Lab Soft News

Continuing Discussion of Personal Health Records

I have written a number of previous notes about personal health records (PHRs).  Here is a link to a set of previous notes on this blog about PHRs. Here is the link to a recent article on the web about PHRs and below is an excerpt from it. Reference is made in the article to the electronic health record (EHR) which differs from the EMR (electronic medical record) in that it is conceptually a longitudinal record of all health events for a patient and family. The EMR, by way of contrast, documents inpatient and outpatient services delivered in a specific health system.

The PHR is the latest addition to the alphabet soup that is health care, and it may actually do a body good...The PHR is the one place where you can accumulate all your health information in a consistent way to reflect your lifetime of care....[A]ny efforts hospitals make now to develop EHRs may well aid individuals building PHRs. And from a legal standpoint, every patient does, in fact, own their own health data....Consumers who collect and control access to their health records have [an advantage, particularly for those] with a chronic disease or a person dealing with cancer....A national survey of about 1,100 adults that Harris Interactive conducted in July found that 7 percent of U.S. adults use online personal health records.

This article makes a number of interesting points regarding PHRs in contrast to EMRs and EHRs:

• EMRs are created by a health system largely as documentation of the services delivered to patients within that system. Although patients are entitled to copies of those documents, they are generally not created with patient needs as a major concern. When articles such as this cite the dogma that "patients own their own health data," the reality is that patients really have the right to receive a copy of their records.
• I suspect that it is only the most obsessive among us who are willing and able to use the various web-based PHR services to laboriously create their own PHRs. However, I do agree with the idea that this is a very good idea, particularly in the event of disasters when hospital records are destroyed.
• As I have stated previously, it will become increasingly common in this era of the EMRs that patients can obtain an electronic copy of their health records from a health system and perhaps even from their private physician. If such records are available in a standard format, it should be possible to upload these records to a web-based PHR service. Ideally, the patient should be able to supplement and amend these documents with additional data about events not documented in the hospital record.
• Also, as noted previously in this blog (link here) and for those people who are the most at risk, these data can be stored on a portable flash memory "key drive" such that they can be presented to hospital personnel in the event of an acute health crisis.
  

Cleveland Clinic Rolls Out Multiple Modules as Part of eCleveland Clinic

Someone at the Cleveland Clinic has their thinking cap on. The health system has spawned multiple different modules from their EMR system that they call eCleveland Clinic. Here is the link to a recent story about their recent IT initiatives and below is a bulleted list of these modules:

• MyPractice is the EMR system for the staff physicians
• MyChart  provide patients access to their own records
• MyMonitoring  enables patients with monitoring devices to transmit data to their medical records
• Dr.Connect is the EMR link for private physicians in the community who wish monitor their patients admitted to the hospital
• eResearch permits investigators to search anonymized patient records for potential subjects who meet study criteria
• Essential Medical Resources allows doctors to lease the Clinic's electronic record system to support their practices

Once a health system has a robust EMR in place, the development of subsidiary components such as those listed above is not that challenging. I was mildly surprised by the eResearch module. In essence, the health system is putting its patients "on display" for investigators to see if any of them meet the acceptance criteria for controlled clinical trials. It's not clear from the article whether this service is available only to internal investigators or also to those outside of the Clinic. I suppose that this can be viewed as an opportunity for those who qualify but I wonder if patients are given the opportunity to opt out of this "opportunity."

I was also mildly surprised about the Clinic's willingness to lease their EMR software to community doctors under the Essential Medical Resources program. This puts the Clinic into the software business which seems to be quite distant from its core competency.

Doctors Slow to Adopt Email with Patients

I have posted a previous note about the use of email in physician practices (link here). Here is the link to a recent article about the topic and below is an excerpt from the it:

E-mail is arguably the most common form of communication in the Information Age, but don't count on getting one from your doctor. Despite growing use of computer technology in almost every other facet of their profession, American doctors are reluctant to use e-mail when interacting with patients.... Barely one-quarter of physicians use e-mail or other electronic communication to reach patients, up from 20 percent four years ago....By comparison, more than half of doctors now use computers to store and access patient notes, up from 37 percent four years ago....In their offices, doctors are under constant pressure to curtail time spent with patients....The last thing most of them want is to give patients another way to get a hold of them....Health plans say they are not averse to paying doctors for electronic consultations, but they would like a more structured setting, such as a secured Web site that screens patients for symptoms and differentiates actual clinical consulting from administrative questions, such as checking whether lab results are in.

This article makes a number of key points and I want to comment on some of them:

• There are multiple barriers to the increased use of email by physicians to communicate with patients such as lack of reimbursement, lack of time, reluctance to adopt new technologies, and concern about a miscommunication with patients in this litigious environment.
• I think that the suggestion above about the need for a more "structured setting" for email and also to differentiate administrative matters from clinical consulting rings true. This may account for the higher adoption rate among physicians for the use of office EMRs (i.e., patient notes) when compared to the use of email.
• Many physicians manage patient email traffic on their own but would never consider answering the telephone in their office, perhaps out of concern for patient privacy. Emails should be handled in the same way as all other communication with the office.
• I think that the key to higher adoption rates for physician office email is to make the process more formal and also integrated into the office EMR systems. Office assistants and nurses need to triage all such communications, handling what they can independently and passing only the more complex clinical questions to the physician.
• This more formal documentation process and audit trail for incoming emails should satisfy health plans enough that physicians can then be compensated for the time spent answering clinical questions electronically.

Transferring EMR Data to Clinical Trial Documents

Although the concept of transferring clinical data from a hospital EMR to the research records generated for the same patient will be of interest only to academic research centers, this is surely an idea whose time has come. This news was reported by Mr. HIStalk (here). Here is a link to the original article and below is an excerpt from it (boldface emphasis mine):

Researchers at the Technical University of Munich and Siemens Medical Solutions have developed a method of electronically transferring clinical data gathered at the point of care for use in prospective clinical trials....Historically, the capture of clinical trial data has been labor intensive and error-prone, requiring data to be transcribed into a paper case report form or, more recently, entered manually into Internet-based data capture systems. This innovative solution offers a scalable, automatic transfer of data between an electronic medical record (EMR) and an electronic data capture (EDC) system, overcoming interoperability challenges associated with systems that operate on different technical standards and work within distinct business environments.... This successful pilot is a crucial step in bringing these two worlds closer for the universal benefit of more timely patient care....Today, the pharmaceutical industry manages many studies looking through the rear view mirror. The availability of real-time data, as demonstrated through this pilot, will transform study management, enabling mid-course corrections and ensuring consistency across study sites, ultimately yielding more timely and accurate outcomes.

Here are some of my thoughts about this approach to record-keeping in support of clinical research projects:

• This approach is obviously most relevant for patients with diseases who are enrolled in clinical trials. Although an electronic medical record will also be created for normal subjects, such records will be scanty when compared to those of patients with diseases who need much more supportive care. This is not to say that subjects should not be carefully monitored when participating in trials. Recall a previous note about normal subjects in a clinical trial who experienced severe reactions to an experimental drug (link here).
• As described above, the electronic data capture (EDC) system would be programmed to extract from the EMR those data required by the FDA when reporting the results of clinical trials. One example would be a detailed description of so-called adverse effects (AEs) that may be experienced by both patients and subjects. The reporting of AEs is an arduous task in many trials and the primary investigators (PIs) would welcome any system that would decrease the FDA reporting requirements.
• What is not described in the article, and what I would find most interesting and relevant, is the funding of this system for replicating a portion of the EMR to the research record. Such a question may not be relevant for Germany with a nationalized health system but it would be an important one for an American research and teaching hospital. In this latter case, research dollars flow through the medical school affiliated with the teaching hospital. Clearly, the funding of the EDC system would fall to the medical school with support from the various grants and contracts awarded to the PIs. I am not sure if grant and contract dollars could or would be allocated for partial support of the hospital EMR that feeds the EDC system.

Best-of-Breed Lab Software Vs. Integrated Clinical Systems

I have a very firm opinion about the choice of best-of-breed lab software versus integrated clinical systems with lab support. Here is the link to the story about this topic and below is an excerpt from it:

The healthcare industry has historically been slow to adopt new information technology, both on the clinical side and the administrative and financial end of the business. This trend is particularly true for providers, who must balance their budget constraints from staffing demands and rising care costs with the demand for new facilities and cutting-edge medical equipment....A recent study by HIMSS Analytics... found that the average 2005 operating budget for IT [in healthcare] was less than 3% of a hospital's overall operating budget....The price tags [for EMRs] are substantial...with hospitals spending between $6m and $12m for a system from a vendor such as Cerner, Epic Systems or Eclipsys. At the lower end of the hospital market, MediTech's $1m to $2m systems are the most common....But recent consolidation in the healthcare software market has narrowed the field to four billion-dollar players: McKesson, Siemens, GE and Cerner. Together these companies comprise about 60% of the healthcare applications market....A similar trend over the past two years,...is hospitals' moves from a best-of-breed, multiple-vendor approach to a single supplier for applications....This movement to a single supplier means more of an emphasis on integrated offerings instead of numerous, different systems, as hospitals look for one or two key software processes and an enterprise system with a common data side....

To quickly get to the point, there is now a very high probability that the CEO/CIO of a health system will move to consolidate systems using a single software vendor and sometimes strong-arm the labs to accept the LIS module of one of the vendors of integrated clinical systems. My position on best-of-breed  lab software versus integrated software can be simply stated in the following way:

Lab professionals should accept the LIS module contained within an integrated clinical package only if and when it approaches the functionality of the best-of-breed LISs available in the market. Such a position, of course, requires that lab personnel have a deep understanding of the functionality of the software offered by the various vendors in the marketplace and can articulate how the offered LIS module falls short of the optimal.

So why not "go along to get along" with the hospital executives who are moving to a single clinical software supplier? Here are the reasons why one shouldn't sacrifice computer functionality in the lab:

• There are two primary initiatives that will account for the success of lab medicine in the future -- automated information management and molecular diagnostics. Lab managers will be running a one-legged race if they compromise on the software that they will be running for a decade or more in the future.
• The success of lab managers will be judged by hospital executive largely on the basis of their adjusted cost per test weighted for test complexity. Lab costs can be hammered down primarily with a high quality LIS and sophisticated analyzers. Customer satisfaction, a secondary factor, can also be directly correlated with LIS functionality.
• It would be foolhardy to trust the assurances of hospital executives about the functionality of the offered LIS module. Most have no detailed understanding about lab functions or software requirements. Their major concerns will be the overall system cost and the functionality of the EMR.
• I have previously blogged about best-of-breed issues (link to links here) and the development of an RFP in anticipation of selecting a new LIS (link to links here). The latter process is critical in creating an audit trail documenting the features and functions that a vendor has promised to deliver to the lab. CIOs will frequently exclude the ancillary departments such as lab and radiology from participation in the RFP process when choosing an EMR vendor. This is not a wise policy given that lab data still comprises the majority of the information contained in the EMR.

Molecular Imaging Center Established at Case Western Reserve

I have published several notes about molecular imaging in this blog (links here, and here, and here). Below is an excerpt from a recent story about the establishment of a molecular imaging research center at Case Western Reserve in Cleveland, Ohio. Here is the link to the original story (boldface emphasis mine).

Case Western Reserve University and the National Foundation for Cancer Research (NFCR) announced today the establishment of a new NFCR Center for Molecular Imaging. This center...will be focused on assisting cancer scientists in identifying specific genes that are ideally suited for molecular imaging and diagnosis improvement that will lead to earlier-stage treatments for cancer patients ....Molecular imaging is being recognized as the next revolution in cancer research and is paving the way for patient-specific imaging evaluation based on molecular and genetic markers....The Case center joins other NFCR Centers established at Oxford, Yale, University of Alabama- Birmingham, University of California-Berkeley, Dana Farber Cancer Center, University of Arizona, Freie Universitat Berlin, and the Institute of Medicinal Biotechnology in Beijing to form a broader international collaborative network of top researchers....Current traditional cancer detection methods such as MRIs and ultrasounds are not effective at detecting small, newly formed tumors in patients. This new center will work to improve these results by identifying cancer-specific genes that are ideally suited for molecular imaging and diagnosis.

I was not familiar with the NFCR. It describes itself as a 501(c)(3) organization that has provided more than $200 million in support of discovery-oriented basic science cancer research and cancer prevention for the past 30 years. Here is a link to its home page. 

Here are my thoughts on this recent announcement about the new molecular imaging center:

• The specialty of radiology has historically focused on imaging as the means to identify and localize disease processes with little attention paid to basic pathobiologic processes and the predisposition to disease. Molecular imaging now provides a vehicle for radiology departments to engage in basic research focused on the molecular and genetic markers of disease. Note the multiple references to these approaches to research in the story above. Armed with this knowledge, the radiologists of the future will possess the tools to both diagnose and treat disease.
• I am impressed with the establishment by the NFCR of molecular imaging research centers in the U.S., Germany, and China. As noted above (link here), multinational imaging companies like GE Healthcare are establishing their own molecular imaging centers abroad for support of clinical trials.
• In a previous note (link here), I discussed how Siemens has extended its knowledge of molecular medicine by the purchase of both DPC and Bayer Diagnostics. Their acknowledged goal is to blend in-vivo diagnostics (i.e., molecular imaging) with in-vitro diagnostics (i.e., molecular diagnostics). They have taken the lead in integrating the two separate disciplines of radiology and pathology.

Networked Textbooks in Pathology and Lab Medicine

You may have seen references to the networked book in the news recently. Here is a definition of the term from the Wikipedia:

A networked book is an open book designed to be written, edited, and read in a networked environment. It is also a platform for social exchange, and is potentially linked to other books and other discussions. Wikipedia is a networked book.

Here is an edgier description of the networked book (link here):

The networked book is here: the new agora/plaza/forum where author, publisher, and readers congregate to ponder, discuss, joke, enjoy, and refer. The book is now searchable, linkable, multimedia-able, commentable, annotatable, previewable, mutable, divisible, aggregatable, correlatable, syndicatable, feedable, emailable, Flickrable, deli.cio.us-able, Diggable. In fact, the book is on the brink of such a huge transformation that we wouldn’t be surprised if it opened its own chain of coffee bars.

Although I am enthusiastic about this new and evolving concept, it occurred to me that in professional fields such as pathology and lab medicine a hybrid form of the networked book will evolve -- the networked textbook. At the theoretical "center" of the networked textbook constellation will be an electronic version of today's paper textbook created by a set of authoritative authors and supported by a medical publishing house.

As soon as this electronic textbook with its authors vouching for its authenticity is launched, evolving around it and linked to it will be a set of relevant web resources. Such resources will include moderated discussion groups, relevant electronic journal articles, blogs, and vendor sites with relevant content. I believe that the authors and the publishing house will help establish the links to these web-based resources from the central text.

Such a networked textbook differs from our previous notion of a textbook, first of all, because it will be accessed as a service rather than purchased as a tangible object. Secondly, the electronic textbook solves one of the major problems of a classic textbook which is that the latter is static and quickly goes out-of-date. The network textbook is dynamic and will be refreshed and updated on a daily basis.

Finally, the networked textbook will be participatory rather than autocratic. Although the authors of the networked textbook will launch the original core material, they will later serve more as guides and referees, sorting through all of the additional materials that are linked to the core material and judging which are important enough to justify inclusion into it.

Vendor Attitudes About FDA Regulation

Two separate comments have been submitted in response to a recent post (link here) which I copy below (boldface emphasis mine):

(1)  From a vendor perspective having to obtain a 510(k) is a costly, time consuming process. It has resulted in many vendors reducing the number and frequency of software enhancements and updates. Once a software product has been declared a "medical device", then the end-user must also take additional steps to document their validation process, which adds to the cost and time required to implement and maintain the product. Obviously, if a vendor has been required to obtain a 510(k), perhaps because that software has been integrated with a medical device, then that vendor should utilize that fact as a marketing tool, but to state that all LIS software should require a 510(k) would increase everyone's cost without any significant benefit.

(2)  At a time when labs face shrinking payments and the threat of competitive bidding, the last thing we need is FDA oversight. Having the FDA involved will do nothing but drive up the costs of systems and support fees. This could result in many labs hanging on to old legacy systems (which I'm sure Mr. Vail wouldn't mind) or having to pay inflated prices for new systems. IMHO this is a solution looking for a problem.

I personally agree with these sentiments as I made clear in my original note about the FDA's interest in what it refers to as IVDMIAs (link here).  I can see little benefit from FDA oversight over LIS software or algorithms used in a lab setting to interpret complex test results. However, the president of a well-known lab middleware company, Data Innovations, has submitted a comment to this blog (copied below) expressing enthusiasm for the 510(k) process for software (link here):

Not only do I believe that laboratory software is regulatible, I believe that the FDA should be regulating it. At Data Innovations, we believe this so strongly, that we have submitted our Instrument Manager product to the FDA for 510(k) clearance.

DI sells its middleware product, Instrument Manager, directly to lab clients. However, it also functions as the original equipment manufacturer (OEM) of middleware for IVD manufacturers. It may be the case, as suggested in the first comment above, that one or more of these IVDs required DI to obtain 510(k) clearance for its software that was then sold by the IVD company. This would be particularly necessary if the DI middleware was viewed as an integral component of the analyzer by the IVD and the FDA as opposed to a software system separate from the device.

Orchard and Atlas on Top of Web Vendor Survey

Jondavid Klipp has launched a newsletter called Laboratory Economics that was discussed in a previous note here. You can read more about this publication here. Below is a copy of an article that appeared in the September issue of this publication that I thought would be of interest to the readers of Lab Soft News.

By and large, lab customers are satisfied with their Web-connectivity vendors, according to our survey. On average, survey participants gave their vendor a rating of 3.52 on a scale of 1 (low) to 5 (high) when asked, “How would you rate the level of support/service that your Web-connectivity vendor provides?” When asked, “Would you buy from your current vendor again?” the overall survey average was 3.58.The combined average rating for these two questions was 7.10 (out of a possible 10).

The Web-connectivity vendor with the highest combined rating was Orchard Software Corp. Fifteen surveyed customers gave Orchard an average rating of 4.33 (out of a possible 5) on the question: “How would you rate the level of support/service that your Web-connectivity vendor provides?” And Orchard received an average 4.08 rating on the question: “Would you buy from your current Web connectivity vendor again?” Orchard’s combined score was 8.41 (out of a possible 10).

Atlas Medical was second with a combined average rating of 7.96 from 28 surveyed lab customers, including 3.88 for “level of service/support” and 4.08 for “Would you buy again?” Other vendors that received a combined score of greater than 7.0 included: CareEvolve at 7.67, Misys at 7.25 and LabTest at 7.07. Meanwhile, despite all the challenges associated with building bridges to practice management systems and EMRs, most survey participants said Web connectivity is now a requirement for competing in the marketplace.

“Web connectivity is becoming a necessity in healthcare. Many physicians like to review reports from their home or office. We are an acute care facility and have several specialty physicians who travel between hospitals spending one day a week in several facilities. It is essential that they can obtain their patient’s results no matter where they work that day,” wrote a hospital lab manager from Wisconsin.“This is an absolute must for hospitals that want to increase outreach business,” said a hospital lab manager from Minnesota. And finally, an independent lab executive from Kansas warned “Connectivity is the total future of the lab industry….Testing capability and associated services are a distant secondary consideration to clients and prospective clients. We are entering a period of ‘connect or die.’”

           Average Customer Ratings on a Scale of 1-5 (with 5 the highest)

# Surveyed Cust.     Level Service/Support     Would Buy Again     Combined Rating

Orchard            15              4.33                      4.08               8.41
Atlas Medical     28              3.88                      4.08               7.96
CareEvolve         6               4.00                      3.67               7.67
Misys                  9               3.50                      3.75               7.25
LabTest              7               3.62                      3.45               7.07
Telcor                 6               3.25                      3.60               6.85
McKesson          13              3.18                      3.45                6.63
4Medica            12               3.18                      3.33                6.51
Cerner              30               3.10                      3.25                6.35
Dr. Chart           7               3.14                      3.14                6.28
Total-Unweighted Avg.

                         —                 3.52                      3.58                  7.10

Source: Laboratory Economics Web-Connectivity Survey (September 2006)

More Discussion About FDA Regulation of Lab Algorithms

I received an email from an experienced R&D manager in the IVD industry about the continuing discussion in this blog about the possible regulation by the FDA of LIS software and algorithms/rules used in the lab. Here are his comments (boldface emphasis mine):

The FDA has always allowed FDA-cleared products to incorporate well known rules.  The FDA states that if the product does no more than save a physician the trouble of opening a text book, then it's OK. Obviously, any rules that confirm with this approach must be available in a textbook! Such software is used all the time in the interpretation of ECGs etc.--no big deal. There is a long list of FDA cleared products on the device side that incorporate rules. Presumably the rules used in LISs ( eg K+ greater than 12 meq/L is not possible) could be traced to text books in most cases--this is not my area.   

In 2005 Bio-Rad received FDA clearance for an algorithm based on pattern recognition (K043341). This algorithm  makes disease associations that are inherent to its training set, at least some of which cannot be found in any textbook. The FDA therefore has a product that they can point to as an example (predicate) for people who wish to make their own associations/predictions and have no textbooks to refer to. By the way, this algorithm will not be commercially distributed until 1Q2007, although the instrument that produces the data is for sale now.  It has been shown at AACC, in papers, posters, etc.   

So I think the FDA does understand the use of rules in LIS systems, and your concern is maybe premature. I personally read this guidance a few times, and it never occurred to me that it would be applied to LISs until I read your blog.  I think the FDA is really more worried about the use of non-validated pattern recognition algorithms.  We used a training set of >1400 patients and a test set of >900 patients at three institutions. You will find very little in the literature at that scope.   

Anyway the really important thing to me about this new guidance is whether it means the end of home-brews, which is a long-term FDA issue. While not an informatics issue, it would really make a difference, especially to AMP folks and the LabCorp/Quest/etc. folks

All of the information in this comment is quite helpful and will contribute to our better understanding of these issues. Let me try to summarize my thinking about this topic thus far, incorporating some of his ideas:

• FDA has a track record of not being concerned with (or willing to approve) algorithms that are straight-forward (i.e., contained in a textbook) or that support well-known disease associations. It has been suggested above that their major concern is with "non-validated pattern recognition" algorithms that are used in a lab setting. My problem is that I am not quite sure what this means or how a "pattern recognition" become validated in way that satisfies the agency.
• I infer from this communication that the issues raised in a previous note (link here) about possible FDA oversight over relatively simple rules embedded in LISs or middleware such as autoverification should not be of concern since all of the rules deployed in this way within a lab will be relatively easy to understand. Moreover, the FDA would theoretically have no objection to lab rules developed locally within a lab.
• I do understand the point that the major target of this proposed FDA guidance is home-brew tests. Nevertheless, the FDA document did make explicit reference to lab algorithms and the problem of their being proprietary and non-accessible by physicians. I therefore do not think that my concern about LIS software/rules/algorithms is misdirected. I think that we need to take their interest in lab algorithms seriously, at least for now.
• LIS software/rules/algorithms form a continuum with substantial overlap. I suspect that some computer experts might have difficulty distinguishing among them. If this is the case, then the FDA needs to precisely define what is meant by algorithms. It also needs to be emphasized that biomarkers in some labs may be analyzed using neural networks to analyze the test results. Below is a definition of this term from the Wikipedia (link here).

Neural network is sometimes used to refer to a branch of computational science that uses neural networks as models to either simulate or analyze complex phenomena and/or studies the principles of operation of neural networks analytically. It addresses problems similar to artificial intelligence (AI) except that AI uses traditional computational algorithms to solve problems whereas neural networks uses neural network as the computational architecture to solve problems. Well-designed neural networks are trainable systems that can often "learn" to solve complex problems from a set of examplars and generalize the "acquired knowledge" to solve unforeseen problems, i.e., they are self-adaptive systems.

Because neural networks do not use algorithms, would they then be excluded from FDA oversight in labs when part of IVDMIAs? Also, because neural networks have the capability of "learning" as they are used, the interpretation of a set of biomarkers by them may change from day to day.