AP and CP reporting, from interfaces to IT wishes

Create PDF

What are your thoughts about reference laboratory testing?
Dr. Carter (Children’s Healthcare of Atlanta): One of the biggest differences is that CP laboratory information systems are good at accepting orders from the outside, performing those tests, issuing a result, and sending results via an HL7 message. For AP lab systems, especially those that are specific to AP or are the AP component of an integrated AP/CP system, those same steps are not done well or at all. Particularly with flow cytometry, molecular, or FISH, any of this ancillary testing that gets ordered on the outside, you’ll often find that the CP system sends the order and gets results back as a text document.

Some AP systems are trying to order FISH or molecular from an outside lab and then bring the result in and add it as an addendum or a procedure to their existing pathology report. There’s no way to do that without attaching a PDF or scanning in a result, or worse, the internal pathologist has to scan in the report and say, “Attached is the report of . . . .” Then they’re signing it, and if that FISH, molecular, or flow report is incorrect, the pathologist who signed out the outside scanned report could be at risk medicolegally. AP systems have a hard time with this, in my experience, to the point that we end up going through the CP system and bypass the AP system completely, other than for the signing pathologist to state that the additional results, such as flow or molecular, are reported separately.

Suren Avunjian, can you comment on what Alexis said?

Avunjian

Suren Avunjian, co-founder and chief executive officer, LigoLab Information Systems: The crux of the issue lies in the limitations of current AP systems to effectively manage and integrate external data. The absence of an electronic interface creates a bottleneck, forcing a reliance on manual processes like scanning and attaching PDFs or creating textual addendums. To mitigate these challenges, we’re advocating for and developing solutions that allow for a seamless electron-ic interface between AP systems and reference laboratories. By building discreteness into subtests and manual mapping capabilities, we can manage and integrate results more efficiently. With full electronic integration, these results can be automatically populated into the pathology report, which preserves data integrity and reduces administrative overhead.

Our goal is to create a more robust and error-proof system, where discrete elements from reference laboratories are captured electronically and attached to the corresponding case in a structured format. The adoption of standardized communication protocols, like HL7 FHIR, could enhance this process, making it easier for systems to exchange information.

Joe Nollar, can you comment on the reference lab problem and solutions for it?
Joe Nollar (XiFin): The main problem is not the system’s inability to do it. It is routine to do an electronic orders and results interface across multiple platforms, from CP to AP or to reference lab customers. Having all that discrete electronic data with the embedded PDF is routine, and there should be no reason for it not to work that way. If a reference lab is doing FISH and flow and other specialized testing, the AP system should be able to provide specific modules for that testing and provide discrete and meaningful results. It comes down to the system’s ability to handle that type of testing and make the financial commitment to get it done.

Beth Eder, a comment from you on the same topic.

Eder

Beth Eder, director of product management, Orchard Software: I agree with my colleagues on the types of interfacing we can do. We interface to the capability of what we can receive. If the reference lab allows us to be able to get that discretely, get an embedded PDF, that’s what we are going to do. The gold standard would be electronic, but sometimes we don’t get that in part of the mix.

Michelle Stram, our forensic medical system suffers from a lack of people and funds, among other things. Generally in pathology now, we have a perfect storm and don’t have people, dollars, or enough ancillary helpers, whether in IT or elsewhere, to solve the problems you’re all articulating. What is your number one wish?
Michelle Stram, MD, clinical assistant professor, Department of Forensic Medicine, New York University Grossman School of Medicine, and medical examiner, New York City Office of Chief Medical Examiner: The number one thing I would ask for is a new integrated AP LIS, one where the data is discrete as much as possible, relies on synoptic reports as much as possible, and is searchable, because our system is not. For example, if I want to review traffic fatalities, I cannot search the text of the reports. The final diagnostic list is stored in our system as uploaded PDFs. All that data is inaccessible to me and everyone in the agency. It makes answering questions we have internally about optimizing practices or from the Metropolitan Transportation Authority or other stakeholders in public health very difficult. It becomes a fully manual process of pulling all the cases and reviewing them.

Ross Simpson, what is your IT wish?
Dr. Simpson (HealthPartners): I hate scanned images and PDFs. They go in a media tab, and I can’t ask the database: Does that exist? I have to go read something. Some of those things are starting to be used in staging for breast cancer, so I wrote the company—which serves the entire U.S.—because we get PDFs from them. I asked if I could get this as discrete data, and they said no, they’re not capable of doing that. We know they’re capable of doing it; they’ve probably been capable since the 1980s. They want to provide their graph, but all I need is the number and I need it in a readable format. If I have to go to a PDF, read it, and stick it in some other place, it can be mistyped. I want everything to be computer understandable. Discrete data is not a barrier to that. It’s part of the solution. The reference lab interfaces are terrible. I try to get them as discrete as possible. We end up with a lot of PDFs, and they’re kind of dead to me.

I’d like to ask the vendors—what is the single advance, or feature, or ease of use of AI you would like to see available in your system and to your customers?
Suren Avunjian (LigoLab): We’re deploying AI across various fronts, such as in decision support, report augmentation, CPT and ICD coding, and a marketplace of AI technologies for digital pathology.

Beth Eder (Orchard): We’d like to examine getting into generative AI for data. Live data is necessary to examine things like workflow and the best, most efficient process or for population management. We’re looking at many things and they seem to be surrounding generative AI.

Nollar

Joe Nollar (XiFin): Our greatest use of AI now is assisting on the billing side. We can look at payer habits and better determine which payers will pay for which test. That gives our clients more information and a better model for how they approach the marketplace.

We’re also integrating with digital pathology AI providers. I was at a digital pathology roundtable last week in Chicago, and by far AI is the number one reason laboratories are rushing into digital pathology. They can monetize the data for clinical trials projects, pharma projects, and overall data sales, not to mention the revenue opportunities in TC/PC [technical component/professional component] and consultations. AI is the future of pathology.

Diana Richard, the use of AI in billing is a reality. Anyone can call you and have that as an immediate benefit. Is that right?
Diana Richard (XiFin): Yes. Payers are shifting the rules so quickly, leveraging AI to do so, that adoption of machine-learning technologies is a must. Keeping up with the trends, the volume of data coming in, and the complexity of that data would be remarkably burdensome for humans to manage alone and ineffective if you’re working with a high volume of transactions. We just released our Payor Rate Transparency Monitor. It’s a conglomeration of payer data that was messy and complex and had petabytes of information. That’s not easy to engineer into a substantive summary, so being able to leverage AI tools to help facilitate that process is necessary. It will become a requirement to be effective.