The Promise of Health Information Technology for Cost, Quality, and Privacy
“We will . . . wield technology’s wonders to raise health care’s quality and lower its cost,” declared President Barack Obama during his inaugural address on January 20, 2009.
Exactly four weeks later, on February 17, he signed into law a $787-billion economic stimulus package that sets aside $19 billion for health information technology (HIT). Of that sum, $17 billion will cover incentive payments from the Medicare and Medicaid reimbursement programs to help hospitals and physician offices adopt electronic medical record systems.
These policies align with recommendations RAND has made since 2005. However, we consider the $19 billion for HIT in the February economic stimulus package to be merely a down payment, although a sizable one, toward fulfilling the promise of HIT.
Our focus is on the costs and benefits of HIT for the United States at a national adoption level of 90 percent, which might take 15 years to attain. Over those 15 years, we project total HIT costs of $115 billion, potential efficiency savings of $628 billion, and thus potential net savings of $513 billion or more. This translates to average annual costs of $8 billion, average annual efficiency savings of $42 billion, and average annual net savings of $34 billion or more. All these estimates are in 2004 dollars.
The benefits should grow as HIT spreads throughout the health care system. After 15 years, the nationwide adoption of electronic medical records and of networking among health care providers could save more than $77 billion each year in terms of efficiency alone and another $4 billion each year in terms of drug safety. Although the potential savings appear large, $81 billion is about 4 percent of the annual $2 trillion cost of health care in the United States. HIT-enhanced preventive care and management of chronic diseases could increase those savings while improving health and providing other social benefits (fewer days lost at work or school, for example).
We emphasize that these figures represent the potential benefits of HIT. These figures also assume that interconnected and interoperable electronic medical record systems are adopted widely and used effectively. The current evidence is not robust enough to allow strong predictions. However, we do not report the best-case scenarios, either, but rather the mean value of potential savings based on the current evidence.
By the end of 2009, the U.S. Department of Health and Human Services plans to develop a set of standards that will determine not only how electronic medical records should work and what they should include, but who should control them and how. To derive the full benefits from HIT, we propose that the nation also create unique patient identification numbers for every person in the country. Such a system will not endanger but rather enhance personal privacy while reducing medical errors, increasing efficiency, and simplifying the use of electronic medical records.
AP IMAGES/CAROLYN KASTER
Infection Control Coordinator Rahn Snyder checks data on a computerized infection monitoring system at Hershey Medical Center in Hershey, Pennsylvania. Pennsylvania health officials view the nascent technology as a critical tool for helping hospitals reduce health care costs by identifying potential problems sooner than is possible by reviewing paper records by hand.
HIT is shorthand for an electronic medical records system that replaces the paper medical record and incorporates such associated functions as clinical decision support, patient tracking, reminders for preventive services, computerized physician order entry to check proper prescribing (and to reduce adverse drug events), and electronic connectivity among health care providers and, in some cases, among providers and patients.
The health records for most people today are still stored on paper, and the transfer of records from one doctor to another is still most often done by phone or fax. These practices are highly prone to errors, such as illegible handwriting; the loss of records of expensive medical tests; and deadly, but preventable, mistakes.
The hope is that the broad adoption of HIT will transform health care by making it more efficient and simultaneously more effective. Greater efficiency would result from reduced test duplication, improved drug utilization, better scheduling, reduced handling of paper records, and expedited processing and billing of claims. Greater effectiveness would result from reduced errors, continual evidence-based decision support, reminders for preventive care, improved management of chronic illness, and improved continuity of care for those patients seeking it when they are away from their primary providers.
But we are a long way from reaching these goals. As of 2005, about 20–25 percent of hospitals and 10–15 percent of physician offices had adopted HIT systems. More recent estimates indicate that hospital-based adoption has improved but that the functionality of many of the adopted systems remains limited. Meanwhile, there has been relatively little change in physician adoption of HIT. Most important, only about 4 percent of physicians have HIT systems that incorporate some of the key supporting functions.
The $115 billion in total costs for HIT over 15 years includes software licenses, hardware and its maintenance, planning, training, implementation, and either reduced revenue or increased provider costs during implementation. Of the $628 billion in total potential efficiency savings throughout those first 15 years, about 75 percent of the savings would be associated with hospitals, and 25 percent would be associated with physician offices.
In hospitals, major savings would come from shorter patient stays (thanks to improved scheduling and more timely and effective care), less nursing administrative time (and therefore lower demand for nurses), less chart administration, reduced drug costs, and fewer laboratory and radiology tests. In physician offices, the savings would come from reduced drug costs, more efficient chart handling, fewer transcriptions, and fewer laboratory and radiology tests (see the centerpiece).
The reduced drug costs would come from aligning prescriptions with their formulary rules, recommending generic drugs when available, advising physicians of the costs and benefits of specific drugs, encouraging providers to discontinue unnecessary or harmful drugs, and encouraging timely conversion from intravenous to oral medications.
We did not estimate the efficiency savings from billing and claims administration, but we expect those to be substantial. Neither did we estimate possible proc- ess improvements that have become the typical consequences of information technology in other industries. Therefore, while our estimates indicate significant potential efficiency savings of $77 billion annually after 15 years, we do not consider our estimates to be overly optimistic.
In addition to the efficiency savings, we project safety benefits of $4 billion per year after the first 15 years, owing to the annual avoidance of up to 2.2 million adverse drug events and their subsequent costs. These events would be averted by reduced handwriting errors, better warnings of allergies and drug interactions, and better dosage monitoring. Because most prescribing occurs in physician offices, the magnitude of this benefit would depend heavily on physician adoption of HIT.
As for health benefits, HIT would enable substantial improvements in managing chronic illness. About 75 percent of the U.S. health expenditure is associated with people with chronic illness. What is needed in such cases is better coordination and communication across providers dealing with the multiple impacts of a chronic illness, along with substantial patient screening, monitoring, and involvement. As exemplified by the electronic medical records system now used by the U.S. Department of Veterans Affairs, networked HIT provides a way to coordinate support, monitor patients, and involve the patient in a team of care.
We simulated the health benefits of improved management of four chronic illnesses: asthma, diabetes, congestive heart failure, and chronic obstructive pulmonary disease. If all eligible patients participated nationwide, the changes in treatment and lifestyle triggered by HIT could yield 20 million fewer inpatient days, 5 million fewer emergency department visits, 9 million fewer office visits, and 20 million added workdays per year.
We did not estimate potential HIT savings from healthier patients, because it is not always true that reducing the incidence of a disease reduces health care costs. In fact, costs can rise or fall depending on the effect of health care on longevity and the occurrence of other diseases.
AP IMAGES/JAMIE-ANDREA YANAK
Physicians Martin Harris and Nancy Pae meet in front of a display of the Cleveland Clinic’s MyChart system, which allows patients to send health information to the clinic’s system from their home computers.
There are four imperatives for government leadership in HIT. The primary one is a market failure: Those who must purchase the HIT systems (the hospitals and physicians) are not the ones who would reap most of the savings (the insurers) or the health benefits (the patients). Of the $115 billion in costs over the first 15 years, about $98 billion would be borne by hospitals, and about $17 billion would be borne by physicians.
Except for health systems that are both insurer and provider, such as Kaiser Permanente, there is little financial incentive for hospitals and physicians to bear the costs of HIT and to disrupt their practices for its implementation. The insurers would garner the savings from reduced duplication of tests, shorter hospital stays, better drug utilization, and so on. Meanwhile, better management of illnesses would mean fewer visits of sick patients to hospital emergency departments and physician offices, reducing hospital and physician revenue. This market failure is a key indicator of the need for government intervention.
Second, larger hospitals and physician practices are adopting HIT at a much higher rate than smaller, less endowed ones. Without any government intervention, the end result of market-based HIT could be increased disparities in health care.
Overcoming these first two constraints will require expanded subsidies for hospitals and physicians. The most appropriate parties to pay the subsidies could be the insurers, which stand to gain most of the benefits. It is thus fitting that the president has assigned Medicare, the largest insurer, to exercise leadership in this regard. Subsidies might also be targeted to smaller hospitals and physician offices and to those serving disadvantaged populations.
Third, there is little market incentive to manage the electronic infrastructure required to reach across providers and to connect them into a network of interoperable systems that could support a patient’s care wherever and whenever it is needed. This is another area in which the government will need to play a larger role. Fourth, America’s current privacy and security protections are inadequate for a national health information network. An important government role will be to establish the privacy protocols for such a network.
To protect patient privacy, most of the U.S. health care system today relies on statistical matching methods that link patients to their health data by use of personal attributes, such as name, address, zip code, and birth date. The problem with these data is that they are often not unique to the individual, they change over time, or they are entered into different systems in different formats, often with errors. The repeated collection, distribution, storage, and use of these data also pose a substantial identity-theft risk.
There are important health, safety, efficiency, and privacy reasons for moving the United States away from the inherent risks of statistical approaches and toward a unique patient identifier for health care. Creating a unique patient identification number for every person in the country would reduce medical errors, simplify the use of electronic medical records, and increase efficiency. Such a system should also enhance privacy, because it would not involve sending large amounts of personal data across the health care network.
It might seem logical to use a Social Security number for this purpose; however, the widespread use of the Social Security number for so many other purposes has led to its being frequently compromised as a secure identifier. Furthermore, the Social Security number lacks the types of protection built into modern identifiers, such as “check digits” that can detect erroneous entries. For a national health information network, it would be better to use a different number or alphanumeric code.
Many privacy concerns related to unique patient numbers could be addressed with laws that severely punish those who misuse information retrieved with the numbers. And in contrast to using personal information, a system that uses a new unique number would simplify the reestablishment of security after any breach of a patient’s health information.
Giving people the choice of whether to acquire a unique patient number could further reduce privacy concerns. Those worried about misuse of a number could simply opt out. A voluntary national system would cost about $25 million for the first five years for issuing the unique patient identifiers. We estimate an additional cost of registering all people in the country to be about $1.5 billion ($5 each for 300 million individuals, based on 5 minutes of health-care-provider office time at $1 per minute).
Most individuals and organizations now addressing this issue are focused on privacy and security, but very few have addressed the distinction between statistical matching and unique patient identifiers. If the methods were debated publicly, the privacy and security risks of statistical matching would likely become an issue. But without this debate, statistical matching has had the advantage of requiring no new national policy and has therefore avoided being judged under the bright lights of public scrutiny.