In-house vs outsourced radiology: costs and trade-offs

"At what point does it make sense to just hire our own radiologist?" In-house vs outsourced radiology turns on one number: somewhere around 50 to 60 studies a day of steady daytime volume, sustained through the year, in the mix one physician is trained and credentialed to read. Below that line, outsourced per-report reading is cheaper almost every time, because a radiologist on payroll costs the same in a slow February as in a record October, and industry compensation surveys put a fully loaded radiologist FTE well past $500,000 a year before you have solved a single overnight hour. Above the line, the in-house hire wins on unit cost inside their own competence.

What the hire leaves behind is the rest of the calendar: nights, weekends, holidays, vacation weeks, and every subspecialty outside one fellowship. AstraRad fills that gap on per-report pricing with no minimums, signs a final report on every study, and publishes turnaround tiers of STAT under 1 hour, urgent under 4 hours, and routine under 24 hours.

In-house vs outsourced radiology: five things to verify before signing

Turnaround measurement, final signed reports, pricing structure, subspecialty routing, and ownership can all be checked in writing before any study moves. Most pages on this topic ask you to trust an adjective; these five will rank any teleradiology option you are looking at, including this one.

Turnaround with the instrument attached. The tiers are STAT under 1 hour, urgent under 4 hours, routine under 24 hours. The clock starts when the last image of the study arrives and stops when the radiologist signs the final report, which is the only interval a clinician experiences. Compliance is measured against those tiers over a trailing 12-month window and published on the SLA page. Ask every vendor for the same three things: the tier, the measurement points, and the window.

Final signed reports, including overnight. Every study comes back as the report of record, signed by the radiologist who read it, licensed in the state where the patient is located and physically located in the United States. That last condition is a payment condition as much as a policy one: the Medicare Benefit Policy Manual, Chapter 16 states that if a radiologist practicing abroad analyzes imaging performed on a beneficiary in the United States, Medicare pays neither the radiologist nor the facility for that work. Nothing comes back to your morning list for a second signature. A preliminary read hands you an interpretation you still have to finish yourself, in your own radiologists' hours.

Per-report pricing with no minimums. No monthly retainer, no platform fee, no volume commitment to grow into. Slow months cost slow-month money. The pricing page explains the model, and the priority multiplier is printed on the rate card itself.

Subspecialty-matched routing on every study. A coronary CTA goes to a cardiac imager. A pediatric abdominal CT goes to a radiologist who reads children all day. The match is made on study type and body region when the study arrives, so the physician who opens it spent a fellowship year on that anatomy and reads it every shift.

Independent ownership. AstraRad is independently owned, and the entity on your signature page is the practice that reads your studies, so the panel you evaluate, the escalation path, and the rate card stay under one owner through your term. Put the same three questions to every vendor, including us: which legal entity signs the contract, who owns it today, and whether that ownership has changed in the last five years.

What does an in-house radiologist cost in 2026?

A fully loaded in-house radiologist costs far more than the salary line, at industry-typical figures that are not AstraRad's. Base compensation is the bulk of it, 65 to 75 percent of the total: industry surveys place radiologist pay among the highest of all specialties, commonly cited above $500,000. Benefits, payroll taxes, and malpractice coverage add 15 to 25 percent of base. Recruiting in a market with a persistent radiologist shortage, which a review in npj Health Systems attributes to rising imaging demand on one side and limited residency positions plus substantial retirements on the other, often runs to a six-figure one-time cost across search fees, signing bonus, and relocation. Call and coverage premiums vary widely. And 10 to 15 percent of paid time is non-productive: vacation, CME, sick leave, and administrative duties, weeks per year when the salary runs and no reports are signed.

Now put the money aside and count hours instead. One radiologist covers something like 45 to 50 reading hours a week, once you subtract protocolling, procedures, phone calls, tumor boards, and the administrative load that lands on the only radiologist in the building. A week has 168 hours. The remaining 120 belong to the emergency department, the after-hours order, and the weekend, which is to say the hours when a study can't wait until morning. You can buy those hours back with call pay, which prices the same physician's fatigue at a premium and eventually costs you the physician. You can buy them with a second hire, which doubles your fixed cost to cover the thinnest volume of the week. You can buy them per report, at the volume in which they arrive.

Training runs the same way: one hire is one fellowship, and a community worklist carries neuro, MSK, body, chest, breast, cardiac, and pediatric studies that no single fellowship spans. A general radiologist reads all of it in most small departments, which is the configuration the subspecialty literature flags when it counts discrepancies.

Take a 2012 AJR series of 773 outside studies reinterpreted by subspecialty radiologists after referral to a children's hospital. Major disagreement with the original report ran 21.7 percent, and where a final diagnosis could be confirmed the second interpretation was the accurate one in 90.2 percent of cases. Those are studies somebody asked to have re-read, so treat the figure as the size of the gap in selected cases rather than a general miss rate. The two problems meet at 3 a.m., when the study handled by the most tired physician on the roster is also the one most likely to sit outside anyone's fellowship.

The recruiting line deserves a source, because the hiring timeline is what the whole cost argument rests on. The ACR's 2026 workforce update reports that US radiologist attrition more than doubled between 2014 and 2022, from 1.1% to 2.5%, with subspecialists 37% more likely to leave the workforce than their peers, set against projected imaging demand growth of 17% for MRI and 25% for CT by 2055 on a radiologist workforce only 20.9% larger. That is the supply side of your search. ACR publishes no figure for how long an open radiologist position stays open, so do not borrow one from a vendor blog: take the elapsed months from your own last posting to that person's first reading day and use that number. Whatever it is, you are buying outside coverage for every week of it, and those weeks belong on the hire's line of the spreadsheet.

None of that argues against hiring. It reprices the hire. What you are buying is a physician in the building for a defined stretch of the week, and everything outside that stretch still needs a signature from somebody.

In-house radiologist vs teleradiology: which one fits your volume?

Teleradiology fits volume that is variable, spread across the whole week, or wider than one fellowship. An in-house radiologist fits volume that is heavy, steady, daytime, and inside the reading competence of the person you employ. The two models are priced on different axes, which is why the comparison stays unresolved for so many facilities: argued in adjectives it has no answer, and argued in hours and case mix it usually has an obvious one. A short primer on the remote model itself is in what teleradiology is.

The table below sets the two side by side on the seven things a staffing decision actually turns on.

What you are deciding	In-house radiologist	Teleradiology panel
Lead time to a first signed report	A search, then credentialing, then a notice period. Measure it from your own last hire, because no workforce survey publishes the figure	First live signed report within 10 business days of countersignature
Fixed or variable cost	Fixed. Salary, benefits, payroll taxes, and malpractice run at the same rate in every month of the year	Variable. One price per signed report, with nothing fixed billed alongside it
Hours covered	Roughly 45 to 50 reading hours out of a 168-hour week	24/7/365 on scheduled US shifts, with nights staffed as a committed shift on a fixed rest interval, every read performed inside the United States
Subspecialty depth	One fellowship, plus whatever that reader is willing to read outside it	240 board-certified subspecialists across ten subspecialties, matched to each study by type and body region
Vacation and sick backfill	Yours to solve, through call pay, a locum premium, or a queue that grows for six weeks	Absorbed by the panel, which carries headroom for 25,000 additional studies a month
Turnaround accountability	An internal expectation, measured only if somebody builds the report	Contractual tiers of STAT under 1 hour, urgent under 4 hours, routine under 24 hours, at 99.4% compliance over the trailing 12 months
What happens when volume drops	The cost is unchanged and the reading day gets shorter	The invoice drops with the volume

Every AstraRad figure in that table is published with its measurement method on the SLA and quality page: the tiers run from last-image arrival to radiologist signature, compliance is 99.4% over the trailing 12 months, the panel is 240 subspecialists across ten subspecialties, headroom is 25,000 additional studies a month, and onboarding reaches a first signed report within 10 business days of countersignature.

Two rows carry most of the decision. Hours is the first: one radiologist covers about a third of the week, and the other 120 hours belong to the emergency department whether or not anybody budgeted for them. Cost shape is the second: a hire converts imaging into a fixed annual commitment, while per-report reading converts it into a unit cost that moves with the work. A facility whose volume is flat all year buys that fixed commitment cheaply. A facility whose volume swings with the season, with a referral pattern, or with one orthopedic surgeon's operating schedule buys it expensively, and pays for the trough twelve months a year.

Where an in-house radiologist is the right answer

An in-house radiologist is the right answer wherever the work requires a physician in the building: procedures, hallway consults, institutional memory, committee work, and heavy steady daytime volume.

Procedures and physical presence. Fluoroscopy, image-guided biopsies, arthrograms, contrast reaction response, and technologist supervision all require a radiologist standing in the department. We sign reports. We cannot hold a needle.

Hallway access. A surgeon who walks into the reading room, pulls the study up on the second monitor, and talks through the plan gets something no callback policy reproduces. Direct phone access to the radiologist who signed the report closes part of that gap, and no teleradiology arrangement closes the rest of it.

Institutional memory. An in-house radiologist learns your referrers, your scanner quirks, your patient population, and where the priors are buried. A panel builds that more slowly, through routing consistency and repeat exposure to your worklist.

Tumor boards and committee work. Multidisciplinary conference presence, protocol committees, and utilization review are jobs for someone on your org chart.

Unit cost at high steady volume. Past break-even, each additional in-house study is close to free at the margin, while each outsourced study still carries its fee.

If two or more of those describe your operation and your daytime volume is heavy and predictable, hire. Then read the next section, because the hire will not have solved your nights.

What are the hidden costs of outsourced radiology?

The hidden costs of outsourced radiology are double reading, communication friction, quality opacity, integration projects, and licensing gaps, and each one has a contract term that closes it. Critical academic coverage of radiology outsourcing documents all five failure modes.

Double reading expense. Preliminary-only overnight reads come back for a morning overread and a second signature, billed in your own staff's hours, which is why this cost stays invisible in a comparison built from vendor invoices. Contract for final signed reports at every tier; AstraRad delivers finals overnight.

Communication friction. Critical findings routed through a call center queue reach the clinician late. Require a documented critical-findings policy with a named escalation path in the SLA.

Quality opacity. Some vendors publish no discrepancy data and no QA method. Ask for measured numbers and their window; AstraRad independently double-reads 1 in 20 signed reports and holds major discrepancies under 0.3 percent.

Integration projects. Platform fees and months of IT engagement can precede the first study. Choose DICOM-native routing: AstraRad accepts studies from your existing PACS or by portal upload and returns reports over HL7 or FHIR.

Licensing gaps. Verify on every report that the signing radiologist is licensed in the state where the patient is located, which is exactly what 42 CFR 482.22(a)(4)(iii) requires of a distant-site physician reading for a Medicare hospital's patients: a license issued or recognized by the state where that hospital sits.

A vendor who will not answer those five in writing has answered them. Our guide on how to choose a teleradiology company turns this into a full evaluation checklist, and teleradiology costs breaks down how the industry prices reads across modalities.

Splitting the week between a core team and a panel

Most facilities converge on a hybrid: an in-house core reads weekday daytime volume, performs procedures, supervises contrast, and owns the clinician relationships, while a per-report panel takes nights, weekends, holidays, overflow, and the subspecialty studies the core group does not read: cardiac MRI, coronary CTA, PET-CT, pediatric cross-sectional work.

Two things make the split work economically. Your in-house team stops taking overnight call, the largest single driver of radiologist burnout and turnover in community practice, so the hire you already made becomes a job a radiologist will stay in. And coverage stops being a fixed purchase: weekend and holiday coverage costs exactly what the weekend volume costs.

The same structure de-risks a staffing transition. When a radiologist retires or gives notice, the panel absorbs the volume for however long the replacement search runs, with no locum tenens premium on top. Capacity is already sitting there: AstraRad carries headroom for 25,000 additional studies a month, and an 8,000-study backlog clears in under 30 days. Facilities running this pattern deliberately use it for overflow radiology reads long before they have a staffing emergency.

When does volume justify hiring your own radiologist?

Your break-even volume is the fully loaded annual cost of the hire divided by your blended per-report rate, then corrected for case mix, nights, and seasonality. Work it in four steps and then test it against the three worked examples underneath. Every figure below is an illustrative industry-typical assumption, none of them an AstraRad price; substitute your costs and the rate card you are quoted.

1. Compute the fully loaded annual cost of the hire. Base compensation, plus 20 to 30 percent for benefits, taxes, and malpractice, plus amortized recruiting, plus the cost of covering that physician's nights, weekends, and time off. In most US markets in 2026 this lands between $600,000 and $750,000.
2. Compute realistic annual capacity. A full-time reader signing a mixed worklist produces on the order of 12,000 to 18,000 reports a year after vacation, CME, and non-reading duties.
3. Compute your blended per-report cost. Weight your modality mix against the vendor's rate card. Published US industry ranges for 2026 run from roughly $10 to $15 for a plain X-ray up to about $99 for PET-CT, broken out by modality in our teleradiology cost breakdown; the professional component in the CMS Physician Fee Schedule is the other public benchmark buyers anchor against. A plain-film-heavy imaging center blends far lower than a CT and MRI-heavy hospital list.
4. Divide. Loaded cost over blended rate gives the annual volume at which the hire breaks even. At $650,000 loaded and a $45 blended rate, that is roughly 14,400 reports a year, about 55 a working day, sustained, in the mix the hire can cover.

Worked examples at 30, 60, and 120 studies a day

The three rows below run the same arithmetic at three volumes, on 250 working days, at an illustrative $650,000 loaded cost per FTE and an illustrative $45 blended per-report rate. Neither figure is an AstraRad price, and both are the first things you should replace with your own.

Steady daytime volume	Reports a year	Per-report cost at $45 blended	Loaded in-house cost	What the arithmetic says
30 studies a day	7,500	$337,500	$650,000, one FTE	Per-report wins by roughly $310,000, and the hire reads under capacity all year
60 studies a day	15,000	$675,000	$650,000, one FTE	Close to a wash on daytime volume alone, and everything outside the daytime decides it
120 studies a day	30,000	$1,350,000	$1,300,000, two FTE	The hires win on unit cost, and the residual volume still goes out

Read the middle row carefully, because it is where most community facilities sit and where the arithmetic stops being the answer. At 60 studies a day the two columns land about $25,000 apart on illustrative figures, which is well inside the error bar of every assumption above them: move the blended rate by $2, or the loaded cost by 4 percent, and the winner flips. What does not move is what each column buys. The $650,000 buys 45 to 50 reading hours a week inside one fellowship. The $675,000 buys every hour of the year across ten subspecialties, and it stops the moment the studies stop.

The 30-a-day row is the clearest case on the page, and it is clear for a structural reason. At that volume no arrangement of a single salary produces a competitive unit cost, because the salary is set by the market for radiologists and not by how busy you keep one. The 120-a-day row is the one that justifies hiring, and it justifies hiring two, since 30,000 reports is past what any single reader signs in a year. Even then the calendar stays open: nights, weekends, holidays, leave weeks, and the cardiac MRI nobody on the roster reads still route out, which is the hybrid split described above. The overnight half of that math is set out on our overnight nighthawk coverage page, where the volume is thinner and the differential is steeper.

AstraRad's side of the arithmetic is one price per signed report, with no minimums, no subscriptions, no platform fees, no per-seat charge, and a fixed priority multiplier on STAT and urgent studies printed on the rate card and itemized on every invoice line. The outsourced column therefore multiplies out cleanly, with no fixed term hiding underneath it and no month in which you pay for studies you did not send. Retainers, monthly subscriptions, per-seat charges, and minimum-volume commitments all behave like a second fixed cost sitting beside the salary you were trying to avoid, so price any quote that carries one at the volume of your slowest month before you compare it. A written rate card priced by study type arrives within one business day of a request.

Three corrections push the practical break-even higher than the raw division suggests. The hire breaks even only on volume inside their own competence, so subspecialty studies keep going out. The hire does not cover nights and weekends, so that line item survives the arithmetic. And the volume has to be steady: a facility that hits 55 studies a day in winter and 35 in summer pays for the gap all twelve months.

Before you post the job

In-house vs outsourced radiology is settled by the calendar before it is settled by the salary. A hire covers roughly a third of the week's hours and one fellowship's worth of anatomy, and the honest question is what you intend to do about the remainder: the 3 a.m. head CT, the Saturday chest pain workup, the cardiac MRI nobody on the roster reads, the six weeks a year the reading room is dark. Answer that with call pay and you buy it back from a physician who is already tired. Answer it with a second hire and you carry a full fixed cost against the emptiest hours on the schedule. Answer it per report and the cost tracks the work.

Run it against your own numbers before you decide. If the gap you are covering is temporary rather than structural, the third option is costed separately in locum radiologist vs teleradiology. Pull your last twelve months of volume by modality, cost your staffing line the way the steps above do, and ask through our contact page for a rate card priced per report by study type. One spreadsheet usually settles the question, and it settles it in a direction most facilities find obvious once the overnight hours are on the page. The rest of the staffing and pricing comparisons live in the teleradiology resource library.