The Costs and Cost-Effectiveness of Allogeneic Peripheral Blood Stem Cell Transplantation versus Bone Marrow Transplantation in Pediatric Patients with Acute Leukemia

Article Outline

• Abstract
• Introduction
• Patients and Methods
  • Patients
  • Study Design
  • Cost Analysis
  • Effectiveness Analysis
  • Descriptive Statistical Analyses
  • Uncertainty Analyses
• Results
  • Patient and Transplant Characteristics
  • Clinical Outcomes
  • Cost Data
  • Cost-Effectiveness Analysis
  • Uncertainty Analysis
• Discussion
• Acknowledgments
• Appendix. Regression Model Cost Equation
• References
• Copyright

In a retrospective study, we evaluated the cost and cost-effectiveness of allogeneic peripheral blood stem cell transplantation (PBSCT) (n = 30) compared with bone marrow transplantation (BMT) (n = 110) in children with acute leukemia after 1 year of follow-up. Treatment success was defined as disease-free survival at 1 year posttransplantation. For patients at standard risk for disease, the treatment success rate was 57.1% for PBSCT recipients and 80.3% for BMT recipients (P = not significant [NS]). The average total cost per treatment success at 1 year in the standard-risk disease group was $512,294 for PBSCT recipients and $352,885 for BMT recipients (P = NS). For patients with high-risk disease, the treatment success rate was 18.8% for PBSCT recipients and 23.5% for BMT recipients (P = NS). The cumulative average cost was $457,078 in BMT recipients and $377,316 in PBSCT recipients (P = NS). Point estimates of the incremental cost-effectiveness ratio (ICER) indicate that in patients with standard-risk disease, allogeneic BMT had lower costs and greater effectiveness than PBSCT (ICER, -$687,108; 95% confidence interval [CI], $2.4 million to dominated). For patients with high-risk disease, BMT was more effective and more costly, and it had an ICER of $1.69 million (95% CI, $29.7 million to dominated) per additional treatment success. The comparative economic evaluation provides support for BMT in standard-risk patients, but much uncertainty precludes a clear advantage of either treatment option in patients with high-risk disease. More studies using larger and randomized controlled trials are needed to confirm the long-term cost-effectiveness of each procedure.

Key Words: Children, Treatment success, Long-term follow-up

Back to Article Outline

Introduction 

Advances in stem cell transplantation (SCT) technology have improved the outcome and increased the availability of the technique, encouraging its use as a front-line treatment for many serious malignant and nonmalignant disorders. Although bone marrow (BM) was initially the exclusive source of stem cells, peripheral blood stem cells (PBSCs) have become an increasingly favored alternative. Indeed, PBSCs have now largely superseded BM as the source of cells in autologous SCT, because of preferable clinical outcomes, such as faster engraftment kinetics and shorter hospitalization times 1, 2, 3. Whereas allogeneic PBSC transplantation (PBSCT) produces similar benefits in the allogeneic setting, these are partially offset by the association of this stem cell source with an increased risk of graft-versus-host disease (GVHD), which remains even with recent improvements in GVHD prophylaxis and molecular techniques for establishing donor–recipient histocompatibility 4, 5, 6, 7, 8, 9, 10. Consequently, whether PBSCs are preferable to BM as an allograft source is unclear. In practice, BM continues to be the main source of stem cells for matched sibling donor and matched unrelated donor transplantations, whereas PBSCs are used more widely for haploidential transplantation, because this source allows for increased stem cell doses, with apparently beneficial effects on engraftment 11, 12, 13. Umbilical cord blood, a third source of stem cells for allogeneic SCT, might be associated with a lower incidence of severe GVHD compared with the alternatives [14]. The use of umbilical cord blood has significant limitations, however, including greatly delayed time to hematologic recovery and increased risk of infection, both of which lead to higher overall costs [15].

Although allogeneic SCT with PBSCs or BM offers the prospect of a permanent cure, few studies have addressed the overall costs and cost-effectiveness of the procedure or compared the cost-effectiveness of these 2 stem cell sources. Previous adult studies have compared the cost-effectiveness of each source, but there are little cost comparison data in the pediatric population 16, 17, 18, 19, 20. This is a significant deficiency, because the disease mix (eg, acute lymphocytic leukemia [ALL] > acute myelogenous leukemia [AML]), the biology of the diseases, and the risks of severe GVHD are all strikingly different in the pediatric and adult populations. These differences might result in differing predispositions to posttransplantation complications and differing overall outcomes, and thus cost-effectiveness assessments.

In the present work, we compared the costs and cost-effectiveness of allogeneic PBSCT and BM transplantation (BMT) in pediatric patients with acute leukemia. We examined the 1-year posttransplantation economic implications of PBSCT versus BMT by stratifying patients' disease status in a retrospective study derived from a single institution in which patients underwent primary transplantation between 2001 and 2006.

Back to Article Outline

Patients and Methods 

Patients 

We studied children and adolescents (age range, 0-18 years) with acute leukemia who underwent allogeneic PBSCT or BMT between January 1, 2001, and September 30, 2006, in the Stem Cell Transplant Unit at Texas Children's Hospital (TCH). We analyzed those patients who received a primary transplantation and had acute leukemia as their primary disease. In our cohort, standard risk was defined as ALL or AML in first or second remission, and high risk was defined as ALL or AML in third or subsequent remission or in relapse or in patients with secondary AML [21]. The study protocol was approved by the Institutional Review Boards at Baylor College of Medicine and TCH.

Study Design 

The data collection period consisted of the transplantation phase (admitted for initiation of pretransplantation chemotherapy until hospital discharge), short-term follow-up (after initial hospitalization to 100 days posttransplantation), and long-term follow-up (100 days to 1 year posttransplantation). Medical records were retrospectively reviewed for demographic data, date of engraftment, duration of hospital stay, onset of acute and chronic GVHD (aGVHD, cGVHD), incidence of infectious complications, and duration of disease-free survival. Data were obtained from the electronic medical records system of TCH's Center for Cell and Gene Therapy using StemSoft (Vancouver BC, Canada) and Logician (Centricity Physician Office EMR, Hillsboro, Oregon, USA) software to ensure thorough and consistent counting of resource use.

The source of stem cells for the allograft was determined by each patient's physician based on availability. The HLA typing method varied by year of transplantation, with high resolution of both class I (HLA-A, -B, -C) and class II (HLA-DRB1) antigens used starting on June 1, 2005. When a potential transplantation candidate lacked an HLA-genotypically identical sibling donor or there was insufficient time to search for a suitable donor, a stem cell graft from a haploidentical related donor was considered. Before 2003, 93% of the patients underwent BMT. After 2003, 30% of the patients underwent PBSCT. BM and PBSC collection was done following institutional standard operating procedures. Myelogenous engraftment and grading of aGVHD or cGVHD were evaluated by standard criteria [22].

Cost Analysis 

Cost data were retrospectively acquired from administrative records, and cost estimates were based on micro-cost information from the internal accounting system. At TCH, the decision support system contains patient demographic and utilization data and can be used along with the general ledger accounting and payroll data to allow financial analysis of each hospital encounter. The general ledger and payroll expenses are input into the system. These costs, along with relative value units or cost weights assigned at the product level, create a cost per procedure, and these data are accumulated and further summarized to create a total cost per visit. The database comprised cost data from October 1, 2003, to September 30, 2007. Data from beyond October 1, 2007, were not validated at the time the data were requested. Among the 140 patients recruited for the study, we initially acquired the actual costs for 57 patients in the BMT group and 19 patients in the PBSCT group. All costs were adjusted to 2008 U.S. dollars according to the medical care component of the Consumer Price Index [23].

The components of costs included days in hospital, outpatient visits, intravenous treatments, nuclear medicine, laboratory and diagnostic services, radiotherapeutic and surgical procedures, blood products, medications, and emergency room visits. The costs of stem cell collection before transplantation were not included in the analysis, because the costs for BM harvest and PBSC apheresis were essentially identical ($9164.40 for harvest and $9285.32 for apheresis). Of the 30 patients who underwent PBSCT, 2 had more than one PBSC collection. The aforementioned procurement costs include laboratory tests and donor collection fees but do not incorporate the costs of processing stem cell products in the laboratory. Indirect costs were excluded, because the perspective of this study was that of health care providers and policy makers.

Because cost data were available only for patients undergoing SCT after September 30, 2003, we developed a regression model with the available cost data to predict the total cost for those patients admitted for SCT between January 1, 2001, and September 30, 2003. Our regression equation included sex, ethnicity, prognostic factors, and length of hospital stay and its quadratic term, along with short-term and long-term study endpoints. (See the Appendix for the regression model and Table A1 for coefficients.) For the transplantation phase, we included donor type, donor/recipient cytomegalovirus (CMV) serostatus, disease risk, and length of stay into the equation. For the short-term follow-up, disease risk, occurrence of aGVHD, and duration of inpatient stay were included in the computation. For the long-term follow-up, disease risk, occurrence of cGVHD, infection or relapse, and hospitalization were included. The equation was used to predict the total costs by each phase of treatment, and the imputation procedure was used separately in the BMT and PBSCT groups. The prediction method assumed that the hospital cost allocation method used after September 30, 2003, was also applicable to the previous period for which cost data were not available. Stability in the institutional allocation of expenses in the transplantation unit during the study period was confirmed by the accounting department. In addition, the cost distribution for outlier cases was examined using a normal probability plot. Outliers were then excluded from the estimation of the cost prediction model [24].

Effectiveness Analysis 

The BMT group served as the reference population for the cost-effectiveness comparison with the PBSCT group. Cost-effectiveness was measured at 1 year posttransplantation. Treatment outcome was defined as treatment success during the 1-year follow-up period up to the last date of contact as of December 31, 2007. Treatment was considered a success if the patient survived disease-free for 1 year. Cost and survival information was censored at the time of disease relapse or death during this 1-year follow-up period. Incremental effectiveness was measured by the 1-year posttransplantation difference in the percentage of cases with treatment success (PBSCT minus BMT). Incremental cost was measured by the 1-year posttransplantation difference in the average cost (PBSCT minus BMT). The incremental cost-effectiveness ratio (ICER) was expressed as the incremental effect divided by the incremental cost.

Descriptive Statistical Analyses 

Descriptive statistics were used to summarize the distribution of each variable and to compare baseline characteristics between the PBSCT and BMT groups, using SPSS 15.0 (SPSS Inc, Chicago, IL) and SAS 9.2 (SAS Institute, Cary, NC). Although the nature of the population surveyed provided unequal numbers in the groups, we allowed for the effects of these discrepancies and used the Mann-Whitney U test to compare the 2 groups on all continuous variables and itemized costs in patients who had actual costs. The discrete variables (eg, frequency of aGVHD and cGVHD and infectious complications) were compared by means of a frequency table using Fisher's exact test. For all statistical tests, P = .05 was considered significant.

Uncertainty Analyses 

Uncertainty was examined by standard sensitivity analysis and probabilistic sensitivity analysis. Standard one-way sensitivity analysis was used to examine the effect of imputing cost data for patients treated before October 1, 2003, with estimates of the mean cost both with and without the imputed data. We considered using bootstrap simulation to examine statistical uncertainty; however, bootstrapping relies heavily on the tails of the estimated sampling distribution, and the smaller the sample, the less likely that all of the relevant characteristics of the population will be represented [25]. Thus, our probabilistic sensitivity analysis of the cost variables used 1000 Monte Carlo simulations to estimate the 95% confidence intervals (CIs), using TreeAge Pro 2009 Suite (TreeAge Software, Williamstown, MA) 26, 27, 28.

Back to Article Outline

Results 

Patient and Transplant Characteristics 

A total of 140 patients were consecutively recruited during the study period, of whom 110 underwent BMT and 30 underwent PBSCT. Patient and transplant characteristics are summarized in Table 1. The PSBCT group included 14 standard-risk patients (46.7%) and 16 high-risk patients (53.3%); corresponding numbers in the BMT group were 76 (69.1%) and 34 (30.9%). In the PBSCT group, 14 patients had ALL (46.7%) and 16 had AML (53.3%); in the BMT group, 74 patients had ALL (67.3%) and 36 had AML (32.7%). In the PBSCT group, 15 patients (50.0%) had an HLA-mismatched related donor, 4 (13.3%) had a matched related donor, 9 (30.0%) had a matched unrelated donor, and 2 (6.7%) had a mismatched unrelated donor. In the BMT group, the stem cell source was an HLA-matched related donor in 38 patients (34.5%), a matched unrelated donor in 49 patients (44.6%), and a mismatched related or unrelated donor in 23 patients (20.9%). A significantly higher percentage of PBSCT patients received CD34 selected allografts compared with the BMT group. This variable was taken into account during multivariate analyses and was not a significant risk factor for acute or chronic GVHD (data not shown). There were no statistically significant differences in terms of patient age, sex, ethnicity, or insurance coverage between the PBSCT and BMT groups.

Table 1. Patient and Transplant Characteristics

Characteristic	BMT	PBSCT	P Value
Number of patients	110	30
Patient age, years, median (range)	8 (0-18)	9 (2-18)
Patient sex, males/females, n	64/46	19/11	NS
Disease risk, n (%)			.021
Standard risk	76 (69.1)	14 (46.7)
High risk	34 (30.9)	16 (53.3)
Diagnosis, n (%)			.033
ALL	74 (67.3)	14 (46.7)
AML	36 (32.7)	16 (53.3)
Ethnicity, n (%)			NS
Caucasian	35 (31.8)	13 (43.3)
Black	13 (11.8)	3 (10.0)
Hispanic	57 (51.8)	12 (40.0)
Asian/other	5 (4.5)	2 (6.7)
Donor type, n (%)			<.001
HLA-mismatched sibling/other relative	10 (9.1)	15 (50.0)
HLA-identical sibling	38 (34.5)	4 (13.3)
HLA-matched unrelated	49 (44.6)	9 (30.0)
HLA-mismatched unrelated	13 (11.8)	2 (6.7)
CD34 selection, n (%)			<.001
Yes	2 (1.8)	13 (43.3)
No	108 (98.2)	17 (56.7)
Medicaid, n (%)			NS
Yes	58 (52.7)	16 (53.3)
No	52 (47.3)	14 (46.7)

ALL indicates acute lymphoblastic leukemia; AML, acute myelogenous leukemia; BMT, bone marrow transplantation; PBSCT, peripheral blood stem cell transplantation.

Clinical Outcomes 

As anticipated, the patients in the PBSCT group had significantly faster neutrophil and platelet engraftment than those in the BMT group (P <.001 for neutrophils; P = .034 for platelets) (Table 2). The frequency of acute or chronic GVHD was not statistically significantly different between the groups; however, the incidence of infectious complications was significantly higher in the PBSCT group (P = .028). At the 1-year follow-up, 61 (80.3%) of the standard-risk BMT patients were disease-free, compared with 8 (57.1%) of the standard-risk PBSCT patients (P = NS). Eight (23.5%) high-risk BMT patients and 3 (18.8%) high-risk PBSCT patients were disease-free at the 1-year follow-up.

Table 2. Clinical Outcomes at 1-Year Follow-Up

Variable	BMT	PBSCT	P Value
Time to engraftment, days, mean (SD)
Neutrophils	17.96 (3.475)	14.48 (3.355)	<.001
Platelets	28.68 (15.069)	23.08 (9.499)	.034
aGVHD grade II-IV, n (%)	17 (15.5%)	2 (6.7%)	NS
cGVHD, n (%)	12 (10.9%)	4 (13.3%)	NS
Infection, n (%)	80 (75.5%)	27 (90.0%)	.028
Treatment outcomes, standard-risk group, n (%)			NS
Success	61 (80.3%)	8 (57.1%)
Failure	15 (19.7%)	6 (42.9%)
Treatment outcomes, high-risk group, n (%)			NS
Success	8 (23.5%)	3 (18.8%)
Failure	26 (76.5%)	13 (81.2%)

aGVHD indicates acute graft-versus-host disease; cGVHD, chronic graft-versus-host disease; BMT, bone marrow transplantation; PBSCT, peripheral blood stem cell transplantation.

Cost Data 

The average costs of treatment and follow-up per category are presented in Table 3. The mean total cost for the initial hospitalization was higher in the PBSCT group compared with the BMT group, but the difference was not statistically significant. Room and board accounted for 50% of the total costs, and pharmacy accounted for ~28.0% of the total costs. Other major cost items during this initial phase included blood products and laboratory and radiology services. The average duration of hospitalization (length of stay [LOS]) was 42.9 days in the PBSCT group and 36.8 days in the BMT group. The total costs over this time period were $282,577 ± $272,344 for the PBSCT group and $208,987 ± $169,554 for the BMT group (P = NS). Three outlier cases (1 PBSCT patient and 2 BMT patients) accumulated >$1 million in costs per patient; all of these patients subsequently died of progressive disease during the initial hospitalization period, leading to large cost and LOS variations in both groups.

Table 3. Treatment and Follow-Up Costs per Patient in the BMT and PBSCT Groups^∗

Variable	BMT, US$, Mean (SD)	PBSCT, US$, Mean (SD)
Costs of the initial hospitalization	n = 57	n = 19
Blood products	$9,392 ($16,560)	$18,335 ($26,383)
Laboratory	$16,503 ($17,914)	$22,543 ($27,522)
Laboratory, pathology	$2746 ($6071)	$3428 ($4043)
Operating room	$3277 ($3035)	$3281 ($2811)
Pharmacy	$59,582 ($62,168)	$76,688 ($97,809)
Radiology, diagnostics	$834 ($1677)	$1317 ($2077)
Radiology, therapeutics	$9544 ($4427)	$9375 ($4823)
Room	$104,052 ($59,444)	$140,126 ($99,975)
Length of stay, days	36.84 (16.42)	42.89 (22.75)
Total costs	$208,987 ($169,554)	$282,577 ($272,334)
Costs of short-term follow up (day 100 posttransplantation)	n = 54	n = 18
Blood products	$8821 ($20,032)	$13,758 ($25,175)
Laboratory	$17,320 ($13,622)	$19,384 ($14,095)
Laboratory, pathology	$3542 ($2176)	$4695 ($2385)
Operating room	$2448 ($3355)	$2206 ($2667)
Pharmacy	$26,365 ($35,668)	$33,509 ($33,040)
Radiology, diagnostics	$868 ($1264)	$1289 ($1503)
Radiology, therapeutics	$3222 ($3690)	$6825 ($9105)
Room	$39,798 ($50,842)	$60,435 ($66,348)
Length of stay, days	18.55 (16.80)	25.06 (20.84)
Total costs	$103,428 ($123,929)	$147,907 ($135,316)
Costs of long-term follow-up (1 year posttransplantation)	n = 45	n = 15
Blood products	$12,234 ($15,903)	$12,952 ($9458)
Laboratory	$23,029 ($25,764)	$15,484 ($11,019)
Laboratory, pathology	$5220 ($3959)	$5746 ($4740)
Operating room	$3097 ($2923)	$2192 ($1554)
Pharmacy	$31,566 ($45,555)	$22,670 ($20,265)
Radiology, diagnostics	$1579 ($2161)	$1641 ($1555)
Radiology, therapeutics	$7471 ($9010)	$1477
Hormone therapy	$1957 ($2884)	$614 ($447)
Room	$43,295 ($57,758)	$39,891 ($32,362)
Length of stay, days	25.07 (26.33)	14.36 (10.32)
Total costs	$124,578 ($147,207)	$106,683 ($76,577)

BMT indicates bone marrow transplantation; PBSCT, peripheral blood stem cell transplantation.

The costs of the short-term follow-up period (after initial hospitalization to 100 days posttransplantation) were higher in the PBSCT group ($147,907 for PBSCT vs $103,428 for BMT). The average LOS was 25 days in the PBSCT group and 19 days in the BMT group. Room and board was identified as the most costly category. During short-term follow-up, PBSCT patients had more blood product and pharmacy usage, consistent with their longer LOS.

We determined costs over the longer-term follow-up period in patients who survived from 100 days posttransplantation to the 1-year follow-up. The average total costs were similar in the 2 groups, $106,683 ± $76,577 for PBSCT patients versus $124,578 ± $147,207 for the BMT group, and the average LOS was 14 days and 25 days, respectively. During the longer-term follow-up, the costs of pharmacy and room and board accounted for about 60% of the total costs in both groups.

The subgroup of patients with standard-risk disease had an average total cost per successfully treated patient of $367,511 ± $164,382 in the PBSCT group and $327,170 ± $156,654 in the BMT group (P = NS) (Table 4). In the subgroup of patients with high-risk disease, the average cost per successfully treated patient was $263,392 ± $129,239 in the PBSCT group and $438,473 ± $256,150 in the BMT group (P = NS). The cumulative cost of treatment intervention was higher in the patients who relapsed or died before the 1-year follow-up compared with those who remained disease-free at 1 year. For these failing patients, the average total cost in the standard-risk subgroup was $705,338 ± $297,292 in the PBSCT group versus $457,459 ± $357,503 in the BMT group (P = .029). In the high-risk subgroup, this cost was $403,606 ± $311,868 in the PBSCT group and $462,803 ± $304,924 in the BMT group (P = NS).

Table 4. Total Costs per Patient in the BMT and PBSCT Groups at 1 Year Posttransplantation^∗

Treatment Outcome	BMT, US$, Mean (SD)	PBSCT, US$, Mean (SD)	P Value
Standard-risk group
Success	$327,170 (156,654)	$367,511 (164,382)	NS
Failure	$457,459 (357,503)	$705,338 (297,292)	.029
High-risk group
Success	$438,473 (256,150)	$263,392 (129,239)	NS
Failure	$462,803 (304,924)	$403,606 (311,868)	NS

BMT indicates bone marrow transplantation; PBSCT, peripheral blood stem cell transplantation.

At the 1-year follow-up, the cost of initial hospitalization for transplantation accounted for approximately 50% of the total costs, and short-term and long-term follow-up costs accounted for 26% and 24%, respectively. In the standard-risk subgroup, patients who did not achieve disease-free survival at 1 year of follow-up had 40%-92% higher costs compared with those who had been treated successfully. In the high-risk subgroup, this figure was 6%-53%.

Fourteen patients (3 in the BMT group and 11 in the PBSCT group) underwent additional transplantation if they relapsed from their primary SCT. The average total cost for these patients was $383,017 ± $143,513 in the PBSCT group and $262,299 ± $65,776 in the BMT group (data not shown). The median survival time was 199 days (range, 10 to >1700 days) in the PBSCT group compared with 150 days (range, 49-391 days) in the BMT group.

Cost-Effectiveness Analysis 

BMT served as the reference when computing the ICERs for standard-risk and high-risk patients (Table 5). For the standard-risk subgroup, the total mean cost per patient was $512,294 ± $280,433 in the PBSCT group and $352,885 ± $214,976 in the BMT group. The treatment success rate was 57.1% in the PBSCT group and 80.3% in the BMT group. Compared with PBSCT, BMT had greater effectiveness and lower costs (ICER, −$687,108) in this subgroup of patients. For the high-risk subgroup, the average cost per patient was $377,316 ± $288,498 in the PBSCT group and $457,078 ± $290,630 in the BMT group. The probability of treatment success was 18.8% for the PBSCT group and 23.5% for the BMT group; 3 patients in the PBSCT group and 8 patients in the BMT group had been treated successfully at the 1-year follow-up. The ICER for the high-risk subgroup was approximately $1.69 million per additional treatment success in the BMT group compared with the PBSC group.

Table 5. Cost-Effectiveness Analysis of SCT for Acute Leukemia in Children^∗

	Total Cost (C)	Incremental Cost (ΔC)	Effectiveness (% treatment success) (E)	Incremental Effectiveness (ΔE)	ICER (ΔC/ΔE)
Standard-risk group
BMT	$352,885		80.3%
PBSCT	$512,294	$159,409	57.1%	<0	−$687,108
High-risk group
BMT	$457,078		23.5%
PBSCT	$377,316	−$79,762	18.8%	−4.7%	$1,697,063

SCT indicates stem cell transplantation; BMT, bone marrow transplantation; PBSCT, peripheral blood stem cell transplantation; ICER, incremental cost-effectiveness ratio.

Uncertainty Analysis 

The ICERs were recomputed to exclude imputed costs to gauge the effect of the cost imputation process on the underlying data (Table 6). In patients with standard-risk disease, BMT remained dominant over PBSCT (ICER, −$969,453). The ICER was reduced to $1.41 million in the analysis of high-risk patients, with BMT more expensive and more effective than PBSCT.

Table 6. Uncertainty Analysis of Cost-Effectiveness Analysis on Actual Cost Data

	ICER (Actual and Imputed Costs)	ICER (Actual Costs)
Standard-risk group
BMT
PBSCT	−$687,108	−969,453
High-risk group
BMT
PBSCT	$1,697,063	$1,415,362

BMT indicates bone marrow transplantation; PBSCT, peripheral stem cell transplantation; ICER, incremental cost-effectiveness ratio.

Results of the probabilistic sensitivity analyses are shown in Figure 1 for the standard-risk subgroup and in Figure 2 for the high-risk subgroup. The mean cost was $515,540 ± $162,969 (95% CI, $258,867 to -$890,989) for the standard-risk PBSCT patients and $357,748 ± $146,874 (95% CI, $132,581 to -$699,985) for the standard-risk BMT patients. The 95% CI ellipse shows that BMT was dominant over PBSCT, and the 95% confidence limit for the ICER ranged from $2.4 million to −$5.5 million. The majority of the points (76.7%) fell in quadrant II, indicating a higher probability that PBSCT was less effective and more costly (Figure 1). There was a 19% probability for PBSCT being less costly, but also less effective (quadrant III).

• 
  • View Large Image
  • Download to PowerPoint

• Figure 1 

  ICER of PBSCT (A) and BMT (B) in the standard-risk subgroup. The ICER was determined using 1000 Monte Carlo simulations to estimate the 95% confidence ellipse. Most points (77%) fell in quadrant II, indicating a higher probability that PBSCT was less effective and more costly than BMT.

• 
  • View Large Image
  • Download to PowerPoint

• Figure 2 

  ICER of PBSCT (A) and BMT (B) in the high-risk subgroup. The ICER was determined using 1000 Monte Carlo simulations to estimate the 95% confidence ellipse. There is a 37% probability that PBSCT would be less costly but also less effective than BMT (quadrant III). PBSCT has an equivalent opportunity (24%) of being the dominant option (quadrant IV) or being the dominated choice (quadrant II) over BMT.

For high-risk patients, the mean cost was $386,193 ± $255,691 (95% CI, $70,297-$1,026,887) in the PBSCT group and $468,244 ± $243,771 (95% CI, $128,543-$1,030,401) in the BMT group. The 95% confidence limits for the ICER ranged from $29.7 million to −$35.2 million. There is a 36.8% probability that PBSCT would be less costly but also less effective compared with BMT (quadrant III), as shown in Figure 2. Furthermore, PBSCT has an equivalent opportunity (24%) of being the dominant option (quadrant IV) or being the dominated choice (quadrant II) over BMT. Overall, there is no clear preference for either treatment method, because of the large degree of uncertainty in the results.

Back to Article Outline

Discussion 

We report a cost-effectiveness analysis comparing PBSCT and BMT in a cohort of pediatric patients. Cost-benefit analyses measure both the costs and the effectiveness of alternative treatments in monetary units and thereby determine the net (social) benefit. Our study measures effectiveness in terms of clinical indicators/health outcomes and is the first to include such cost-effectiveness endpoint analyses out to 1 year of follow-up. The clinical outcomes indicate faster engraftment, but a higher frequency of infectious complications, in patients receiving PBSCT compared with those receiving BMT. Between-group differences in LOS and hospital costs were not statistically significant. At the 1-year follow-up, patients in the BMT group had a higher treatment success rate and lower costs within the standard-risk subgroup but not within the high-risk disease group, in which both the treatment success rate and the cumulative costs were lower in the PBSCT group.

Previous cost-effectiveness analyses of SCT versus chemotherapy or no transplantation mostly studied adult populations with heterogeneous diseases 29, 30, 31, 32, 33, 34, 35. Several of these studies found that SCT was cost-effective, with a cost of intervention of <$50,000 per quality-adjusted life-year 32, 33, 34, 35. Comparisons of the cost-effectiveness of PBSCT versus BMT have been limited. Only one exclusively pediatric study analyzing the cost-effectiveness of these 2 stem cell sources in children with hematologic malignancy has been reported. This single-institution study (Hospital Nino Jesus, Spain) of 25 patients accrued over 9 years found lower overall costs at 100 days posttransplantation for PBSCT compared with BMT [36]. Although more adult cost-effectiveness analyses have been published, these investigated only acute costs 20, 29, 31, 33.

In a separate study, we analyzed the impact of stem cell source on disease-free survival (DFS) in the pediatric transplantation population at 1 year and 3 years of follow-up, respectively. In our series, Kaplan-Meier analysis demonstrated a significant difference in the cumulative probabilities of DFS in patients undergoing BMT and those undergoing PBSCT. However, multivariate Cox regression analysis showed no effect of stem cell source on treatment-related mortality, relapse, or treatment failure overall, with HLA typing, disease status at transplantation, the impact of CD34 selection, and occurrence of aGVHD or cGVHD also included as independent variables. In the subgroups of patients with standard-risk and high-risk disease, the estimated probabilities of survival did not retain statistically significant differences between the PBSCT and BMT groups. Of note, it was mostly stem cells from haploidentical donors that were CD34-selected, whereas nonselected PBSCs were used mainly for matched unrelated donors. Our subsequent multivariate analysis to adjust for the independent effects of potential risk factors found that CD34 selection was not associated with overall outcome. We also analyzed the effect of age on outcome by stem cell source and found no significant correlation between graft type and age distribution. Ultimately, only pretransplantation CMV seropositivity and differing severity of underlying disease remained significant risk factors for DFS.

By dividing our patients into standard-risk and high-risk subgroups, we found that despite the faster hematopoietic recovery observed in PBSCT recipients after initial transplantation, there was no difference from BMT recipients in LOS or costs incurred. In addition, during initial (short-term) follow-up after hospital discharge to 100 days posttransplantation, the PBSCT group had higher costs and longer LOS compared with the BMT group, although these differences did not reach statistical significance. Our results also showed a large difference in mean cost depending on disease risk, a measure not evaluated in the previous pediatric study [36]. Our analysis of the ICER showed a significantly different treatment success rate in the standard-risk and high-risk subgroups. Although there was great variability in costs and the group sizes were relatively small, our results appear to be robust in sensitivity analyses of the observed between-group differences.

We developed a cost prediction model from detailed accounting data to estimate costs for patients undergoing SCT between January 2001, and September 2003. We confirmed that there were no changes in institutional allocation of expenses for the transplant unit during the study period. Thus, we have no reason to believe that any possible changes in hospital accounting practices would bias our cost comparisons. Based on the regression model that we developed to impute cost data, all of the adjusted R² values were well above 0.90, with <10% of the variance explained by additional variables, including possibly outcome-relevant variables. We found that HLA-matched related donors (donor 2) had a lower negative coefficient than HLA-mismatched sibling donors (donor 1), because the latter are perceived as being more costly. This internal inconsistency likely results from the small number of subjects in this study, providing an idiosyncratic outcome resulting from random effects in a small data set.

During the first year of follow-up posttransplantation, patients mainly returned to the primary transplant clinic for standard of care. Inclusion of outpatient costs outside of the institution and for home care utilization would be expected to raise the total costs of follow-up. A previous study of patients who returned to the care of a local physician after 100 days found that costs incurred between 3 and 6 months posttransplantation represented 3.8% of the total cost [29]. Although these data were not available at our institution, we would not expect the exclusion of those relatively minor cost estimates to result in a significant differential increase in costs. Because the intent of our study was to analyze the total direct costs from the perspective of health care providers, we did not include the costs resulting from the time loss of patients or their families in the transplantation program.

Although differences between institutions and in health care structures and pricing make direct comparisons with our study difficult, our data can be adjusted to factor in such differences. Precise estimates of hospital costs are essential to an accurate evaluation, so we used micro-costing to measure all of the direct medical costs of allogeneic SCT for up to 1 year. These resource-intensive transplantation services may vary at regional and even local levels, and our results may not be representative of other areas, even within the United States. Nonetheless, our detailed micro-costing and consistent cost categories allow valid comparisons of resource requirements for the alternative therapies.

We investigated the total direct costs for up to 1 year in all patients who underwent primary transplantation at our institution. Fourteen patients in our study had multiple transplantations and were excluded from our primary analysis, because these additional treatments are one of several salvage options for primary treatment failure and carry different, and not comparable, expectations of cost and complexity, regardless of the stem cell source. In those patients who had multiple transplantations after they relapsed from their initial transplantation, we subsequently measured costs and outcomes in separate analyses. Because this was a retrospective observational study, accrual was subject to selection bias; however, treatment groups were similar in terms of age, sex, ethnicity, and insurance coverage. We appreciate that fewer PBSCT recipients than BMT recipients were studied, a difference that reflects clinical practice in pediatric SCT during the study period. It certainly would be valuable to reanalyze our data with a larger, more balanced cohort, which likely would increase our confidence in our conclusions. Similarly, we followed our patients for 1 year, an intermediate health care outcome, and a longer study would be valuable to compare the longer-term cost-effectiveness of PBSCT and BMT.

The foregoing limitations of our study notwithstanding, cost-effectiveness analyses likely will become increasingly important as health care policies change. Our current ICER and analysis of uncertainty suggest that allogeneic BMT is a more cost-effective treatment option than PBSCT in patients with standard-risk childhood acute leukemia. For high-risk patients, our data are less clear, because the between-group differences were more limited and the range of costs was much wider. Our comparative economic evaluation provides support for BMT in standard-risk patients, but the high degree of uncertainty in the data limits any clear advantage for either treatment option in patients with high-risk disease. A larger and randomized controlled trial, especially in high-risk patients, is needed to definitively demonstrate the long-term cost-effectiveness of allogeneic PBSCT and BMT in the pediatric population.

Back to Article Outline

Acknowledgments 

The authors thank the nursing staff of TCH's Bone Marrow Transplant Unit for their excellent care; Myrlena Lee, Carolyn Smith, Bernadette Burttchell, Bonnie Byrne, and James Arce for their assistance with data collection; and Dr Jonathan S. Bloomer for his critical reading of the manuscript. This work was supported by National Institutes of Health Grant P30 CA 125123. H.H. is supported by a Dan L. Duncan chair, and M.B. is supported by a Fayez S. Sarofim chair.

Financial disclosure: The authors declare no conflicts of interest.

Back to Article Outline

Appendix. Regression Model Cost Equation 

Table A1. Parameter coefficients and SEs of the Independent Variables Included in the Cost Regression Model

BMT indicates bone marrow transplantation; PBSCT, peripheral blood stem cell transplantation; SCT, stem cell transplantation; aGVHD, acute graft-versus-host disease; cGVHD, chronic graft-versus-host disease; LOS, length of stay; CMV, cytomegalovirus.

Ref*, Reference group; BMT phase 1, initial hospitalization for SCT; BMT phase 2, after discharge to day +100 post-SCT, short-term follow-up; BMT phase 3, from day +100 to 1 year post-SCT, long-term follow-up.

Back to Article Outline

References 

1. Schmitz N, Linch DC, Dreger P, et al. Randomised trial of filgrastim-mobilised peripheral blood progenitor cell transplantation versus autologous bone-marrow transplantation in lymphoma patients. Lancet. 1996;347:353–357
   • View In Article
   • Abstract
   • CrossRef
2. To LB, Roberts MM, Haylock DN, et al. Comparison of heamatological recovery times and supportive care requirements of autologous recovery phase peripheral blood stem cell transplants, autologous bone marrow transplants and allogeneic bone marrow transplants. Bone Marrow Transplant. 1992;9:277–284
   • View In Article
   • MEDLINE
3. Roberts MM, To LB, Gillis D, et al. Immune reconstitution following peripheral blood stem cell transplantation, autologous bone marrow transplantation and allogeneic bone marrow transplantation. Bone Marrow Transplant. 1993;12:469–475
   • View In Article
   • MEDLINE
4. Nademanee A, Schmidt GM, Parker P, et al. The outcome of matched unrelated donor bone marrow transplantation in patients with hematologic malignancies using molecular typing for donor selection and graft-versus-host disease prophylaxis regimen of cyclosporine, methotrexate and prednisone. Blood. 1995;86:1228–1234
   • View In Article
   • MEDLINE
5. Nash RA, Piñeiro LA, Storb R, et al. FK506 in combination with methotrexate for the prevention of graft-versus-host disease after marrow transplantation from matched unrelated donors. Blood. 1996;88:3634–3641
   • View In Article
   • MEDLINE
6. Anasetti C, Etzioni R, Petersdorf EW, et al. Marrow transplantation from unrelated volunteer donors. Annu Rev Med. 1995;46:169–179
   • View In Article
   • MEDLINE
   • CrossRef
7. Goodrich JM, Mori M, Gleaves CA, et al. Early treatment with ganciclovir to prevent cytomegalovirus disease after allogeneic bone marrow transplantation. N Eng J Med. 1991;325:1601–1607
   • View In Article
8. Schmidt GM, Horak DA, Niland JC, et al. A randomized, controlled trial of prophylactic ganciclovir for cytomegalovirus pulmonary infection in recipients of allogeneic bond marrow transplantation. N Eng J Med. 1991;324:1005–1011
   • View In Article
9. Rooney CM, Smith CA, Ng CY, et al. Use of gene-modified virus-specific T lymphocytes to control Epstein-Barr virus-related lymphoproliferation. Lancet. 1995;345:9–13
   • View In Article
   • Abstract
   • CrossRef
10. Heslop HE, Ng CY, Li C, et al. Long-term restoration of immunity against Epstein-Barr virus infection by adoptive transfer of gene-modified virus-specific T lymphocytes. Nat Med. 1996;2:551–555
    • View In Article
    • MEDLINE
    • CrossRef
11. Handgretinger R, Klingebiel T, Lang P, et al. Megadose transplantation of purified peripheral blood CD 34(+) progenitor cells from HLA-mismatched parental donors in children. Bone Marrow Transplant. 2001;27:777–783
    • View In Article
    • MEDLINE
    • CrossRef
12. Anderlini P, Rizzo JD, Nugent ML, et al. Peripheral blood stem cell donation: an analysis from the International Bone Marrow Transplant Registry (IBMTR) and European Group for Blood and Marrow Transplant (EBMT) database. Bone Marrow Transplant. 2001;27:689–692
    • View In Article
    • MEDLINE
    • CrossRef
13. Körbling M, Anderlini P. Peripheral blood stem cell versus bone marrow allotransplantation: does the source of hematopoietic stem cells matter?. Blood. 2001;98:2900–2908
    • View In Article
    • MEDLINE
    • CrossRef
14. Barker JN, Wagner JE. Umbilical cord blood transplantation: current state of the art. Curr Opin Oncol. 2002;14:160–164
    • View In Article
    • MEDLINE
    • CrossRef
15. Gluckman E, Locatelli F. Umbilical cord blood transplant. Curr Opin Hematol. 2000;7:353–357
    • View In Article
    • MEDLINE
    • CrossRef
16. Couban S, Dranitsaris G, Andreou P, et al. Clinical and economic analysis of allogeneic peripheral blood progenitor cell transplants: a Canadian perspective. Bone Marrow Transplant. 1998;22:1199–1205
    • View In Article
    • MEDLINE
17. Bennett C, Waters T, Stinson T, et al. Valuing clinical strategies early in development: a cost analysis of allogeneic peripheral blood stem cell transplantation. Bone Marrow Transplant. 1999;24:555–560
    • View In Article
    • MEDLINE
18. van Agthoven M, Groot MT, Verdonck LF, et al. Cost analysis of HLA-identical sibling and voluntary unrelated allogeneic bone marrow and peripheral blood stem cell transplantations in adults with acute myelocytic leukaemia or acute lymphoblastic leukaemia. Bone Marrow Transplant. 2002;30:243–251
    • View In Article
    • MEDLINE
    • CrossRef
19. Majhail NS, Mothukuri JM, Brunstein CG, et al. Costs of hematopoietic cell transplantation: comparison of umbilical cord blood and matched related donor transplantation and the impact of posttranplantation complications. Biol Blood Marrow Transplant. 2009;15:564–573
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (393 KB)
    • CrossRef
20. Costa V, McGregor M, Laneuville P, et al. The cost-effectiveness of stem cell transplantations from unrelated donors in adult patients with acute leukemia. Value Health. 2007;10:247–255
    • View In Article
    • Abstract
    • Full-Text PDF (179 KB)
    • CrossRef
21. Hongeng S, Krance RA, Bowen LC, et al. Outcomes of transplantation with matched-sibling and unrelated-donor bone marrow in children with leukemia. Lancet. 1997;350:767–771
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (83 KB)
    • CrossRef
22. Glucksberg H, Storb R, Fefer A, et al. Clinical manifestations of graft-versus-host disease in human recipients of marrow from HLA-matched sibling donors. Transplantation. 1974;18:295–304
    • View In Article
    • MEDLINE
    • CrossRef
23. Bureau of Labor Statistics. The Consumer Price Index (CPI), 2008. Available from: http://www.bls.gov/cpi/. Accessed January 12, 2009.
    • View In Article
24. Weisberg S. Outliers and influence. In:  Weisberg S editors. Applied Linear Regression. New York: Wiley; 2005;
    • View In Article
25. Mooney CZ, Duval RD. Conclusion. In:  Mooney CZ,  Duval RD editor. Bootstrapping: A Nonparametric Approach to Statistical Inference. Thousand Oaks, CA: Sage; 1993;
    • View In Article
26. Critchfield GC, Willard KE, Connelly DP. Probability sensitivity analysis methods for general decision models. Comput Biomed Res. 1986;19:254–265
    • View In Article
    • MEDLINE
    • CrossRef
27. Doubilet P, Begg CB, Weinstein MC, et al. Probability sensitivity analysis using Monte Carlo simulation: a practical approach. Med Decis Making. 1985;5:157–177
    • View In Article
    • MEDLINE
    • CrossRef
28. Manning WG, Fryback DG, Weinstein MC. Reflecting uncertainty in cost-effectiveness analysis. In:  Gold MR,  Siegel JE,  Russel LB, et al. editor. Cost-Effectiveness in Health and Medicine. New York: Oxford University Press; 1996;
    • View In Article
29. Lee SJ, Anasetti C, Kuntz KM, et al. The costs and cost-effectiveness of unrelated donor bone marrow transplantation for chronic phase chronic myelogenous leukemia. Blood. 1998;92:4047–4052
    • View In Article
    • MEDLINE
30. Uyl-de Groot CA, Hagenbeek A, Verdonck LF, et al. Cost-effectiveness of ABMT in comparison with CHOP chemotherapy in patients with intermediate- and high-grade malignant non-Hodgkin's lymphoma (NHL). Bone Marrow Transplant. 1995;16:463–470
    • View In Article
    • MEDLINE
31. Welch HG, Larson EB. Cost-effectiveness of bone marrow transplantation in acute non-lymphocytic leukemia. N Eng J Med. 1989;321:807–812
    • View In Article
32. Messori A, Bonistalli L, Costantini M, et al. Cost-effectiveness of autologous bone marrow transplantation in patients with relapsed non-Hodgkin's lymphoma. Bone Marrow Transplant. 1997;19:275–281
    • View In Article
    • MEDLINE
33. Barr R, Furlong W, Henwood J, et al. Economic evaluation of allogeneic bone marrow transplantation: a rudimentary model to generate estimates for the timely formulation of clinical policy. J Clin Oncol. 1996;14:1413–1420
    • View In Article
    • MEDLINE
34. Beard ME, Inder AB, Allen JR, et al. The costs and benefits of bone marrow transplantation. N Z Med J. 1991;104:303–305
    • View In Article
    • MEDLINE
35. Kouroukis CT, O'Brien BJ, Benger A, et al. Cost-effectiveness of a transplantation strategy compared to melphalan and prednisone in younger patients with multiple myeloma. Leuk Lymphoma. 2003;44:29–37
    • View In Article
    • MEDLINE
    • CrossRef
36. Madero L, González Vincent M, Ramirez M, et al. Clinical and economic comparison of allogeneic peripheral blood progenitor cell and bone marrow transplantation for acute lymphoblastic leukemia in children. Bone Marrow Transplant. 2000;26:269–273
    • View In Article
    • MEDLINE
    • CrossRef