"machine learning-based clinical decision support" This is just begging for Medical Device approval.
The enthusiasm generated for EHRs by the 2009 ARRA legislation is almost palpable and hospitals across the country are scrambling to install systems at a breakneck pace. Behind the enthusiasm, however, are two issues, related yet disparate, that have been the confounding factors of EHR adoption in the past and will continue to be so in the foreseeable future.
EHRs offer the promise of data aggregation which can be used to refine clinical treatments for both improved quality and, possibly, lower costs, but this aggregation is dependent upon standardized dictionaries and, importantly, standardized data entry. EHRs currently offer standardized data via the use of templates, boilerplates, and pre-defined order structures. But the standardized data entry model often (usually?) does not completely and precisely conform to the observed signs, symptoms, and problems displayed by patients in the physician’s office, and therein lies the rub.
Patient care, especially when dealing with complex problems, requires the clinician to differentiate subtle distinctions among less than obvious alterations from normal physiology. Shortness of breath, one of the most common problems encountered in the emergency room, can result from problems with the lungs, with the heart, with the vascular system, with the blood, from medications, or simply from pollution or toxins breathed in by the patient, and those are the direct causes. Indirect causes such as intra-abdominal pathology, skeletal deformity or muscle weakness must also be considered.
While there is a high statistical likelihood that shortness of breath will result from one of a relatively small number of potential pathologies, assuming a diagnosis based on statistical likelihood will lead to poor or even dangerous patient care. The reason a pulmonologist trains for 12 or 13 years, and a nurse practitioner for six or seven, is to allow the pulmonologist to learn not only the underlying basis of the more rare causes of disease, but also to be able to discern the subtle differences that those more unusual pathologies may display. The use of template- or boilerplate-driven clinical notes negates the benefits of the more refined knowledge and experience of the pulmonologist. Requiring the use of such standardized data inputs is antithetical to quality medicine, yet allowing free text entry is equally antithetical to the as yet unrealized potential of the EHR. It is this contradiction which has slowed adoption of EHRs and will continue to hinder their use.
The challenge is for IT designers to work out a way for experienced clinicians to be able to commit to the record the sometimes subtle thought processes and observations that lead to their diagnoses, while maintaining enough control and/or discipline over the input to allow the potential of data aggregation to be realized. Monetary issues, regulatory compliance, and usability are important as well, but the paramount concern of the EHR must be to ensure that the best quality patient care can be delivered. If the cost of the input restrictions needed to allow data aggregation is the loss of ability to place nuance and subtlety into the record, the EHR fails that most primary of tasks.
Al Davis, MD is in private practice in Elmhurst, IL.
A Meaningful Ruse?
By Frank Poggio
At the risk of being a called a Cassandra, or at best a contrarian, I will attempt to explain why the federal government’s HITECH Act and Meaningful Use (MU) incentive program is a wolf in sheep’s clothing and why the better response for a provider would be to run, don’t walk, from this wolf.
First let’s review the basics. When a hospital or physician’s practice purchases and implements an electronic medical record (EMR) or Computerized Physician Order Entry (CPOE) before 2011 and files with the federal Department of Health and Human Services (DH&HS) the yet-to-be-developed regulatory documentation to declare their meaningful use (MU), then starting in 2011 that provider will be potentially eligible for an MU bonus payment. For physician practices, that could amount to a total of $44,000 over three years. For hospitals, depending on the number of discharges, somewhere between $2million to $3.8 million total. These incentive amounts are to be paid over three stages, or years, starting in 2011.
On the other hand, if a provider does not implement an EMR or CPOE, or purchases and implements a system but cannot show meaningful use, then a penalty will be incurred on Medicare payments in years 2015 thru 2017. This penalty will be in the form of a reduction to the legislated increase in Medicare payments for that year. Note: this is not a reduction in overall Medicare payments, but a reduction on the yearly Medicare inflationary adjustment factor. The first year the penalty is a 33% reduction of the adjustment, the second 66%, the third 100% (or in effect, you will get no adjustment at all).
Before I explain why I believe there is a wolf at your door, let me say I am a believer in the benefits of EMRs and CPOEs. There can be significant benefits in both, but not unless they are incorporate a sound work flow re-engineering processes prior to installation. Unfortunately there are very few if any MUs that are workflow-focused.
There are at least four major reasons why I believe your facility will never see an MU bonus.
1) MUs are, by the DH&HS’s own admission, a moving target. As stated in the Interim Final Rule (IFR) published in the Federal Register, December 30, 2009, on page 314, “We expect to issue definitions of meaningful use on a bi-annual basis beginning in 2011”. Hence, MUs will evolve over time. That will allow DH&HS to make them as easy or as onerous as they choose. How can you predict you will hit a moving target that you can’t even describe today? And if you believe the Feds may try to make it easier to foster participation, read on.
2) If you hit all but one MU, will you get the full bonus, or 95%, or 50%? Nobody knows and the question is not addressed in any IFR or other documents. I am willing to wager you will get nothing, and my reasoning follows.
3) The federal government has stated they are funding the HITECH program with $34 billion for MU bonuses. They also have stated repeatedly they expect to save over $200 billion to help fund the new national health plan. That’s about a seven-to-one expected payback in only a few years. When was the last time you had a seven-to-one ROI on any IT project over three years? If the feds do not see the seven-to-one payback in time, how many providers do you think will get to cash an MU check?
4) Our government is under extreme pressure to cut the federal deficit. In the President’s recent State of the Union Address, he stated he will freeze the government budget for ‘non-essential’ items to save $250 billion, to alleviate the trillions of dollars in deficits predicted by the OMB. Essential is currently defined as Social Security payments, interest payments on debt, entitlement programs, Medicare benefits, and the defense budget. These taken together make up over 80% of the total government expenditures. So the freeze has to come from ‘non-essential’ departments and programs. Medicare payments to providers are not considered part of Medicare benefits, they come under the DH&HS /CMS department operating budget. So, although the benefits to the seniors will not be reduced, the payments to the providers are fair game. And therein lays our wolf.
I noted earlier that if you fail to purchase and install an EMR / CPOE, you will be penalized by a reduction in the increase in Medicare inflationary adjustment in future years. Based on the above reasons, I believe there will be little or no adjustment increase in future years. If you don’t think this will happen, look at what Congress and DH&HS had allocated for the adjustment ‘increase’ in 2010 for physician Medicare payments. DH&HS wants to apply a -21% adjustment for physician payments. Yes, that’s minus twenty-one percent. Then, to get the AMA on board with the national health initiative, the Administration and Congress was going to delay this adjustment, but now even that agreement is up in the air.
On the hospital side of the world, look at what the Medicare adjustment increases have been over the last five years. The most they have been is 2% and the average is around 1%. If you run those numbers for a typical 200-bed community hospital with a Medicare utilization percent of 50%, the one percent increase amounts to about $300,000. Hence, reduce it by a third and you will miss out on $100,000 that year. Again, and that’s assuming there is any increase at all in future years.
Lastly, let history be your guide. I have worked in the healthcare world for 35 years as a CFO, CIO and multitude of other roles. As a CFO, I saw Medicare renege on many case mix adjustments, TEFRA adjustments, and DRG adjustments,all in the name of national budget deficits and health care cost controls. At one point, they set up a Medicare Payment Advisory Committee, then disbanded it when the Committee disagreed with too many DH&HS adjustment policies. I doubt the future will be much different, in fact probably worse.
So, run the numbers again, in future years if the Medicare adjustment increase is zero – because the feds and DH&HS say we can’t afford an increase due to overall deficits and budget freezes, then reducing the zero adjustment increase by 33% will incur how many penalty dollars?
What’s a shepherd to do?
The bottom line is there is no need to “horse in” a new EMR/ CPOE regardless of what vendors say. Secondly, horsing in a system as complex and far-reaching as EMR/CPOE and while hitting the expected glitches along the way is going to cost you far more than any Medicare adjustment penalty.
My advice … take your time, do it right ,and install components that will give you the most ROI the fastest. And watch out for the wolves.
Frank L. Poggio is president of The Kelzon Group.
Accurate Patient Identification and Privacy Protection – Not an “Either/Or” Proposition
By Barry Hieb, MD
Whether you support federal government funding of HIT or not, it can’t be denied that healthcare is undergoing a major revolution as more and more clinical automation capability is being adopted. Funding of HIE projects, building toward the Nationwide Health Information Network (NHIN), will further these efforts. And clearly progress is being made, as noted in the recent KLAS report that verifies 89 active HIEs across the US. The ultimate vision of regional clinical information exchange crosses political, operational, and geographic boundaries using the NHIN’s network of health information exchanges.
However, we’re not addressing one of the most significant challenges that must be overcome for this scenario to work: the ability to accurately identify patients whose information may be scattered across a number of providers using disparate HIT applications and platforms.
The current state-of-the-art approach for patient identification centers on EMPIs that identify patients using demographic matching techniques. But industry experience indicates that EMPI matching techniques are only accurate 90 to 95% of the time, introducing a variety of potential errors in care delivery within and across provider organizations.
We know the answer to the problem — issue each patient a unique identifier that would be used to label their information across all participating provider locations. In fact, the 1996 HIPAA legislation mandated just such individual healthcare identifier. But, in 1998, Congress reversed itself on the patient identification issue based on valid concerns about the inability to protect the privacy of this data, and forbade the expenditure of federal funds on further pursuit of this essential component for accurate patient identification and data exchange. Since that time, there has been virtually no progress on this issue at the federal level, although recently a number of states have begun to pursue state-wide identifiers to support their HIE projects.
Since I left Gartner in 2008, I’ve been working with Global Patient Identifiers Inc. to build out the Voluntary Universal Healthcare Identifier (VUHID) system under the umbrella of a non-profit, private enterprise. The VUHID system is based on over 20 years of patient identification standards work done by the ASTM international E31 medical informatics group, and proposes a solutions that is both inexpensive and effective.
The VUHID system communicates with the EMPI system at the heart of each HIE. It issues identifiers upon request and maintains a directory indicating the sites that have information for each identifier. The VUHID system has been specifically designed to enhance the privacy of clinical information because it has no identifiable patient data — only the locations where each identifier is recognized.
VUHID identifiers are globally unique and are designed to support activities that the patient or others indicate need to be handled with privacy. The VUHID system represents a secure, cost-effective, currently available solution to enable error-free patient identification that extends across political and organizational boundaries.
Barry Hieb, MD is chief scientist for Global Patient Identifiers, Inc.