Blog: Lou Agosta Subscribe to this blog's RSS feed!

Lou Agosta

Greetings and welcome to my blog focusing on reengineering healthcare using information technology. The commitment is to provide an engaging mixture of brainstorming, blue sky speculation and business intelligence vision with real world experiences – including those reported by you, the reader-participant – about what works and what doesn't in using healthcare information technology (HIT) to optimize consumer, provider and payer processes in healthcare. Keeping in mind that sometimes a scalpel, not a hammer, is the tool of choice, the approach is to be a stand for new possibilities in the face of entrenched mediocrity, to do so without tilting windmills and to follow the line of least resistance to getting the job done – a healthcare system that works for us all. So let me invite you to HIT me with your best shot at LAgosta@acm.org.

About the author >

Lou Agosta is an independent industry analyst, specializing in data warehousing, data mining and data quality. A former industry analyst at Giga Information Group, Agosta has published extensively on industry trends in data warehousing, business and information technology. He is currently focusing on the challenge of transforming America’s healthcare system using information technology (HIT). He can be reached at LAgosta@acm.org.

Editor's Note: More articles, resources, and events are available in Lou's BeyeNETWORK Expert Channel. Be sure to visit today!

November 2009 Archives

AMB logo.JPG

I had the opportunity to sit down with Steve Meister, President, and Paul Henkins, Director, of AMB and get the update on the data quality software innovations in the pipeline - and in production at client sites. Although these guys have been in the data quality market for about ten years, a few years ago they the company bet on new development using web services and C# when the market was still skeptical about the long-term viability of the language. Today they are repeating the rewards in terms of leap-frog capabilities that deliver comprehensive, integrated metadata-driven instream discovery (profiling), pattern analysis, probabilistic (fuzzy) matching, drill back to source, repository-based reporting, and anomaly correction.

AMB has been making solid inroads in the public sector and healthcare in both the provider and payer markets. Insurance companies, including healthcare payers, have been clients from the start. In addition, its partnering relationships have gotten it "out there" on a global basis through internal applications at large software providers. For example, AMB is "under the hood" with Sartori Software's postal and list cleansing services, which gets it out to over 5000 clients without their necessarily knowing about it. PDM is integrated with Microsoft SQL Server Integration Services (SSIS).

Any data quality project quickly surfaces a wealth of tactical details about valid values, tolerances, and metadata. However, the payoff occurs at the mission critical business level. Business people "get" data quality when it is expressed as policy-based governance - it is about all the key master data dimensions - customers, patients, providers, products, services, diagnoses, procedures, and all the related transactional details.

In any market - including the challenging ones we face today - information is that which reduces uncertainty. Data whose quality is suspect increases uncertainty, and that is a situation incompatible with enterprise success. Much of the AMB product demo is necessarily about features and functions that make the technology easy to use and powerfully productive for stake holders such as data stewards and business analysts. However, the management at AMB understands that it will succeed by automating data governance and information management at the level of the enterprise. One lesson learned? Quality attracts buyers and prospects, and one of the challenges faced by AMB will be to manage its growth, choose its clients wisely, and build for long-term success, even while hitting the number quarterly.

Based on a live, interactive demo, this is one of the most usable interfaces and products that I have seen in years. Fuzzy, probabilistic matching is now mainstream and the products delivers it in high performance algorithms that meet the need of real time and near real time web services. Note this runs as a web services engine, not an API, which means performance benefits and a flexible architecture that accommodates the demanding interactive environment. Of course, it also works well for less demanding batch processing.

Although the current pricing is aimed at "getting a footprint" in the site, PDM supports enterprise environments (and enterprise pricing where applicable) for direct sourcing for profiling and quality, including DB2/UDB, DB2/400, DB2/z-OS Mainframe, Flat files, MS SQL, Oracle, and Teradata. PDM is one of the first supported products for SQL Server 2008 (as source, repository, and SSIS) with Version 6.0.5. PDM delivers the Profiling/Analysis results as a standard relational database repository rather than a vendor proprietary database. Repository database options include the de facto open options - MS SQL Server, DB2/UDB and Oracle. An actual open source option such as MySQL would be a nice to have as the open source movement marches on.

The "predictive" aspect of the "PDM" branding is based on the results of profiling that builds a rule to flag outliers that exceed the tolerance from a base-line. One healthcare client - call them Acme Dental for the time being - discovered a highly suspect $400k reimbursement that in itself would pay the cost of the entry level starter-kit 40-times over. Whether an inadvertent error or something more sinister, such outliers still requires follow up by a human auditor. So standby for update.


Posted November 24, 2009 1:33 PM
Permalink | No Comments |

To be sure, the USA is not the Sudan. The USA does have a system. Medical innovations are redefining the limits of what is possible in healthcare - biologic "miracle drugs," minimally invasive fiber optic procedures, robotic surgery, powerful magnetic resonance imaging (MRI) machines, and emerging prospects of personalized medicine. These innovations are enormously valuable, incredibly expensive, and enjoy wide public support. Count me in! Yet there are significant "buts"... Much of the healthcare technology is isolated, under utilized, and mis-directed by being located behind information technology (IT) systems that are years if not decades behind the curve. Even when the information technology is upgraded - say, in order to capitalize on federal incentives - it is sometimes used to "pave the cow path." In other words, lay down a veneer of GUI-based modernity over a workflow that is otherwise distinctly 1950s in its organizational and informational design.

When the patient looks at the doctor taking handwritten, paper-based notes; when she tries to get an x-ray sent from the facility to a specialist - and finds it is easier to walk it over in person; when a member of the family is out of town and requires a record from his healthcare file; when a patient or provider tries to get an accurate, committed answer about whether a treatment is reimbursable by a payer - then the modern operating room and laboratory seem less modern and indeed are held in check by a back office, workflow, and organization design that is distinctly prior to the invention of automation. In some cases, these are symptoms of a non-existent information infrastructure, and, in others, of one firmly rooted in the past. Unless explicitly addressed, they represent the default future - more of what we are already doing. We have highly trained doctors and equipment trapped on islands of information separated by manual processes, IT systems with robust 1980s architecture, policy-based turf delimiters, and sneaker networks - remember that? Get on your sneakers and carry the diskette (now a CD) over the connecting bridge between buildings on foot. A bold statement of the obvious: The IT system is due for an upgrade. IT is a lightening rod for policy and organizational transformation that is on the horizon and closing fast. However, the emphasis in this article is on the IT aspects. You will know we are making progress when the IT system supports -

Coordination of care (COC): Note that here "coordination of care" refers to multiple medical specialists in different locations accessing the same patient information. It is important to make this distinction since "coordination of care" also came to mean "reduction or denial of service" by payers, taking on a politically-charged meaning that it should not really have. This time COC does not mean reduced payments for doctors or services for patient like it did in the 1990s. This time it means doctors and care-givers work together using high bandwidth networks and accurate databases to mange complex conditions presented by individuals in context. In a commercial business context, this kind of coordination would be called "customer relationship management" (CRM), including a single view of the customer. Doctors and healthcare providers are smart people - very smart - and this means they will readily learn and adapt when a new possibility is presented that enables improved patient care, accountability, transparency, and reduced risk.

Computational resources available at the point of service: The goal is to access and use information to reduce uncertainty in the examining room, patient bedside, or ambulatory facility. Nevertheless, where even a single life is at stake, automation requires supervision and approval by human (medical) intelligence. This is the value of the collaboration between bar-coded prescriptions, nurse identifiers, and clinical decision support software "double check" prior to administering drugs. A few high profile cases where potentially fatal medication mix ups were caught and prevented - or, unfortunately, not prevented - is a powerful justification. The drill requires that patients, medications, and providers be unambiguously identified and authenticated by a unique, barcode-like id. Then when the nurse goes to change the intravenous drip, he or she must scan the medication, scan their own id, and scan the patient's id - as easy as 1-2-3. If potassium chloride has been mistaken for sodium chloride, then the system catches the error, and the lethal mistake is avoided. Note also the upstream implications for physician order entry and physician accountability. The doctor (or her or his assistant) must have entered the order for the medicine into the clinical management system and it must be available in a real time or near real time way. Anything that reduces risk in a system already stressed out by the need for tort reform in an overly litigious society will be embraced and exploited.

The value of database resources is recognized and implemented. This is taken for granted in many business contexts. Thinking in healthcare is still catching up. A case in point. When Hurricane Katrina put the city of New Orleans under nine feet of water, including the medical records, the need for disaster recovery was demonstrated in spades. Of course, both paper and electronic records were submerged. However, in some cases - such as the Veterans Administration - an off site, distributed back up of the electronic records was available. Those veterans were able to receive up-to-the-minute care in Houston or other evacuation site, since the electronic version was actually located in a different city or it was restored and available within days. So some things are working right, albeit intermittently across the system. The recommendation? Get to a standard relation database. A centralized repository or a distributed one that appears and functions in a centralized way is critical path for coordination of care. Maybe I need to get out more, but the number of so-called modern, run-your-hospital systems that implement the admittedly powerful b-trieve database, Mumps, is a significant symptom - and data point. Mumps is not bad, but it is a curiosity. There is a clear cost reduction opportunity in surmounting the technology lock in such a "one off," proprietary (legacy) choice in favor of an open approach where "open" extends to de facto standard market offerings from IBM, Microsoft, Oracle (to name just the "big three").

Personalized medicine gets traction. In a commercial business context, this would be called "business intelligence," using data to track trends, influence markets, and create opportunities. I hasten to add that the challenge is to use comparative effectiveness research (CER), which is related to pay-for-performance, in an accurate and timely way to identify winning treatments rather than to find an excuse to deny service. For example, Genentech/Roche's Avastin costs $50-$100,000 per year of treatment but works in fewer than 50% of patients. A relatively simple upfront test for Avastin response could save as much as $6 billion per year if all nonresponders could be identified and not targeted. The potential cumulative savings are substantial. (Data taken from - http://www.genengnews.com/specialreports/sritem.aspx?oid=64921206 ) This is a high bar, and there is still a lot of data modeling, data collection, data cleansing, and system integration to be done before meaningful predictions can be made about treatment effectiveness.

Social factors and usability are integrated. Short cuts for expert and advanced users of the system to fast forward through steps that are already familiar are required. Expect healthcare providers make for demanding users - and they should be. They require support accordingly. We are learning a lot from practices at the Veteran Administration (VA) and at the Mayo Clinics, including the stunning insight that automation is most useful only if it is used; that automation can be inflexible, requiring doctors and providers to accommodate the rules. The latter can be misperceived as an affront to the sovereign authority of physicians operating in the "command and control" practice model. Usability testing and user (physician) buy in remain on the critical path.

Patients and consumers take responsibility, and do this regardless of the work-in-progress nature of the new and existing IT systems. Lest the reader imagine that IT is a silver bullet, there are lessons here for healthcare patients (consumers) too. Regardless of what comes out the US Congress, this quarter or in the future, consumers are held accountable for their health. The default is to be held accountable by life - getting sick. We simply cannot afford unhealthy lifestyles. Eating right, regular exercise, following the advice of one's doctor (apparently about one half of patients actually do so), and participating in wellness and treatment are the order of the day. Obviously this is easier said than done, especially for stressed out consumers, working two jobs - or none at all. Yet it is critical path. You don't feel like exercising? Instead of going to the refrigerator, get on your sneakers and get going! Even then there are no guarantees - people who "live right" still get sick or have accidents. That is why insurance was invented - get everyone into the pool and spread the risk. Still, do I smell an economic opportunity for gyms, work out shoes and gear, and fruit and vegetable merchants? Obviously there's lots more to be said - but let me acknowledge this article is already way too long for a blog post, and will be republished as a standalone article. However, I see value in launching a "prototype" - just to "get it out there" and get instant feedback - so let me hear from you. The title is intentionally provocative ... meanwhile, I've got to get up out of this chair, follow my own advice, and get some physical exercise.


Posted November 23, 2009 1:56 PM
Permalink | No Comments |

The career path to becoming Chief Information Officer is as diverse as the individuals who occupy the role. Some have come up through the information technology (IT) ranks, installing networks, coding software, backup up data warehouses, designing and implementing applications. Others cross over from the business side of the house - marketing, customer service, finance, operations - once the individual realizes that all business processes have software as an essential part of their operation.

The Medical Doctor (MD) as CIO adds a new wrinkle to the mix. It also extends the black humor that says "CIO" really means "career is over." It becomes an end point, albeit one with privileges and (some) power. By the way, "CIO" has yet another meaning that we will explore shortly.

If we take a step back and look at the methods and practices that inform each role - MD and CIO - there are some engaging fits and misfits. At a high level and when the training is at its best, the MD is trained to

  • handle symptoms in relation to underlying causes, making difficult differential diagnoses,

  • thinking organically about systems as a complex interacting whole

  • communicate with the customer [patient] as an end in her- or himself, deserving respect

This seems like a good fit for the role of CIO. It includes the necessary skepticism that the presenting error message thrown up by the application has anything to do with the underlying cause. The network is slow so the application times out and "thinks" the connection is down without saying anything about the network. Things are not what they seem.

Organic metaphors are increasingly common in an information technology context. Computer viruses are pervasive on the Internet, and virus detection software embraces the metaphor of inoculation, though it is inevitably a step behind the pathogen. Human engineering factors are critical path, and best practices indicate caution about the vector with whom one exchanges bodily fluids - or email attachments. The goal of "self healing" IT systems is on the horizon, and ongoing monitoring of "heart beat" in high availability systems is implemented by mechanisms that are still primitive but increasingly powerful.

In comparison with the "human biocomputer," our IT systems are primitive at best, using kludgey mechanical and serial processes where the wetware is able to perform complex protein synthesis. This makes our most sophisticated business intelligence systems seem like child's play. So the metaphors and analogies are likely to remain limited, conditioned, and qualified for the foreseeable future.

The "people skills" of the CIO - whether MD or information techology (IT-trained) - are as diverse as the individuals who occupy the role. Given the pressure of primary care medicine, clinical practice does not always promote listening, empathy, and education to good health to the extent we all might wish. The myth (and misfit) of the physician as member of a "sovereign profession," whose word is law, has been explored and (to an extent) debunked in the wake of the HMO revolution of the 1980s. This is not new news (and for those for whom it is, Paul Starr's The Social Transformation of American Medicine is as timely today as ever.) It is not that the role of CIO is intrinsically more humbling than that of the MD; but both require mastery of the distinctions between knowing what you know and knowing what you don't know - and knowing there is a third area - you don't know what you don't know - the area of "blind spots," out of which most system disasters and malpractice actions arise. Of course, there is no easy answer for the latter - since by definition one does not know what it is that one does not know - but the prophylaxis, if not the antidote, is to be resolutely open to inquiry into one's own limitations and those of the systems on top of which one is riding. In short, CIO also means "chief inquiry officer" - always inquiring, always asking "what if," and creating possibilities for the future.


Posted November 14, 2009 12:08 PM
Permalink | No Comments |