About Aleris Ultragyn

Ultragyn Sabbatsberg is a gynaecology clinic within Aleris, one of the Nordics' largest specialist care providers. The clinic handles more than 2,000 patient visits a month, all documented in TakeCare. During the pilot, 27 doctors had access to Tandem's Coding Assistant, and 20 used it actively.

Like many specialist clinics, Ultragyn relied on a single coding controller who reviewed visits each week, after the fact, and corrected the diagnosis and procedure codes. The process kept billing accurate. It also made the quality of the documentation dependent on one person.

The problem was missing codes, not wrong ones

Coding at Ultragyn followed a pattern familiar across specialist care. Doctors focused on the patient in front of them and left the coding for later. Sometimes for days. Sometimes not at all. The coding controller picked up what fell through: going through visits, adding diagnoses, fixing procedure codes.

The data showed something the clinic hadn't expected. Her work was almost never about correcting codes that were wrong. It was about filling in codes that were missing. Around 77 per cent of her changes added missing codes to visits that were already partly coded. Around 13 per cent were visits with no codes at all. None of her changes involved removing a code.

So the issue wasn't accuracy. It was completeness. And it was getting worse. In the months before the pilot, coding at the clinic was steadily becoming less complete: fewer diagnoses per visit, fewer procedure codes, and more visits left entirely uncoded.

Tandem suggests codes as the note is written

In early 2026, Aleris switched on Tandem's Coding Assistant at Ultragyn. The tool reads the clinical documentation in real time and suggests diagnosis and procedure codes as the note is being written. The doctor reviews each suggestion and confirms it before anything goes further. Nothing is coded without the doctor's sign-off.

The pilot ran for two months, February to March 2026, measured against a baseline period from October to November 2025 that captured how coding looked before Tandem was introduced.

Doctors coded more completely, and the corrections stopped

The clearest result comes from comparing doctors with themselves. Same doctors, same kinds of patients, different coding depending on whether Tandem was active. With Tandem, they recorded 0.259 more diagnoses and 0.232 more procedure codes per examination visit. Both results are statistically significant (p<0.001). A paired test per doctor holds up too: 75 per cent of the doctors with enough visits in both periods improved (p=0.047).

This wasn't doctors simply getting better at coding over time. It was the same clinicians, seeing the same types of visit, coding more completely when Tandem was there to prompt them. The tool only suggests codes that are already supported by the conversation or the note.

Uncoded visits disappeared

For doctors working without Tandem, 1.2 per cent of visits went out with no codes at all. For visits that went through the Coding Assistant, that dropped to zero.

The coding controller had nothing left to correct

Of the 841 examination visits that went through Tandem with confirmed suggestions, not one needed a change from the controller. Visits that didn't go through Tandem carried a correction rate of 0.8 per cent.

Without the tool, coding went backwards

The most telling finding is what happened to doctors who didn't use Tandem. Over the same period, in the same clinic, the two groups moved in opposite directions:

• Diagnoses per visit: Tandem users +0.259 (p<0.001); non-Tandem visits −0.082 (p<0.001)

• Procedure codes per visit: Tandem users +0.232 (p<0.001); non-Tandem visits −0.223 (p<0.001)

Another Aleris clinic, running the same medical record system and the same coding rules but without the Coding Assistant, showed the same decline. The gains at Ultragyn weren't a wider trend. They were specific to the tool.

Confidence in the suggestions grew as the pilot went on. The share of visits where a doctor confirmed at least one Tandem suggestion rose from around 27 to 44 per cent at baseline to 76 to 84 per cent by the later weeks.

For Minna Westerberg Ekman, a specialist registrar in gynaecology at Ultragyn Sabbatsberg, the difference was immediate:

"With the coding module, I felt for the first time that I'm saving time on coding... Click, click, click, and it's done."

Complete documentation no longer rests on one person

When a visit goes uncoded, the clinic risks receiving no reimbursement for care that was actually delivered. Many specialist clinics depend on one person, often a coding controller, to make the documentation complete after the event. When that person is away, the gaps stay open.

At Ultragyn, the pilot showed what changes when the support arrives during the documentation itself rather than a week later. Visits that went through Tandem came out complete. The weekly correction round, once a fixed part of the workflow, was no longer needed for those visits.

This doesn't make coding expertise redundant. It reduces the reliance on after-the-fact review, and it means complete documentation no longer hits a ceiling set by one person's capacity.

Following the pilot, Aleris is continuing to use Tandem's Coding Assistant at Ultragyn, with discussions under way about rolling it out to more clinics.

Strongest results where Tandem was used most

The pilot was short: two months with full access to the features. Some coding capabilities, including DRG-specific code packages, only went live partway through (16 March), so their full effect isn't visible in the data yet.

Usage also varied between doctors. The most active reached 80 to 91 per cent coverage of their visits, while some never used the tool at all. The per-doctor result is significant, but the coding gains were concentrated among the clinicians who used Tandem consistently. The tool was used mainly for examination visits; coverage of procedure visits was lower, which the clinic flagged as an opportunity for the next phase.

See how the Coding Assistant fits your clinic

Book a 20-minute demo with our clinical team and see how well Tandem works with your medical record system and workflow.