every route a report takes belongs in your HIPAA risk analysis.

Structured reporting: common questions