Research Agenda for Pediatric Gastroenterology, Hepatology and Nutrition: Transplantation: Report of the North American Society for Pediatric Gastroenterology, Hepatology and Nutrition for the Children's Digestive Health and Nutrition Foundation

McDiarmid, Suzanne V.; Bucuvalas, John C.; Cox, Kenneth L.; Freese, Deborah K.; Rosenthal, Philip; Vanderhoof, Jon A.

Journal of Pediatric Gastroenterology & Nutrition:
Author Information

Los Angeles


Palo Alto

Rochester, MN

San Francisco


Address requests for reprints to: Executive Director, Children's Digestive Health and Nutrition Foundation, PO Box 6, Flourtown, PA, 19031, U.S.A. (e-mail:

Article Outline
Back to Top | Article Outline


In 1998, a total of 522 liver transplants, or approximately 12% of such operations in the US, were performed in patients younger than 18 years of age (1). For these children as for adults who underwent liver transplantation, 1-year survival rates approached 85%, the result of improvements in surgical technique, immunosuppression and antiviral therapy.

The findings of studies in adult populations cannot be generalized to children because of differences in the indications for transplantation, as well as differences in surgical, infectious and developmental complications. A number of factors hamper pediatric studies. Even at the largest centers, only 30 to 40 children undergo transplantation each year and the population is heterogeneous. Consequently, individual transplant centers do not care for populations of sufficient size to identify interventions that improve outcome. Furthermore, with pharmacotherapeutic and surgical advances, the standard of care has evolved. The interpretation of studies performed over time at any single center is subject to the biases introduced by changes in care practices.

Three major areas of research are considered of primary importance: tolerance induction, evaluation of outcome after liver transplantation, and post-transplant lymphoproliferative disease. In addition, there are three major research areas that may be considered a rank below: nonimmune graft injury, intestinal graft rejection and hepatocyte transplantation.

Back to Top | Article Outline


Evaluate Strategies to Induce Tolerance
Research Goals

Tolerance is classically defined as donor-specific immuno-nonresponsiveness and is manifest by long-term allograft function, without evidence of immunologic injury, maintained in the absence of immunosuppression (2). Full immunoresponsiveness to non-donor-derived antigens is preserved: animal studies have demonstrated the ability to fully reject a graft from a different donor as well as acceptance, without immunosuppression, of a second graft from the original donor. To date, true tolerance has been achieved in some small rodent models, but it is proving difficult to achieve consistently in large primate models.

Current immunosuppressive strategies are nonspecific and, as such, associated with significant long-term risks of malignancy and infection. The nonimmunologic toxicities of current therapeutic modalities are substantial. With prolonged use, debilitating and life-threatening complications, including nephrotoxicity, neurotoxicity, bone disorders and cardiovascular disease, may occur. This has spurred research efforts to explore the mechanisms and clinical applications of tolerance induction. Furthermore, the problem of chronic rejection, which most likely is mediated by immune pathways different from those of acute rejection, has not been prevented by current immunosuppressive regimens (3).

With successful tolerance induction strategies, long-term immunosuppressive drugs can be avoided. Such avoidance is of particular urgency in pediatric transplant recipients, who currently face the prospect of many decades of immunosuppressive drug use. Efforts to decrease long-term immunosuppression, or even consider discontinuation of immunosuppressive agents, are severely handicapped by lack of a reliable test to measure the recipient's degree of immunoresponsiveness to the graft. Without such a tolerance assay, random discontinuation of immunosuppressive therapy in stable patients is fraught with uncertainties.

Back to Top | Article Outline
Research Goals

The principal mechanisms of tolerance—anergy or depletion of alloreactive T cells, immuno-regulation and chimerism—are now providing promising strategies for clinical application. T cells are an absolute requirement for the rejection response. Cell surface markers that are capable of initiating co-stimulation are prime targets for new blockade approaches to modulate the immune response. Monoclonal antibodies to these targets are being developed, many of which are humanized to avoid induction of a neutralizing human antibody response.

To advance this critically important field, the following areas of research must be targeted and supported:

Increased application of immunomodulatory strategies to induce tolerance in large primate models

Clinical trials in pediatric populations. Even early human trials of tolerance-inducing strategies must include pediatric recipients, while recognizing that the immune response and its regulation may be different in children

Development of biologic markers to measure donor-specific immune response. Such markers will not only assess outcome in studies designed to induce tolerance but also, in the short term, provide important information for the tailoring of current immunosuppressive regimens to individual patients, to avoid over- and under-immunosuppression

Back to Top | Article Outline
Research Strategies

Critical to these efforts are basic science investigations that will further our understanding of T-cell signaling and activation. Clinical trials will require the participation of multiple transplant centers.

Back to Top | Article Outline
Projected Timetable and Funding Requirements

Basic science and clinical research initiatives to achieve tolerance will require a substantial and ongoing financial commitment. These initiatives provide an ideal opportunity for partnering between federal agencies and private groups, including the pharmaceutical industry. This research area is an extremely important frontier for young investigators.

The infrastructure needed to support such studies includes data-gathering and analysis mechanisms to allow multicenter trials to be conducted. Funding for such collaborative multicenter databases, specific to the special outcomes and requirements of pediatric liver and intestinal transplant recipients, is critical.

Back to Top | Article Outline
Prospectively Evaluate Outcome Measures After Pediatric Liver Transplantation

Advances in immunosuppressive therapy and surgical techniques have improved graft and patient survival rates as well as expanded access to donor organs for pediatric liver transplant recipients. The improved rates have, in turn, resulted in greater acceptance of the procedure by patients, parents and physicians. The number of centers performing pediatric liver transplantation has increased as a direct result of greater patient and physician enthusiasm for the procedure, while the number of pediatric transplants has remained constant. Although the increased number of transplant centers likely has improved access to the procedure and proved convenient for patients and their families, the experience at each center has been substantially diluted.

Increasing donor demand adversely affects donor availability for pediatric recipients. In addition, there is increasing pressure at present from public and private payers to raise efficiencies and cut costs. These factors provide a substantial impetus for pediatric transplant centers to examine the outcomes achieved in order to provide adequate stewardship for scarce donor resources and utilize dwindling financial resources most effectively (4). Unfortunately, the steady dilution in pediatric transplant experience precludes all but the crudest analysis of transplant outcomes.

Back to Top | Article Outline
Research Goals

It is recommended that a large multicenter transplant registry be developed that would prospectively collect data from all pediatric transplant centers. Such a database is essential for the accurate analysis of patient outcomes and for the development of innovations that might improve these outcomes in the future.

Data from the multicenter pediatric transplant registry would be analyzed to determine the:

A) Long-term graft and patient survival for pediatric transplantation, stratified by disease. Data on survival are currently available only for a few pediatric liver diseases. However, less common liver diseases that are considered potential indications, such as metabolic diseases, collectively account for about 30% of pediatric transplants. For many of these disorders, only anecdotal experience of short-term outcomes at a single center has been published. Establishment of a registry detailing the results of transplantation would allow for a more accurate assessment of the effectiveness of transplantation. These data, for example, would help determine the optimal timing of transplantation and the appropriateness of transplantation as a treatment option.

B) Best methods of surgical and medical management. With dispersion of the pediatric transplant experience, local variations have developed in both the surgical procedure and postoperative care, including the use of immunosuppression. Some of these local idiosyncrasies may add to the overall cost of the procedure. Since there is currently no method to track the outcome of these various approaches, their effectiveness cannot be determined. The analysis of outcomes resulting from these management strategies, with comparison to outcomes in the entire data set, would be a first step in determining whether there is an optimal approach to operative and postoperative management.

C) Long-term growth potential and long-term development potential of patients undergoing liver transplantation. Information regarding these issues is currently lacking and requires the collection of data to track somatic growth, reproductive capabilities, and educational and occupational achievements relative to the clinical characteristics of patients undergoing transplantation. Acquisition of this information would help to determine the optimal timing of the transplant procedure, and would allow for comparison of transplantation with other management strategies (5). For example, if it were shown that otherwise stable patients with cholestatic liver disease suffered irreversible compromise of growth or intellectual achievement over time, early transplantation would be justified.

D) Long-term effects of immunosuppressive therapy. Several complications of standard immunosuppressive medications have been documented in children and adults. These include increased lipid levels, hypertension, decreased renal clearance, altered glucose metabolism, compromised growth and diminished bone accretion (6–8). The impact of these adverse effects is substantially greater in children than in adults because of the longer period of exposure to the drugs. The incidence of severe injury leading to potentially life-threatening conditions such as renal failure, heart disease or osteoporosis is not known. A determination of the incidence of these adverse effects, and the factors that predispose to their development, would be an important initial effort in devising strategies to reduce drug-induced organ damage.

E) Long-term quality of life of children and their families after liver transplantation. The goal of transplantation is to improve the quality, as well as the quantity, of the patient's life. Recently, accurate, validated tools have been developed to assess pediatric quality of life. A large multicenter study utilizing these tools to evaluate the effectiveness of liver transplantation in reaching this goal is essential. As part of this study, children and their families would be followed over the long term, data would be stratified by specific diseases, and management strategies could be identified that yield the best outcomes.

F) Long-term costs of transplantation. Anticipated costs for individual liver transplant recipients are currently difficult, if not impossible, to assess. Factors influencing cost include regional differences in overall cost of medical care, the United Network for Organ Sharing (UNOS) status of the patient at the time of transplantation, the number of comorbid conditions, and the type of graft. A database that tracks expenditures in addition to medical variables would be a first step in developing an accurate projection of costs for individual patients. Long-term follow-up of patient outcomes and cumulative associated costs would enable development of realistic cost-benefit analyses for pediatric transplantation.

Back to Top | Article Outline
Research Strategies

To achieve the above goals, a registry encompassing the majority of North American pediatric liver transplant recipients needs to be developed. Data must be collected in a prospective, standardized manner and analyzed in a timely and statistically valid fashion. Individual patients must be followed until adulthood. The ultimate goal of such a registry would be to determine the expected outcomes of liver transplantation for specific recipients and to identify factors that would influence the likelihood of achieving these outcomes. Validated tools need to be utilized or developed to assess some of these outcomes. Comparable experience in the pediatric oncology community has demonstrated the value of such a registry for determining therapeutic outcomes and developing new strategies to improve outcomes.

An industry-funded pediatric liver transplant registry called SPLIT (S tudies in P ediatric L iver T ransplantation) currently collects data from 34 centers in Canada and the US, representing approximately 25% of the procedures performed annually. We propose expanded funding of the existing database to enable recruitment of additional centers to capture a minimum of 75% of the transplants performed each year. In addition, prospective studies evaluating specific outcomes are recommended. It is only through acquisition and analysis of these data that true measurements of the long-term effectiveness of liver transplantation will be achieved.

Back to Top | Article Outline
Projected Timetable and Funding Requirements

Funding for individual centers is needed to expand the existing SPLIT database to encompass the majority of pediatric liver transplant centers. Funding would largely be directed at support for transplant coordinators who gather the large amount of data required and for data entry personnel responsible for inputting the data. A small part of the funds would be allocated to a central data collection agency.

Back to Top | Article Outline
Evaluate Interventions to Prevent and Treat Post-Transplant Lymphoproliferative Disease

Post-transplant lymphoproliferative disease (PTLD) occurs in up to 11% of pediatric liver transplant recipients and up to 25% of pediatric intestinal transplant recipients. The associated mortality rate can be as high as 20% to 60%. In pediatric patients, more than 85% of PTLD is related to Epstein-Barr virus (EBV) infection and PTLD presents as a spectrum of disease ranging from benign B-cell hyperplasia to malignant lymphomas (9).

Patients who are EBV naive and receive an organ from an EBV-positive donor, especially those being treated with increased levels of immunosuppressive agents for resistant rejection, are at high risk of developing PTLD (10). Infants and toddlers, who constitute 50% of the pediatric liver transplant population, are usually EBV naive. Up to 15% of high-risk liver transplant recipients will develop PTLD. More than 75% of high-risk patients acquire the virus within the first year of life. For children, especially those younger than 2 years of age, PTLD not only can be lethal but also can critically affect quality of life and graft function.

It has been hypothesized that the outcome of EBV infection in pediatric transplant recipients reflects a balance between EBV-driven B-cell proliferation and the activity of EBV-specific cytotoxic T cells. If this hypothesis is true, then therapy that enhances the EBV-specific T-cell response or decreases B-cell proliferation should prevent PTLD. Prevention and preemptive treatment strategies include polymerase chain reaction (PCR) monitoring of the peripheral blood for the EBV genome, combined with antiviral therapy and reduction of immunosuppression with the first evidence of EBV infection. Treatment requires stopping T cell-directed immunosuppression so that immune surveillance by EBV-specific cytotoxic T cells is restored. Transplant physicians also use medications that inhibit viral replication and high-titer cytomegalovirus (CMV) globulin to prevent and treat PTLD. The efficacy of antivirals and immunoglobulin is difficult to assess because reduction of immunosuppressive therapy is almost always initiated simultaneously. However, the enhanced immune response that results from reduced immunosuppression is nonspecific and may precipitate allograft rejection.

Back to Top | Article Outline
Research Goals

Interventions need to be tested that can prevent or treat PTLD in pediatric liver transplant recipients by shifting the balance between B-cell proliferation and activation of EBV-specific cytotoxic T cells. Basic research is needed to develop:

A reproducible method to measure EBV-specific cytotoxic T-cell activity; and

A method for in vitro activation of recipient-derived EBV-specific T cells, including T cells from EBV-naïve recipients. The T cells can then be reinjected into the recipient at the time of diagnosis of PTLD to restore EBV-specific cytotoxic T cell competence (11).

Clinical trials are needed to:

Evaluate preemptive therapy that enhances EBV-specific cytotoxic T-cell activity and prevents development of PTLD

Study the role of serial EBV PCR monitoring of the peripheral blood in preemptive therapy

Determine the efficacy of standard treatment approaches (reduced immunosuppression, antivirals, hyperimmune globulin) in patients with PTLD

Determine the efficacy of treatment with monoclonal antibodies directed against B cells or chemotherapy in restoring the balance between B-cell proliferation and EBV-specific T-cell response in patients who fail to respond to preemptive or standard therapy.

Back to Top | Article Outline
Research Strategies

An analysis of the SPLIT registry shows that 120 children annually meet the criteria for entry into the registry. It is known that up to 80% of high-risk patients are infected with EBV each year and that as many as 12% develop PTLD. We predict that 50% of patients with PTLD will not respond to standard therapy. A multicenter trial to determine optimal novel treatment is necessary because the number of patients at even the largest transplant centers is too small to achieve adequate power.

Back to Top | Article Outline
Projected Timetable and Funding Requirements

One or two funded investigators will be required to address each basic research goal. The clinical research goals will be aided by creation of a clinical trials network linking multiple centers.

Back to Top | Article Outline
Evaluate Treatment Strategies for Minimizing Nonimmune Graft Injury

Liver allografts are a precious commodity since demand far exceeds supply. To use existing resources efficiently, initial graft function must be optimized. When the graft does not function, or initial function is poor owing to reperfusion injury, there is an increased risk of perioperative morbidity and graft loss, as well as greater health resource utilization. It is increasingly recognized that nonimmune injury to the allograft is associated with significant short-term and long-term consequences. Nonimmune graft injury results from the effects on the donor organ of brain death itself, ischemia/reperfusion injury and preexisting injury in the graft. With a better understanding of the underlying mechanisms, preventive strategies can be designed.

There are several compelling reasons why progress in this area of research is important. First, preventing or ameliorating the injury induced by brain death and ischemia/reperfusion is likely to improve early cadaveric graft function and reduce the need for retransplantation due to primary nonfunction or poor initial function. In view of the ongoing cadaveric donor shortage, advances in this area are of particular importance. In addition, improved graft protection may allow the successful use of more marginal donors, which would further expand the cadaveric donor pool.

Second, it is now known that the nonspecific inflammatory response induced by nonimmune injury itself up-regulates the immune response to the graft (12). The risk of acute rejection may be increased, but of greater importance is recent evidence suggesting that chronic rejection may be linked to early nonimmune injury.

Third, an understanding of the regenerative response of the liver after injury is critically important with the increasing use in children of segmental liver grafts, particularly those from cadavers. This research is also relevant to liver donor grafts. Interleukin-6 is a key cytokine involved in activating transcription factors, such as JAK kinase and STAT3, which activate hepatocyte cell division. Current immunosuppressive drugs may be detrimental to some of these responses; for example, steroids inhibit liver regeneration.

Back to Top | Article Outline
Research Goals

Research efforts are needed to develop strategies for minimizing nonimmune injury and thereby optimizing early graft function. Such efforts will improve the cadaveric donor supply by allowing more cadaveric organs to be split. It is also important to promote the regenerative response of the segmental liver graft for a successful outcome after pediatric split transplantation.

Back to Top | Article Outline
Identify Markers and Develop Diagnostic Tests for Rejection Following Intestinal Transplantation
Research Goals

Isolated intestinal grafts constitute about one half of the 50 to 100 intestinal transplants that are performed each year in the US. Recent reports have indicated that acute rejection of the isolated graft is virtually universal and an important cause of graft loss. The incidence and severity of rejection, although somewhat less in cases of combined liver-intestinal transplantation, is greater than occurs with isolated liver grafts. Intestinal allografts are also susceptible to chronic rejection, which occurs rarely with liver grafts.

Rejection of intestinal grafts is difficult to diagnose and treat. Unlike with liver and kidney transplantation, there is no simple blood test to detect when the small intestine initiates the process of rejection. Tissue diagnosis is often difficult as the process can be patchy. A late diagnosis often results in loss of an intestinal graft. Studies are needed to:

Evaluate histologic markers for rejection. Special staining techniques may be used based on an understanding of the mechanisms underlying rejection.

Develop functional tests to assess changes in intestinal function that correlate well with rejection. Permeability studies have been evaluated for this purpose, but results were nonspecific (13).

Identify serum proteins, enzymes, and other markers that might be elevated early during the course of rejection (14).

Back to Top | Article Outline
Research Strategies

Several complementary approaches may be required. Studies can be conducted only at centers where large numbers of intestinal transplant procedures are performed.

Back to Top | Article Outline
Projected Timetable and Funding Requirements

These studies are likely to be an ongoing project. The cost of part-time technical assistance, specialized nursing support for clinical aspects of the research, and data analysis may be significant. Supplemental immunologic studies, if included, could easily raise estimates of the total annual cost.

Back to Top | Article Outline
Broaden the Clinical Applications of Hepatocyte Transplantation

Hepatocyte transplantation (HTX) continues to evolve as a potential therapeutic adjunct to liver transplantation. By providing normal hepatocytes to patients with a liver-related metabolic defect, HTX can improve metabolism (15). HTX can also serve as a lifesaving “bridge” in liver failure patients awaiting liver transplantation, providing a cellular mass sufficient to temporarily carry out metabolic functions (16).

The development of HTX as a treatment modality came after decades of basic research in liver cell biology. Technical issues concerning the isolation of hepatocytes and cryopreservation needed to be resolved. With successful cryopreservation, hepatocytes can be stored and shared among centers for use in the treatment of patients with liver failure.

Back to Top | Article Outline
Research Goals

To broaden the applications of HTX, further studies are critical to determine:

Optimal methods of infusion/transplantation of hepatocytes. Is a threshold number of cells needed for successful transplantation? Is a single infusion sufficient, or are repeated infusions preferable? What is the optimal route of infusion, intrasplenic or intraportal?

Guidelines for patient selection. Are patients with metabolic disease appropriate candidates for HTX? In considering HTX, do criteria differ for patients with acute versus chronic liver failure?

Optimal time of infusion/transplantation in relation to a patient's clinical course. What principles might apply during advanced stages of liver failure? During early stages?

In addition, studies are needed to address the issue of availability of hepatocytes. New methods must be developed that enable hepatocytes to proliferate in culture. Once that technology has been secured, the number of hepatocytes available for transplantation will increase independent of the availability of new livers.

Back to Top | Article Outline
Research Strategies

To achieve these research goals, multicenter clinical trials as well as basic research are needed. It is important to support the creation of regional centers for the storage/banking of cryopreserved hepatocytes. Future research will also be enhanced by the development of animal models of liver cell transplantation.

Back to Top | Article Outline


The population of long-term survivors of liver transplantation has grown, exceeding by tenfold the number of transplant procedures performed each year. In adults, the mean cost of the liver transplant procedure and associated hospitalization alone is estimated at about $150,000.17 For each year after a successful liver transplantation, direct health care costs are estimated to reach 7% to 10% of the mean cost of transplantation and recovery. Consequently, after 10 years, the cumulative cost of maintaining graft function and wellness in a population of long-term survivors is equal to the cost of the liver transplantation procedure. Based on these assumptions, it is estimated that $154 million is spent each year on liver transplantation in the pediatric population: $77 million for the procedures and an equal amount to maintain graft function and wellness in survivors. Moreover, the impact of pediatric transplant survivors on total health care costs will become increasingly relevant, since the potential for years of life gained for a 2-year-old undergoing transplantation is much greater than that for a 40-year-old.

The economic and psychosocial cost to the family is more difficult to estimate. There is loss of productivity when patients care for their children through long periods of hospitalization and frequent follow-up visits. Parents may lose their jobs and health insurance while meeting the intense medical needs of children during the peri-transplant period. Quality-of-life studies are virtually nonexistent in children and parents after liver or intestinal transplantation. The learning abilities of children who successfully undergo transplant procedures will directly affect their ability to become productive and independent members of society; the impact of liver and intestinal transplantation has not been studied.

Back to Top | Article Outline


1. Annual Report of the US Scientific Registry for Transplant Recipients and Organ Procurement and Transplantation Network: Transplant Data: 1989–1998. US Department of Health and Human Services, Health Resources and Services Administration, Office of Special Programs, Division of Transplantation. Richmond, Va: US Government Printing Office; 1999.
2. Dong VM, Womer KL, Sayegh MH. Transplantation tolerance: the concept and its applicability. Pediatr Transplant 1999; 3:181–92.
3. Harlan DM, Kirk AD. The future of organ and tissue transplantation: can T-cell costimulatory pathway modifiers revolutionize the prevention of graft rejection? JAMA 1999; 282:1076–82.
4. Goss JA, Shackleton CR, McDiarmid SV, et al. Long-term results of pediatric liver transplantation: an analysis of 569 transplants. Ann Surg 1998; 228:411–20.
5. Wayman KI, Cox KL, Esquivel CO. Neurodevelopmental outcome of young children with extrahepatic biliary atresia 1 year after liver transplantation. J Pediatr 1997; 131:894–8.
6. McDiarmid SV, Gornbein JA, Fortunat M, et al. Serum lipid abnormalities in pediatric liver transplant patients. Transplantation 1992; 53:109–15.
7. Bartosh SM, Alonso EM, Whitington PF. Renal outcomes in pediatric liver transplantation. Clin Transplant 1997; 11:354–60.
8. McDiarmid SV, Gornbein JA, DeSilva PJ, et al. Factors affecting growth after pediatric liver transplantation. Transplantation 1999; 67:404–11.
9. Sokal EM, Caragiozoglou T, Lamy M, Reding R, Otte JB. Epstein-Barr virus serology and Epstein-Barr virus-associated lymphoproliferative disorders in pediatric liver transplant recipients. Transplantation 1993; 56:1394–8.
10. Green M, Michaels MG, Webber SA, Rowe D, Reyes J. The management of Epstein-Barr virus associated post-transplant lymphoproliferative disorders in pediatric solid-organ transplant recipients. Pediatr Transplant 1999; 3:271–81.
11. Rooney CM, Smith CA, Ng CY, et al. Infusion of cytotoxic T cells for the prevention and treatment of Epstein-Barr virus-induced lymphoma in allogeneic transplant recipients. Blood 1998; 92:1549–55.
12. Lu CY, Penfield JG, Kielar M, Jeyarajah DR, Vazquez MA. Does the injury of transplantation initiate acute rejection? Graft 1999; 2:S36–43.
13. Kaufman SS, Wisecarver JL, Ruby EI, et al. Correlation of mucosal disaccharidase activities with histology in evaluation of rejection following intestinal transplantation. Pediatr Transplant 1998; 2:134–8.
14. Kuo PC, Morris J, Marks WH, et al. Human syngeneic intestinal transplantation: evaluation of enterocyte viability with serum levels of intestinal fatty acid binding protein. Clin Transplant 1996; 10:282–6.
15. Fox IJ, Chowdhury JR, Kaufman SS, et al. Treatment of the Crigler-Najjar syndrome type I with hepatocyte transplantation. N Engl J Med 1998; 338:1422–6.
16. Strom SC, Fisher RA, Thompson MT, et al. Hepatocyte transplantation as a bridge to orthotopic liver transplantation in terminal liver failure. Transplantation 1997; 63:559–69.
Showstack J, Katz PP, Lake JR, et al. Resource utilization in liver transplantation: effects of patient characteristics and clinical practice. NIDDK Liver Transplantation Database Group. JAMA 1999; 281:1381–6.
© 2002 Lippincott Williams & Wilkins, Inc.