|
|
| |||||||||||||||||||||||||||||||
Background Transcatheter implantation of the Rashkind PDA occluder is an alternative to conventional surgical closure of isolated patent ductus arteriosus. Neither the clinical outcomes nor the costs of these procedures have been formally compared.
Methods We performed a retrospective cohort study to evaluate the clinical outcomes within a seven-month period for comparable patients with patent ductus arteriosus who underwent either placement of an occluder or surgical closure. The patients were treated between 1982 and 1987 at 14 major North American centers where patent ductus arteriosus was closed predominantly by a surgical procedure or by the occluder technique. To estimate inpatient and follow-up costs, we multiplied the observed use of resources by 1989 unit costs based on hospital-accounting and physician-reimbursement data.
Results On the basis of cardiac auscultation at follow-up, the initial procedure resulted in closure of the ductus arteriosus in 77.3 percent of 185 patients in whom the occluder was implanted (95 percent confidence interval, 70.6 to 83.1 percent) and 99.8 percent of 446 surgical patients (95 percent confidence interval, 98.8 to 100.0 percent). Second procedures increased the percentage of successful closures to 87.6 percent (95 percent confidence interval, 81.9 to 92.0 percent) and 100.0 percent (95 percent confidence interval, 99.3 to 100.0 percent) for patients in the occluder and surgical groups, respectively. There were no deaths. Major complications occurred in 2.7 percent of the patients in whom the occluder was implanted (95 percent confidence interval, 0.9 to 6.2 percent) and 0.2 percent of the patients who underwent surgery (95 percent confidence interval, 0.0 to 1.2 percent); moderate complications in 16.8 percent (95 percent confidence interval, 11.7 to 22.9 percent) and 15.0 percent (95 percent confidence interval, 11.8 to 18.7 percent), respectively; and minor complications in 11.4 percent (95 percent confidence interval, 7.2 to 16.8 percent) and 24.9 percent (95 percent confidence interval, 20.9 to 29.2 percent). Including the cost of follow-up care, the mean estimated cost per case treated surgically was $8,838 (in 1989 U.S. dollars), as compared with $11,466 per case treated with the occluder technique. Sensitivity analyses based on our data identified no plausible situations in which the costs of surgery and of implantation of the occluder would be equal.
Conclusions The more effective and less costly surgical procedure was superior to transcatheter placement of the occluder for closure of isolated patent ductus arteriosus. Consequently, our results do not support the widespread dissemination of the occluder procedure for the management of this common congenital lesion.
|
In this retrospective cohort study, we enrolled patients admitted to 14 referral centers in the United States and Canada (see Appendix) between June 9, 1982, when a double-umbrella device was first used, and December 31, 1987. The study was restricted to patients for whom both implantation of the occluder and surgery were appropriate. Eligible patients were less than 19 years old at admission and had isolated patent ductus arteriosus without concomitant cardiac disease or clinically important noncardiac disease. Because of potential problems with the insertion of the sheath and with permanent implantation of the device,10 patients weighing less than 6 kg and those in whom the morphologic features of the patent ductus arteriosus were inappropriate -- that is, those for whom the occluder procedure was abandoned after diagnostic catheterization but before loading of the device because of unfavorable morphology and those in whom the minimal internal diameter of the ductus was more than 10 mm measured angiographically or the average external diameter, as noted at surgery, was more than 15 mm -- were ineligible. We also excluded patients in whose cases documentation of outcome might be incomplete, such as foreign nationals.
Patients who received the occluder were treated at the six U.S. and Canadian institutions at which at least 15 device placements had been attempted before 1988 ("occluder centers"). Although the choice of treatment reflected the physician's judgment, policies at the six occluder centers generally favored use of the device in eligible patients. To meet our requirements for sample size,11 the surgical group consisted mainly of patients treated at eight comparable referral centers -- classified as "surgical centers" -- that used surgery exclusively, were distant from the occluder centers, and were characterized by specific treatment approaches (for example, the use of surgical division in some centers). Surgical patients treated at the facilities classified as occluder centers were potentially eligible if they were operated on less than three years before the first use of the occluder at that institution and if the surgery was performed for nonmedical reasons (for instance, the parents' preference).
Approval was obtained from the institutional review board at Harvard University and at the individual centers if necessary. A single investigator was responsible for all on-site patient selection and data collection from August 1988 through May 1990. The eligibility of potential subjects identified through computerized screening, review of operative or cardiology-department logbooks, or both, was determined on the basis of referral letters and admission notes, generally without knowledge of the subsequent outcome. Although patients' treatment was known, criteria for patient selection and assessment of outcomes were applied as consistently as possible.
Treatment success was determined on the basis of the results of cardiac auscultation performed at the "target" follow-up contact -- the last visit to the cardiologist or cardiothoracic surgeon within seven months after the initial procedure. For patients without documented follow-up by a specialist, we used auscultatory findings from readmissions or from general pediatric or emergency room visits. Outpatient follow-up could not be documented in the chart or through supplementary phone calls to the referring physician in 12.4 percent of the cases treated by implantation of the occluder and 37.8 percent of the surgical cases. In these cases, we used the results of auscultation during hospitalization or, rarely, inpatient echocardiographic or angiographic findings.
Outcomes were evaluated for the initial procedure and for the overall treatment strategy, which included the initial procedure and any subsequent attempts at closure within seven months. Attempts at closure were classified as unsuccessful if continuous or diastolic murmurs were described at the target follow-up visit. Initial procedures were also considered unsuccessful if they were followed by separate closure procedures performed during the initial admission or the follow-up period.
Complications, defined as untoward consequences of the attempt at closure, were classified as major, moderate, or minor on the basis of the risk or discomfort posed by the complication or its treatment. They were distinguished from repeat closure procedures, separate diagnostic catheterizations, and consequences intrinsic to each type of treatment (for example, scarring at the site of the procedure). Trivial complications (such as transient radiographic abnormalities or discomfort) at the site of thoracotomy or occluder insertion were not analyzed. Follow-up care, including treatment provided at other sites, was included in the analysis only if it was noted at follow-up in the cardiology department.
Our cost-effectiveness analysis12,13 compared the costs of inpatient and outpatient care in relation to the clinical success of the two treatment strategies: implantation of the occluder and surgical closure. In this analysis we considered only the costs of medical care and not social costs. We itemized the costs of hospital care and care by the physician during the initial admission and the 7-month follow-up period for five categories of care: (1) all closure procedures performed, (2) management of complications and incidentally identified abnormalities, (3) all cardiology follow-up visits, surgical follow-up after the first visit, and all tests ordered during cardiology or cardiac-surgery visits, (4) visits to the emergency room or hospitalizations for any reason within 30 days after any closure procedures, and (5) diagnostic catheterization before or within 7 months after the closure procedure. We excluded inexpensive, frequently performed tests (such as measurement of serum electrolytes) and investigations that preceded both treatments (such as echocardiography). We did not consider opportunity costs associated with the need to have surgical backup available for occluder placement in some centers.
We used the cost-accounting system of Transition Systems, Inc. (TSI),14 to estimate average unit costs (including overhead) for components of hospital care (such as the hourly cost of operating room time) in undiscounted 1989 U.S. dollars. Data on unit costs were obtained for pediatric patients with patent ductus arteriosus who were treated surgically in 1988 and 1989 at the New England Medical Center, where the TSI system was developed. The occluder was not used at the New England Medical Center. Therefore, we used the TSI costs for the cardiac catheterization, angiography, and balloon dilation performed during analogous established procedures (for example, pediatric pulmonary valvuloplasty). We replaced the estimated costs of specialized valvuloplasty equipment (such as balloon catheters) with the 1989 U.S. purchase price for the occluder and its delivery system.
Unit-cost figures were available to estimate the costs of inpatient ward care by physicians, but not for the costs of physicians' performing specific procedures such as pulmonary valvuloplasty or ligation of patent ductus arteriosus. Therefore, we estimated these physicians' costs on the basis of maximal charges allowed in fiscal year 1988-1989 by Blue Cross and Blue Shield of Massachusetts. For the few physicians' services not specified in these data, we used 1988-1989 billed charges from the pediatric cardiovascular surgery, cardiology, and radiology groups at Children's Hospital (Boston) and New England Medical Center, adjusted for payer-specific collection rates. On the basis of Children's Hospital policies, we included one postoperative visit in the standard surgeon's fee for ligation or division and used this fee for the cost of surgery involving the retrieval of an occluder.
The frequency of outcomes, with exact 95 percent confidence intervals, was calculated15. To simulate total costs per patient, we multiplied the unit costs of various components of care by their observed use and then summed the products. For example, to calculate operating room costs, we multiplied hourly costs for operating room materials and facilities by the duration of the procedure. We then averaged costs for all patients cared for according to each treatment strategy. Sensitivity analyses16 performed with standard decision-analytic software17 were used to assess the implications of changing the frequency of specific clinical outcomes and patterns of care.
Results
Study Population
The age, weight, and clinical characteristics of the 185 patients in the occluder group and the 446 surgical patients we studied were similar (Table 1). Of 278 patients who underwent placement of a double-umbrella occluder from 1982 through 1987, 93 were deemed ineligible for the study because of concomitant cardiac disease (29 patients) or noncardiac conditions (5 patients), age or weight outside the eligibility criteria (26 patients), foreseeable absence of follow-up data (13 patients), inappropriate morphologic features of the patent ductus arteriosus (6 patients), or more than one of these reasons (14 patients). We did not compile information on the many neonates and other surgical patients excluded (often without review of their records) because they did not meet the study criteria.
|
Clinical Outcomes
The surgical approach produced more successful closures of a patent ductus arteriosus with fewer clinically important complications than did the use of the occluder (Table 2). There were no deaths and no ageor weight-specific variations in rates of successful closure in either treatment group. Five patients in the occluder group had the following major complications: a cerebral infarction documented by computed tomography with moderate residual neurologic sequelae; a persistent perfusion deficit documented by left-lung scanning after embolization of the device accompanied by an electrocautery burn that occurred during the surgical removal of the device; embolization requiring operative removal of the device from the right ventricle; Staphylococcus aureus bacteremia after closure, which was treated with intravenous antibiotics; and probable malignant hyperthermia, which was successfully managed. The only major surgical complication was a large chylothorax requiring operative repair. The overall treatment strategy was successful in all six patients who had major complications.
|
Cost-Effectiveness Analysis
The resource-use patterns we observed are summarized in Table 3. Table 4 describes the differences in the unit costs of major components of care. Table 5 summarizes the total costs associated with observed use of resources and outcome scenarios, resulting in mean costs of $8,838 per surgical case and $11,466 per occluder placement. In 1987, improved outcomes and increased use of outpatient procedures reduced the mean hospital stay to 1.9 days and the cost per occluder case to $10,679. For surgical patients, the mean length of the hospital stay fell to 5.0 days and the mean cost to $8,586.
|
|
|
Eliminating the five patients in the occluder group who had the most costly complications reduced the mean cost per case of implantation by only $219. The mean costs of successful outpatient occluder procedures still exceeded the mean costs of successful, uncomplicated surgical closures not preceded by catheterization by $1,941, and the costs of outpatient occluder placement exceeded the mean costs for the overall surgical strategy by $882. Therefore, increasing the frequency of outpatient occluder procedures would not eliminate the cost advantage of surgery.
Partly because of the requirements of the FDA, patients who received the occluder device underwent more extensive follow-up than did surgical patients. Therefore, we examined costs associated with a less intensive pattern of outpatient care (including one visit to a cardiologist, an electrocardiogram and chest film, but no echocardiogram) for all successful initial closure procedures. These costs were combined with those for all observed inpatient care and for care provided after initially unsuccessful closure procedures. This simulation increased the mean cost per surgical case to $8,924 and decreased that of occluder placement to $11,337.
We hypothetically required the use of surgery without preoperative catheterization, the least expensive option, to eliminate the residual flow from a patent ductus arteriosus that was still present after seven months in 23 patients in whom occluder implantation had been attempted. Increasing the success of the occluder strategy to the surgical figure of 100 percent allowed us to compare costs at equivalent rates of success. This simulation raised the mean cost per case of the occluder strategy to $12,553. The base-line figure of $11,466 reflects the observed strategy of leaving some occluder failures untreated during the observation period.
In multiway threshold analyses, we altered the probability of various outcomes in the scenarios described above. These simulations also required unrealistic combinations of scenarios to eliminate completely the base-line cost difference of $2,628. Consequently, the occluder was not preferable to surgery under any plausible combination of outcome probability and cost generated on the basis of our data.
Discussion
In this study, we investigated one instance of the increasingly common conflict between conventional treatment and a less invasive alternative. New procedures are infrequently evaluated in randomized trials,18 and the trials that are conducted rarely include cost analyses19. No randomized trial comparing the two strategies for closure of patent ductus arteriosus has been performed. Variations in or limited description of patient-selection procedures, outcome assessment, resource use, and costs have precluded meaningful comparisons of the results of published series of cases in which occluder implantation and surgery have been used.
Under some circumstances the results of observational studies may approximate those of randomized trials20. We sought to minimize bias21 and confounding22 through features of the study design that restricted entry to patients free of concomitant conditions that might influence the choice of treatment, outcome, assessment of closure, or costs. The rapid, almost universal adoption of the occluder device in some centers coincided with the continued exclusive use of surgery in comparable institutions. Thus, for a cohort of patients with isolated patent ductus arteriosus treated by placement of the occluder, we could identify prognostically similar surgical patients treated primarily at centers or during periods in which the occluder was unavailable.
Other reports23,24 support our finding that age, weight, and clinical presentation did not predict outcomes in otherwise healthy patients with patent ductus arteriosus. Consequently, minimal differences between the occluder and surgical groups in the distribution of these variables (Table 1) could not account for the observed differences in outcome. Data on the learning curve for the occluder device25 indicate that, over time, patients with more favorable morphologic features were increasingly referred for occluder placement. To the extent that the remaining patients, with larger-diameter ducts, were therefore treated surgically, the uniformly successful operative results described here are even more impressive. Since only one initial surgical ligation in our series was unsuccessful, a hypothetical subgroup of surgical patients matched with the occluder group for relevant covariates23 would have had essentially the same success rate as the surgical group as a whole.
Documentation of complications and inpatient use of resources in patients' charts was generally satisfactory, and the more extensive follow-up data available for patients who received the occluder device are unlikely to account fully for our finding less desirable outcomes in this group. Rates of closure based on cardiac auscultation do not reflect residual flow through a patent ductus arteriosus, which can be documented only by Doppler ultrasonography. Such inaudible shunting may persist after placement of the occluder or after surgery,10,24 but its clinical importance is unclear26,27. The turbulence that may contribute to the development of endarteritis should produce a characteristic murmur. Rates of closure based on auscultatory findings were more readily interpretable. Avoiding the use of Doppler results, which were inconsistently available, eliminated potential bias due to the use of tests performed selectively because of suspected abnormalities28. Incomplete follow-up precluded consideration of outcomes more than seven months after the initial procedure, but the effect of such later events on our conclusions is unclear.
The unexpectedly higher costs of the management strategy based on use of the occluder reflect several factors. The occluder is a costly device that must be implanted after a cardiac-catheterization procedure is performed with expensive imaging equipment; straightforward surgical closure incurs neither cost. Heretofore, policies of reimbursement to cardiologists have permitted separate billing for catheterization, angiography, interventional procedures (such as valvuloplasty), and follow-up, so that physicians' estimated costs for the placement of the occluder, based on these figures, exceeded those of surgery by a considerable margin. Furthermore, higher costs were incurred in treating the incomplete closures and clinically important complications that were seen more frequently in patients who had received the occluder device.
The results of our study changed little when we conducted sensitivity analyses involving different clinical outcomes and patterns of care, and our threshold analyses generally required unrealistic scenarios for use of the occluder to be favored. On balance, features of this study related to institutional accounting conventions and other issues tended to favor the occluder strategy. These factors all served to strengthen our conclusions.
In our analysis we did not equate hospital charges with costs29,30. Our figures are useful for comparative purposes, but they do not approximate the actual costs incurred by study patients. Instead, they reflect the average unit costs based on specific accounting conventions, reimbursement policies, and observed patterns of resource use that were in operation in 1988 and 1989. They are not necessarily applicable elsewhere (for example, in centers with catheterization laboratories used at maximal capacity). Even among similar referral centers, variations in hospital practices (such as the routine observation of stable patients in the intensive care unit after the procedure) within treatment groups did produce significant cost differences among institutions after uncomplicated procedures.
As is consistent with an analysis focusing on the costs of medical care, we did not explicitly consider the desirability of avoiding surgery or indirect economic costs (for example, for child care), which might be higher for postoperative recovery than for care following the implantation of the occluder device. However, in a companion decision analysis,31 we found that surgery was slightly favored over use of the occluder when physicians and parents compared outcomes observed over the entire study period.
Evaluating the outcomes of concurrently available treatment options requires comparing the results of an established procedure with those of an evolving alternative. Outcomes of occluder use may improve with accumulating experience and as the device itself, the procedure for implantation, and patient-selection criteria are refined32,33,34,35,36. Independent of outcome, costs associated with new procedures may decline over time37. In this case, costs should decrease with increasingly efficient provision of care -- for example, implanting the occluder as an outpatient procedure38 -- and if the allowable fees for the cardiologist fall in response to market pressures or changes in reimbursement patterns.
Nonetheless, our results indicate that, because of factors that include intrinsic differences in the use of resources, the appreciable cost advantage of surgery is unlikely to disappear even if outcomes with use of the occluder improve further. Surgical costs may decrease as length of stay and the need for preoperative catheterization decline and as modifications in technique39 reduce operative morbidity. Although relevant differences in unit costs changed little between 1989 and 1993, a substantial increase in the price of the occluder followed government approval of the device for use in Canada in 1989. Similar cost increases could occur in the United States if the FDA approves the device. The clinical value and cost effectiveness of implantation of the occluder as compared with surgical treatment for isolated patent ductus arteriosus have yet to be demonstrated. Despite the inherent promise of the new procedure, our results do not provide support for its widespread dissemination.
Supported by the Harvard Pharmacoepidemiology Teaching and Research Fund; by a grant (88-24) from the Hospital for Sick Children Foundation, Toronto, a grant (88-4-10) from the Alfred P. Sloan Foundation, and a grant (72-060-2842-2) from the Wellington Fund, Harvard Medical School; by a National Research Service Award (5T32 ESO7069-11/0021) from the National Institute of Environmental Health Sciences; and by a grant (1 R03 HS 06302-01) from the Doctoral Dissertation Grant Program of the Agency for Health Care Policy and Research.
We are indebted to Robert Beekman, M.D., M.P.H., of the Section of Pediatric Cardiology, University of Michigan-C.S. Mott Children's Hospital, Ann Arbor, Lee Benson, M.D., of the Division of Cardiology, Hospital for Sick Children, Toronto, and James Lock, M.D., of the Department of Cardiology, Children's Hospital, Boston, for their initial interest and support; to Frederick Mosteller, Ph.D., of the Harvard School of Public Health, and Lee Benson, M.D., for assistance in obtaining funding for the study; to the cardiologists, cardiac surgeons, and ancillary staff members at the collaborating centers for their cooperation; to Mary Anne Thadeu of the Department of Finance, New England Medical Center, Boston, and Thomas Byrne of Blue Cross and Blue Shield of Massachusetts, Quincy, Mass., for providing sensitive data on hospital costs and third-party reimbursement; to Frank Ho, Colin Joe, and Murray MacKinnon of the University of British Columbia, Vancouver, for computing assistance; to James Lock, M.D., for valuable comments and criticism; to Alice McKinney of the Mayo Clinic, Rochester, Minn., for the preparation of the figure; and to Sondra Buehler and Karen Tennison of the Mayo Clinic for assistance in the preparation of the manuscript.
Source Information
From the Department of Health Sciences Research, Mayo Clinic, Rochester, Minn. (D.T.G.); the Department of Cardiology, Children's Hospital, Boston (D.C.F.); and the Departments of Epidemiology (D.T.G., A.M.W.), Biostatistics (M.C.W.), and Health Policy and Management (M.C.W., T.C.C.), Harvard School of Public Health, Boston. Investigators and institutions participating in the Patent Ductus Arteriosus Closure Comparative Study Group are listed in the Appendix.
Address reprint requests to Dr. Gray at the Department of Health Sciences Research, Mayo Clinic, 200 First St. S.W., Rochester, MN 55905.
References
The following investigators and centers participated in this study: Occluder centers -- L. Benson, Division of Cardiology, Hospital for Sick Children, Toronto; W. Hellenbrand, Department of Pediatric Cardiology, Yale University School of Medicine, New Haven, Conn.; L. Latson, Department of Pediatric Cardiology, University of Nebraska Medical Center, Omaha; J. Lock, Department of Cardiology, Children's Hospital, Boston; C. Mullins, Section of Cardiology, Texas Children's Hospital, Houston; and J. Murphy, Department of Cardiology, Children's Hospital of Philadelphia, Philadelphia. Surgical centers -- J. Bass, Department of Pediatric Cardiology, University of Minnesota Hospital, Minneapolis; R. Beekman, Section of Pediatric Cardiology, University of Michigan Hospitals, Ann Arbor; J. Finley, Department of Cardiology, Sir Isaak Walton Killam Hospital for Children, Halifax, N.S.; C. Hardy, Department of Cardiology, Children's Hospital of Northern California, Oakland; G. Sandor, Department of Cardiology, British Columbia Children's Hospital, Vancouver; S. Stamm, Department of Cardiology, Children's Hospital, Seattle; P. Stanger, Department of Pediatric Cardiology, University of California-San Francisco Medical Center, San Francisco; and R. Williams, Department of Pediatric Cardiology, UCLA Medical Center, Los Angeles. D. Fulton, Division of Pediatric Cardiology, Floating Hospital for Infants and Children, Boston, provided data on unit costs.
| |||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
