Lab Soft News

Pathology and Radiology Collaborate with a Concordance Conference

The University Hospitals Cleveland Medical Center (UHCMC) is in the process of deploying a digital pathology system and therefore exploring ways to educate and prepare their pathologists for this new technology (see: Digital Pathology Pilot Predicts Prosperity: Pondering Pathology’s Pivot). As part of the process, a pilot study was undertaken with a focus on frozen sections. In the study, pathologists’ diagnoses using digitized slide images were compared with diagnoses using standard microscopy. Frozen sections were chosen as the focus of this study because digital pathology allows remote access by a specialized pathologist if necessary. The results showed an average overall concordance rate of 95% for the two methods across 11 organ systems.

In the article referenced above, brief mention was made about a routine rad-path conference in the hospital that was greatly improved by the use of digital images. The article went on to discuss the purpose of this rad-path concordance conference: to resolve issues of concordance between imaging findings and the microscopic findings to also determine whether or not patients need additional surgical intervention or can be followed with imaging. In the support of this goal, it was deemed necessary for all conference participants to examine both the radiologic images and the microscopic images. 

In October, 2006, I first proposed close collaboration (or even, shockingly, merger) of the two medical specialties of pathology and radiology into a new one called "diagnostic medicine" (see: Ten Reasons for Merging Pathology/Lab Medicine with Radiology). I think that a concurrence conference similar to the one at the UHCMC would be an important first step in promoting collaboration between radiology and pathology in any hospital. One major goal that could be accomplished by such a conference would be to determine whether reports generated for a particular patient by both departments were in agreement and/or consistent with each other.

To put the matter more bluntly, the continuing goal of the conference would be to determine whether a significant pathologic finding by one group of diagnostic specialists was overlooked or deemphasized by the other group. Another important goal would be for pathologists and radiologists to better understand the reporting conventions and medical vocabulary of the other group and, in so doing, promote better concurrence in the future. In this same spirit, it would be possible to determine which pathologic changes in an image or specimen noted by one group would prompt a telephone call to to the other group rather than communicating solely by "formal" reports that may not always highlight subtleties or uncertainties.

Diagnostic Management Teams in Multi-Hospital Health Systems

I have posted two notes recently about diagnostic management teams (DMTs) (see: A Major Change Now Occurring in Both Clinical Pathology and Anatomic Pathology; DMTs as an Example of a Team Approach to Clinical Lab Diagnostics). Generally speaking, DMTs have usually been deployed in a single hospital with the diseases covered by the team limited to the special expertise of that team. With such a scenario, communication among the DMT members does not pose much of a challenge, given that all of the team personnel work in close proximity.

I think that the next major hurdle relating to the broader adoption of DMTs is their deployment across all of the hospitals within a large system. System-wide DMTS will definitely be an attractive target for the leadership of such large systems because of the major goals of DMTs: reduce the number of incorrect diagnoses, reduce the number of erroneous lab orders, particularly for expensive tests, reduce the length-of-stay for complex patients, and increase patient satisfaction. However, system-wide DMTs pose a new challenge not seen within a single hospital, which is to say communication among team members who are physically separated.

Modern IT does provide solutions for communication across physical space -- multi-party video conferencing using apps like Skype and BlueJeans. One additional feature available with these apps is the option to record and archive the discussions of the DMT for later reference and quality control purposes.Therefore, a multi-hospital DMT would probably function in the following way:

• A DMT consult is ordered for a patient in one of the hospitals within the health system
• Personnel in the hospital where the consultation is ordered launch a DMT app that generates a worksheet with clinical and lab findings for the patient based on his or her diagnosis category (see: A Major Change Now Occurring in Both Clinical Pathology and Anatomic Pathology). 
• The relevant content expert(s) for that diagnosis category are notified and a time established for the relevant DMT to convene a video call to collaborate on the case. This assumes that the necessary members of the health system DMT are scatted across physical space.
• The worksheet that has been generated can be screen-shared by all parties simultaneously or can be sent to DMT members via email.
• At the end of the process, one of the team members will write and publish the DMT report with possible review by others, if deemed necessary.

Which Healthcare Jobs Are Safe from Replacement by Automation or Robots?

One facet of modernity that is getting more attention is that many of today's jobs will be encroached upon (i.e., replaced) by automation and robots. As one current example, you will not find many people working in modern hospital clinical labs because much of this human labor has been replaced by automated equipment. That's why lab testing is one of the biggest bargains in healthcare today. A recent article addressed the question of whether healthcare jobs in particular are safe from replacement by AI (See: Are healthcare jobs safe from AI?). Below is an excerpt from the it:

Across the country and across industries, workers are nervous that automation and artificial intelligence will eventually take over their jobs. For some, those fears may be grounded in reality. Healthcare, however, looks like it will be largely safe from that trend, a new report from the Brookings Metropolitan Policy Program finds. Examining a chunk of time from the 1980s to 2016, the piece tracks the historical evolution of the technology and uses those findings to project forward to 2030. The verdict? AI will replace jobs in various industries, but not so much in healthcare.... Overall, though, only about 25 percent of U.S. jobs are at a high risk of replacement by automation. That translates to about 36 million jobs, based on 2016 data. A higher percentage, 36 percent, are at medium risk (52 million jobs) while the largest group is the low-risk group, at 39 percent (57 million jobs). Most of healthcare belongs in the medium-to-low categories, largely driven by the complexity of healthcare jobs....Medical assistants have what the report calls "automation potential" of 54 percent, but home health aids have just an 8 percent automation potential. Registered nurses sit somewhere in between, at 54 percent.

The emerging growth areas in pathology and the clinical labs, from my perspective, will be based on the conversion to digital pathology (see: Seeking Funding from Hospital Executives for Digital Pathology Platforms; Integration and Interoperability Are Essential for Growth of Digital Pathology), the automation of anatomic pathology, and the explosion of analytics in all of the clinical labs. The topic of analytics in healthcare was discussed in more detail in a recent post (see: Diagnostic and Predictive Analytics and Their Possible Link to the Future of the LIS). Analytics was described there as bring comprised of the following: descriptive analytics, diagnostic analytics, predictive analytics, and prescriptive analytics. I have also blogged about what I envision as the future of LISs as being highly related to analytics such that I think they should be referred to LIS-A's (see: Predicting the Future Functions of the LIS-Analytic (LIS-A)).

Proposed Classification for Health Information Systems: Provider and Individual

Recent events and some of my recent blog notes have caused me to question whether we need a new, high-level classification that encompasses all health information systems. I am going to give this idea a try. The goal of such a large scale systematic plan is to understand the functionality of current and future health information systems. Details are listed below:

1. Provider health systems (PHS's): The common feature of PHS's is that they are designed as tools for use by healthcare providers. The largest and most complex of these systems are EHR's which are deployed in hospitals and health systems. Also included in this category are office health systems (OHS's) designed to support the office functions needed by physicians and other providers like nurse practitioners. PHS's provide all the necessary IT functionality in health systems ranging from hospital inpatient units to the PCP's managing ambulatory patients.

Subcategories of PHS's include departmental systems deployed in hospitals such as LIS's, RIS's, and PAC's. In recent blog notes, I proposed that standalone LIS's, as opposed to LIS's that are part of the software suites of EHR vendors, could perhaps evolve into LIS-A's (see: Predicting the Future Functions of the LIS-Analytic (LIS-A); Continuing Discussion about the LIS of the Future; Analytics as a Key Feature?). Such systems would change from classic, best-of-breed LIS's to systems that support descriptive, predictive, and diagnostic algorithms. By developing such LIS-A's, LIS companies may cede ground to the LIS's provided by large EHR vendors as part of their enterprise-wide offerings. 

2. Individual (or personal) health systems (IHS's): These are systems designed primarily for consumers. I blogged recently about the Rally system of UnitedHealth Group that initially provided coaching and other health services (see: UnitedHealth Group to Launch Its Own Variant of an EHR in 2019). It's similar to Apple Health which is an app that runs on iPhones and the Apple Watch. Analytics will be a major thrust in the consumer healthcare market and delivered via IHS's (see: The Future of the LIS: Reflections about Consumers Ordering Their Own Lab Tests) particularly when PHS/EHR clinical data can be downloaded by consumers from their EHR accounts. Apple is already providing such a feature for iPhones in collaboration with selected health systems (see: Apple Has Plans to Copy EHR Records to iPhones and Apple Watches).

3. Hybrid health systems (HHS's): These would be systems such as the one being developed by UnitedHealth Group mentioned above that span physican office health systems (OHS's) and individual health systems. The advantage of such systems is that they fulfill the desire of patients to share health measurements obtained in their homes such as EKG tracings recorded by their smart watches or home lab testing performed by devices that plug into their iPhones. 

As to the future of health information systems, most of the development (and consumer market) is going to be related to IHS's and perhaps relatively simple hybrid systems. This is where consumers will be operating as well as health wearables and home testing devices. EHR's are expensive and accessed only via complex user interfaces. Hospitals will continue to use them because they have no substitute products. Consumers will never use them directly because of their complexity and because of the security concerns of the health systems that own and operate them.

New CMS Reimbursement Guidelines: Support for Provider-to-Provider Consults

Here's a link to a high quality webinar from AmericanWell about telehealth (see: A Guide to Integrating Cross-Continuum Telehealth into your Health System). Starting at about 11:00 minutes in the recording, the discussion turns to new CMS telehealth reimbursement guidelines for FY2019 that include new services such as remote patient monitoring (see: CMS to Reimburse Providers for Remote Patient Monitoring Services). Another new type of service that will be reimbursed is provider-to-provider consultations (see: CMS wants to expand telehealth coverage. Here's what that means for you). Here's the key passage from this latter article relating to provider-to-provider consults:

Internet-based inter-professional consultations: New reimbursement for provider-to-provider consultations via online messaging applications or portals. Both the initiating physician and the consulting physician would be eligible for reimbursement.

I have previously covered diagnostic management teams (DMTs) in Lab Soft News (see, for example: Deploying a DMT at the Front-End of Lab Orders for a Complex Patient). One of the challenges for DMT's in the past has been the lack of reimbursement.There is now a ray of light with CMS-approved reimbursement for provider-to-provider consultations which accurately describes the DMT process. A team of expert diagnosticians provides advice to a clinician caring for a patient about the ordering and interpretation of complex laboratory tests. There are, of course, additional benefits attributable to such consults such as decreased medical errors and decreased length of stay but at least some degree of reimbursement can provide an incentive for launching a DMT team in a health system. 

Does a DMT consultation fall under the general rubric of telemedicine? First of all, I think that collaboration between the members of a DMT, even within the confines of a single hospital, will often take place via telephone or a video call. Secondly, discussion of the case between members of the team and with the consult-ordering clinician may often be more effective using video calls. Plans are now underway in collaboration with Sunquest and Epic to develop an app that will work in the following way after a DMT consult has been ordered: (1) enter the name of a patient into the LIS; (2) select the organ and/or disease state driving the DMT consultation; (3) the app will then interrogate the lab and EHR databases and generate a report with all of the relevant lab test results and clinical findings for the patient that will form the basis for the DMT report.

Roche Diagnostics Shows Keen Interest in Diagnostic Management Teams

I was slightly taken aback by a recent email blast I received from Roche Diagnostics with links to a current and planned complementary articles about Diagnostic Management Teams (DMTs) (see: DMT series). I have posted a number of blog notes in the past about DMTs (see, for example: First Diagnostic Management Team Conference Scheduled for Next February; Deploying a DMT at the Front-End of Lab Orders for a Complex Patient). The concept of the DMT was pioneered by Dr. Mike Laposata in 1984 as a means to assist clinicians in the ordering and interpretation of coagulation tests about which he was a national expert. In the intervening years, DMTs have gained increasing support as a means to avoid diagnostic errors, decrease the length of hospital stay, decrease the cost of care, and as a tool in support of value-based care.

You might then ask why I was surprised by a broad endorsement of DMTs by a large in-vitro diagnostics company like Roche. This requires reference to some historical issues in clinical pathology. I started my career as a blood banker who was keenly interested in blood utilization. I published an article in 1976 about the maximum surgical blood order schedule (MSBOS) whereby a predetermined number of units of blood were crossmatched for a particular operation. The goal was to reduce blood outdating. I posted a blog note about this topic ten years ago (see: Hospitals Seek to Limit Blood Transfusions as a Cost-Saving Measure). In those days, blood reagent companies such as Ortho Clinical Diagnostics did not usually promote blood conservation strategies because it was not good for their business -- it resulted in lower use of testing reagents. The same approach generally applies to other IVDs regarding strategies to reduce test volume.

So, why the sudden interest on the part of Roche in DMTs? Theoretically DMTs could be "bad" for business and result in fewer tests being performed and less reagent usage. I think that the obvious answer is that we have now arrived at a point in lab medicine where some tests are so expensive and difficult to interpret that DMTs will play a critical role in lab diagnostics. Special talent and expertise is required to interpret some test results and the necessary talent is not always available among clinicians and even the majority of pathologists who tend to focus on surgical pathology because it is more remunerative. Hence, local, regional, and national DMTs need to be organized that have special expertise and can be called upon to help render timely and accurate diagnoses. Information technology can be used to bring the lab diagnostic experts, wherever they are physically located, into the local diagnostic process.

Registration Almost Full for the Second Annual API Digital Pathology Workshop

The second annual API Digital Pathology and AI Workshop will be presented on December 7-8. 2018 (see: Digital Pathology and AI Workshop II). The location will be Ohio State University College of Medicine in Columbus, Ohio. You can register for the workshop here. The capacity for the workshop has been doubled over last year's event, now with a maximum of 80 people. Only nine spaces remain open at this time. A waiting list will be created when these places are filled.

Available on the API web site are the schedule and the sponsoring vendors to date for the workshop. Here is a list of the workshop faculty and moderators: Ulysses J. Balis, M.D,; Michael D. Feldman, M.D; Wendy L. Frankel, M.D.; Chris Garcia, M.D.; Stephen M. Hewitt, M.D., Ph.D.;  Keith J. Kaplan, M.D.; Anant Madabhushi, Ph.D. Robert McGonnagle; Liron Pantanowitz, M.D.; Anil Parwani, M.D.;S. Joseph Sirintrapun, M.D.; and J. Mark Tuthill, M.D. Below is a quote from the workshop summary page on the API web site describing the workshop in more detail:

This workshop will focus on the practical issues surrounding the deployment of digital pathology and AI in a hospital setting. Ten lectures by national experts will be punctuated by Q and A sessions including industry representatives and a guided tour of the Ohio State University's Department of Pathology. Topics covered will include the fundamentals of digital pathology, the integration of the technology with LISs, the enhancement of AP workflow, the development of an RFP for purchasing digital pathology equipment, and current perspectives about AI within pathology.

This workshop has been planned with busy surgical pathologists and departmental leadership in mind. It lasts only one and a half days and takes place on a Friday and Saturday morning. The pathology department at OSU is the first department in the country providing primary surgical pathology diagnoses using digital pathology so a tour of the imaging facility will be highly educational. A dinner is planned for the Friday night for networking and relaxation and is included in the registration fee.  

The second annual API Digital Pathology and AI Workshop will be presented on December 7-8. 2018 (see: Digital Pathology and AI Workshop II). The location will be Ohio State University College of Medicine in Columbus, Ohio. You can register for the workshop here. The capacity for the workshop has been doubled over last year's event, now with a maximum of 80 people. Only nine spaces remain open at this time. A waiting list will be created when these places are filled.

Available on the API web site are the schedule and the sponsoring vendors to date for the workshop. Here is a list of the workshop faculty and moderators: Ulysses J. Balis, M.D,; Michael D. Feldman, M.D; Wendy L. Frankel, M.D.; Chris Garcia, M.D.; Stephen M. Hewitt, M.D., Ph.D.;  Keith J. Kaplan, M.D.; Anant Madabhushi, Ph.D. Robert McGonnagle; Liron Pantanowitz, M.D.; Anil Parwani, M.D.;S. Joseph Sirintrapun, M.D.; and J. Mark Tuthill, M.D. Below is a quote from the workshop summary page on the API web site describing the workshop in more detail:

This workshop will focus on the practical issues surrounding the deployment of digital pathology and AI in a hospital setting. Ten lectures by national experts will be punctuated by Q and A sessions including industry representatives and a guided tour of the Ohio State University's Department of Pathology. Topics covered will include the fundamentals of digital pathology, the integration of the technology with LISs, the enhancement of AP workflow, the development of an RFP for purchasing digital pathology equipment, and current perspectives about AI within pathology.

This workshop has been planned with busy surgical pathologists and departmental leadership in mind. It lasts only one and a half days and takes place on a Friday and Saturday morning. The pathology department at OSU is the first department in the country providing primary surgical pathology diagnoses using digital pathology so a tour of the imaging facility will be highly educational. A dinner is planned for the Friday night for networking and relaxation and is included in the registration fee.  

Continuing Discussion about the LIS of the Future; Analytics as a Key Feature?

In a recent note, I discussed how LIS's as components of EHR's were slowly gaining favor in the market over interfaced best-of-breed LIS's (see: Predicting the Future Functions of the LIS-Analytic (LIS-A)). Bill Grolly, a reader of the blog, submitted the following comment in response:

I think that the best of breed LIS's trying to find a way to move into this space is probably going to be very difficult for them. Why would the EMR vendors (Cerner, Epic), who have a built in lab system want to interface that data to a standalone LIS when they can provide the soup to nuts solution, including that analytics component? And why would an IT department want to spend the time/money to do that integration, given if the lab would just use the "built in" LIS, no integration would be needed, or since the lab results are sent back to the EMR, do the analytics there... I certainly think those best of breed systems do need to continually evolve and add better functionality, I'm just not sure this is a path they would want to tread in an attempt to add value.

Pathology departments thrived for decades by installing so-called best-of-breed LIS's that were interfaced to the HIS's (hospital information systems) and later EHR's. However and as emphasized in the comment above, it's become harder to compete with the "soup to nuts" solution offered by EHR vendors. Early in their development arc, EHR/LIS modules like Beaker suffered in terms of functionality compared to best-of-breed LIS's but that gap has decreased over the years. Many hospital executives now seem to prefer to work with a single hospital software vendor and pursue an enterprise-wide-solution.

Another point raised in the above comment by Bill Grolly is whether the leading EHR vendors could come up with an "analytics component" utilizing lab test data just as easily as the vendors of best-of-breed LIS's. This is certainly possible. Cerner was an LIS vendor in its initial software offering, PathNet, in 1984. The company later developed Cerner Command Language (CCL) which was similar to Structured Query Language (SQL) and was used to write LIS-based rules. These rules were an early incarnation of the more sophisticated algorithms that we deploy today based on deep learning and neural networks. As to which vendor would bring a broad set of diagnostic/predictive analytics to the hospital market, It could certainly be an EHR vendor like Cerner or Epic (see: Diagnostic and Predictive Analytics and Their Possible Link to the Future of the LIS).

We already know of one instance where researchers from Massachusetts General are developing a web site that will calculate a disease risk score after a consumer has uploaded his or her DNA analysis results (see: Upload Your DNA to a Web Site to Generate Risk Scores for Five Diseases). Such risk scores certainly fall into the family of diagnostic/predictive analytics. Look for more such projects from academic medical centers. On a longer term basis, I anticipate the development of a large number of diagnostic and predictive algorithms that will be licensed to run on hospital-based EHR's or LIS-A's. However, the FDA will certainly exercise regulatory oversight regarding the use of software like this. It's already active in regulating diagnostic algorithms for detecting wrist fracture in radiology by scanning images (see: FDA permits marketing of artificial intelligence algorithm for aiding providers in detecting wrist fractures). It therefore seems highly likely that the FDA will be very important in terms of the sale or license of imaging algorithms in digital pathology.

Predicting the Future Functions of the LIS-Analytic (LIS-A)

In a recent blog note, I stated that the LIS modules of EHRs such as Epic's Beaker are gaining traction in the market and thus performing the functions of the previously dominant best-of-breed LISs like Sunquest or SCC. I went on to suggest that the vendors of these latter systems should be aggressively working on a new type of LIS to protect their market share. Such a product would undoubtedly include support for lab-initiated diagnostic and predictive analytics (see: Diagnostic and Predictive Analytics and Their Possible Link to the Future of the LIS). In this and future blog notes about this topic, I will use my personal term to refer to such a future LIS -- the LIS-Analytic (LIS-A).

One of the first things that comes to mind regarding the functionality of the LIS-A of the future is that it will be both interdepartmental and collaborative. This is by way of contrast with the best-of-breed LIS's of today which are intra-departmental. Most lab test results are generated currently within the clinical labs without knowledge of the clinical findings of the patients being tested. These results are then been reported back to the clinicians without any feedback from them. I do not say this in any critical way -- the nature of the lab test transaction has always been that clinicians order lab tests and pathology responds to these orders. The clinicians are then responsible for interpreting the results in the context of a particular patient.

Diagnostic and predictive analytics, which I think will be one of the major functions of the LIS-A of the future, will need to take into consideration test results but will also require the monitoring patient outcomes subsequent to a testing episode (see: Appraising laboratory quality and value: What's missing?). This consideration of clinical outcomes is important in order to measure the value of lab testing. It is logical to assume that the acquisition of patient outcome data will be based on an interface with the EHR. I might also add that, in this era of precision medicine, the clinical labs will also be able to generate outcome data. One example would be the absence of circulating cancer DNA in a patient post treatment (see: Circulating tumor DNA).

My crystal ball is cloudy at this time as to how and when this interdepartmental and collaborative data will be collected and integrated into the LIS-A. I do understand, however, that the process will be highly automated and reliant on AI because for the following reasons: (1) test-ordering clinicians will be too preoccupied with patient care to provide such data to the LIS-A; and (2) there may be a substantial delay from the time that any particular test result is generated and the subsequent knowledge of outcome data for that same patient. As a result, I suspect that the LIS-A will be collecting and analyzing patient data for a period after the patient is discharged from the hospital. 

It's Not Too Late to Register for a Free Pathology Informatics Webinar Tomorrow

The Association for Pathology Informatics (API) and Sunquest Information Systems are co-sponsoring a webinar tomorrow, October 23, which is the second in a series of two focusing on pathology informatics challenges in large health systems. You can view the first webinar in this series from September 25, 2018, here. The time for tomorrow's presentation will be 1:00 p.m. ET. The speakers are Mark Tuthill, M.D., Division Head, Pathology Informatics Henry Ford Health System, and  Brad Brimhall, M.D., Medical Director of Healthcare Analytics & Bioinformatics, University of Texas Health Science Center, San Antonio.

Mark Tuthill will discuss LIS challenges in hospital to hospital system integration. Brad Brimhall will speak on the definition, development, and deployment of algorithms in pathology. Both of these topics are of vital interest in large multi-hospital health systems. It's not too late to register for this second webinar here.  You can also view the entire archive of recordings of the API/Sunquest webinars dating back four years here.