Advances in the delivery of healthcare have allowed Americans to live longer, healthier lives than ever before, but costs are out of control and medical errors are dangerously common. Such is the universal assessment of healthcare in the United States and it is widely acknowledged that healthcare information technology (health IT or HIT) can help. What’s largely missing from the literature in the healthcare field is precise and actionable advice to IT staff and the clinicians who work with them to make the health IT transformation a reality. This book was written to start filling the void.
About the same number of people die each year from medical errors as from automobile accidents. Heart disease and cancer kill the most people in the United States, more than 500,000 each year. But stroke and lung diseases are each responsible for about 100,000 deaths each year—and scandalously, so are medical errors. Medical errors are notoriously difficult to track, given our litigious society, so we really do not know how many deaths that statisticians attribute to cancer or heart disease were also related to medical errors. But given the high likelihood that errors are implicated in some of these deaths, it is possible that medical errors could be the third leading cause of death in the United States.
In 2000, the release of a report titled “To Err is Human” by The National Academies (the country’s leading research institute in medicine) highlighted the astonishing rate of medical errors. It was a wake-up call to the healthcare industry, but the problem is still little known among the public, and in the absence of organizational change and technical adoption, little has been done to fix the problem.
Cost now dominates the news about healthcare. Some estimates put healthcare in the United States at one sixth of the total national economy. Healthcare insurance costs that go up sometimes 15% or 20% a year are threatening to bankrupt many local governments and forcing them to cut back services in a poor economy. Other wealthy nations spend much less on healthcare, but still have similar or better levels of healthcare quality.
HIT, or more colloquially “software for clinicians,” promises to address these two fundamental problems: to lower healthcare costs and improve patient safety.
The Veterans Affairs (VA) hospitals in the United States offer the most substantial example of systemic improvements in quality using health IT. Since the 1970s, the VA has gone from a system with a reputation as a low-quality provider to a system widely regarded as the safest and most effective healthcare delivery system in the world. VA hospitals almost obsessively measure the quality of the healthcare they deliver, and they have the numbers to back up the assertion that they are tops. The quality of the VA system, and its focus on health IT to deliver quality, is documented in the book The Best Care Anywhere by Phillip Longman. Rather than quote all of the quality statistics in that and many other books, we will relate two simple cases that show the power of leveraged health IT systems at the VA.
In 2004, the drug company Merck voluntarily withdrew Vioxx from the market. Vioxx had been used to treat chronic pain, but it had become clear, over time, that Vioxx had a dangerous side effect: fatal heart attacks. Evidence also emerged that by 2000, Merck had evidence that Vioxx was dangerous. The fact that Vioxx was approved by the Food and Drug Administration (FDA), and that it was used so long after it was known to be dangerous, has been the subject of intense scrutiny.[2]
But years before the healthcare profession as a whole was aware of the dangers of Vioxx, the VA discovered on its own that it was a dangerous medication. Data from the VA’s electronic healthcare record, VistA, had alerted the VA that something was amiss with Vioxx. The VA took steps to ensure that Vioxx was prescribed only with careful monitoring and only in special circumstances, a drug of last resort. By doing so, the VA saved thousands of lives.
The second case is the level of integration experienced by VA hospitals. If a veteran receives treatment in one VA hospital for a decade and then moves to another hospital, even another state, he can expect a decade’s worth of VA records to be available at the new hospital on his first visit. The VA has achieved near-complete health data liquidity for its covered veterans. In comparison, most other healthcare systems typically still use fax machines to exchange health information.
This is the stage that has been set for health IT. Medical errors are too common, costs are out of control, and effective deployment of computerized records and workflow can dramatically reduce these errors and lower costs. This book will discuss preventable medical errors in detail, and show how many different health IT functions, from health data exchange to different types of reporting, can help to address healthcare quality and reduce medical errors.
Most of those who are deeply involved in healthcare IT have chosen this field as a mission or vocation, rather than merely a career. Many health IT professionals have historically taken substantial pay cuts compared with IT professionals in other areas (although this is improving now). Many of us work in this industry because we lost a loved one to a simple medical error, or some other failure of the healthcare delivery system. For many of us, this is our life’s passion. To us, “reducing the costs and improving the quality of healthcare” is a dry and frail description of our ambition for health IT. To paraphrase Steve Jobs, we want to make a dent in the human condition.
Before we can talk about what that next stage will be like, we should acknowledge that it will not be anything like past medical advances. Pasteur’s microorganism model of disease, Darwin’s theory of evolution, Florence Nightingale’s redefinition of nursing, Roentgen’s X-ray, or perhaps even the discovery of DNA by Watson and Crick are examples of game-changing insights. These are classic examples of massive improvements in healthcare delivery that come from a new fundamental insight. The improvements to healthcare that happen because of computerization will not be a revolution, but an evolution.
Fundamental to the ambitions in the health IT community is a humble acknowledgment that these huge game-changing insights are rare. We can expect fewer and fewer of them as the science of medicine progresses. Instead, medicine must now begin the difficult work of chronicling the immense complexity of a single cell’s DNA, proteins and other structures, and how that cell cooperates with other cells in the human organism. We can no longer expect that individual insights will leap medical science forward, but instead the medical community will make hundreds of thousands of small incremental advances on tens of thousands of diseases.
If we hope to continue the rate of improvement in healthcare we must find a way to coordinate the contributions of countless clinicians, researchers, and patients. To make any sense out of the genotype, we must have a understanding of phenotype —the manifest characteristics of individuals, such as their age, weight, medical symptoms, mental status, and many other measurable traits —than is several orders of magnitude deeper than it is today. We must be able to gather and parse a hundred times more data about each patient than we do today, and we must be able to compare that rich data among millions of patients. Today, the sciences and the software that support clinical trials, genomics, and standard clinical operations are separate and distinct, with infrequent overlap. Tomorrow, these disciplines will merge into a single enormous effort to improve healthcare. Science on this scale is impossible without mass high-quality computerization. There is no reason why all of this cannot be accomplished while respecting patient privacy and other basic notions of human dignity.
We hope to use technology to improve every aspect of healthcare. We hope to create information systems that help to turn medicine into a higher art and a higher science.
As you can imagine, with such ambitions, the health IT community frequently has delusions of grandeur. But we also suffer from frequent and stifling disillusionment. Although most of us agree that health IT has tremendous potential, progress in the field has been far too slow. We have a few good examples, like the VA with VistA, demonstrating that massive improvements to healthcare delivery are possible by leveraging technology. But we must admit that although we have a few good examples, we have countless examples of failure.
The authors of this book believe both that health IT has tremendous potential and that health IT is surprisingly difficult. As we discuss its difficulties, and the methods that have been used to successfully overcome them, we hope to avoid the pessimism that is all too common in health IT. Having said that, when pessimism and discouraging voices abound, it is often for good reason. There are real pitfalls in health IT, and this book should show you how to avoid many of them.
Health IT has changed tremendously over the last few years. The biggest change in the United States has come from the simple phrase “meaningful use.” The term is now solidly entrenched as the catchphrase for health IT in the United States. Most important, meaningful use represents reasonable first steps toward the long-term potential for health IT. For better or worse, the dreams and ambitions of the healthcare informatics industry are tied to the concept.
The phrase first appeared in the Health Information Technology for Economic and Clinical Health (HITECH) portion of the American Recovery and Reinvestment Act of 2009 (ARRA). ARRA defined that a substantial portion ($20 billion) of the money set aside by Congress to stimulate the United States economy after the financial and foreclosure crisis would go to doctors and hospitals who “meaningfully use” clinical software. The HITECH act was the first step in President Barack Obama’s comprehensive plan for healthcare reform. Clinicians would receive the stimulus money to pay for software to improve the delivery of patient care.
The bill referred to that software by the currently popular term electronic health record (EHR) software. But software designed to improve the delivery of clinical care has been around for decades, under different names. Such software has been called computerized patient records (CPR), electronic medical records (EMR), electronic health records (EHR), and countless other similar names with corresponding abbreviations. Even more confusing, there was no set definition of what this class of software was supposed to achieve. Unlike software products such as word processors, spreadsheets, or database storage engines, which all have well-understood definitions, the software category of EHR has meant very different things to different people. Passionate users would assume that EHR software meant the features that they wanted. Developers assumed that EHR meant the set of features in the software that they had developed. Of course, different users and different developers rarely agreed on which features were the most important. Dr. Ignacio Valdes (a medical doctor possessing a master’s degree in computer science with a stellar reputation in the health IT community) has frequently said, “For decades, doctors had no idea what they wanted, and software developers have given it to them.”
For clinicians, these terms served as a source of confusion and frustration. It was totally unclear what different names implied about the functionality of the software. Even a few years ago, when a doctor would say “I want an EHR!”, the right response from a health IT software vendor would have been “Fine, exactly what do you mean by EHR?”
The meaningful use EHR certification requirements have finally dictated exactly what EHR software needs to do for the doctors, hospitals, and other eligible providers who purchase, use, and deploy the software to receive payments from the ARRA-HITECH stimulus plan. In fact this has made the meaningful use requirements even more important than the term EHR. What is an EHR? That which can be used by a clinician to achieve meaningful use.
IT experts, as well as the general public, often fail to ask one simple question that will help focus any discussion of healthcare informatics: Why did the United States healthcare industry need to be paid to computerize? Every other industry computerized when, and to the degree that, computerization held intrinsic competitive advantages for members of that industry. Market forces compelled computerization, and companies that refused or resisted the move to computerization were squeezed out by competitors who were leaner and faster as the result of automated processes.
This has not happened in healthcare. Why not? It seems like such a simple and obvious step! Almost all hospitals and clinics already have some computers. They use them to type letters and send emails, to research on the web and coordinate schedules. They also automate some clinical tasks, most notably medical imaging, which is almost entirely computer-based. But, with few and notable exceptions, clinicians have not computerized the most central information resource they possess, the patient chart. The patient chart remains a paper record, usually a set of papers wrapped in a simple manila envelope.
For most information professionals (or clinical professionals with good information instincts), the use of computers to achieve standardization in data and work processes is a mantra. It is almost beyond question that computerized automation of processes and record-keeping would dramatically improve the performance of any industry. Still, healthcare has resisted computerization for decades.
In the answer to the question “Why hasn’t this happened on its own?” we will find the heart of meaningful use. The reasons that healthcare has not computerized can be summarized as screwy incentives and a difficult domain. Specifically:
Healthcare is orders of magnitude more complex than most other industries. This complexity has generated extensive clinical specialization. In some cases this specialization calls for extensive changes to the “normal” health IT workflow. It also means that each medical specialization has its own diagnostic categories, terminology, and procedures.
Healthcare is constantly changing. Treatments and best practices that are even two or three years old are frequently so out of date as to be nearly unethical. (Remember Vioxx.)
Attempts to computerize healthcare facilities have typically failed badly. Many explanations have been offered for this problem, and will be discussed throughout this book.
Even successful efforts are typically only partial automations, resulting in half-paper and half-computerized workflows that have the benefits of neither system and the drawbacks of both.
Healthcare providers in the United States, for the most part, are paid for their time. EHR typically slow doctors down, ensuring that they are paid less for the same work.
Computerization has typically been a very expensive proposition. Healthcare providers have been saddled with this cost, despite the fact that most of the benefits of computerization accrue to either patients or insurers. Essentially, until now, EHR systems have been disincentivized.
These factors, and others like them, have resulted in an abysmal value equation for EHR systems. For the average doctor or hospital, before ARRA funding, buying an EHR system was risky and expensive, with little benefit. To say that adoption of EHR software has been sluggish would be generous. It was widely known that fewer than 20% of small practices in the United States had purchased an EHR before the HITECH act. For a time, conventional wisdom held that although small practices had not purchased EHR systems, they were common (at least over 50%) in hospitals. But a study published in the New England Journal of Medicine in 2009 largely destroyed that illusion.[3] The study compiled a list of the 30 or so tasks that an EHR should do that actually improve the quality of healthcare, and found that less than 10% of all hospitals have installed software that accomplished even a few of these important tasks. The health IT industry was treating EHR deployment as a yes/no question, but if you asked “Is this software actually helping doctors deliver improved care in the hospital?” rather than just “Is something electronic in the hospital?” it became clear that hospital EHR software was typically underperforming.
One of the authors of that study was David Blumenthal. It was not an accident that he was chosen as the National Coordinator for health IT during a critical phase, when meaningful use would first be defined. You might call Dr. Blumenthal an expert in nonmeaningful use.
But it is not enough to merely pay doctors to use EHRs. The two counterexamples to systemic failure in health IT are Kaiser Permanente and the VA. The key word here is “systemic,” because many other, smaller organizations have succeeded with health IT deployments but they are not comprehensive systems of healthcare. The VA and Kaiser are systems with large numbers of both clinics and hospitals that use advanced health IT systems to effectively coordinate care between locations. Two commonalities between these systems should be mentioned at the outset of any intelligent discussion of perils and potential of health IT.
This book is all about the first item. Many health associations and experts deal with the second item, but the first must also get our attention for the whole package to work. We will focus on making, deploying, and leveraging good health IT software, but it is useless to think about the purchase and use of EHR software in an environment that disincentivizes those activities. These two concepts form what we call the VistA effect, which has turned the VA into the best hospital system in the world. If you do not believe us, I again recommend Longman’s book.
Meaningful use is an attempt by the U.S. government to define the baseline for what a clinician using an EHR should be able to accomplish. But it’s not just one set of requirements. Its definition will change every year. It will become more and more stringent as time goes forward, until it encompasses most of what the health IT research community agrees is needed for improved clinical care. Meaningful use is a moving target. The criteria do not explicitly include the deployment of flexible, software that conforms to clinicians’ needs as a goal, but without the ability to change gracefully, the same software that meets today’s meaningful use requirements will not be able to meet tomorrow’s.
Thankfully, meaningful use includes the financial incentives mentioned in the previous section. Its payments occur only when eligible entities prove that they are meaningful users of EHR systems, usually by reporting the details of how they use an EHR system certified for meaningful use. As a result, the meaningful use standards will always focus on specific measurable (and reportable) details of how an EHR system can operate. You need to install a certified system, and use it in valuable ways.
The incentive schedule for HITECH is almost as important as the funds themselves. The ARRA/HITECH incentives are only the beginning. Institutions that adopt systems and use them in certified ways in 2011 and 2012 could getting about $50,000 per doctor in total payments by being meaningful users of certified EHR technology over the period from 2011 to 2016. Those payments could come from either Medicare or Medicaid as bonus payments. Late adopters of the technology get less money, but have the luxury of watching what works among the early EHR adopters. But the real change occurs sometime around 2017. In that year or soon after, Medicare and Medicaid will no longer provide bonus payments for those who have adopted EHR systems, but instead will cut payments to everyone who has not adopted EHR systems. The meaningful use criteria are with us to stay, but the financial process behind meaningful use will serve as carrot for a short time, then it will become a stick.
One of the main limitations of the meaningful use funding is that it applies only to doctors, hospitals, and other eligible providers who bill either Medicaid or Medicare. Taken together, these two programs make the U.S. federal government the largest healthcare purchaser in the United States. But many healthcare providers do not take Medicare or Medicaid at all. There are many more who have threatened to stop taking Medicare or Medicaid if the planned reimbursement cuts are implemented. Many doctors and hospitals will not have the opportunity to get payments under HITECH at all.
But meaningful use will still matter to doctors who forgo Medicare and Medicaid dollars, eventually. Soon after the U.S. federal government starts to penalize healthcare providers who fail to provide EHR-derived healthcare quality data, it is reasonable to assume that private insurance companies will follow suit. In fact this might even happen before then. As soon as EHR systems become pervasive, they will become fair game for discriminatory payments from private insurers. The private insurers in the United States will not join Medicare and Medicaid in paying more for EHRs, but they will certainly join them in paying less for the lack of EHR systems. It is after the bonus payments are entirely gone that meaningful use will truly become a mandate.
At the time of this writing, it is hard to predict the course of more fundamental efforts to reform healthcare. The Obama administration has outlined ambitious plans to change the way healthcare in the United States is financed from the ground up. These reform efforts face several legal challenges that will ultimately be resolved by the Supreme Court of the United States, with unforeseeable results.
Not all of these reforms are relevant here, but those that focus on changing healthcare from a system that rewards the quantity of care to one that rewards quality of care are directly related to meaningful use. Most notable in the evolution of meaningful use is the concept of the accountable care organization (ACO).
ACOs are conceptualized in many different ways, but most of their incarnations focus on capitated or partially capitated care. Capitation means that providers are paid a set monthly fee for covered patients, whether they are sick or not. In theory, capitation incentivizes providers to keep patients healthy, thereby reducing the amount of money spent on their healthcare over the long term.
Capitation was at the heart of the health maintenance organization (HMO) movement that began in the United States during the 1970s and continued into the 1980s and 1990s. Eventually HMOs earned a notorious reputation for simply not paying for healthcare to keep costs down. Instead of responding to the incentives in capitation by improving the quality of healthcare delivery, they cut corners and abandoned patients to save money.
What is the difference between the hated HMO model and the currently popular ACO model? Only the capacity to detect whether the care delivered as part of the system is of high quality. The HMO model used statistics to provide models of what treatments were appropriate at given times, and denied treatments that fell outside the model. These simplistic models tended to second-guess healthcare providers about what patients needed, frequently denying coverage for sensible treatments. An ACO, however, should be capable of measuring the end result of a provider’s actions to determine whether treatment was successful, making second-guessing unnecessary. An HMO meddled with doctors’ methods, whereas an ACO focuses only on the doctors’ results. How will an ACO accomplish this? By leveraging data from meaningful use certified EHR systems.
ACOs and concepts like them are an attempt to reform the crazy incentives in healthcare, but those reforms will not work unless it is possible to truly measure the performance of providers through accurate data on the health of the patients in their care. With the data that an EHR provides, an ACO could actively seek out difficult patients, like diabetics, knowing that they would be compensated more for a patient with diabetes. By using smart information systems to customize care, they could treat the diabetic more effectively with lower costs. Everyone wins: the individual diabetic gets better care, the provider gets paid more for providing that care, and the system as a whole pays less for the treatment of that particular diabetic. All of this is made possible by a highly capable EHR system.
In short, meaningful use, paired with the other healthcare reforms, has the potential to initiate the VistA effect, where healthcare organizations constantly measure the quality of care they are delivering and use flexible software to enforce higher and higher levels of patient safety and quality care.
Meaningful use will be at the heart of healthcare reform in the United States for the next several decades, making it one of the most significant components of healthcare reform. To the degree that the United States is a worldwide health IT leader, meaningful use will also have implications internationally.
Happily, the meaningful use requirements are relatively short and to the point. The initial version of meaningful use includes features such as demographics, medication list, problem list, and vital signs. These features are trivially intuitive for clinicians, but end up being far more complicated than they seem to implement in software. These core health information constructs become far more complex when we consider just how much we want to do with the underlying data. The first version of meaningful use requires one test of health information exchange, but later versions make it clear that meaningful use will ultimately require providers to securely share patient data with other providers who treat the same patient. Tracking health data and tracking health data in a way compatible with other health data are very different things. Ensuring that health data is liquid is much more complex then just gathering it together.
Things like demographics become tremendously complex in the context of health information exchange. For instance, different healthcare providers in a given city might have electronic records for:
| Fred Trotter — 12/31/1975 |
| Frederick Trotter — 12/31/1975 |
| Freddy Trotter — 12/30/1975 |
| Fred Trotten — 12/31/1975 |
Are these all the same person? If an EHR accepts data sent to it from an outside source, but under a given name that is not the same as one in the EHR database, how should the system react? Should it react differently depending on whether a bill to a insurer under the original given name has been accepted and paid? (Demographic details are a frequent reason for rejection of the payment requests that providers make to third-party payers.) What if the insurance company has the wrong name, but for whatever reason, is unwilling to change it? Do you keep the name you know is wrong for billing purposes, and if so, how do you keep it from polluting data exchange for clinical purposes?
To understand what an item like problem list or demographics truly means in terms of EHR systems, you have to understand a tremendous amount of healthcare-related context, as well as a few sticky points of software design. Things in healthcare IT often work counterintuitively to the normal workings of IT. This is because things in healthcare work very differently than in any other industry. The issues associated with medical billing alone are usually enough upend typical IT approaches. At the end of this book you should be able to read the meaningful use requirements and have an understanding of what it will take to execute them. You should be able to recognize which issues in health IT are open, difficult problems, and which issues have already been solved by industry best practices. You will be able to see through those pundits who frequently present health IT molehills as mountains or mountains as molehills. We also believe that you will begin to see the meaningful use requirements just as we do: a reasonable set of standards that are simple enough to actually fulfill, with enough punch to still make a difference in healthcare. Like many health IT experts, your authors can tend to be jaded, but we see the meaningful use requirements as fundamental evidence of good government. It is difficult to strike any kind of balance with health IT standards and the Office of the National Coordinator has done a good job of this with the meaningful use requirements.
It has taken software professionals about a decade to get up to speed on health IT, a decade that was frequently spent in confusion and frustration. We remain confused and frustrated with health IT. The authors still think, however, this industry also holds the highest hopes for IT: to make a real difference in peoples lives. We have seen far too many IT professionals leave health IT because the frustrations with the daily grind outweighed the hope for change. We believe that by reading this book, you can skip part of that process that we went through, and be confused, frustrated, and hopeful at a much higher level.
Get Hacking Healthcare now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
