The Creative Destruction of Medicine

EVENTS LIKEWHEN President John Kennedy was assassinated in Dallas, Texas, in 1963 or when the airplanes crashed into the World Trade Center towers on 9/11 become index, vivid, lifelong memories. Some, like these, are shared by almost everyone who was alive for the event. Some are more personal. For me, one such memory of significance occurred on November 29, 1999. The Institute of Medicine, a division of the National Academy of Sciences, representing our most prestigious scientists and physician researchers, published the report To Err Is Human, which proclaimed that “at least 44,000 people, and perhaps as many as 98,000 people, die in hospitals each year as a result of medical errors that could have been prevented,” but which arose because “faulty systems, processes, and conditions” led people either to make mistakes or to fail to prevent them. Beyond the human toll, these errors cost between $17 billion and $29 billion.³ The report created a media frenzy that sent shock waves throughout the public. Medical errors are far more common and serious than had generally been perceived, and certainly far greater than what I had ever estimated. The errors killed more people than highway accidents and breast cancer combined, and more than six times greater than the deaths from AIDS at the time, which is particularly noteworthy since many people believed then that AIDS was by far the leading cause of death. This alarming exposé of medical errors raised awareness of how poorly and chaotically patient data is recorded and how it is notoriously inaccessible—perhaps representing the root cause of the problem. Until this point, most physicians had considered medical record keeping exceptionally mundane and inconsequential.

An alarming Washington Post editorial on this subject, “A Medical Enron,” declared that “these various errors reflect the arrogance of the medical priesthood” and “thousands will continue to die needlessly with no one held to account.”⁴ A New York Times article on the day the report was released described a possible remedy: the report called for a new Center for Patient Safety and the minimum goal of reducing medical errors by 50 percent in the following five years.⁵ There was a surprising omission in the plans, however: the computerization of medical records.

But the next Institute of Medicine (IOM) report on this subject—Crossing the Quality Chasm: A New Health System for the 21st Century, published in March 2001—zeroed in on health information technology: “Health care organizations, hospitals, and physician groups typically operate as separate ‘silos,’ acting without the benefit of complete information about the patient’s condition.” With medication errors especially prominent as a leading cause of hospital related deaths and harms, the IOM advisory committee pointed out the need for embracing automated computer systems for ordering medications, monitoring for proper dosing, averting mistaken prescriptions, and allergic reactions. It trumpeted the need for a frank change of the medical environment summarized by a couple of key sentences: “The committee calls for a nationwide infrastructure to support health care delivery, consumer health, quality measurement and improvement, public accountability, clinical and health services research and clinical education. This commitment should lead to the elimination of most handwritten clinical data by the end of the decade.”⁶

Now a decade later, there is little to show for this shout-out. There is minimal hard evidence for either the reduction of medical errors or the adoption of electronic medical records. Forty-two percent of Americans report that someone in their family has been a victim of a medical error.⁷ In a report in late 2010 in the New England Journal of Medicine, ten well-regarded North Carolina hospitals described how, having actively pursued quality improvement measures to reduce errors, they tracked 2,341 admissions between 2002 and 2007 to see whether there was evidence of improvement during that time. The data showed none. There were more than 60 patient injuries per 1,000 patient days both at the beginning and at the end of the study. Roughly two-thirds were deemed preventable. Overall, some form of harm to patients accompanied 25 percent of hospital admissions.⁸ In 2011, the journal Health Affairs published a special issue, “Still Crossing the Chasm of Quality,” with multiple articles on the persistent and serious problems related to medical errors. One group calculated the yearly cost of the errors in the United States was more than $17 billion, another found that the actual number of errors could be ten times greater than the estimates, and an expert in this topic from Johns Hopkins, Peter Pronovost, wrote, “For the past decade, health care quality has largely sought quick fixes and run from science; the results are evident.”⁹

Other reports have been grimmer. In 2002, Dr. Barbara Starfield published her perspective in the Journal of the American Medical Association. She claimed hospitals killed roughly 225,000 patients per year: 113,000 by medication errors and 80,000 by hospital-acquired (nosocomial) infections, many attributable to unnecessary surgeries and procedures.¹⁰ Later, in 2007 in the British Medical Journal, the Netherlands reported that approximately 30,000 patients each year suffer avoidable harm in the hospital, leading to about 1,700 deaths.¹¹ These and many other publications reinforced the notion that the IOM was not hyping the problem. What’s more, these reports discussed only hospitals; many serious errors occur during office or clinic visits or while phoning in prescriptions or filling them at pharmacies.

The ultimate solution to this enormous problem is electronic record keeping. Of more than 3,000 American hospitals surveyed in 2009, only 1.5 percent had fully electronic health records and health information technology (HIT) systems, and these were largely confined to large teaching hospitals in big cities.¹² Only 4 percent of clinics and physician’s offices were using fully electronic medical records.¹³ The situation is not so bad elsewhere; in fact, the United States is one of the slowest countries to adopt HIT in the developed world. Countries like Denmark have fully integrated hospital and clinic HIT, connecting virtually every citizen, physician, clinic, and hospital. Even in India, the Apollo hospital chain is a pacesetter for adoption and streamlining advanced electronic health records.¹⁴ Among countries ranked for health information technology capabilities, the United States ranked eighth after New Zealand, Australia, United Kingdom, Italy, Netherlands, Sweden, and Germany.¹⁵

An alternative or complementary strategy to EHRs is the electronic personal health record (PHR), in which all of the data is coalesced and made available to the patient via a flash drive or a login to cloud-based information or both.¹⁷ The advantage of the PHR is that the patient is the mechanism for collecting each piece of data as he or she travels from doctor to doctor, test to test, and one health care system to another. It sounds attractive and would well serve the individual who is on vacation in a remote place and has an unanticipated medical problem. Getting the data to populate the PHR, as I will discuss, represents a formidable challenge.

Truly powerful heath information technology (HIT) is more than just records, however.¹⁸ For example, it might alert a physician writing a prescription for a medication allergy. Or it might provide decision support to inform the physician that a patient needs a vaccination or that an imaging test that was just ordered is not supported by the latest evidence from the medical literature. HIT also includes radio-frequency identification tags or bar codes to track medications given in the hospital setting.¹⁹ Without these capabilities, EHRs alone would not be enough to reduce medication errors. The aggregate database of HIT would allow for patients to check their own laboratory or test results by logging into the system. And far beyond this, HIT could be used for population surveillance, such as detecting the earliest signs of an impending epidemic or the adverse effects of a newly released prescription medication. Aggregating HIT systems throughout the country—and in developed countries, or even worldwide—would make it very effective indeed.

In 2006, in the Annals of Internal Medicine, a group of researchers from Los Angeles combed 257 studies to see just how effective electronic patient data could be in avoiding those 100,000—or 225,000, or more—deaths by medical care. HIT was shown to improve compliance to guidelines, decrease medication errors, and improve monitoring for such preventive tactics as flu vaccination. However, they concluded “a disproportionate amount of literature on the benefits that have been realized from a small set of early-adopter institutions that implemented internally developed health information technological systems.”²⁰ In fact, one-fourth of the studies came from just four academic teaching hospitals, and only 9 of the 257 studies used commercially developed EHRs. In 2009, a group of University of Minnesota researchers followed four years of Medicare data to determine whether EHRs had any improvement in patient safety, and they found minimal evidence of support—such as two infections fewer each a year at an average hospital. They concluded, “Health IT’s true value remains uncertain.”²¹

In late 2010, the evidence that EHRs reduce errors was summarized as follows: “EHR users overwhelmingly report improvement in the quality of care they provide. On the other hand, despite experts’ optimism, there is currently no evidence that the use of EHRs reduces diagnostic errors.”²² Yet another study in 2011 on outpatient care concluded there was “no consistent association between better quality of care and electronic medical records.”²³ Indeed, one of the problems is connecting the dots. Just collecting data, without processing it into actionable information and providing vital feedback to physicians, nurses, and patients, is not enough. Automating a broken process won’t provide the fix. As Steve Lohr asserted in a recent review of the topic in the New York Times, “what is also beyond doubt is that the promise of digital records will be unfulfilled if doctors refuse to adopt them, because they regard the technology as cumbersome, time-consuming and possibly dangerous.”²⁴

Another important dimension of the problem, which will be approached by enhanced digitization of humans in the future, was articulated by my colleague Paul Yock, a cardiologist at Stanford, as quoted in the Economist: “The dirty little secret about medicine is that we physicians make decisions all the time based on woefully incomplete information.”²⁵ This brings us back to the fundamental tenet of Chapter 2—incomplete information. Comprehensive digital medical records for each individual, it is hoped, can provide critical scaffolding to alleviate this deficiency.

EHRs may not be saving large numbers of lives yet, but they do already provide significant improvements to medicine. Think of the illegible handwriting of doctors, which, many have joked, is a requirement for graduating from medical school. Eradicating that alone would be a worthwhile digital objective, but when you add real-time abilities to capture, store, exchange, access, and analyze vital medical information—theoretically, from anywhere in the world—it adds up to be irresistible. We can give patients the record of their visit immediately, with recommendations and medications spelled out. HITs can, as I outlined above, provide guidance or evidence for medical decision making, or even make the correct diagnosis in the first place.²⁶ They can certainly reduce costs, improve productivity and quality of care, and reduce malpractice liability. They can also eliminate the unnecessary duplication of blood tests, X-rays, and other diagnostics; indeed, at least 10 percent of such tests are estimated to be redundant, which amounts to billions of dollars of waste each year.²⁷

The American think tank the RAND Corporation projected that there could ultimately be $77 billion a year saved from efficiency if 90 percent of doctors and hospitals adopted HIT.²⁸ With the anticipated reduction of medication errors and other harms, RAND forecasted the savings figure could easily double. A follow-up report from the IOM on medication errors provided some quantitative context of the problem: “The use of medications is ubiquitous. In any given week, more than four of five U.S. adults take at least one medication and almost a third take at least five different medications.” According to the IOM, there are 1.5 million people injured by medication errors each year, and hospital related medication errors alone result in a cost of over $3.5 billion per year.²⁹ The Society of Actuaries estimated the overall cost of medical errors, for which medications represent just one component. Their study found that there were 6.3 million medical injuries in the United States in 2008 and that the average cost per error was $13,000.³⁰ More than ten million excess days of missed work and short-term disability were part of the calculated $19.5 billion nationwide cost for 2008. The discrepancy between the financial estimates of the RAND Corporation and the Society of Actuaries is noted, but both groups come up with very big numbers for economic impact.

Of course, this won’t all necessarily go without a hitch—the potential problems are neatly captured by a terrific parody website of EHRs known as Extormity. Extormity’s tagline is “expensive, exasperating, exhausting.” Here are some notable quotes on the Extormity site: “Our slow and painful change process significantly interrupts patient volumes and revenues, and this cumbersome transformation can only be appreciated in hindsight and with the aid of prescription medication,” and “Extormity service and support is available on an hourly fee basis, with a minimum initial commitment of 225 hours at an hourly rate generated by a confusing algorithm.” Under the “Perpetual Investment” tab:

Operating the Extormity Bundle requires a phalanx of servers, which of course need to be replicated for redundancy. Fortunately, Extormity acts as a value-added reseller of these servers, which we pre-load with operating software. This allows us to mark-up the cost of the servers and charge for server configuration. In addition, the server software carries with it steep annual license fees. In addition to investments in servers and software, these servers generally have additional physical, communication, power and environmental requirements. Planning for this additional infrastructure can be provided by the Extormity Strategic Consulting unit, with implementation provided by the Extormity Solutions and Services Business Unit. These Extormity business units operate in silos, ensuring that you receive and pay for duplicated services. Of course, the Extormity EMR Software Suite must be integrated with other systems. Extormity software development engineers create custom, one-of-a-kind interfaces for each and every system or piece of equipment. In the event that we have already built an interface with a particular vendor, we employ a ‘reinvent the wheel’ framework so that each of our customers receives a solution customized for their special needs, no matter how similar to work we have already performed for another client.³¹

The other model is Kaiser Permanente, a very large health system comprising nine million individuals, 14,000 physicians, 431 medical offices, and thirty-six hospitals in nine states and the District of Columbia. Kaiser spent $4 billion in 2003 to develop the largest civilian installation of EHRs. Beyond providing integrated EHRs for all of its hospital and office visits, it has used the aggregated data to propel quality of care initiatives, such as the assured use of a preventive medication for specific conditions or providing feedback on how practitioners are performing against each other for productivity or patient satisfaction metrics. The Kaiser HIT system alerts its doctors and nurses on screening practices, reminds them to schedule follow-up appointments when they’re necessary, provides some decision support and relevant guideline information for many conditions and diagnoses, and can monitor side effects of medications. Kaiser’s patient database served as one of the first means to recognize the higher than expected incidence of heart attacks that occurred as a result of Vioxx. One of the most impressive features of its HIT is to enable more than 3 million of its patients to access their data and fully communicate with their physicians via secure email, leading to a 26 percent decrease in office visits over the past four years in a recent study. On any given day, more than 100,000 patients access their data. All Kaiser patients can tap into their laboratory data, which is an efficient way to transmit the information, because all too often physicians in practice do not have the time to follow through in getting the results back to patients in a timely fashion or at all. But like the VHA, it does not communicate well with non-Kaiser facilities or heath care practitioners; it is an effective but closed loop. There are efforts to fix that; in San Diego, for example, the VHA and Kaiser are exploring methods to combine their databases and make information transferable between their systems.³³

In contradistinction to the virtues of EHRs and these successful, albeit rarified models, there are many limitations and challenges. The absence of interoperability is a huge problem, and because physicians and hospitals are too often not connected to start with, and frequently at odds, collaboration in developing EHRs and HIT is difficult. The most serious concerns, however, lie at the level of the individual physician-patient encounter.³⁴ Rather than looking the patient in the eye, the physician is looking at a screen and typing in the data. Most doctors are uncomfortable typing and are slow and prone to making errors. Instead of the customary pre-EHR era, in which the doctor freely narrated particular thoughts about the patient’s symptoms, condition, or treatment, everything is point and click. Dr. Danielle Ofri recently wrote about this challenge, including 1,000-character limits to the typed assessments of patients and an untoward influence on how doctors think: “The system encourages fragmented documentation, with different aspects of a patient’s condition secreted in unconnected fields, so it’s much harder to keep a global synthesis of the patient in mind.”³⁵

So in the mind of the physician, a lot of insight is missing as a result of less direct communication with the patient and less ability to freely express one’s thoughts. Symmetrically, the patient feels less direct contact and is often disturbed by the doctor’s pecking on the keyboard and looking at a screen. The sense of not being heard or understood is often prompted by this distracted, electronically fettered encounter. Such feelings are exacerbated with the brief time that the doctor and patient come together, typically less than ten minutes, and frequently this occurs after more than an hour of waiting.

One result of this bilateral dissatisfaction has been hiring scribes to enter the information in an EHR as the doctor and patient have a real visit. The reaction of using EHRs in one health system in Portland, Oregon, is telling. The 140 physicians, nurses, and physician assistants there faced a crisis in morale because they were spending three hours after a shift working on EHR data entry. They hired three scribes to help them and radically improved the situation. The president of the group said, “It allows our doctors to see more patients, physician satisfaction is up a lot, and patients are much more satisfied because the doctors can spend time on bedside interaction. Now, if you don’t have a scribe it feels like you’re showing up without a stethoscope.”³⁶ The fast pace of seeing patients in the emergency department, along with the requirement to enter EHR data, has also made the use of scribes there particularly popular.

The scribes are mostly young people, often premed or prenursing students, who get paid only $8 to $10 per hour.³⁷ There are already three big companies that train and coordinate the hiring of scribes: Scribe America in California, which has fifty programs; PhysAssist in Texas with forty; and Emergency Medicine Scribe Systems in California with thirty. So much for bringing costs down and efficiency up—EHRs have managed to spawn a new profession.

This compensatory mechanism is just one small feature that introduces the extraordinary expense of building EHRs and highly functional HIT systems. For hospitals, it is estimated to cost at least $100,000 per bed, and this led to Jonathan Bush, CEO at Athena Health, an HIT company, and cousin to ex-president George W. Bush, to declare, “Hospitals will enter a financial crisis on the scale of the subprime mortgage crisis.”³⁸ The cost estimates for converting outpatient practices to EHR vary considerably, and part of this relates to the legacy paper records. There is no magic on-off switch to digital records; to do this right it involves digitizing records of each patient going back many years. With these high cost considerations, many groups have seen little to no incentive to go digital.

That changed, in 2009, when the Obama administration came up with HITECH, the Health Information Technology for Economic and Clinical Health Act. This was part of the economic stimulus bill and allocated potentially over $100,000 for each doctor ($44,000 through Medicare and $63,750 through Medicaid) and between $2 million to $10 million for each hospital to become “meaningful users” of EHRs.³⁹ This program represents an investment of more than $36 billion over ten years to accelerate adoption of EHRs and HIT.⁴⁰ As David Blumenthal, the initial leader of HITECH put it, “this funding will provide support to achieve liftoff for the creation of a nationwide system of EHRs.”⁴¹ Note the national integrated characterization that is intended, even though the bare bones of just getting offices and hospitals off the ground is where we sit today. Although President Obama said in a speech in January 2009, before taking office, that “we will make the immediate investments necessary to ensure that within five years all of America’s medical records are computerized,” the goal will likely prove elusive. Only 10 percent of hospitals and 20 percent of physicians were using these systems in 2011. Meaningful use has now been defined by HITECH; there are very specific, conservative criteria to fulfill, such as having more than 50 percent of patients with vital signs data, smoking status, and demographic data (e.g., sex, date of birth, ancestry) and more than 80 percent with a complete and accurate medication and allergy list. Furthermore, more than 50 percent of patients must be able to get a copy of their EHR within three business days.⁴² The criteria from the HIT side were even easier to meet (for example, performing at least one test of data submission to a public health agency or demonstrating capability of electronically sharing some information among providers).

Meaningful use looks easy to me, but many health systems called the HITECH requirements unrealistic. Notably, Intermountain Healthcare based in Salt Lake City, Utah, stated that it “could not meet 36 of the 48 meaningful use requirements.” The reason this is particularly important is that Intermountain, like the VHA and Kaiser, is considered a model system of HIT and has been frequently praised by President Obama. Dr. Thomas Lee, president of Partners HealthCare physician network, which includes some of the prestigious Harvard teaching hospitals in Boston, concurred with Intermountain’s position.⁴³ Nevertheless, they need to do it; in addition to financial support, the law also includes penalties, including decreased reimbursement from the government if meaningful use criteria are not met by 2015.

The initial period after installation of EHRs and HIT systems often sees an increase in errors. In 2010, the FDA received more than two hundred fifty reports of HIT bugs, flaws, or crashes, with multiple deaths and injuries as a result.⁴⁴ Examples included incorrect data on allergies and blood pressure, and it is widely thought that EHR errors have been grossly underreported. That assertion is substantiated by many isolated reports at a variety of health systems. For example, at Geisinger Health System in Danville, Pennsylvania, $35 million was spent in 2005 to purchase and install the Epic EHR system, but incompatibility between the pharmacy database and the Epic system led to several major medication errors per week.⁴⁵ Children’s National Medical Center in Washington, DC, had an eightfold increase in dosage errors for high-risk medications after spending $30 million to purchase the Cerner EHR system; it switched back to paper records until the glitches could be worked out.⁴⁶ At the University of Pittsburgh Children’s Hospital there was a doubling of patient deaths in the five-month period following installation of a computerized order-entry system, which was mainly related to marked delays in administration of the prescribed drugs and attributable to the then-new Cerner software.⁴⁷ In 2011, a group from the RAND Corporation published a study in the American Journal of Managed Care on the process of health systems going electronic. The researchers told the Wall Street Journal that “trying to introduce an EHR system to an already complex health-care workplace causes a myriad of unintended consequences in terms of workflow and communication.”⁴⁸

On the other hand, integrating EHRs and a bar code of all medications has recently been shown to reduce medication errors substantially. At Harvard’s Brigham and Women’s Hospital, with over 14,000 bar-coded medications and more than 3,000 physician orders transcribed, there was a 41 percent reduction in error rate and a 51 percent reduction of adverse drug events compared to traditional, non–bar coding of medication prescription and administration to patients. Interestingly, 39 percent of the errors were at the level of physician ordering, and 12 percent were from transcribing the physician orders. Dispensing by the pharmacist accounted for 11 percent of the errors, and nurse administration of the drug was responsible for the remaining 38 percent.⁴⁹ The encouraging findings from this study suggest that when an EHR is integrated with other electronic tagging and surveillance, a marked improvement in quality may be achievable.

Cloud computing could be ideal for HIT with relatively inexpensive, seemingly limitless places to store, process, and maintain patient and medical information. Nevertheless, the reluctance to embrace EHRs is matched by a reluctance to place the data in the cloud. A 2011 Economist article on “Heads in the Cloud” highlighted the “reflexive conservatism and technophobia of medical folk” and that the medical industry can be seen as anti-innovative. Still, there hasn’t been total rejection of the idea. One big chain of rehabilitation centers has incorporated both cloud computing and connected mobile devices to transmit all health data. Some U.S. health systems are embracing the use of private cloud computing to lower costs of HIT. Rather than spending $80 million on a new data center, the University of Pittsburgh health system, representing twenty hospitals, 50,000 employees, 4000 physicians, and over $8 billion annual revenue, has moved most of its computing operations to a private cloud. ⁵⁰

The basis for comparison is paper records, which are much less apt to be lost to a large-scale breach of security or an electronic “phishing” scheme.⁵² But a single individual’s medical record is relatively easy to access, and there is no way to tell if the record has been read or copied. An EHR’s metadata—the data of the data, in the form of tags or descriptors for different data elements—serves as a permanent electronic footprint that can track when and where a record has been accessed or modified.⁵³ It can also enable partial transmittal of the information in an EHR. For example, if you are brought to an emergency room, should the hospital have complete access to your medical information? If you have a history of mental illness, should that data be accessible? Are the data for certain conditions like diabetes fair game but other information on previous cancer to be withheld? If other physicians are consulted in your care, should they have complete or partial access to your EHR? If you go for a second opinion, do you want that physician to know the first opinion?

Besides customization of the data flow, the tagged data elements and metadata properties are essential to building optimal privacy and security. The patient information data elements are encrypted or unreadable, when stored or transmitted, and ideally not ever stored on the same computer system as the one that holds the encryption key. The metadata and the patient data itself are inseparable and are typically protected by a digital signature. To maximize security, accessing the data can require two-factor authentication, with not only a password but also a smartcard, physical credential, or biometrics, such as a fingerprint. Predetermined customized authorization by the patient and extensive audit processes are further safeguards that are part and parcel of secure HIT systems.⁵⁴

Ironically, although EHRs are felt to diminish medical errors and reduce the liability of malpractice, the metadata feature can also be seen to facilitate malpractice claims.⁵⁵ The recording of all electronic transmissions from the input of medication or other orders to time stamps of activity creates discoverable evidence and can be used to establish a doctor’s culpability. If the EHR was modified at a later and inappropriate time after the treatment was rendered, this can be tagged and used against the physician. Such examples have already been tested in the courts and have demonstrated the game-changing potential of metadata in substantiating physician and hospital malpractice.⁵⁶ The other looming and intriguing feature of malpractice and EHRs as this field evolves is the anticipation that a hospital or doctor who has not adopted digital records or HIT will be liable for deviating from the standard of care.

Although EHRs will give patients easy access to their full medical records, the legal right to access has only been assured in the United States since 1996, with the passage of the Health Insurance Portability and Accountability Act (HIPAA).⁵⁷ Nevertheless, very few patients have actually seen their records. The obstacles to access include a charge for copying the records, the ability to review records only when a doctor is present, and inordinate delays following patient requests. The long-standing lack of access to the patient’s medical information reflects the traditional paternalistic views of medical professionals; moreover, until recent years, when a patient requested a copy of his or her medical records, the first thing a physician would think of was a lawsuit. Under HIPAA, the only legal exception to the right for full access is in the case of mental illness, if a psychiatrist believes that access to the medical record could be detrimental to the patient’s condition.

Without question, there are marked potential advantages to full patient access to records. Not only are patients more engaged in their care with enhanced understanding of their condition, but such involvement may foster improved compliance, self-care, patient-doctor communication, and prevention of medical errors by having another pair of eyes (in this case the most important person of all) involved. As Morris Collen, a Kaiser Permanente physician—now in his nineties—and the father of HIT, with pioneering efforts dating back to the 1960s, points out, “the patient has lived with his medical problem and often knows it better than the doctor.”⁵⁸

But there are also several potential problems. One example is interpretation of medical jargon, such as the frequent use of “SOB,” for shortness of breath, which could be taken by the patient to mean something altogether different in reading the statement “the patient appears SOB.” And then there’s NERD, which actually means “no evidence of recurrent disease.” Or patients may take offense by descriptors of their appearance with common terms such as “obese” or the insinuation that the patient is “somatisizing” or is a “hypochondriac.” Review of the notes may actually engender more confusion and anxiety for the patient and increased time for the physician to explain the note contents and clear up any misunderstandings. The impact of full access to medical records for patients is being prospectively studied by an initiative called “Open Notes,” involving 25,000 patients in four health systems with well-developed EHR and HIT systems, funded by the Robert Wood Johnson Foundation. In the study, over one hundred primary care physicians prompt their patients to review their records through secure electronic patient portals. ⁵⁹

The potential drawbacks of EHRs—usability, cost, security, privacy, medico-legal liability—are also counterbalanced by the big picture, or macro effect, of ultimately creating a national health network that can be used for several meaningful purposes. For a new medication or device that is being released for commercial use, a coordinated national system using de-identified patient data can detect and quantify unanticipated or low-frequency side effects. Already a Swedish national registry has detected blood-clotting problems with drug-coated stents and problems with liver toxicity after the release of a new drug intended to prevent clotting.⁶⁰ In both cases the frequency of the side effect was less than 1 percent, but the large sample size of individuals in the population database made the detection possible.

The nationwide HIT would be ideal for the earliest possible detection of an impending epidemic of the flu or another pathogen, or the emergence of antibiotic resistance to a bacterial strain.⁶¹ Such public heath electronic monitoring transforms the population to a clinical research platform. Instead of the random sampling, voluntary reporting, and incomplete methods that are currently used by government agencies like the Centers for Disease Control and the FDA, there is potential for comprehensive assessment and the capability of studying the effect of different treatments on large segments of the population.

Patient-centered health records provide a potentially powerful tool for improving patient care as well as their ability to control their own health information. One study has shown significantly better use of preventive health services like flu shots, mammography, and colonoscopy.⁶⁴ They could have other powerful effects as well. For one, they might provide an escape from the problem posed by the myriad different EHR systems in use at different hospitals and practices. A new U.S. government program, known as Blue Button, enables every individual in the Medicare and VHA databases to download his or her data for personal records (although the quality and utility of the Blue Button–derived PHR remains unclear).⁶⁵

Nevertheless, the embrace of PHRs is notably weak and in stark contrast to the enthusiasm and clear rationale for their potential benefit. While the Markle Foundation and American Medical Association survey indicated that only 44 percent of physicians are willing to use the patient’s PHR as part of their medical effort, it may be even more surprising that only 2.7 percent of individuals have established a PHR. To date there have been only three major PHR products developed in the country—Microsoft’s HealthVault, Google Health, and WebMD’s Health Manager—and all are struggling, even though they are free applications. In mid-2011, Google Health was shelved. Google Health was good for printing out or emailing a complete or partial record to share, but it didn’t have tools to evaluate one’s health. Health Vault has self-evaluation tools, and you can invite someone to view all or part of your records, but printing out parts of the PHR can be challenging. The WebMD product has the health evaluation features and is fine for printing, but it doesn’t allow exporting the file in either of the two main industry formats.⁶⁶

The real problem with the PHRs of today is not the presence or absence of particular features. That can be readily fixed. The central issue is populating all of the data fields, which is more than an arduous task. It requires dozens of hours of effort only after requesting and successfully retrieving all of one’s medical records from a multitude of sources and providers. When we offered Health Vault as part of our research program on consumer genomics, out of more than 3,000 participants in the study, only 30 collated their data within a PHR. This problem will likely only get resolved when the importation process is automated, such as through a mechanism derived from hospital and physician EHRs. Since part of the HITECH definition of “meaningful use” includes transfer of data to patients, there has been some groundwork and financial incentive toward this worthy objective. A large survey conducted by United Health in 2011 suggested that 77 percent of individuals want a PHR to track medications, test results, and their medical data, and are not worried about cost or privacy issues. So it’s chiefly a matter of getting a system that is user-friendly, interoperable, and hassle free.⁶⁷ With so many groups and institutions working on this goal, the prospects are encouraging.

FIGURE 7.1: The size of various files and pieces of data that would be a part of any electronic and personal health record of the future.

There are concerns about incorporating genomic data, since the GINA bill (discussed in Chapter 5) does not protect abuse of these data for life insurance or long-term disability coverage.⁶⁹ However, using tagged data elements should provide the means to keep everything together but partitioned to control access to DNA sequence and other “-omic” data as required. ⁷⁰ This is for the protection not only of the individual but of any relatives. It is worth pointing out that no matter how granular and comprehensive the EHR and PHR are developed for an individual, this does not provide a guarantee of avoidance of medical errors or infallibility of the HIT system. But having all the essential and relevant health data from digitizing a baby, child, or adult, and having it be fully accessible to the individual or the parents, will undoubtedly foster better care—at both the individual and the population levels.

While some may consider the topic of electronic medical records prosaic, it should now be abundantly clear that their ultimate adoption and full interoperability will prove fundamental to the future of medicine. Only via full electronic convergence can all the tools of digital medicine be in sync and immediately useful. With the torrent of individualized data flow that is coming from whole-genome sequencing, remote physiologic monitoring, and medical imaging, electronic information storage and processing will become more essential than even envisioned today.

UNBUILDING A TOWER OF BABEL

MODEL EHR HEALTH SYSTEMS

CHALLENGES WITH EHRS

PRIVACY AND SECURITY OF THE DATA

OPEN NOTES TO PATIENTS

ESCAPING THE SILO WITH PERSONAL HEALTH RECORDS

THE EHRS AND PHRS OF THE FUTURE