| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | |
Commentary & Perspective | ||||||||
Commentary & Perspective on Over 475,000 total hip and knee replacements are performed annually in the United States at a cost exceeding ten billion dollars1. Population-based surveys of patients who have arthritis of the hip or knee document a large untapped need for these procedures2, suggesting that the rates of total knee and hip arthroplasty should and likely will increase further. Two- thirds of these procedures are insured by Medicare, which provides hospitals with capitated, prospective payment. Thus, hospitals have strong incentives to control the costs of total hip and knee replacement. Surgeons also often have financial incentives to reduce hospital costs. Dr. Healy and colleagues have worked at the cutting edge of research on containment of the costs of total hip and knee replacement. These investigators developed the Lahey Clinic knee-implant standardization program, based on a patient-type scoring system that is used to evaluate the physical demand that a patient is expected to place on the implant. The most expensive, sophisticated implants are suggested for patients who are likely to place the greatest demands on the prosthesis. Healy and colleagues also developed a clinical pathway program for total knee replacement. Clinical pathways (sometimes called critical pathways or care maps) are treatment algorithms developed to standardize the sequence and timing of the various stages of patient care. Pathways are intended to minimize variations in the process of care and thereby reduce costs while maintaining or enhancing quality3. During the 1990s, hospitals nationwide developed and implemented critical pathways for a wide range of high-cost procedures associated with high rates of prevalence. In the current issue of The Journal, Healy and colleagues evaluated whether these two programs—knee-implant standardization and a clinical pathway—influenced a range of economic and clinical outcomes of total knee replacement at the Lahey Clinic. The study compares a cohort of 103 patients who had a total knee replacement in 1995, when both programs were operational, with a control cohort of fifty-six patients who had a total knee replacement in 1992, before the two programs were implemented. The epidemiologic term for this type of comparison group is "historical controls," since the patients in the control group were treated well before those in the intervention group. In comparison with the 1992 historical controls, the 1995 cohort had a lower average length of hospital stay (4.2 versus 6.8 days in 1992), lower mean inflation-adjusted hospital costs ($8747 versus $10,805), and 35% lower implant costs. The two cohorts had similar clinical outcomes, evidenced in the assessment that included follow-up clinical knee scores, range of motion, and the need for revision or manipulation under anesthesia after the replacement. Thus, costs and the average length of hospital stay were reduced with no significant change in clinical outcomes. Much work aimed at quality improvement and cost containment is never evaluated rigorously. Healy and colleagues should be applauded for taking the high road and evaluating the success of their efforts to reduce costs and improve quality. Evaluation of these programs presents a range of methodological challenges. The fundamental problem is that programs designed to ensure quality improvement are implemented at the level of the hospital, not at that of the patient. Therefore, one cannot simply adapt the clinical trial paradigm and randomly allocate half of the patients undergoing total knee replacement at one hospital during a certain time period to an intervention, such as the clinical pathway or the knee-implant standardization program, and use the other half as controls. If the interventions are successful, providers will tend to treat all patients in a more standardized manner, and contamination (the spillover of intervention effects into the control group) would be rampant if patients in the same hospital were randomized to intervention and to control groups. A practical alternative is the use of historical controls, which is the approach chosen by Healy and colleagues. The problem with historical controls is that changes that are independent of the intervention may have occurred between the control period and the intervention period. The length of hospital stay for total joint replacement procedures decreased from a mean of 6.3 days in 1993 to 4.2 days in 1999 in hospitals across the country1. In the Boston area, the length of hospital stay for patients undergoing a total knee replacement was reduced by 25% over a two-year period in the mid 1990s4. Thus, even without use of the clinical pathway, the average length of hospital stay would probably have diminished considerably at the Lahey Clinic. While it seems likely that the knee-implant standardization program achieved cost savings that would not have occurred otherwise, this conclusion would also be strengthened considerably with supporting data from a control group. Examples from the literature in this area underscore these concerns. Quality improvement projects launched in cardiovascular surgery programs in northern New England and in New York in the late 1980s and early 1990s were associated with annual reductions of 3% to 7% in patient mortality following coronary artery bypass surgery. However, an analysis of data from Massachusetts hospitals (not involved in the northern New England project) demonstrated reductions in patient mortality following coronary artery bypass surgery similar to those documented in northern New England and New York5. In the same vein, a hospital that introduced a clinical pathway for total knee replacement experienced reductions in the length of postoperative hospital stay identical to those at three control hospitals that did not implement clinical pathways4. The use of concurrent controls (subjects who receive the control treatment at the same time that the experimental subjects receive the study treatment) poses a different set of challenges in data evaluation. As mentioned earlier, assigning some patients to an intervention group and others in the same institution to a control group invites contamination. Study protocol could include randomization of hospitals that implement cost-reduction interventions with those who do not, but it would be difficult to distinguish the effects of intervention from fundamental differences in institutional practices and culture that may drive costs and other outcomes. Adjustment could make the comparisons more valid, but some factors are hard to quantify and adjust for, including the underlying institutional concern for cost and patient safety. These observations suggest two somewhat contradictory conclusions. First, efforts to improve quality of care should be evaluated with use of controlled studies, as uncontrolled observations may reflect only secular change. On the other hand, the protocols of controlled trials are susceptible to contamination and results may be confounded by differences in institutional culture that are difficult to measure. In short, the research agenda is methodologically daunting. That said, we should applaud efforts, such as the work of Healy and colleagues, to bring science to bear on the evaluation of quality improvement, and we should encourage further research aimed at overcoming the methodological problems inherent in this important work. *In support of his research or preparation of this manuscript, the author received grants or outside funding from NIH (NIAMS) [P60 AR 47782 and K24 AR 02123] and the Arthritis Foundation. The author did not receive payments or other benefits or a commitment or agreement to provide such benefits from a commercial entity. No commercial entity paid or directed, or agreed to pay or direct, any benefits to any research fund, foundation, educational institution, or other charitable or nonprofit organization with which the author is affiliated or associated. References 1. HCUPnet, Healthcare Cost and Utilization Project. Agency for Healthcare Research and Quality, Rockville, MD. http://www.ahrq.gov/data/hcup/hcupnet.htm (accessed 12/26/2001; 2/08/2002). | ||||||||
| ||||||||