Wednesday, June 29, 2011

Obligatory Post on Google Health

Not that anybody needs another post on the seemingly shattering news of Google Health’s recent entrance into palliative care hospice, but I think we may be missing something. Opinions out there range from Mr. Histalk’s summary of how Google Health is not offering anything of value to consumers, to the passionate cry for how the Google Health model is vital to health care, from Matthew Holt, and everything in between, including Jon Mertz’s view that Google Health’s Personal Health Record (PHR) was ahead of its time and should, or will be resurrected someday, with the usual comparisons to online banking and its one aggregator Mint. So which one is right? Perhaps all…..

Online banking’s main attraction is online bill pay. Online banking, which commenced in its current form in the mid ninties, would have probably languished without electronic bill pay. Paying bills is something people had to do since the days when this activity entailed hunting down a buffalo and carrying its various parts across one’s shoulder to pay for those cool moccasins with the lovely colored stitching. It continued to the day when you had to pull out a few banknotes and coins from the secret jar and run them over to the corner store to pay your tab, culminating with writing a check, placing it in an envelope, turning your house upside down to find a stamp and running it over to the mailbox, and of course you had to order checks and buy envelops and stamps in advance. Online banking allows you to pay for life’s necessities and its pleasures without moving an inch from your couch. It’s all in the wrist, and it takes 5 seconds at the most. What required a few thousands calories in the buffalo days and at least a hundred or so in the paper checks days, only takes about 5 calories of your energy today. The contribution of online banking and online shopping in general, to the obesity epidemic must be quite substantial.

The point here is that online banking is an evolution in the process of transacting business between human beings, which is as old as humanity itself. Online banking is just one more stepping stone between bartering for goods and the eventual disappearance of the archaic Gold calibrated valuation of human effort ant output. Online banking did not need to create a new market and did not need to convince customers to buy something they didn’t know they needed. As a byproduct, online banking took the intellectual effort out of balancing a checkbook. Not everybody was balancing their checkbooks, but those that didn’t were acutely aware that they should and that dire consequences are inevitable, and you could see the victims everywhere. Would people lineup to sign on to an electronic check balancer? Perhaps, but they would not log in very often, if at all. Enter the online banking aggregator. Mint was launched in 2007, well over a decade after online banking was introduced, to solve the unique problems of those with too many accounts, at too many banks, to keep track of. It’s a nice problem to have and obviously this aggregate service did appeal to a niche market. Mint says that they have 4 million registered users, but no numbers on how many registered users actually use the system. Either way, this is a rather small fraction of online banking users.

Patient Portals are to the health care sector, what online banking is to the financial industry. These are web utilities made available by the institution providing you with services (health care or banking), that allow you to efficiently perform transactions from your couch, transactions which you are forced to perform anyway. Patient Portals are rather new, certainly not a decade old just yet, and are slowly being adopted by consumers. As health services are being consolidated into larger and more comprehensive systems, just like banks historically were, Patient Portals are becoming more capable of providing efficiency and actually solving consumer problems. And just like online banking did not require interoperability between banks to become useful, Patient Portals have no such requirement either. You don’t need interoperability to make an appointment or pay a bill to Kaiser. You just need to log into Epic’s MyChart. And the same is true for many other large health systems running on Epic or Cerner or smaller technology platforms. In a few short years, most people will have accounts on Patient Portals (some more than one), and gone will be the days when you had to call the office for an appointment, call for a copy of your immunization record for school, or look for a tattered stamp to send your doctor payment in.

Google Health is not like direct online banking. Google Health is like Mint, and just like Mint had to wait over a decade after the introduction of online banking to be born, Google Health would have to wait until online health becomes a reality, examine its shortcomings and see if aggregation can solve problems for those who have too many health accounts at too many health facilities, and are wanting to balance their healthbooks. Most people don’t have enough ongoing financial concerns to actually need a Mint, just like most people don’t have enough health concerns to see value in an aggregated PHR. Sure, there are those who want to keep track of self-generated measurements of their health, just like some folks need to use Quick Books to keep financial records independent from, and comingled with, banking records. Most folks don’t feel compelled to use Quick Books in addition to bank accounts and most folks don’t have the urge to keep records of their health status outside a medical facility. Unless these market drivers change, either by popular culture shift or by regulation, Google’s business model will not be a good fit with aggregating health records any more than it is for aggregating financial records. Since Intuit, who makes Quick Books, was interested enough to acquire Mint, perhaps Apple, who has more self-measuring gadgets on its platform than any other vendor, may find an interest in health data aggregation. Call it Coconut, which is also a fruit and goes fabulously well with Mint.

Monday, June 27, 2011

Data - Free At Last!

Next time you visit a doctor’s office, try to peek beyond the front desk and you will usually get a glimpse of dozens of metal shelves loaded with thousands of paper charts neatly stacked in conventional library stacks. If you have a chance to walk into the medical records department of a hospital, you will see the same, multiplied by many orders of magnitude, including folks walking around with carts full of paper charts busily putting stuff into the shelves and taking more stuff out, just like librarians do. If you close your eyes and try to imagine what health information technology is doing now, you can almost hear a huge sucking sound pulling all those pieces of paper off the shelves in a big tornado-like swirl, and into the lone computer at the front desk. Play the Walt Disney music score for Fantasia in the background while the steel shelves are swiped clean and medical records are transformed into zeros and ones and swiftly pulled into cyberspace as Data. Then an eerie silence sets upon the room and all you can hear are keyboards clicking away. No more shelves, no more carts, no more paper and very few people remaining. Somehow the computer, which has morphed into dozens of terminals scattered across the rooms, looks a lot bigger now. This is how Data becomes free.

If you stick around a bit longer, a faint trickling sound will start appearing here and there. It will grow stronger and more ominous very quickly. The Fantasia movie score has music for that too. It now sounds like a raging torrent, a waterfall of sorts, and it is coming from the computer. You can practically see the deluge of zeros and ones furiously exploding out of computer screens faster and faster until the rooms are filled with free Data. Free Data comes and goes as it pleases, and for every Byte you released into cyberspace, trillions of other Bytes have been released by others, and they are all coming your way now. In computer parlance, you are experiencing a “stack overflow” – too many zeros and ones to process correctly. As computer folks know all too well, in order to fix the problem, you must first identify its source. So where is newly freed Data coming from and why is it coming your way at all?

The first source of free Data should be very familiar to any health care provider. It consists of previously paper-bound clinical information from other facilities of care. These are the old lab reports, imaging results, prescription renewal requests and the occasional consult note, referral or discharge note. These used to make their way into an office via fax, telephone, regular mail or hand delivery by patients. Instead of being stacked on your desk for review and sign-off, the Data is now neatly arranged in to-do lists or dashboards on your computer, marked with colored icons for Abnormal, Panic or Must See, depending on your software and its particular configuration. Not to worry, the lab will still call in critical/panic results. This is one of the nicest features in EHR software and a true time saver, since you can quickly go through all results, sign-off, forward to staff and release to patients. Electronic prescription renewals, if done right, are also a big improvement on the old fax requests and endless phone calls. Various progress and consult notes travelling uninhibited into your charts can be regarded as a positive development. True, these notes are currently bloated and consist of stilted, computer generated prose, but they should get better given enough time. All in all, having professionally generated clinical Data flowing freely between medical facilities has the potential to be a huge improvement on the status quo.

And then there is patient generated Data, either the manually created trickle, or the automatically created and released flood building up behind millions of mobile devices. No one can argue that obtaining as much information as possible from a patient is a bad thing. The carefully honed art of taking histories and the thorough examination that usually follows, are both aimed at collecting as much information as possible about the patient as a whole person to facilitate diagnosis and treatment. When medical records were chained down by their paper existence, patients’ access to their own records was always mediated by a staff member and required physical proximity to the chart location. As the Data in medical records becomes free, the Data currently stored in patients’ heads becomes free as well, and the two are yearning to be joined.

Several leading EHRs are already allowing patients to access their electronic records and add their own Data, such as histories, and many more will do so in the very near future. The most obvious use of such access is the replacement of the endless intake forms before checking patients in, but patients could also update their Data at will, when things change or when they remember something new. For example, Jane Doe working with the care team on weight loss and fitness could go update the frequency of workouts or note her new all grapefruit diet plan. Someone should check Jane’s long medications list and call her right away. The grapefruit is now part of your chart and part of your responsibilities. So should you have a clinician monitor patient entries into their records on a daily basis and take necessary action? It may save a life here and there, and it may also prevent a lawsuit here and there. [Note to EHR developers: when you provide patients with write access to the chart, you should also provide the clinic with an assignable task list of all updates made in the last 24 hours, with an option to accept, deny or pend the change based on further actions.]

But the most ominous deluge of patient originated data is still gearing up for its ultimate release into the land of the free Data. There are iPhone glucometers, and iPhone blood pressure cuffs and an iPhone EKG device is almost ready to enter the market. There is even an Android application that compiles and analyzes such mundane Data as where you go and who you call or text to figure out if you may be on the verge of becoming sick. All this Data, and probably much more from a variety of sensors and labs-on-a-chip developing faster than mushrooms after the rain, will eventually be set free and find its way into your chart. We are not talking about a few hundred entries a day here, which could be monitored, albeit at major cost, by a human staffer. We are looking at terabytes of data continuously streaming up into the cloud and down into your medical records. It gives new meaning to the term “ambulatory”, since you would be presiding over an entire patient panel connected to FDA approved, ICU-like machinery wherever they go. How many EKGs can you look at every day? What if you miss one and that’s “the one”? What is the value of looking at 10 EKGs from the same “worried well” person every single day? [Note to gadget developers: without a mechanism to review and analyze gadget generated data, and without accepting legal responsibility for generating medically correct credible alerts, you will have a very hard time convincing any medical facility to open their charts to your free Data.]

It is very rarely that when a problem presents itself, it comes with a solution standing right behind it, but this is exactly the case for free Data. If iPhone apps and sensors are about to create a flood of Data, the next generation of gadgets is waiting in the wings to use this Data and divert all headaches from practicing physicians to their patients. The X-Prize Foundation is developing two new challenges. The bigger one is the famous Tricorder X PRIZE which “will award $10 million to the team that develops a mobile solution that can inexpensively diagnose patients by combining expert systems and medical point-of-care data—such as lab-on-a-chip or wireless sensors, provide a recommended course of treatment, and upload all relevant data to the cloud”. The second one is the Digital Doctor X CHALLENGE which “will award $1 million to the team that builds a low-cost, point-of-care expert diagnostics and treatment system based on a computer platform that can be operated by a minimally-trained person to accurately diagnose a range of common, regional diseases”. A combination of IBM Watson software wired to a bunch of the newest iPhone gadgets should pretty much do it. Not only patients will be able to measure everything, but for $9.99 or $99.99, they could purchase a shiny “doc-in-a-box” to take home. I assume that pharmacies will have to honor prescriptions from those devices eventually, and I assume that there will be plenty legal disclaimers to make patients fully responsible for any mishaps.

And if that’s not enough to alleviate your concerns regarding terabytes of free Data floating out there, please know that the X Prize folks have it all planned out. They are considering an X Prize for a Brain Computer Interface, which should make all the difference on how Data is set free and how free Data is processed: “The winner of the Brain Computer Interface competition will be the team that successfully demonstrates a bi-directional non-verbal brain-computer communication. Teams may use an invasive or non-invasive brain transmission device, and may train the human participant in any way they desire.” The future is here.

Thursday, June 23, 2011

Excelling at Average Results

Fee-for-service (FFS) is the method by which you pay your lawyer, your barber, your waiter, your cleaning lady, your airline, your plumber,  your dry-cleaner, and your mechanic, to just name a few. If you expand services to include tangible products, then you pay fee-for-product (FFP) at your grocery store, bookstore, hardware store, car dealer, farmer market and any other purveyor of goods. Sometimes, you don’t pay FFS or FFP. When you purchase a subscription service like cable TV or New York Times online news, or a membership for a gym or a professional society, you pay a capitated rate for a predefined package of services that you have unlimited access to for a certain period of time, usually a year or more. A third way of paying for services is through extended warranty contracts. Here you pay a fixed fee, in advance, and the seller is committing to keep your product in functioning order, or replace it, no matter how much labor goes into this task. Warranties usually have exclusions, in very small print, spelling out exactly what repairs will not be made and usually if you are negligent in your care for the covered product, they won’t fix it under the warranty contract. You’ll have to pay extra if you want it fixed. It is easy to buy warranties for shiny new products and almost impossible to obtain coverage for older items that are more likely to break and may not be worth much anyway.

When you go to a doctor or a hospital, and you have no insurance, you will pay on a FFS basis for each encounter. You will receive an itemized bill for all services and materials used in your treatment, with the obligatory shock inducing grand total at the bottom. This is very similar to taking your car to a mechanic, except that nobody will give you an estimate before doing the work. Many times you will get separate bills from several physicians involved in your care, if they are not employed by the same facility, which is very confusing. If you have insurance, which is similar to having purchased an extended warranty on your persona, they will still send you the same bills “for your records”. If all services fall under the contractual terms of your warranty, your insurer will pay the bills, and worst case scenario, you will have to pay a small percentage of the total, which is much lower than the original asking price. If you happen to be insured by an HMO, your doctor won’t get payment for the particular bill. Instead your HMO has a fixed price contract with your doctor, which in effect means that your HMO is purchasing a third party warranty from your doctor to cover the warranty they sold you in the form of insurance. For services that are outside the warranty terms, you will have to pay full price.  If you visit a subscription based concierge practice, it’s pretty much like going to the gym. You won’t have to pay anything unless you order one of those high energy smoothies at the bar.

The current common wisdom is that the FFS model in medical care is not working as well as it does in every other part of the economy mainly because most consumers are covered by those pesky warranties and therefore completely desensitized to price of services. People that buy those extended warranties at the electronics store, somehow don’t exhibit the same levels of gluttony for repair services and those warranties prove to be very good business for the seller as evident in the incessant and sometimes downright aggressive efforts to make you buy one before you leave the store. So why is it that folks don’t regularly throw their iPhones down the toilet, and don’t send their notebooks back to Dell on a weekly basis, but seem to constantly be seeking more “free” medical services? The most common explanation is that those doctors, who are not selling third party warranties to insurers and are therefore paid under the FFS model, are using their professional status to encourage the superfluous usage of their medical services for the sole purpose of personal financial gain. On the other hand, those physicians who do provide warranties on their patients under an HMO model have been often accused of withholding necessary services with the exact same purpose in mind.

The proposed solution to this quandary is to require physicians and the entities they work for to sell third party warranties augmented by Service Level Agreements (SLA) to ensure adequate quality of service. Since, if you run a small shop, it is rather hard, and very risky, to sell comprehensive warranties for something as complex as a person, most care providers are consolidating to create large shops and networks capable of providing soup to nuts services in-house. SLAs are not a new concept. In the computer industry, for example, maintenance and support agreements usually include specified uptimes, first response times and resolution times based on problem severity. They also include financial penalties for all of the above. If humans could be kept on locked racks in temperature controlled rooms, connected to adequate energy supplies and be equipped with diagnostic monitoring devices (like in the Matrix), we could easily apply very good SLAs to capitation contracts. Since this is not yet the case, and since we are talking about many millions of units, the best we can do is formulate medical care SLAs mostly in terms of percentages of evidence-based processes leading to good enough outcomes at a reasonable cost per head. So for example, if 80% of diabetics can be controlled at a cost not to exceed an average of, say $10,000 per diabetic per year, the care provider will be considered to provide good value for the dollar and therefore rewarded with some sort of performance bonus. If you can maintain 99% of diabetics controlled at an average cost of $15,000 per diabetic per year, you’re out of luck, and the same goes for achieving only 50% control regardless of the price. This is called Value Based Purchasing (VBP) and this is how insurers will be contracting with increasingly larger provider organizations to supply medical care to increasingly larger populations.

This model of paying for services is not necessarily bad, since we probably have a long way to go before most providers are able to meet the currently proposed SLAs. We can also keep pushing the SLAs up incrementally, and by tightening controls on both consumers and providers, we should be able to reduce variability of processes and standardize enough to achieve Six Sigma on measures shown to best contain overall costs of care. Of course, in this zero sum system, there is no room for outliers. Just like spending all day trying to fix one stubborn server can derail maintenance on a farm of thousands of machines, and may even get you fired, providing heroic and very expensive medical care to one individual is directly detrimental to assigned population needs and may cause serious financial loss to the provider. Every rookie computer engineer knows that in large enough server farms, it is best to ditch a malfunctioning server. You won’t even feel it. That’s the whole point of having server farms. Now if you work for a small business and all you have is one lonely rack of servers, and they all have names (the Oracle box, the NAS, the web server, etc.), you’ll work all day and all night to keep each and single one alive. You’ll get fired if you don’t.

Monday, June 20, 2011

Fulfilling the PROMISe

Dr. Larry Weed
A brand new EMR is being rolled out in a midsize hospital. The EMR is exclusively based on touchscreen technology, with devices strategically placed on the floor. It provides concurrent access to medical records for all team members (physicians, nurses, pharmacists, radiologists, dieticians, secretaries) wherever they may be. Patients are also accessing the EMR. They enter their own histories and describe symptoms in detail through the same touchscreen devices. This patient-centered EMR, built by a team of clinicians and technologists working together, is taking a huge step forward in Clinical Decision Support (CDS). Physicians are not only shown differential diagnoses based on what patients and other team members entered into the system, but are also presented with individualized care plans, possible side effects, dosage recommendations and drug-drug-interaction alerts, all referencing evidence available in medical literature. Longitudinal records, test results and narratives are available by problem and by patient, and the response time is never more than half a second between the thousands of screens available. The place is Vermont, and the year is 1970.

Half a century ago, when work on this EMR was taking place, Healthcare IT was on the cutting edge of technology. The Problem Oriented Medical Information System (PROMIS), the brainchild of Dr. Lawrence Weed, was pushing the envelope on every technology from hardware to operating systems, to network communications, database design and programming languages. By the time this government funded project was finally shut down, the PROMIS team dealt with such issues as mass storage, federated or single database, high availability, human interface design and networking between geographically dispersed locations. It will take several decades for the rest of the world to catch up with Dr. Weed’s, now defunct, innovation and produce something like IBM’s Watson software package, which is yet to be adapted and tested in health care. Somewhere, somehow, we took a wrong turn in Healthcare IT, and it wasn’t the much maligned billing influence, since PROMIS from day one, attempted to integrate billing in its software, with no ill effects.

But something of that brilliant era did survive. PROMIS was essentially an early attempt to computerize the Problem Oriented Medical Record (POMR) proposed by Dr. Weed in 1964, which is almost exclusively used today in clinical documentation, and better known as the ubiquitous SOAP note (Subjective-Objective-Assessment-Plan). Practically every EHR in existence today is based on Dr. Lawrence Weed’s SOAP note format. Whether small or large, client-server or browser, free text or all template based, once you open the encounter note, you are presented with the familiar structure of History of Present Illness (HPI), followed by Social, Medical and Family Histories, Review of Systems (ROS), Exam, Assessment and Plan. Since this is perceived to be the heart of the EHR you will find much “innovation” and “secret sauce” added to the electronic SOAP note, with the singular purpose of speeding up documentation and ensuring that the finished note is a proper clinical, legal and financial document. And as most of us know only too well, we are not there yet.

Interestingly, the folks working on PROMIS faced the same hurdles we are facing today, albeit their tools and technologies were pretty much stone-age compared to present day technology. It is fascinating to see how much effort and concern went into selecting just the right user interface, ensuring that response time was measured in fractions of a second and in keeping the system up 24x7 and as error free as possible. For the clinical staff that took the time to enter data through the state of the art user interface of a  Cathode Ray Tube (CRT) monitor with little touch-sensitive strips attached in just the right places, PROMIS delivered serious value. It offered differential diagnoses for each problem, patient specific care plans, collaboration and real time access to medical records. It obviously wasn’t enough though, since the project did get canceled in the early eighties. Today’s EHRs can, and do, offer collaboration and real time access to records much more efficiently then PROMIS ever dreamed possible. However, only very few EHRs are capable of coming up with differentials, and care plans consist mostly of order sets that you can create yourself (if you wish). As to efficiency of data entry and pertinent information retrieval, we are only slightly better off than the PROMIS pioneer users were. You would expect that every EHR vendor, big and small, would be feverishly working on exactly these problems, trying to bring more value to their customers and differentiate themselves in a crowded market. Well, they were, until very recently.

As Meaningful Use is pushing, shoving, enticing and coercing everybody, by any means necessary, to abandon paper medical records and adopt EHRs, it is also redefining the nature and construct of those records and it is imposing a new set of priorities on all EHR builders. Meaningful Use is about collecting certain data and moving all data out of the originating system to all sorts of other systems, including patients, care providers, governments and other facilities, and more than anything else, Meaningful Use is about measuring clinical quality or lack thereof. There are 113 clinical quality measures proposed for Meaningful Use Stage 2 and there are 65 measures in the Accountable Care Organization (ACO) proposal, which Meaningful Use is also endeavoring to support. Although users are only required to report on a handful of measures for now, those who build decent EHRs must write code for all measures, and this is not a simple thing, since a single measure can require complex computations over many data elements which may or may not even exist in the software. Add to this the remaining Meaningful Use measures and you have a big problem, amplified by orders of magnitude due to the very short timelines between publications of new mandatory requirements. The result is that there are no development cycles left for such things as enhancing user experience, or adding features that customers routinely ask for. Granted, Meaningful Use is considering taking on usability of EHRs as well, but at least initially, this will be through the narrow lens of patient safety, and not so much related to direct value to the actual paying customer.

Are we then doomed to repeat the PROMIS disappointment on a grander scale because the value to the customer is not readily recognizable? Not quite. The differences between our effort to computerize medical records and those led by Dr. Weed forty years ago are many, and none is larger than the fact that today’s copiously funded campaign for EHR adoption is firmly anchored in a much larger effort to change the health care delivery system as a whole (for better or worse). While Dr. Weed was experimenting in one remote hospital, today we are moving full steam ahead on a national level in thousands of hospitals with hundreds of thousands of physicians in both hospital and ambulatory practice, and the point of no return (to paper) has been passed a long time ago. We do however run the risk of making the entire process unnecessarily painful, slow, expensive and fraught with unintended consequences, if we continue to prioritize the political needs of the project itself above and beyond the needs of the customer.

Thursday, June 9, 2011

Meaningful Use Stage 2 – Horses, Camels & Signals

On June 8, the HIT Policy Committee at ONC has approved the Workgroup recommendations for Meaningful Use Stage 2. Before diving into the details, it is worth noting that the time crunch for moving from Stage 1 to Stage 2, for those seeking incentives in 2011, was proposed to be resolved by postponing Stage 2 for these early adopters for one year. As I noted before, if you are able to attest and obtain incentives in 2011, go ahead and do that. You will be rewarded by having the opportunity to stay at Stage 1 for 3 consecutive years. The final Stage 2 ruling is not expected to occur until June 2012 and judging by previous experience with Stage 1, the recommendations approved today will be significantly relaxed by the CMS process of proposed rulemaking and public comments. So although analyzing (rejoicing or bemoaning) the various measures on this long list is a bit premature, it may be helpful to look at the general principles embedded in this new stage of Meaningful Use.


Many of the Meaningful Use more pedestrian measures have remained unchanged, have increased in intensity, or have been moved from menu to core (more on this later). These measures include such items as recording patient demographics, maintaining medications, allergies and problem lists, recording of vitals, running reports, electronic prescribing, incorporating structured lab results, medications reconciliation, using formularies, enabling clinical decision support, reporting to state and federal agencies and ensuring privacy and security of medical records.

Other measures were slightly expanded in scope. CPOE, for example, was restricted to medications in Stage 1. Stage 2 is adding laboratory orders and capability for radiology ordering. While Stage 1 required that physicians send clinical reminder notices to young children and the elderly, the current proposal is to send those reminders to at least 10% of all patients, and patient education materials which were required “if appropriate”, are also extended to all patients in Stage 2. If you were not a hypochondriac before, you will be encouraged to become one in the near future. The requirement to document the existence of advance directives, which was limited to hospitals, is now also applied to physicians of all types. To even the burden, electronic prescribing, a strictly ambulatory measure, will now be required for 25% of hospital discharges.

And then there are several benign brand new measures:
  • Both office visits and hospital days are required to have electronic clinical notes (30%). The measure is not very prescriptive on what constitutes a note other than forbidding scanned documents and requiring that notes should be searchable.  I can’t wait to see the certification test for “searchable”. It seems the committee is not aware that with very few and limited exceptions, EHRs are not very much like Google. This new measure ought to single handedly disqualify 99% of currently certified EHRs. The searchable part will probably be dropped during rule making.
  • For hospitals, use of an electronic Medication Administration Record (eMAR) will be required in at least one department.
  • Physician practices will need to record patient preferences for methods of communications, but hospitals strangely don’t need to do the same.
  • There will be new clinical quality measures added to the current list to choose from, but this item alone warrants its own separate dissertation.

As one of the committee members astutely remarked, along with the horses a few camels are also being created by this proposal. Hopefully, most of these creatures will be reshaped and/or eliminated in the final rule, but right now here are the problematic recommendations:
  • No menu items – Stage 1 allowed hospitals and clinicians some freedom in choosing which measures best apply to their situation. While 15 measures were mandatory (core), physicians and hospitals were free to select 5 additional measures from a menu of 10 choices. Meaningful Use Stage 2 is an all or nothing proposition.
  • Hospital labs should send structured electronic clinical lab results to outpatient providers for at least 40% of electronic orders received – This particular triple humped camel elicited much discussion in the committee and exclusions may be provided for small hospitals. Considering that the overwhelming majority of hospitals, small and large, are not currently capable of receiving electronic orders, a fact that seemed to have escaped the committee, this measure amounts to pretty much nothing, since 40% of zero is still zero. It remains to be seen how many of the few hospital labs that can currently receive electronic lab orders will turn that capability off just to escape this particular requirement. Bi-directional lab interfaces are not without cost to build and maintain.
  • Measures to be fulfilled by patients – There are two types of measures that hold providers accountable for patients viewing medical records online (10%) and for patients sending electronic communications to physicians (25 individuals). It is not clear who the messages should be sent to, but Meaningful Use is applicable to clinicians individually. I guess, practices and hospitals can make medical records available online and advertise the ability to communicate electronically, and hope that enough patients take advantage of that. Since we are dealing with all or nothing Meaningful Use compliance, it is more likely that patients will discover one more step in the check-in process – logging into the portal. The other issue here is that while certifying for Stage 1 Meaningful Use, many small vendors utilized third party PHR products (usually Microsoft HealthVault) to make medical records available to patients automatically. These vendors have no patient portals and no way to verify that patients actually viewed their medical records, without asking the patient to be kind enough and send in a copy of their PHR utilization reports.
  • For care transitions, summary of care and care plan are sent electronically to receiving facility or physician, and care team members’ list is available for electronic exchange – This was probably the most debated set of measures and the results as they appear in the proposal are difficult on many levels. First and foremost, the care plan in Stage 2 is envisioned to be just free text attached to the standardized summary of care record. Most advanced EHRs have separate sections, some structured, where care plans are directly documented and/or automatically assembled based on orders. It is not clear how that free text is to be created to satisfy the measure, other than requiring physicians to document the plan a second time. It is also not clear how you hold one party responsible for sending things electronically while it has no control on the abilities of the receiving entities (look for major exclusions here). And it is completely unclear what exactly should be done with that free text list of team members, once it is available, and who should be on it (Other specialists? Hospitalists? Community resources?)
  • The last two new items are just suggestions for CMS to evaluate, and both deal with expanding ambulatory reporting requirements to include syndromic surveillance and cancer reporting. Not sure if there’s anybody out there willing and able to accept such data.

As expected Meaningful Use Stage 2 is significantly upping the ante for 2013 (or 2014 now), but more importantly, the committee’s recommendations are also “signaling” to the more distant future of Stage 3. Possibly the most frequently used term in workgroup and committee meetings, “signal’ (used both as a noun and a verb) refers to providing hints to users and software vendors, indicative of intended future requirements of Meaningful Use – one if by land, two if by sea. So here are the brightest lanterns shining towards Stage 3:
  • Have nationally endorsed lists of drug-drug-interactions (DDI) and ability to record reason for overriding alert – Your DDI alerts, the ones that you were able to adjust and configure to your liking, will be coming from a federal designated source. Hopefully they will come in the form of friendly advice. Either way, they will want to know why you are overriding things.
  • Second hand smoking status will be added to just plain smoking status, and more refined race and ethnicity data will need to be captured.
  • In Stage 3, EHRs will need to be able to store and retrieve those advance directives, not just indicate that they do or do not exist, which is pretty useless in and of itself.
  • Family History recording will become a requirement as soon as they devise a proper standard for it. Considering that most EHRs already have structured templates for documenting all histories, and most of them are pretty much identical, it would be interesting to see what the new standard will look like and how much rework and data migration will be necessary.
  • That care team field mentioned above will no longer be just free text. National Provider Identifiers (NPI) will be required. It seems that only licensed clinicians can be on a care team.
  • Patient-generated data will be submitted to public health agencies - I am not going to even venture a wild guess regarding this one.
  • Compliance with the Nationwide Health Information Network (NWHIN) governance policies should be included in Stage 3 certification criteria - Here is where the future is beginning to look awfully hazy to me.
If the proposal to extend Meaningful Use Stage 1 through 2013 is accepted by ONC and CMS, and it should be, Stage 3 will most likely be pushed out by one year as well, so these various signals are directed to the year 2016 through heavy fog and precipitation. It is too soon to alarm the country. The most immediate order of business is to smooth those camel backs, and we have an entire year to do just that.

Saturday, June 4, 2011

Coordinating Care Coordination

Care coordination is one of the four pillars of Meaningful Use, one of the six NCQA Patient Centered Medical Home (PCMH) standards and one of the main goals of Accountable Care Organizations (ACO). Care coordination, particularly for patients with multiple chronic conditions, is expected to reduce unnecessary repetition of laboratory testing or imaging and the number of avoidable admissions. Other than reducing overall costs, care coordination is also supposed to improve quality of care. According to experts like Joe Flower, “Lack of care coordination is at the core of the mess healthcare is in”, and nobody in their right mind would argue that it is best that medical care remains disorganized and uncoordinated, if it is indeed so. It seems that our fee-for-service, fragmented and fractured (lots of f-words here) health care system is not conducive to care coordination. When patients float around in a sea of hospitals, physicians, nursing homes and other facilities, each care provider gets paid, and is responsible for the piecework performed at their independent entity and nobody is minding the handoff of patients to the next provider of care, and nobody is assembling a comprehensive picture of the entire care process, let alone orchestrating, or coordinating, the progression of patients between stages of care and the overall needs of patients in transit. What would it take then, to see that the bits and pieces of health care we now have, become a safe and affordable continuum of care?

CMS is taking the lead, as it should, in an all-out effort to encourage health care coordination through various carrot-stick initiatives, aligned to ultimately base payment for medical care on value to the patient, as measured on a population level, instead of fee-for-service and no accountability for outcomes. These initiatives fall into three general categories:
  1. Health Information Technology to assist with documentation, information exchange and measurements as required in any coordination effort.
  2. Incentives and penalties for providers based on measures thought to be influenced by care coordination (e.g. preventable hospitalizations, readmission rates, etc.)
  3. Financial and structural encouragement for vertical integration of the delivery system (e.g. ACOs, consolidation, employed physicians, etc.)
Of all three categories, only Health Information Technology (HIT) is foundational.  HIT is supposed to provide a toolbox for simplifying and even automating many of the tasks associated with coordination of care. The other two categories are based on series of assumptions, hope, belief and lots of other magical thinking. A brief scan of Meaningful Use regulations, which is quickly becoming prerequisite to both PCMH and ACO, reveals the following proposed care coordination enhancing features:
  • Health Information Exchange (HIE) – To paraphrase the quality measurement mantra, you cannot coordinate that which you don’t know. Most care coordination efforts today are based on faxes, phone calls and using the patient as courier for paper based artifacts. If there is one thing that computers excel at, it must be timely exchange of information. Aggregating and placing complete health information at the fingertips of all clinicians, in real time and with minimal loss of fidelity, would in and of itself eliminate most effects of geographical and organizational diversity and fluidity of care providers for any given patient. Most, if not all, remaining features are just elaborations on the information that needs to be exchanged, the actions to be taken following the exchange and the tracking of both the actual exchange and the activities surrounding such exchange. 
  • Planning Care – Care plans are not a new idea, but now they will have to be composed with patient participation, and shared between physicians, hospitals and other care givers. The vision is to eventually have one collaborative plan of care accessible to all involved.
  • Managing Transitions of Care – Here you find the multitude of documents to be shared when care is transferred between facilities and/or physicians. There are care summaries, discharge notes and instructions, medication lists that need to be reconciled, and the PCMH standard adds careful tracking of referrals, appointments, test results and consultation notes.
  • Measuring Results – Meaningful Use lists 48 clinical quality measures, and more are being developed, ranging from weight measurement and dietary counseling to inpatient medication administration. At this point, this is mostly about measuring care process metrics.
What we have here is nothing more than classic Project Management: Planning, Collaboration, Change Management and Measuring. The project is health and the project owner is the patient. There are millions of such small projects executing every day around the country. In the current system, many are executing with no project management. Sometimes the owner assumes the role of project manager, sometimes a conscientious primary care physician takes on that role, but most often we all step in and out of a management role as the project heats up during events such as birth, acute disease or trauma, chronic disease diagnosis and terminal illness.  Most of us, including physicians, have no resources, no training and no decent tools to manage our health projects. In the business world, professional project management is part of every (successful) project, with purpose built tools and adequate budget allocations for this coordinating function.

Perhaps this realization is the main driver behind the expert advocated, and government endorsed push to consolidate health care into large business entities. The viewing resolution from Washington, and from equally removed academic departments, renders all our pixel-size individual health projects indistinguishable from each other in the large picture of population health. This is one case where the forest is obscuring the trees. So with the best of intentions, the government is proposing to create a handful of Project Management Offices (PMO), which is what ACOs really are, and make them fiscally accountable for operational profit and loss (P&L) at a population level. Instead of dealing with hundreds of thousands of small contractors, the government will be stepping into its familiar role of contracting with large corporations as the prime contractors for goods and services. These prime contractors, similar to their counterparts in the defense industry, will be at liberty to subcontract with smaller entities or employ their own resources, and together they will completely dominate all aspects of the market.

This massive reengineering of the health care delivery system is deemed necessary in order to begin eliminating the excesses and fragmentation brought on by the centuries old fee-for-service model of medicine. But if we learned one lesson from the fee-for-service model, we certainly learned that if we pay by the yard, we get more yards. So common sense begs the following question: if we want more care coordination, why not pay for care coordination (by the yard)? Unlike quality of care and outcomes, care coordination processes are easy to define, easy to measure and widely understood by all stakeholders. If CMS could define a set of CPTs for Evaluation & Management which entail counting all body parts being examined and accounting for all diseased relatives, couldn’t it define a set of Coordination CPTs to track care planning, exchange of care summaries, referrals, results, instructions and whatever else is deemed to contribute to care coordination?  Computers, HIT and the Internet are greatly enabling these activities, both inside and across organizations of all sizes, and are automating their documentation in the medical record. If physicians and hospitals were allowed to bill these Coordination CPTs at various levels for each encounter, and they were paid at attractive rates, we would see a flurry of activities leading to HIT adoption, health information exchange, collaboration and ultimately coordination of care.

Paying for care coordination on a fee-for-service basis will allow a unified definition of services and individual customization per patient. Some patients will need more, others will need less, and different people will need different coordination services at various times. Unlike a capitated fixed fee per patient per month, there will be no temptation to skimp on services and it is really hard to argue that too much coordination is even possible. Indirectly paying for a finished product, such as paying for outcomes, is very difficult in health care since the product is only finished when we are dead, and retrospective measurement of expenditures for dead people is nothing more than crying over spilled milk. If we maintain the granular fee-for-service system of levers, we can recoup coordination expenses by carefully reducing payments here and there for services we want less of.
So instead of punishing hospitals for failing to coordinate care, instead of penalizing doctors for lack of technology, instead of incenting everybody to do the right thing in roundabout ways, and instead of faith-based reengineering of the entire system from the top down, why not try to achieve our goals the American way, by paying a fair fee for an honest service?