What is a radiology reading service? A plain explainer

Somewhere in your operation, imaging is being produced faster than it is being read: an urgent care X-ray queue at 7 p.m., a weekend CT list waiting for Monday, an ultrasound backlog behind one part-time radiologist. A radiology reading service exists for precisely that arithmetic; it interprets the studies your facility produces and returns signed reports, without you employing another radiologist.

This explainer covers what the service actually is, the coverage shapes it comes in, and, most usefully, how it plugs into an operation you already run, because the plugging-in is where buyers' real questions live. It stays plain and operational; the deeper commercial anatomy of the category has its own article.

Since this site is itself operated by one: AstraRad is a subspecialist reading service, 240 fellowship-trained radiologists across ten subspecialties, final signed reports at every tier, per-report billing with no minimums. The worked examples below use those terms where a concrete number helps, and a request today puts a written rate card, priced line by line for your mix, in your inbox by tomorrow.

What is a radiology reading service?

A radiology reading service is a contracted radiology practice that supplies interpretations for hire. Your facility keeps everything physical: the scanner, the technologists, the patient encounter, the facility billing. The service supplies the physician work, the interpretation and the signed written report, which Medicare's framework calls the professional component of the study, defined in the CMS Claims Processing Manual, chapter 13.

Mechanically, nearly every reading service today is a teleradiology operation: studies travel by DICOM over encrypted connections, a radiologist licensed in your state interprets on a diagnostic workstation, and the report returns electronically. The vocabulary split is historical; "reading service" stuck among plain-film-heavy buyers like urgent care, orthopedics, and chiropractic, while hospitals adopted "teleradiology," and the pillar guide to how it all works covers both under one roof.

Three boundaries define the service honestly. It is a physician service: a named, licensed radiologist signs every report, which is what distinguishes the category from software. It is diagnostic only: procedures stay with on-site physicians. And it is bounded by the contract's product definition: whether the reads are final signed reports or preliminary interpretations your own radiologists must finalize, the single term that most changes what you are buying, unpacked in STAT vs preliminary vs final reads.

What does a reading service cover?

Coverage has two dimensions: which studies, and which role the service plays for them.

On the study dimension, the modality list runs from plain-film X-ray through CT, MRI, ultrasound, nuclear medicine, and PET, with mammography reading available on displays meeting FDA MQSA requirements. Services differ on the edges: cardiac MRI, coronary CTA, and advanced oncologic imaging need scarce subspecialty readers, and generalist services either decline them or read them anyway, which is worth knowing before you send one.

The role dimension is where contracts actually differ:

Coverage shape	The service's role	Typical buyer
Primary reads	The interpreting radiologist of record for the covered volume; the signed report is the diagnostic document	Facilities with no radiologist, or none for that modality or window
Over-reads	A radiologist's read layered on an initial clinical impression, confirming or refining it	Urgent care and orthopedics, where a treating clinician acted on their own review first
Overflow and window coverage	Defined slices: nights, weekends, holidays, leave, volume above group capacity	Facilities and groups with radiologists who cannot cover everything

The over-read row deserves one caution from the reading side: an over-read is a full interpretation with a signed report, done with the same care as a primary read, and the discrepancies it surfaces against the initial impression need a documented communication path back to the treating clinician. The ACR practice parameter on communication of diagnostic imaging findings treats discrepant findings as one of the three scenarios demanding expedited, documented communication, so ask any service how an over-read that changes management gets phoned, and how fast.

Most real contracts combine shapes. An urgent care network buys primary reads on everything it shoots; a community hospital buys nights and weekends around its employed radiologist; an imaging center buys overflow above its contracted local group plus subspecialty sends for advanced MRI. The combinations are the point, and a service that prices per study lets the combination follow your actual volume.

How does a radiology reading service plug into your operation?

The integration question is really three questions: how studies get out, how reports get back, and how the humans connect. None of them requires changing how your technologists work.

Studies out. Your PACS or the modality itself pushes by DICOM to the service, over site-to-site VPN or TLS. Priors and order data travel with the study, and that clause carries more report quality than any other technical detail: a reader without priors and history writes a thinner report, every time. Sites under a few studies a day skip the interface and upload through a secure portal, which works from day one and converts to interfaces later; the mechanics live in PACS integration for teleradiology.

Reports back. Signed reports return as HL7 ORU messages into your RIS and EMR, with PDF delivery anywhere you want a copy: fax paths for referrer offices still exist in 2026, and good services still feed them. Delivery is the clock's endpoint that matters clinically, so confirm reports land where your clinicians actually look, in the format your referrers already read.

Humans connected. The phone tree for critical findings, with names and direct lines, walked before the first live study. The escalation path when a STAT read is aging. The route by which a clinician questions a signed report and reaches the interpreting radiologist. These three lists decay as your staffing changes, and re-walking them quarterly is the cheapest quality control in the whole arrangement.

Time-wise, the pattern repeats across the industry: the technology connects in days, and licensure verification plus any hospital credentialing set the real go-live date, typically two to six weeks. AstraRad's onboarding commitment inside that pattern is a first signed report within 10 business days of countersignature where licensure is in place, with the workflow agreement, tiers, phone tree, escalation, settled in one session with a named lead; the fuller sequence is on how we work.

What changes on day one, and what does not

Buyers overestimate the disruption, so here is the honest ledger.

What does not change: image acquisition, technologist workflow, your PACS, your viewer, your archive, your patient billing for the technical component, and your referrer relationships. The service is downstream of all of it.

What changes: who reads the covered volume and how fast. A defined slice of your worklist now returns on a contractual clock, STAT under 1 hour, urgent under 4, routine under 24 on AstraRad's published tiers, and the reports carry an outside radiologist's signature. Your morning worklist changes shape if you were running preliminary overnight coverage before, because finals eliminate the re-read pile. And your quality program gains an input: the monthly report, turnaround by tier, SLA compliance, the discrepancy log, that a measured service delivers and your imaging committee can actually read.

One change to plan deliberately: your own radiologist's role, if you have one. The stable configurations give the employed or local radiologist the work that benefits from presence, procedures, consultation, the studies they prefer, and route the rest; the unstable ones let the service's coverage creep into the work the radiologist counts on, which is a staffing conversation, and better had at contract design than discovered at renewal.

Three operations, three configurations

The abstractions above become concrete quickly when mapped onto real facility shapes.

A six-site urgent care network, roughly 60 X-rays a day plus occasional CT sends. Configuration: primary reads on all plain films with same-day finals, urgent-tier turnaround so dispositions happen inside the visit, and subspecialty routing for the CTs. Connection: DICOM push from each site's modality, portal upload as the fallback, reports back into the EMR the clinicians chart in. The number that matters most in the contract is the urgent tier ceiling, because a patient in a waiting room is the clock; the urgent care radiology use case works this profile in detail.

An outpatient imaging center, 900 studies a month, cross-sectional heavy, one contracted local radiologist three days a week. Configuration: the local radiologist keeps the days and studies they prefer; the service takes overflow, the other weekdays, and the advanced MRI the radiologist would rather not read alone. Priors transmission is the make-or-break technical item, since elective imaging is comparison imaging. The quality report earns its keep here too: referrer-facing centers live on report consistency, and the monthly discrepancy log is the early warning.

A 40-bed community hospital with one employed radiologist. Configuration: the employed radiologist owns weekday daytime, procedures, and the medical staff presence; the service owns nights, weekends, leave, and STAT around the clock, delivering finals so mornings start clean. Credentialing by proxy carries the administrative load, and the phone tree gets walked with the ED charge desk, since the 2 a.m. critical finding is the whole reason the night half exists. The rural overnight coverage use case extends this shape to its hardest version.

Three different operations, one common property: in each, the reading service wraps around what already works instead of replacing it, and the contract's coverage definitions are just these paragraphs written formally. The configurations also evolve in a predictable direction; nearly every facility that starts with a narrow slice, weekend coverage, say, or CT sends only, widens it within the first year once the turnaround and report quality are measured, because the marginal study costs one report and no renegotiation. Design the contract so widening is an email, and never an amendment.

Who watches the watchers: quality oversight of a reading service

Plugging in a reading service adds an external physician group to your quality perimeter, and the oversight is simpler than committees fear if the service measures itself. The inputs you should receive without asking: a monthly report with turnaround by tier and SLA compliance, the discrepancy log with case-level review outcomes, and the peer review program's sampling rate and blindness in writing. AstraRad's versions are 1 in 20 reports blind double-read by a second subspecialist and major discrepancies under 0.3 percent, reviewed monthly and closed with the reader; whatever service you use, the structural demand is identical, and RADPEER scoring decodes the numbers.

Your side of the oversight is lighter but real: route the monthly report to a named person who reads it, feed your own clinicians' report complaints back through the discrepancy process instead of the grapevine, and sample a handful of reports quarterly with your medical director. A reading service that resists any of this is telling you how it will behave when a real discrepancy surfaces.

What it costs, and how to start

Per-study pricing dominates the category. Published market context: NDX Imaging lists starting rates from $12 per study, checked September 2026, and industry-typical bands run $10 to $15 per X-ray, around $28 for ultrasound, $30 to $50 for single-region CT, and $45 to $75 for single-part MRI, with STAT adding 20 to 50 percent; all third-party figures, none of them AstraRad prices. The structure around the rate, minimums, platform fees, surcharge stacking, moves your effective cost more than the rate itself, and the full anatomy is in teleradiology cost per read.

Starting is cheaper than evaluating, which inverts most procurement instincts. A reading service priced per report with no minimum lets you run a twenty-study pilot for the cost of twenty reports: send a representative slice, put the returned reports in front of your medical director, score them against the clinical questions asked and the clock they came back on. The evaluation framework for the companies themselves covers the paperwork half of the decision. For the pilot half, AstraRad's terms are the ones already stated: no minimums, final signed reports, and a written per-report rate card for your study mix within one business day of a request.