Nighthawk radiology services in the US: how overnight preliminary reads actually work

What a nighthawk service is, and what it is not

A nighthawk service reads the studies your radiologists would otherwise read at 3 am. The emergency physician gets a preliminary interpretation within minutes and treats on it; your radiologist reviews the study in the morning and issues the final. The preliminary is a clinical work product for the night. The final is the official, signed, billed interpretation.

The name comes from NightHawk Radiology Services, founded in Coeur d'Alene, Idaho in 2001. Its 2005 SEC filing describes the design: American Board of Radiology-certified, US state-licensed radiologists in Sydney and Zurich providing US off-hours reads "during our radiologists' local daylight hours"; by June 2005 it served 693 hospitals. The Bangalore lineage runs in parallel. Teleradiology Solutions was started in Bangalore in 2002 by Arjun Kalyanpur and Sunita Maheshwari, began as a pilot with Yale and took early business from NightHawk itself (Business Today, 2011). The model was pioneered from Bangalore as much as from Idaho, and it has been continuous US practice for more than twenty years. The ACR's 2013 White Paper on Teleradiology Practice draws the line: "In contrast to international teleradiology, in which the interpretations are preliminary, domestic teleradiology often provides final interpretations."

What it is not: a final-read service, a replacement for your radiology group, a route to billing offshore reads, or, when run from India, an FDA-cleared AI reading your scans.

The regulation, plainly

Medicare payment: 42 CFR 411.9. The rule reads: "Except as specified in paragraph (b) of this section, Medicare does not pay for services furnished outside the United States." The exceptions concern inpatient care in foreign hospitals. So a read produced in India cannot be the billed professional service; the US radiologist's final interpretation is, and the offshore preliminary is a non-billed input to it. That one sentence is why every offshore nighthawk contract is written as preliminary reads.

Patient data: HIPAA and the Business Associate Agreement. A vendor handling protected health information for a covered entity is a business associate, and 45 CFR 164.504(e) requires a written Business Associate Agreement fixing permitted uses, safeguards, breach reporting, subcontractor flow-down and termination. HIPAA does not prohibit PHI being processed outside the United States (McDermott Will & Emery, July 2025). State overlays exist: Section 6505 of the Affordable Care Act bars state Medicaid payments to entities located outside the US, and Texas managed-care contracts, Florida's Electronic Health Records Exchange Act and Ohio executive orders add data-location limits (Sheppard Mullin, August 2024). These are checked against your payer mix before go-live.

Licensure: the ACR position. The ACR's 2013 white paper states that "medical practice is considered to occur at the location of the patient" and endorses the 2012 ACR Technical Standard's requirement that radiologists know the licensure rules at both transmitting and receiving sites and "obtain licensure as appropriate"; under current law that typically means the transmitting state, where the patient was scanned. The same standard says physicians interpreting a hospital's images "should be credentialed and obtain appropriate privileges at that institution". Both duties sit with the physician who signs the final. 5C Network's radiologists in India hold no US state licence, which is exactly why their output is a preliminary read and why 5C Network never issues the final report for a US patient.

Credentialing by proxy: 42 CFR 482.22 and Joint Commission MS.13.01.01. Under 42 CFR 482.22(a)(3) and (a)(4) a hospital may rely on a distant-site hospital's or telemedicine entity's credentialing, provided the practitioner "holds a license issued or recognized by the State in which the hospital whose patients are receiving the telemedicine services is located". Joint Commission Standard MS.13.01.01, EP 1 mirrors this and since 2021 accepts a distant site that is Joint Commission-accredited or Medicare-enrolled (Foley & Lardner, March 2021). Proxy credentialing covers the US-licensed radiologists who render your finals; it does not extend to an offshore preliminary reader, who is a non-signing input outside the credentialing perimeter.

The timezone table

India Standard Time is UTC+5:30 and has no daylight saving. Figures are US standard time; during US daylight saving (second Sunday of March to the first Sunday of November) every row except Hawaii moves one hour earlier in IST, so 10 pm Eastern becomes 7:30 am IST.

Read the Pacific row: a Los Angeles trauma CT at 2 am reaches Bangalore at 3:30 pm. The radiologist is mid-shift, not woken. That is the whole case for offshore preliminary reads, and it is geography, not stamina.

The economics

The Medscape Physician Compensation Report 2026 puts average US radiologist compensation at $571,000 for 2025, up 9% from $520,000 (The Imaging Wire, April 2026). One overnight seat, every night, takes more than one radiologist once weekends, leave and CME are counted, and recruiting for it is hard: the ACR Bulletin counted more than 1,400 open positions on the ACR Career Center in March 2024, and its February 2026 update reports attrition more than doubled from 1.1% in 2014 to 2.5% in 2022.

Locum cover is the usual fallback. Wellhart's radiologist salary guide cites diagnostic radiology locum rates of up to around $330 an hour; a ten-hour overnight seat every night is 3,650 hours a year, roughly $1.2 million at that rate before agency fees, travel and malpractice. The Neiman Health Policy Institute's 2025 JACR projections say why this does not ease: radiologist supply grows about 25.7% by 2055 while imaging demand grows 16.9% to 26.9%, so the shortage is projected to persist.

The preliminary-read model changes the shape of the cost, not just its size. Your radiologist keeps the billable final and reviews a structured preliminary in the morning instead of reading cold; the overnight work becomes a per-study charge that scales with volume. 5C Network prices US preliminary reads per study, quoted against your modality mix and the hours you want covered, with no minimum volume and no licence fee. There is no published US rate card because a Level I trauma centre and a two-scanner community hospital are different workloads.

What overnight coverage should include

• A 15-minute stat preliminary on CT head, CTPA and trauma CT. 5C Network's measured emergency turnaround is 15 minutes across 15,000+ scans a day; write that number into the SLA.
• A phone call for critical findings. Haemorrhage, large-vessel occlusion, pulmonary embolism, free air, tension pneumothorax: the ordering physician hears it before the written preliminary lands, and the call is logged.
• A structured preliminary. Findings, impression and an explicit "preliminary; final by your radiologist" line, so the morning review is a comparison, not a rewrite.
• Discrepancy tracking against a definition agreed in advance. Benchmarks exist and are not interchangeable. Pfeifer and Dinh (Acta Radiologica Open, 2021) found a 1.6% actionable discrepancy rate across 8,778 after-hours paediatric teleradiology preliminaries. Chong et al. (AJR, 2022) logged 8,444 major and 17,208 minor discrepancies in 5.88 million acute exams at one national provider, about 0.14% and 0.29%. Storjohann et al. (Healthcare, 2021) found clinically important changes in 5.8% of teleradiology on-call CTs against 6.1% in-house, not significantly different. Fix the definition, then measure.
• AI as workflow infrastructure, stated honestly. Bionic, 5C Network's AI, is regulator-approved in India and not FDA-cleared, so in the US it runs triage, worklist orchestration and automated QC and never reads for the radiologist.
• Subspecialty routing and an audit trail. Neuro, MSK, body and paediatric studies go to matching radiologists, and every study records who read it, when, what changed at final and which findings were called.

How 5C Network runs it

1. A HIPAA Business Associate Agreement is executed before any PHI moves. Where a state Medicaid or payer term requires it, 5C Network deploys with US-resident data hosting.
2. Your facility connects over secure DICOM/PACS, typically live within about 72 hours.
3. Overnight and overflow studies route to 5C Network, where AI-driven triage and worklist orchestration prioritise urgent cases and flag studies for QC.
4. An NMC-registered radiologist reads every study during India's working day and produces a structured preliminary, which passes 5C Network's automated QC pipeline (every study, not a sample) before release.
5. Preliminaries land on your US radiologist's worklist by morning with full study context and audit trail. Your US-licensed, credentialed radiologist performs the final interpretation, signs it and bills for it, as 42 CFR 411.9 requires.

The scale behind it: 15,000+ scans a day, 20M+ studies reported, 2,000+ hospitals and diagnostic centres in India, and around 400 NMC-registered specialist radiologists on the network, which is what gives real subspecialty depth at 3 am Eastern. 5C Network is certified to ISO 27001, 27701, 13485 and 9001. The country page is 5C Network in the United States.

For the Indian side of the same service, where 5C Network signs finals under Indian registration, see nighthawk radiology. Day-time and subspecialty reporting sits under teleradiology services, and the radiology reporting software that runs the worklist, QC and audit trail has its own page.

Questions to ask a nighthawk provider

1. Who signs the final report, and in which state are they licensed? If the answer is anyone outside the US, stop.
2. Will you sign our Business Associate Agreement as drafted, and can you host data in the US if a payer contract requires it?
3. What is your measured, not promised, turnaround for stat CT head, CTPA and trauma CT, and over how many studies?
4. How are critical findings communicated, to whom, and how is the call logged?
5. What is your preliminary-to-final discrepancy rate, how do you define "major", and who adjudicates?
6. Are studies routed by subspecialty, and can we see the credentials of the radiologists who will read ours?
7. Is your AI FDA-cleared? If not, what exactly does it do, and does any AI output reach our clinicians unreviewed?
8. What does the audit trail contain, and can we export it for peer review and payer documentation?

Frequently asked questions

Sources

1. eCFR, 42 CFR 411.9: Services furnished outside the United States
2. Cornell LII, 42 CFR 411.9 (mirror text)
3. Cornell LII, 45 CFR 164.504(e): Business associate contracts
4. US Department of Health and Human Services, HIPAA Business Associates guidance
5. Cornell LII, 42 CFR 482.22(a)(3)-(4): telemedicine credentialing and privileging
6. Foley & Lardner (National Law Review, 15 March 2021), Credentialing by Proxy: Joint Commission Standard MS.13.01.01
7. Silva et al., ACR White Paper on Teleradiology Practice, JACR 2013;10:575-585
8. NightHawk Radiology Holdings, Form S-1 (SEC, 2005)
9. Business Today (May 2011), India's teleradiology companies thrive
10. McDermott Will & Emery (1 July 2025), US healthcare offshoring: patient data privacy laws and regulations
11. Sheppard Mullin (National Law Review, 8 August 2024), Navigating the Waters of Offshoring and Patient Data
12. Harvey L. Neiman Health Policy Institute, radiologist supply and imaging demand projections to 2055 (JACR, February 2025)
13. JACR, Projected US Radiologist Supply, 2025 to 2055
14. ACR Bulletin (5 February 2026), The Radiologist Shortage: A Workforce Update from HPI
15. ACR Bulletin (1 March 2024), How Will We Solve Our Radiology Workforce Shortage?
16. The Imaging Wire (16 April 2026), Radiologist Salaries Grew 9% in 2025, reporting the Medscape Physician Compensation Report 2026
17. Wellhart, Radiologist Salary Guide (locum tenens hourly rates)
18. Pfeifer & Dinh, Discrepancy rates of preliminary and final reports for after-hours pediatric teleradiology interpretations, Acta Radiologica Open 2021
19. Chong et al., Discrepancy Rates Among 5.9 Million Examinations From a National Teleradiology Databank, AJR 2022;218:738-745
20. Storjohann et al., The Accuracy of On-Call CT Reporting in Teleradiology Networks in Comparison to In-House Reporting, Healthcare 2021;9(4):405

5C Network figures are the company's own operating data as of September 2026. Regulatory summaries are for orientation, not legal advice; confirm state-specific requirements with counsel.