|
PERSPECTIVESavings In Electronic Medical Record Systems? Do It For The Quality
National policymakers are considering whether to make major long-term investments in electronic medical record (EMR) systems. The matter of rising health care costs is never far from any health care debate, and the prospect for EMR systems to decrease costs is a potential selling point. The paper by Richard Hillestad and colleagues presents a well-documented analysis of the potential costs, savings, and other benefits of widespread adoption of interoperable EMR systems. It focuses on the potential savings such systems could yield. Here I examine the main components of their argument and question whether such savings could ever be realized.
Richard Hillestad and colleagues build a carefully documented, thorough, and transparent case for the potential costs, savings, and other benefits of widespread adoption of interoperable electronic medical record (EMR) systems.1 The capstone of their analysis estimates potential savings in national health spending. This commentary examines the main components of Hillestad and colleagues’ argument and questions whether any such savings could ever be realized.
The authors start by presenting a top-down approach to estimating the potential savings of widespread EMR adoption, then move to the main focus of their analysis, a bottom-up approach. The top-down approach applies productivity gain rates in other industries that adopted information technology (IT) during the 1990s. The authors apply these rates to estimated national health spending during the fifteen-year period 2002–2017. At the low end, the 1.5 percent per year productivity gain attributed to IT in the retail/wholesale sector would translate into an average annual decrease of $346 billion. At the high end, a 4 percent per year productivity gain, just half of that attributed to IT in the telecommunications sector, would translate into an average annual decrease of $813 billion. If the authors’ estimated average savings were reached by, say, year ten, they would amount to 11 percent and 26 percent, respectively, of projected 2012 national health spending. That would be real money. As noted by the authors, a truly transformational change—such as the 8 percent or more in annual productivity gains experienced in the telecommunications and securities industries—would require market conditions and other changes that are virtually absent in health care today.
The authors move to a bottom-up approach, using a simulation model of health IT adoption and scaling literature-based effects. The savings are built up from efficiency savings in inpatient and outpatient care and safety benefits (primarily attributable to using computerized physician order entry, or CPOE, to reduce adverse drug events). Additional savings are derived for using HIT for short-term preventive care, which actually results in a small net increase in spending, near-term chronic disease management, and prevention of chronic disease in the long term. The authors project that effective EMR implementation and networking could save almost $82 billion in health care efficiency and safety per year by year fifteen. This figure assumes savings of $77.4 billion from efficiency in inpatient and outpatient settings, plus another $1.0 billion from reduction of inpatient adverse drug events and $3.5 billion from reduction of ambulatory adverse drug events. The ramping up to $77.4 billion in efficiency savings assumes eventual 90 percent adoption in inpatient and outpatient settings, averaging $42 billion during the adoption period. Cumulative savings from efficiency and improved safety would amount to $628 billion, including $468.5 billion on the inpatient side and $159 billion on the outpatient side. These savings would be partially offset by the costs of implementing EMR systems over those fifteen years. The cumulative additional costs would amount to $98 billion, averaging $6.5 billion, for hospitals, and $17.2 billion, averaging $1.1 billion, for physicians. As opposed to the later-accruing efficiency savings, these costs would be more front-end-loaded, followed by flat maintenance costs. The simulation assumes that implementation of EMR systems lasts two years in the outpatient setting and four years in the inpatient settings, followed by annual maintenance costs that are 20 percent and 30 percent, respectively, of implementation costs. The authors note that these savings have not been discounted and do not account for inflation in health spending.
At a time when national policymakers are weighing the potential costs and benefits of making substantial financial commitments to broad implementation of HIT, the prospect of realizing dramatic reductions in national health spending could be persuasive and perhaps decisive. But is it viable? Contrary to their assertion, the fine work of Hillestad and colleagues suggests that a better rationale for widespread adoption of EMR systems lies elsewhere. The projected annual savings in efficiency and safety would reach $82 billion by year fifteen. Assuming the 2004 start used by the authors, that would be by 2019, when the midpoint baby boomers are turning sixty-five and annual health spending will be past $5 trillion. Annual efficiency savings of $82 billion achieved by 2019 would amount to 1.6 percent of health spending that year. Cumulative net savings (accounting for costs) that would accrue over the fifteen years for combined efficiency and safety would be $513 billion, including $371 billion in hospitals and $142 billion in physician practices. This would amount to a 1.05 percent dent in the cumulative $48–$50 trillion in health spending over that span. (The cumulative net savings are provided only for efficiency and safety, not for the disease management and lifestyle savings. Although the paper offers that improvements in prevention and management of chronic disease could double the $82 billion in annual savings to be achieved after fifteen years, the figures presented appear to indicate a 50 percent increase, to around $120 billion by then.) Even fifteen years out, $82 billion sounds like real money, as does $513 billion accrued over that time. But the case for investing now in widespread adoption of EMR systems based on efficiency and safety savings that would eventually rise to an annual 1.6 percent clip and that would chip away 1.05 percent from aggregate health spending by 2019, does not look dramatic from here.
How useful is it to project how HIT savings would trim national health spending? It might help to gain a rough feel for the relative magnitude of the potential impact of HIT and how gradual this effect would be, even under optimistic scenarios. But then the usefulness of the exercise starts to fade. Because the savings estimates themselves, particularly those regarding efficiency, are derived largely from the effects reported in the literature, they are scaled up from the effects in pockets of the health system as we have recently known it. Furthermore, these potential savings are superimposed onto projected national spending of a ballooning—not fundamentally changing or transforming—version of the same system, adjusted mostly for demographic and macroeconomic assumptions. As the authors note, the potential savings assume that "interconnected and interoperable EMR systems are adopted widely and used effectively." Although this does not represent the type of transformed health system that could achieve productivity gains seen in the aforementioned other industries, it certainly would be a different system than a bigger version of the one we have now. Perhaps most important is that none of the potential savings of widespread HIT adoption are likely to reach the bottom line of national health spending. This is not a system whose demand is currently being satisfied. Whether in its current form, evolving with widespread HIT adoption, or entirely transformed, the U.S. health system will find ways to reallocate the money. As the authors note, "It is possible that the efficiencies will be used to improve health care quality rather than to reduce costs." Other optimistic reallocations would be expanding access to Medicaid or funneling savings into pay-for-performance programs. In any case, it is unrealistic to hold out effective, widespread adoption of HIT as a net cost saver. The extraordinary IT-enabled productivity improvements experienced in telecommunications, securities, and other sectors did not result in less spending in those industries, but they do exemplify creation of previously unimaginable forms of benefit for burgeoning markets.
One of the many illuminating aspects of Hillestad and colleagues’ paper is how it points to the need for content and smart analytical software. Transformation will not just be a matter of plug and play. There will also be the need to develop predictive modeling algorithms linked to pharmacogenomic databases and other emerging resources to identify and optimally manage patients with particular conditions. Other new and emerging tools will have to be not just upgraded but uploaded to higher-order systems to realize the HIT potential. For example, CPOE provides benefits in the form of warning about potentially harmful interactions of a new drug with other drugs. But as a component of an HIT system, it can avoid upstream errors in drug order and execution. Across a large population, a health plan, or a country, CPOE data can do even more for postmarketing surveillance, epidemiology, and market research. Accomplishing this will require extending and creating entirely new algorithms and software whose applications, including some that would otherwise be inconceivable, will be enabled by widespread adoption of HIT. The potential of HIT adoption offers a direct means of addressing the inexcusable quality deficit in U.S. health care. The potential efficiencies and improvements described by Hillestad and colleagues are well within the reach of the technology that we already have. A major commitment by the federal government with the active support of the private sector is necessary to resolve the short-term disincentives and immediate market barriers to HIT participation. The capacity for transformation will arise when this system enables new forms of high-speed, broadly integrated data collection, analysis, and knowledge development and transfer in a value-based health care market.
Clifford Goodman (clifford.goodman{at}lewin.com) is vice president of the Lewin Group in Falls Church, Virginia.
This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Top-down approach.
Bottom-up approach.
