PERSPE C T I V E Defining Safe Use of Anesthesia in Children R.D.M., A.S.) and the Center for Drug Evaluation and Research (J.W.), Food and Drug Administration, Silver Spring, MD. This article (10.1056/NEJMp1102155) was published on March 9, 2011, at NEJM.org. 1. Slikker W Jr, Zou X, Hotchkiss CE, et al. Ketamine-induced neuronal cell death in the perinatal rhesus monkey. Toxicol Sci 2007; 98:145-58. 2. Brambrink AM, Evers AS, Avidan MS, et al. Isoflurane-induced neuroapoptosis in the neonatal rhesus macaque brain. Anesthesiology 2010;112:834-41. 3. Paule MG, Li M, Allen RR, et al. Ketamine anesthesia during the first week of life can cause long-lasting cognitive deficits in rhesus monkeys. Neurotoxicol Teratol 2011 January 15 (Epub ahead of print). 4. DiMaggio C, Sun LS, Kakavouli A, Byrne MW, Li G. A retrospective cohort study of the association of anesthesia and hernia repair surgery with behavioral and developmental disorders in young children. J Neurosurg Anesthesiol 2009;21:286-91. 5. Wilder RT, Flick RP, Sprung J, et al. Early exposure to anesthesia and learning disabilities in a population-based birth cohort. Anesthesiology 2009;110:796-804. Copyright © 2011 Massachusetts Medical Society. implemented. The FDA is com­ mitted to pursuing these an­ swers with the medical and sci­ entific communities and will take the steps necessary to ensure that the benefits of anesthetic use in children continue to outweigh any potential risks. Disclosure forms provided by the authors are available with the full text of this arti­ cle at NEJM.org. From the Division of Anesthesia and Analgesia Products, Office of New Drugs (B.R., How CER Could Pay for Itself — Insights from Vertebral Fracture Treatments Adam G. Elshaug, M.P.H., Ph.D., and Alan M. Garber, M.D., Ph.D. T he pain and disability caused by osteoporotic vertebral frac­ tures have long motivated the search for effective therapy. Two procedures designed to restore vertebral body height and func­ tion have been widely adopted: percutaneous vertebroplasty, in which cement is injected into the vertebral body to support the fractured bone; and kyphoplasty, a variant of vertebroplasty in which a balloon is inserted and inflated in a collapsed vertebral body, restoring the bone’s height before the cement injection. Ini­ tial studies suggested that these procedures were superior to con­ ventional symptomatic treatment. But when later studies cast doubt on those favorable findings, health care funding agencies sought to curb their use. The story of these procedures offers a glimpse of the ways in which comparative­effectiveness research (CER) may influence medical practice and health care expen­ ditures. Early studies of these proce­ dures were neither randomized nor blinded, and because the symptoms of compression frac­ tures often abated over time, the lack of adequate controls made it impossible to know whether im­ provements that followed treat­ ment would have occurred even without surgery. Furthermore, neither procedure was risk­free; reported complications included compression fractures, cement leakage, pulmonary complica­ tions, paraplegia, and death.1 In a scenario that’s likely to be re­ peated frequently as CER gains greater acceptance and support, randomized trials eventually fol­ lowed the observational studies that had fostered the initial en­ thusiasm.2 If the full conse­ quences of that research are not yet fully apparent, their potential importance is. Were the results of better­designed studies trans­ lated into practice, the reduction in U.S. health care expenditures would be considerable. CER treats effectiveness as a balance of benefits and harms; when the risks associated with a procedure outweigh its clinical benefits, it is appropriate and ethical to limit its use. Both the clinical need and the desire to avoid wasteful expenditures were part of the rationale for subject­ ing these procedures to compar­ ative studies. Furthermore, con­ sensus that these procedures were promising but unproven led sev­ eral countries to make them available on an interim­coverage basis. These arrangements, in ef­ fect from 2006 through 2010, al­ lowed the procedures to be per­ formed in everyday practice while further evidence was generated. Trials conducted during that period suggested that kyphoplasty did not improve outcomes. The studies of vertebroplasty produced varying results, but the highest­ quality trials cast doubt on the benefit and raised additional safety concerns. In a randomized but non­blinded trial by Kallmes et al.,3 patients who underwent vertebroplasty and controls had similar reductions in disability and pain scores, with a trend to­ ward a higher rate of clinically meaningful improvement in pain 1390 n engl j med 364;15 nejm.org april 14, 2011 The New England Journal of Medicine Downloaded from nejm.org at HARVARD UNIVERSITY on December 16, 2013. For personal use only. No other uses without permission. Copyright © 2011 Massachusetts Medical Society. All rights reserved. PERSPECTIVE How CER Could Pay for Itself Vertebroplasty and Kyphoplasty Volumes and Costs in the United States, with Potential Savings from Decreased Use* Total Estimated No. of Procedures (% Accounted for by Specified Payer) Annual Savings with a 50% Decrease in Utilization dollars Medicare (NIS) Medicare (SASD) Medicare total (NIS + SASD) Private insurance (NIS) Private insurance (SASD) Vertebroplasty Kyphoplasty Vertebroplasty Kyphoplasty Vertebroplasty Kyphoplasty Vertebroplasty Kyphoplasty Vertebroplasty Kyphoplasty Private insurance total (NIS + SASD) Vertebroplasty Kyphoplasty Combined total of principal procedures — 11,253 (82.2) 19,397 (82.7) 6,260 (77.7) 11,684 (76.2) 17,513 (80.5) 31,081 (80.1) 1,522 (11.1) 2,992 (12.8) 1,341 (16.6) 2,913 (18.9) 2,863 (13.2) 5,905 (15.2) 57,362 (94.8) 3,024 14,932 653 2,979 78,950 11,411 14,336 4,451 7,328 — — 14,514 11,968 4,865 9,945 — — — 11,411 14,336 14,514 11,968 — 128,407,983 278,075,392 27,863,260 85,620,352 156,271,243 363,695,744 22,090,308 35,808,256 6,523,965 28,969,785 28,614,273 64,778,041 613,359,301 34,506,864 214,065,152 9,477,642 35,652,672 907,061,631 64,203,992 139,037,696 13,931,630 42,810,176 78,135,622 181,847,872 11,045,154 17,904,128 3,261,983 14,484,893 14,307,137 32,389,021 306,679,652 17,253,432 107,032,576 4,738,821 17,826,336 453,530,817 102,726,386 222,460,314 22,290,608 68,496,282 125,016,994 290,956,595 17,672,246 28,646,604 5,219,172 23,175,828 22,891,418 51,822,433 490,687,440 27,605,491 171,252,122 7,582,114 28,522,138 725,649,305 Annual Savings with an 80% Decrease in Utilization Payer (Data Source) Procedure Mean Cost per Procedure Aggregate Costs in 2008 Medicare, secondary procedure (NIS) Vertebroplasty Kyphoplasty Private insurance, secondary procedure (NIS) Grand Total, Principal + Secondary Vertebroplasty Kyphoplasty — * Data are from the 2008 Nationwide Inpatient Sample (NIS) and the State Ambulatory Surgery Databases (SASD) of the Healthcare Cost and Utilization Project, Agency for Healthcare Research and Quality (AHRQ).5 Complete SASD data were analyzed and provided by Claudia Steiner of AHRQ. Primary procedures not paid for by Medicare or private insurance were provided to patients with Medicaid or no insurance. The estimates of budget impact are conservative, in part because we used costs (as reported to the Centers for Medicare and Medicaid Services) instead of charges or actual payments, and in part because the SASD covers only 28 states. We conservatively estimate that 17,000 (80%) of the NIS procedures coded as secondary were actually the major procedure performed; SASD coded fewer than 3000 procedures as secondary, and these have been excluded from the analysis. (30% decrease from baseline) in the vertebroplasty group that neared statistical significance (64% vs. 48%, P = 0.06). In a ran­ domized, blinded trial by Buch­ binder et al.,4 vertebroplasty did not have a statistically significant advantage over placebo in any measured outcome over 6 months, although pain diminished in both groups. These studies illustrate the dif­ ficulty of inferring the effects of treatments for a condition with a variable time course, particular­ ly when its manifestations are strongly influenced by placebo effects. But the studies at best cast doubt on the magnitude of any benefits from these proce­ dures and at worst established their ineffectiveness. The find­ ings led U.S. and other payers to revisit their interim funding de­ cisions. To improve safety and quality and to respond to pres­ sures for fiscal responsibility and efficiency in health care, payers are deciding to limit or withdraw coverage for these procedures. In late 2010, the Blue Cross Blue Shield Association’s Medical Ad­ visory Panel confirmed its deci­ sion that neither procedure met its criteria for established effec­ tiveness, and in Canada, the On­ tario Health Technology Advisory Committee ruled that vertebro­ plasty should not be considered n engl j med 364;15 nejm.org april 14, 2011 1391 The New England Journal of Medicine Downloaded from nejm.org at HARVARD UNIVERSITY on December 16, 2013. For personal use only. No other uses without permission. Copyright © 2011 Massachusetts Medical Society. All rights reserved. PERSPE C T I V E How CER Could Pay for Itself the standard treatment for osteo­ porotic vertebral fractures. Any CER agenda strives for improved safety and quality of care. By identifying relative inef­ fectiveness, CER may also im­ prove the health care system by freeing up resources to be used for safer and more effective forms of care. Savings from limiting the use of care whose effectiveness is unproven can be substantial, Centered Outcomes Research In­ stitute, created by the Affordable Care Act (ACA), is fully opera­ tional, its budget is expected to reach $500 million annually, or just two thirds of the potential savings each year from dimin­ ished use of just two apparently ineffective procedures. The savings might be reduced if patients who don’t receive one of these procedures end up being Without randomized trials, ineffective and costly treatments with risks and complications would continue to be administered largely because the alternative treatments are disappointing. whether the intervention is new or has already been disseminated. According to our analyses of data from the Healthcare Cost and Utilization Project,5 in 2008 the cost of kyphoplasty and vertebro­ plasty was approximately $1 bil­ lion. The table shows the estimat­ ed savings in the United States, by insurance type, under alterna­ tive assumptions about reduc­ tions in utilization. A 50% reduc­ tion in utilization would deliver annual savings of $450 million; an 80% reduction would save about $725 million annually. Since these figures are based on costs rather than charges or payments, they are highly conservative. And although these figures appear small relative to U.S. health care expenditures, the procedures are not among the most common. Furthermore, savings are large in relation to the $1.1 billion that Congress allocated to CER in the 2009 American Recovery and Re­ investment Act. When the Patient­ treated more aggressively with other forms of care. For example, patients who do not undergo ver­ tebroplasty might receive more pain medications or physical ther­ apy than patients who undergo the procedure. However, such off­ sets in savings would be substan­ tial only if the procedures greatly diminish symptoms for an ex­ tended period. Furthermore, the cost­savings estimates don’t take into account expenditures for the treatment of adverse effects of the procedures. CER won’t always yield defini­ tive conclusions about a therapy’s effectiveness; individual patients might benefit despite disappoint­ ing results in randomized trials. But the adoption of a procedure in routine practice, if not part of a well­designed study, probably won’t reveal the characteristics of the patients likely to benefit. If observational studies are well de­ signed and build on clinically de­ tailed data, they can often eluci­ date information about subgroups that were not studied in a trial. But the limitations of convention­ al observational studies for a con­ dition with fluctuating symptoms and whose main manifestation is pain apply here: without double blinding and closely matched controls, it will be surpassingly difficult to distinguish the effects of the intervention from the nat­ ural history of the condition. Thus, without randomized trials, ineffective and costly treatments with risks and complications would continue to be adminis­ tered largely because the alterna­ tive treatments are disappoint­ ing. If nothing else, well­designed studies demonstrating ineffective­ ness can help redirect research toward the development of alter­ natives. Of course, savings will be de­ rived from CER only if practice changes. In the United States, it’s unclear whether these studies are powerful enough to overturn cov­ erage decisions or cut utilization of established procedures. The status quo plays a large role in de­ termining the burden of proof for interventions: if a procedure has spread widely, large, well­designed studies must show that it’s clearly ineffective or harmful before pay­ ers and providers will abandon it; for a new procedure, the assump­ tion is that effectiveness has not been established, so good studies demonstrating effectiveness are required for its adoption. Increas­ ingly, funding agencies and poli­ cymakers aim to subject estab­ lished practices to greater scrutiny, since often interventions adopted without strong evidence are later found to be ineffective or not as effective as initially thought. ACA features such as bundled 1392 n engl j med 364;15 nejm.org april 14, 2011 The New England Journal of Medicine Downloaded from nejm.org at HARVARD UNIVERSITY on December 16, 2013. For personal use only. No other uses without permission. Copyright © 2011 Massachusetts Medical Society. All rights reserved. PERSPECTIVE How CER Could Pay for Itself represent those of the Commonwealth Fund or its directors, officers, or staff, or those of the Department of Veterans Affairs or Stanford University. Disclosure forms provided by the authors are available with the full text of this arti­ cle at NEJM.org. From the International Program in Health Policy and Practice, Commonwealth Fund, New York (A.G.E.); the School of Population Health and Clinical Practice, University of Adelaide, Adelaide, SA, Australia (A.G.E.); the Veterans Affairs Palo Alto Health Care System, Palo Alto, CA (A.M.G.); and the Center for Health Policy and Center for Primary Care and Outcomes Research, Stanford University, Stanford, CA (A.M.G.). 1. Manufacturer and User Facility Device Experience Database (MAUDE). Silver Spring, MD: Food and Drug Administration. (http://www.fda.gov/MedicalDevices/ DeviceRegulationandGuidance/Postmarket Requirements/ReportingAdverseEvents/ ucm127891.htm.) 2. Weinstein JN. Balancing science and informed choice in decisions about vertebroplasty. N Engl J Med 2009;361:619-21. 3. Kallmes DF, Comstock BA, Heagerty PJ, et al. A randomized trial of vertebroplasty for osteoporotic spinal fractures. N Engl J Med 2009;361:569-79. 4. Buchbinder R, Osborne RH, Ebeling PR, et al. A randomized trial of vertebroplasty for painful osteoporotic vertebral fractures. N Engl J Med 2009;361:557-68. 5. Healthcare Cost and Utilization Project home page. Rockville, MD: Agency for Healthcare Research and Quality. (http:// hcupnet.ahrq.gov.) Copyright © 2011 Massachusetts Medical Society. payments, shared savings pro­ grams, and outcomes­based pay­ ments offer mechanisms for stim­ ulating the adoption of practices that are supported by CER and the abandonment of practices that CER calls into question. The ben­ efits for patients are large, as are the potential savings. Support for CER, reinforced by appropriate payment changes, is likely to rep­ resent a very good investment for the federal government and U.S. taxpayers. The views expressed in this article are those of the authors and do not necessarily n engl j med 364;15 nejm.org april 14, 2011 1393 The New England Journal of Medicine Downloaded from nejm.org at HARVARD UNIVERSITY on December 16, 2013. For personal use only. No other uses without permission. Copyright © 2011 Massachusetts Medical Society. All rights reserved.