Controlling the escalating cost of health care is a major issue for physicians. Voluntary practice guidelines, financial incentives, clinical restrictions, benchmarking (providing information on the practice patterns of peers), and educational efforts have had variable success in changing prescribing behavior and controlling costs associated with physicians' clinical decision making [1-5]. Techniques from industry quality management science, such as total quality management, are among the newer approaches to managing the delivery of health care. Clinical pathways, also known as "care pathways" or "critical pathways," are an application of industrial quality management science to health care. Hospitals have begun introducing clinical pathways to cut costs and to reduce variation in care that does not influence outcome [6,7]. As perioperative physicians, anesthesiologists may be asked to participate in pathway development, thereby producing value for the health care system .
Clinical pathways outline a recommended sequence and timing of interventions by surgeons, anesthesiologists, nurses, and other staff for the production of care for a group of patients with a particular diagnosis or procedure. Unlike practice parameters, which are generated by governmental agencies or national specialty societies, clinical pathways are developed by a multidisciplinary group of providers to create an optimal regimen of care in their own institution. Clinical pathways address variability in practice (e.g., use of hospital support systems and resources for patients with similar conditions) by having providers agree prospectively on a common regimen of clinical interventions.
The effect of clinical pathways on the economics of patient care and surgery is largely unknown, although a few studies have been completed. Clinical pathways have decreased hospital costs for patients admitted with chest pain  and decreased hospital and intensive care unit lengths of stay for patients treated for congenital heart disease . The goal of this study was to determine whether implementing a perioperative clinical pathway for patients undergoing knee replacement reduces total hospital costs.
This prospective trial with historical and concurrent control cohorts was approved by the Stanford University Medical Center (SUMC) human subjects committee. All patients underwent surgery at SUMC, a 660-bed, university- and community-affiliated, tertiary care medical center with 22,000 surgical cases per year.
We enrolled all patients who underwent elective total knee replacement (International Classification of Diseases, 9th rev. ed., Clinical Modification [ICD-9-CM] procedure 81.54) between September 11, 1995 and March 31, 1996. These patients were cared for under a new clinical pathway described below. The patients undergoing knee replacement were operated on by one of four orthopedic surgeons. We compared total hospital costs in this prospective cohort with a historical control group of patients undergoing the same procedure (but without the clinical pathway) by the same four surgeons between September 1, 1994 and August 31, 1995. The prospective study of knee replacement patients ended when, based on a preliminary study of 34 patients, enough patients (n = 63) were enrolled to give the study 90% statistical power (with a type I error of 5%) to detect a 10% change in total hospital costs. All patients undergoing knee replacement surgery were included in each of the two (before and after) groups. Patients undergoing bilateral knee replacement, revision of a knee replacement, or unicompartmental knee replacement were excluded from the study.
To control for changes in cost levels (inflation or deflation) and physician behavior not related to the clinical pathway, we studied patients undergoing surgical procedures that did not have a clinical pathway. These control groups consisted of all patients who underwent either a) total hip replacement (ICD-9-CM procedure 81.51) with the same surgeons as the knee replacement group or b) radical prostatectomy (ICD-9-CM procedures 60.4 and 60.5) with different surgeons than for the knee replacement cohort during similar time periods (September 1, 1994 to August 31, 1995 and September 11, 1995 to March 31, 1996) as the study patients undergoing knee replacement surgery. During the study periods, there were no initiatives to change aspects of clinical management of patients undergoing prostatectomy or hip replacement surgery.
Development of a New Clinical Pathway (the Intervention)
SUMC developed a perioperative clinical pathway for knee replacement surgery because a preliminary cost study of patients undergoing knee replacement suggested that a) the mean hospital cost for patients having a knee replacement was greater than the mean reimbursement and b) hospital costs and length of stay for these patients were greater than benchmarks in the literature and in the community .
A detailed description of the pathway development has been previously reported . A multidisciplinary team, consisting of orthopedic surgeons, anesthesiologists, and nurses, developed detailed flow charts of treatment processes for patients undergoing knee replacement. The team collected clinical and cost data on existing practice patterns, ranging from preoperative testing to postoperative care, and presented these data to other team members to initiate discussion on practice patterns. Experts in related specialties joined the team on an ad hoc basis to provide special expertise (e.g., transfusion service regarding preoperative autologous blood donation). The final pathway required consensus by team members on a written, reproducible, day-by-day clinical care plan. The pathway defined multiple interventions in the perioperative period (Table 1). For instance, patient care and time guidelines within the operating room were clearly defined. The new clinical pathway incorporated a preexisting postoperative nursing protocol. After the pathway was initiated, monthly meetings were held to assess performance and to make adjustments.
The medical records of patients were abstracted to obtain demographic data, including age, gender, and type of insurance. We also recorded the patient's severity of coexisting disease as measured by using the ASA physical status, the principal diagnosis responsible for the hospitalization and surgery, and the hospital length of stay.
To measure changes in cost patterns before and after pathway implementation, we computed total hospital costs for each patient. We examined actual hospital costs instead of patient charges because charges are an inaccurate measure of hospital resource use . Cost data were extracted from the hospital's existing data management systems (e.g., materials management, patient billing, laboratory, financial) to determine actual supply costs, compensation rates, and labor effort as measured by hospital department managers. The institution's cost management and decision system software (Transition Systems Inc., Boston, MA; [TSI]) combines clinical and financial data to determine actual costs of patient care. TSI cost standards are calibrated to actual cash expenses on a quarterly basis. TSI has become a well accepted standard tool for hospital cost management and has been used in analyses published by ourselves and investigators at other institutions [14,15]. Physician professional costs were not included in our analysis.
We also examined changes before and after implementation of the clinical pathway in cost centers directly affected by the pathway, such as the patient ward (length of hospital stay) and the operating room (OR). Total hospital costs were the sum of fixed and variable costs attributed to 11 hospital departments: patient ward, radiology, OR, anesthesia, pharmacy, postanesthesia care unit (PACU), surgical admission unit, intensive care unit, laboratory, and blood bank. For example, OR costs included facility costs associated with OR time necessary for each case and costs related to labor and supplies (e.g., staples, drapes, sutures, electrocautery, knee implants, bladder catheters, cement, drains, splints, saw blades) involved in producing a specific patient service .
Cost analyses can be performed from multiple viewpoints. In this study, measurement of costs for pathway development and implementation was performed from the hospital's perspective. We recorded attendance at organizational meetings to calculate labor (physician and nursing) costs of pathway development. Hourly wages were based on actual compensation rates including benefits. We included only actual expenses paid by our institution (e.g., nurse time dedicated to pathway development) and excluded the cost of physician time not paid directly by the hospital.
Statistics and Data Management
To evaluate the possibility that the composition and characteristics of the patients undergoing knee replacement surgery changed over time, we compared patient demographic factors, severity of illnesses, and surgeon mix in both the historical control group and the prospective intervention group.
Changes in costs were calculated by using percent changes in mean costs. The percent decrease in the total hospital costs of patients undergoing knee replacement was compared statistically with that of patients undergoing radical prostatectomy and hip replacement. Confidence intervals and the median relative changes were found with the use of the bootstrap percentile method, using 4000 simulations . Changes in variations in costs were calculated by using percent changes in coefficient of variation of costs.
The three groups were similar with regard to demographics, severity of illness as measured by the ASA physical status, and surgeon mix (see Table 2 for patients undergoing knee replacement). We studied 183 patients undergoing knee replacement: 120 in the historical control group and 63 in the clinical pathway intervention group.
Hospital costs for knee replacement surgery decreased from $21,709 to $17,618 after implementation of the clinical pathway (Table 3). The percent decrease in hospital costs was 1.56-fold (95% confidence interval 1.02-2.28) greater in patients undergoing knee replacement surgery than radical prostatectomy, and 2.02-fold (95% confidence interval 1.13-5.22) greater than patients undergoing hip replacement. Results of cost analyses may be overly sensitive to the effects of a few costly patients. We therefore repeated the analyses after trimming (i.e., deleting) the values for one and then two of the most costly patients from each of the groups. Thus, the percent decreases in hospital costs are 1.62-fold (95% CI 1.17-2.31) (one patient trimmed) and 1.62-fold (95% CI 1.15-2.30) (two patients trimmed) greater in the patients undergoing knee replacement than radical prostatectomy and 2.01-fold (95% CI 1.21-4.79) (one patient trimmed) and 2.05-fold (95% CI 1.24-4.35) (two patients trimmed) greater than in the patients undergoing hip replacement.
Of the total hospital cost savings after pathway implementation, 54% resulted from decreasing OR costs by 22%. OR costs decreased because of a mean reduction of 18 min in operative times, as well as widespread reductions in resource utilization (e.g., instruments, blood retrieval systems). Of the total hospital cost savings after implementation of the clinical pathway, 16% resulted from patient ward costs (length of hospital stay decreased by 15%) decreasing by 15%. The other 30% of savings occurred in pharmacy (8%), physical therapy (6%), blood bank (4%), anesthesia (3%), PACU (2%), radiology (3%), and miscellaneous (4%). Although anesthesia and PACU costs were not primary end points, mean (+/- SD) anesthesia costs decreased from $336 +/- $132 to $240 +/- $95, and PACU costs decreased from $357 +/- $115 to $314 +/- $118 (P < 0.02). Coefficients of variation of total hospital costs, operating room costs, and length of hospital stay were reduced 56%, 42%, and 57%, respectively.
We studied 332 patients undergoing radical prostatectomy: 168 in the historical control group and 164 contemporaneously with the prospective data collection on patients undergoing knee replacement surgery. During the period of pathway implementation for knee replacement surgery, total hospitalization costs for patients undergoing prostatectomy decreased 12%, although there were no initiatives to change management of prostatectomy patients (Table 4). No one cost center accounted for the majority of this decrease. OR costs decreased 6%. Length of hospital stay did not change. Coefficients of variation of total hospital costs, OR costs, and length of hospital stay decreased by 0%, -8%, and 5%, respectively.
We studied 251 patients undergoing total hip replacement surgery: 144 in the historical year and 107 in the same time period as patients undergoing knee replacement surgery. Total hospital costs, OR costs, and length of stay decreased during the period of knee replacement pathway implementation by 13%, 10%, and 17%, respectively, although there were no initiatives to change management of hip replacement patients (Table 5). Coefficients of variation of total hospital costs, OR costs, and length of hospital stay decreased 24%, 4%, and 60%, respectively.
The pathway cost the hospital $21,350 to develop (Table 6). This did not include physician time that was not paid directly by the hospital.
Before developing clinical pathways, our institution's attempts to control patient care costs with case management programs, postoperative nursing protocols, and quality improvement initiatives for clinical support systems yielded limited cost savings. This may have been because clinical processes involving physician decision making were unaffected. As perioperative physicians, we were interested in studying the economic results of our institution's first perioperative clinical pathway, which provides an integrated plan for physician-directed interventions for patients undergoing knee replacement surgery.
The clinical pathway reduced total hospital costs by 19% per patient undergoing knee replacement. This was a 1.6-fold decrease compared with the concurrent control cohort of patients undergoing prostatectomy and a 2.0-fold decrease compared with hip replacement surgery without a clinical pathway. Most of the cost reduction for knee replacement surgery occurred because of shorter lengths of stay in the OR and in the hospital. The pathway specifically addressed physician-managed resource utilization in the OR because this is where most hospital costs accrue for surgical services . Costs for hip replacement and prostatectomy surgery also decreased during the study period (but to a lesser extent), probably from overall institutional cost control efforts related to marketplace pressures. Based on this experience and assuming outcomes are constant, we believe that anesthesiologists, surgeons, and hospitals can use clinical pathways for incremental improvements in the cost of hospital care for other surgical inpatients.
Before we implemented our knee replacement pathway, there were no consistent, standardized processes of care against which to measure deviations in process or cost. Clinical pathways facilitate multidisciplinary communication, data collection, data analysis, and feedback to providers. Similar cost reduction strategies have led to simplification and standardization of production processes, eliminating waste and duplication of effort. It seems that this, too, is achievable in medicine.
Shorter OR times resulting from pathway implementation may yield budgetary savings only if variable costs are reduced. We have previously shown that 44% of OR costs are variable . Hourly compensation of nurses at our institution permits OR time savings to reduce costs because nursing personnel are not paid when surgery is not being performed. In hospitals in which this does not occur, cost savings may not be realized from time savings achieved via a clinical pathway.
Work at Intermountain Health Care, Salt Lake City, UT suggests that teams of physicians, nurses, and other staff have little knowledge of the practice of others . Clinical pathways are difficult to design and implement because physicians' practice patterns vary . Although the pathway for knee replacement surgery was a hospital initiative, physicians actively participated with other providers in the process. Clear, scientifically grounded, continuously reviewed treatment strategies (i.e., clinical pathways) are essential to measure and reengineer the processes of care .
The pathway may also incrementally reduce variation in hospitalization costs for knee replacement surgery (i.e., by 32% [coefficient of variations decreased 28% to 18% for knee surgery versus 31% to 25% for hip surgery]). Variation in practice both among providers, and even by the same provider, may adversely affect quality. Reducing unnecessary variation in care may be increasingly important in a capitated system in which the financial risks of patient care are transferred to the provider. In this setting, although most of the risk may depend on the numbers of procedures performed, clinical pathways may reduce the portion of financial risk due to provider variability.
Many physicians react to pathways with skepticism. Unlike the manufacturing industry, physicians in most hospitals are not employees. Physicians question the effects of clinical pathways on the costs of care, and may also fear that pathway emphasis on reducing variation will threaten their autonomy, their ability to vary care to meet the needs of particular patients, and their ability to innovate. However, more than a year after pathway adoption, physicians are cooperating because the pathway represents a consensus on the best current method for a more uniform, efficient process. Physicians override the pathway as necessary, and variations from specified interventions become the basis for updating the pathway.
The costs of pathway development were not allocated to patient management costs. Who and what is needed for pathway development may depend on the surgical procedure and the hospital. In our case, a physician leader trained in total quality management, as well as nursing time and support staff, were necessary. The cost savings related to pathways may depend on performing a minimal (or threshold) number of procedures. Clinical pathways are most likely to yield cost savings for high-volume, high-risk procedures. For instance, with the measured incremental savings of knee replacement relative to hip replacement surgery, had we performed only a few knee replacements, developing the pathway would have cost more money than it saved. Studies must ascertain under which conditions pathways are likely to reduce costs.
We performed the cost analysis of pathway development from the perspective of the hospital. Because the taxonomy used to describe costs affects the results of economic evaluations of health care interventions , a different methodology for measuring and quantifying costs may have yielded different results. However, the results are reasonable and expected. Although the hospital did not pay an incremental amount for physician involvement in pathway development, physicians may benefit from pathways if this increases physicians' ability to acquire contracts for patient care.
Cost containment strategies need to control the rising costs of technologies such as the knee prosthesis . The present analysis did not include the effects of a competitive bid process for knee prosthesis vendors that was instituted after pathway development. Implant systems may comprise up to 25% of the total hospital cost for knee replacement surgery . To reduce knee replacement surgery costs, we recommend that hospitals focus on implant costs, OR time and resource utilization, and length of hospital stay.
In addition to the monetary savings, there may be nonmonetary benefits of pathway implementation that are difficult to quantify. Knowledge by providers of quality improvement principles and of costs (e.g., operating room time, supplies) may have an impact on other hospital procedures. For example, we observed an overall decrease in hospital costs (less than that for knee replacement patients) for the two surgeries that did not have a pathway.
Whereas practice parameters are produced by governmental agencies or national specialty societies, clinical pathways are completed by a group of providers from one hospital to create a recommended regimen of care. Clinical pathways standardize practice in the unique culture and environment of individual hospitals. Whether the pathway developed at our institution can be applied directly to another institution to reduce cost requires further study. We believe that the active participation by providers in pathway development is necessary for cooperation with and adherence to the pathway, although the experience of other institutions can be helpful.
Although a prospective, randomized, clinical trial (RCT) is the "gold standard" for evaluating new interventions, they are expensive and may not apply to the general population. A process intervention such as a clinical pathway is more difficult to study using an RCT design because the nonpathway group may be "contaminated" by pathway development processes. We studied clinical pathways in the actual, routine care of the general population, excluding complicated surgical procedures. We did not choose an RCT design because it was not practical to randomize patients or to blind providers as to whether a patient was being treated under the clinical pathway or the control group. We also did not include patient satisfaction or outcome measures in the control and intervention groups.
The next generation of clinical pathways may address the management of the broader continuum of patient care from the initial office encounter through hospitalization and the postdischarge period. Future studies must determine whether cost reductions from pathway implementation are applicable to the outpatient setting and whether the observed decrease in inpatient costs resulted in poorer posthospitalization outcomes (e.g., an increase in complication rates or in unscheduled medical clinic visits) or higher outpatient costs. We are now evaluating the posthospitalization period to measure patient outcome and satisfaction using measures of quality of life such as the Medical Outcomes Study Short Form - 36 . Both generic and disease-specific health-related quality of life measures are necessary to assess patient outcomes related to knee replacement . Based partly on this study's results, our institution is developing perioperative pathways for spine surgery, coronary artery bypass grafting, and intracranial aneurysm clipping.
1. Schroeder S, Myers L, McPhee S, et al. The failure of physician education as a cost containment strategy: report of a prospective trial at a university hospital. JAMA 1984;252:225-30.
2. Johnstone R, Jozefczyk K. Costs of anesthetic drugs: experiences with a cost education trial. Anesth Analg 1994;78:766-71.
3. Grumbach K, Bodenheimer T. Mechanisms for controlling costs. JAMA 1995;273:1223-30.
4. Lubarsky D, Glass P, Ginsberg B, et al. The successful implementation of pharmaceutical practice guidelines. Anesthesiology 1997;86:1145-60.
5. Cohen M, Rose K, Yee D. Changing anesthesiologists' practice patterns: can it be done? Anesthesiology 1996;85:260-9.
6. Laffel G, Blumenthal D. The case for using industrial quality management science in health care organizations. JAMA 1989;262:2869-73.
7. Coffey R, Richards J, Remmert C, et al. An introduction to critical paths. Qual Manage Health Care 1992;1:45-54.
8. Saidman L. The 33rd Rovenstine Lecture: what I have learned from 9 years and 9000 papers. Anesthesiology 1995;83:191-7.
9. Weingarten S, Riedinger M, Conner I, et al. Practice guidelines and reminders to reduce duration of hospital stay for patients with chest pain: an interventional trial. Ann Intern Med 1994;120:257-63.
10. Turley K, Tyndall M, Roger C. Critical pathway methodology: effectiveness in congenital heart surgery. Ann Thorac Surg 1994;58:57-65.
11. Healy W, Finn D. The hospital cost and the cost of the implant for total knee arthroplasty. J Bone Joint Surg Am 1994;76:801-6.
12. Horne M. Involving physicians in clinical pathways: an example for perioperative knee arthroplasty. Joint Commis J Qual Improve 1996;22:115-24.
13. Finkler S. The distinction between cost and charges. Ann Intern Med 1992;96:102-9.
14. Macario A, Vitez T, Dunn B, McDonald T. Where are the costs in perioperative care? Analysis of hospital costs and charges for inpatient surgical care. Anesthesiology 1995;83:1138-44.
15. Mark D, O'Neill W, Brodie B, et al. Baseline and 6 month costs of primary angioplasty therapy for acute myocardial infarction: results from the primary angioplasty registry. J Am Col Cardiol 1995;26:688-95.
16. Dexter F, Macario A. Application of information systems to operating room scheduling. Anesthesiology 1996;85:1232-4.
17. Mooney C, Duval R. Bootstrapping: a nonparametric approach to statistical inference. Newbury Park, CA: Sage University Press, 1993:36-7.
18. Classen D, Evans R, Pestotnick S, et al. The timing of prophylactic administration of antibiotics and the risk of surgical wound infection. N Engl J Med 1992;326:281-6.
19. Wennberg J, Gittelsohn A. Small area variations in health care delivery. Science 1983;182:1102-8.
20. Berwick DM. Continuous improvement as an ideal in health care. N Engl J Med 1989;320:53-6.
21. Hlatky M, Lipscomb J, Nelson C, et al. Resource use and cost of initial coronary revascularization. Circulation 1990;82(Suppl IV):208-13.
22. Schwartz W. The inevitable failure of current cost-containment strategies. JAMA 1987;257:220-4.
23. Zuckerman J, Kummer F, Frankel V. The effectiveness of a hospital based strategy to reduce the cost of total joint implants. J Bone Joint Surg Am 1994;6:807-11.
24. McHorney C, Ware J, Lu R, et al. The MOS 36-item short from health survey (SF-36). III. Tests of data quality, scaling assumptions, and reliability across diverse patient groups. Med Care 1994;32:40-66.
25. Bombardier C, Melfi C, Paul J, et al. Comparison of a generic and a disease-specific measure of pain and physical function after knee replacement surgery. Med Care 1995;33:AS131-44.