What event could bring together four major leaders of health care within the Administration — Surgeon General Regina Benjamin, Secretary Kathleen Sebelius of the Department of Health and Human Services, National Coordinator David Blumenthal, and Donald Berwick, who has just been sworn in yesterday over the objections of Republican Congressmen as Administrator of the Center for Medicare & Medicaid — for a press conference?
It was the release of the final “meaningful use” criteria for the adoption of electronic health records by doctors’ offices and hospitals. This initiative, part of the American Recovery and Reinvestment Act (ARRA), offers several thousand dollars to each health care provider as a motivation for adopting electronic records.
The catch is that they can’t just install the electronic system, but have to demonstrate that they’re using it in ways that will improve patient care, reduce costs, allow different providers to securely share data, and provide data to government researchers in order to find better ways to care for patients. That’s what “meaningful use” means.
The recovery act bets a lot of money on the ability of doctors and hospitals to convert to electronic records–including small and rural providers whose offices still look a lot like a Norman Rockwell painting. The pressure on everyone is high, and the rewards are substantial.
The press conference offered hardly any details about the final rules, but you can find a summary on the web site of the New England Journal of Medicine. There’s now also a meaningful use microsite at the HHS’s Health IT site. (You can even get the whole 864 pages as a PDF from the Federal Register or view it in the embedded link further down this blog.) Experts in the health field are going to put in late nights figuring out the impacts of the final standards–which are considerably relaxed from the proposed standards–on health care providers, and O’Reilly Radar will run a longer article on meaningful use later this week. Right now I’ll focus on areas of interest to technologists: what elements of digital records will be required to make use meaningful.
Getting data out of electronic systems
To be useful, data has to be fluid. In the case of patient data, it must be:
Available to the patients themselves, as described in a heart-rending story by a guest in today’s press conference.
Available to pharmacies for electonic prescriptions (Computerized Provider Order Entry, or CPOE).
Available to government agencies responsible for ensuring public health and safety, regulating institutions, and supporting public research into better cures.
All these requirements call for the collection, storage, and sharing of different combinations of data. They obviously presents security issues that I won’t cover here–let’s just start with the basic problem of getting data out of the electronic systems in which it is stored.
To say that current commercial health record systems are closed is an understatement. Doctors have been known to buy the same systems from the same manufacturers as the hospitals at which they practice, and still have trouble exchanging certain kinds of data. To exchange data among entities that have no formal relationship, they have to hook up with a health information exchange (HIE) and sometimes pay large sums to do data transfers.
Part of the barrier lies in the security and privacy requirements I mentioned before, but much of it lies in the incompatible formats used by manufacturers of electronic systems–even when they claim to follow the same standards.
A few providers now have APIs. Google Health and Microsoft HealthVault, which are targeted at patients, have APIs, and a few of the physician systems (such as Practice Fusion and Eclipsys, to whom I spoke at conference four months ago) have joined the movement. Some electronic systems have been enhanced to release data directly to the Google or Microsoft systems, at the patient request. But most patients still prefer to leave their records with their doctors. And each API is different.
HHS is doing its best to make sure electronic record systems support its goals. In parallel with the meaningful use criteria, it has been developing a system for independent organizations to certify the electronic records systems. Today, HHS also released an “Initial Set of Standards, Implementation Specifications, and Certification Criteria for Electronic Health Record Technology” that lay out their expectations. Significantly, open source developers are explicitly encouraged to submit their systems.
A number of the meaningful use criteria deal with giving patients their own information in electronic form. As the previous section explained, this is not a trivial requirement. For stage 1, a criterion of success is that the patient gets the data within three business days. Most of us will be out of bed and walking by then.
One could well ask how a patient can read the electronic record even when it’s provided–does the patient need the same $15,000-per-seat system as the hospital to get access to it? Or is the data in an “open” format, so that the patient must simply be comfortable with an XML parser?
The criteria speak repeatedly of “structured formats,” loosely described with examples such as “the electronic transmission of orders entered using computerized provider order entry (CPOE) and the electronic transmission of diagnostic test results” (p. 35 of the final rules). People in the computer field, particularly those who are used to ecommerce systems, may assume that data will be stored in a database following canonical criteria, but in the health care field chaos tends to rein.
It’s easy for an electronic format to store data in a state just as unstructured as a hand-scrawled progress note. If data is scanned from paper or just entered casually into the computer, a free-from text section may contain critical information the patient’s condition and treatment.
When hospitals do adopt standards, they are are often unsuitable for the kinds of processing required to answer such basic questions as, “How many of our patients with diabetes smoke tobacco?” The standards have evolved over decades, and their ambiguity is compounded by inconsistent data entry by thousands of people over many years.
To take one example, most hospitals store diagnostic information in the International Classification of Diseases, Ninth Revision, ICD-9. This is universally understood to be an inadequate format–lacking in detail, ambiguous, hard to parse–but it will take years to adopt1 the much superior ICD-10 and migrate existing records.
Similar evolution and mismatches take place in every aspect of patient records: medications, allergies, etc. I don’t want to present a hopeless scenario, because some data is actually in good, helpful formats, such as lab results. But the most established formats are the codes used by physicians to bill insurance companies, because those companies demand the codes. We’ll see the results of this standardization a bit later.
Electronic systems can facilitate immense improvements in patient care. A simple procedure such as giving each patient a bar-code and checking that code before administering medicine can prevent medicines from being given to the wrong patient.
The meaningful use criteria mention several quality and decision support issues, such as medication reconciliation, which ensures that a patient continues getting the meds she needs after going from the hospital to the rehab facility or some other transition. Similarly, clinicians will have to start reporting data to the government on three core quality measures in 2011: blood-pressure level, tobacco status, and adult weight screening.
But this is sad case of garbage in, garbage out. For instance, doctors often shoehorn their patients’ conditions into bizarre codes in order to get payment from the insurance companies. This is not fraud, but simply playing the game the way the insurers want it. But if a researcher tries to draw clinically relevant conclusions from these diagnoses, the results are ludicrous.
One researcher in the health care field told me that researchers never use the standard formats found in electronic record systems for research, because they don’t fit. Each researcher invents his or her own ad hoc format, and doesn’t spend much time on the task because they’re in a rush to process the data. This is a lot of reinventing the wheel, and makes data sharing more difficult.
I think data reporting is great, but government departments have to establish formats and manufacturers have to honor them. This will no doubt be part of the certification process to designate health record systems as suitable for meaningful use.
Making a dream come true
In her introduction to today’s press conference, Secretary Sebelius said that doctors who implement electronic records tell her they “can’t imagine practicing health any other way.” Her words were echoed by ONC chair Blumenthal. What I’ve heard in the field is a good deal different.
Many doctors have spent enormous amounts of money installing electronic systems and converting parper records, only to find no increase in efficiency. In fact, studies have questioned the value of using existing systems. No doubt one has to expect a loss of efficiency during the conversion and training period, and most of the blame for poor results lie in doctors’ workflows rather than the electronic products. As administrator Berwick said today, moving From paper to electronic records requires a “new culture.” In any case, satisfaction is by no means guaranteed.
Meaningful use is the answer to that problem. It is supposed to make doctors conscious of the steps they need to take to get the benefits promised by electronic records. HHS has devoted considerable effort to choosing criteria that can really make a difference, and to tweaking the implementation schedule to strike a balance between the need to give providers a push and the need to recognize that change is hard. On the side of the technologies, the issues described in this blog will have to be solved to reach our goal.
Later this week a longer article will appear on Radar about the implications of the final meaningful use rules and their effects in clinical environments.