Phantom Scans: The POCUS Revenue You Already Earned and Never Billed

A consensus process involving 107 respondents across 63 emergency ultrasound fellowship programmes, representing 52% of US fellowships, adopted phantom scan: a study that has not been documented in the medical record or has no archived images, so that when audited it does not exist (Nomura et al., POCUS Journal 2022).

“Ghost scan” was the alternative on the ballot; in the second voting round 21 of 30 respondents preferred “phantom”. Both terms remain in use.

How much of it there is

At a Level I trauma centre with more than 110,000 annual emergency department visits and around 4,500 POCUS studies a year, researchers matched documented bedside scans against PACS over two years. Among resuscitations where POCUS was documented on the run sheet:

• 70.5% of cardiac arrest scans had no archived images (232 of 329)
• 86.5% of trauma scans had no archived images (347 of 401)

Boivin et al., POCUS Journal 2023.

A follow-up study across four academic Level I trauma centres examined 6,078 trauma patients, of whom 2,182 received an eFAST. Ghost scan rates among those eFAST examinations differed sharply by site: 21.4%, 33%, 65.5% and 93.2% (Boivin et al., Am J Emerg Med 2025).

The four-fold spread between the best and worst site matters as much as the headline. These are academic trauma centres rather than under-resourced departments, and the lowest rate among them is still one scan in five.

The attrition funnel

The most detailed accounting comes from a study of cutaneous abscess POCUS at a large urban academic emergency department with both an emergency medicine residency and an ultrasound fellowship, using 2017 data. Of 710 eligible encounters:

Of scans actually performed, 23.3% produced payment. The department collected $1,400.69 against $6,006.00 achievable on the scans it had already performed, which works out at $11.67 per billed study against a potential $21.22.

Two caveats belong with those numbers. This is one clinical application rather than a whole POCUS programme, and it is priced at the 2017 conversion factor. What it demonstrates is the shape of the attrition, and that shape is consistent with the archiving figures reported elsewhere.

Baseline archive rates from other settings support that. A community emergency department in Toronto covering 72,986 visits over nine months began from entirely unarchived POCUS scanning (Aspler et al., CJEM 2022). At a Canadian internal medicine teaching unit, none of 94 scans performed by first-year residents was saved before intervention (Kolbenson et al., POCUS Journal 2025).

Why it happens

Reading these numbers as a discipline problem leads to a memo, and memos have been tried. Individualised email feedback on scan volume improved documentation adherence from 60.1% (484 of 805 scans) to 71.7% (521 of 727) (Lahham et al., J Med Ultrasound 2022). That is a real improvement, and it requires someone to keep sending the emails indefinitely.

The structural explanation is more useful. POCUS tends to be performed mid-resuscitation or mid-consultation, with the clinician holding a probe in one hand and making a decision within the next half-minute. The scan answers a question and the clinician moves on. Documentation then becomes a separate task, performed later, on a different device, requiring recall of a medical record number and navigation of a workflow designed for scheduled radiology studies.

This explains why phantom rates are highest in cardiac arrest and trauma, where the clinical value of POCUS is greatest and the opportunity to document it is smallest.

It also explains why manual patient identification fails so persistently. A single mistyped medical record number produces a scan sitting in the archive attached to nobody. It will never be billed, and it will never appear in denial reporting, because it never became a claim.

What closing the gap is worth

Three published before-and-after studies give a range to model against.

• An academic emergency department that implemented web-based archival, electronic reporting and automated coder notification saw faculty participation in billing rise from 30% to 75%, billed exams increase 5.1-fold from 857 to 4,449, and approximately $350,000 in net profit in the first year (Adhikari et al., Am J Emerg Med 2014).
• A follow-up at the same institution using bedside documentation with automatic image and report transfer found technical billing rising from 32% to 61% and professional billing from 37% to 65%, both p<0.0001 (Flannigan & Adhikari, J Ultrasound Med 2017).
• An EHR-integrated workflow with automatic demographic retrieval at Baystate Health increased fully documented POCUS from 6.4% to 10.3% of all ED visits, a volume-adjusted gain of 265 scans per month (95% CI 150.60 to 380.09) (Thompson et al., Acad Emerg Med 2023).

Two of those three come from the same institution, and none is a randomised comparison. Note also that these are billing and documentation rates rather than collection rates, and the three studies use different denominators: faculty participation in billing, technical and professional billing capture, and documented studies as a share of all ED visits. They are not directly comparable, and none of them is a collection figure.

What they support is a billing capture range of 60% to 75% of scans performed, against Alerhand’s 42% baseline. Each of those studies measured a workflow in which clinicians still had to initiate documentation, so the top of that range is the reasonable target where capture happens automatically at the point of scan. What this converts into depends on how well your revenue cycle collects on the claims it submits, which is a separate measurement and the subject of the benchmarking article.

The design principle behind all three

Each intervention moved the creation of the billable artefacts to the moment of the scan rather than reconstructing them afterwards.

The least conspicuous version of this is among the most effective. An encounter-based auto-populated worklist at UC Irvine, replacing manual medical record number entry, reduced patient identifier errors from 141 to 90 over matched 138-day periods, a 36% reduction, and recovered approximately $1,800 in professional and $5,600 in technical charges at zero implementation cost (Rowland et al., BMC Health Serv Res 2025).

Scans attached to the wrong record, or to none, never become denials. They cease to exist.