Short-Term Costs Associated With Primary Prophylactic G-CSF Use During Chemotherapy

, , ,
The American Journal of Managed Care, February 2013, Volume 19, Issue 2

Granulocyte colony-stimulating factor therapy reduces hospitalizations and improves chemotherapy administration in elderly breast cancer patients, but increases overall Medicare costs during first year of therapy.

Background:

Chemotherapy is vital for breast cancer treatment, but early-onset toxicities like neutropenia hinder its administration. Neutropenia also increases costs due to hospitalizations and aggressive systemic antibiotic administration. Primary prophylactic (PP) use of granulocyte colony-stimulating factor (G-CSF) helps prevent neutropenia. However, evidence supporting the cost-effectiveness of PPG-CSF is inconclusive, and American Society of Clinical Oncology guidelines state the need for performing cost analyses in high-risk groups like the elderly.

Objectives:

To examine the effect of PPG-CSF administration on neutropenia hospitalization costs and overall Medicare costs during the year following chemotherapy initiation.

Methods:

A retrospective observational study of patients newly diagnosed with breast cancer between 1994 and 2002 was performed using the linked SEER-Medicare database. To account for the nonrandom nature of observational data, a covariate matching technique was used to preprocess the data before performing parametric analysis estimating the effect of PPG-CSF on costs.

Results:

Administration of PPG-CSF during the first course of chemotherapy was associated with a 57% increase in overall Medicare costs during the study period, despite a drop in neutropenia hospitalization costs. Forty-two percent of the increase in costs was due to increase in chemotherapy costs during the year after chemotherapy initiation.

Conclusions:

A significant part of the increase in immediate medical costs in breast cancer patients receiving PPG-CSF is due to improved chemotherapy administration. It is important to determine whether these short-term cost increases lead to long-term health benefits and savings. Cost analyses with longer follow-ups are crucial for chronic diseases like breast cancer.

(Am J Manag Care. 2013;19(2):150-159)Our past studies show granulocyte colony-stimulating factors (G-CSFs) are effective in preventing neutropenia and neutropenia-related hospitalizations, and improving chemotherapy administration in elderly breast cancer patients.

  • The current study shows that G-CSF use increases overall Medicare costs during the first year of breast cancer therapy by 57%.

  • This increase is chiefly due to improved chemotherapy administration in the first year. Hence, G-CSFs might improve long-term health benefits and reduce long-term costs for Medicare beneficiaries suffering from breast cancer.

Chemotherapy is vital for breast cancer treatment because it prevents recurrence and reduces mortality.1-8 Despite the demonstrated efficacy of first-course chemotherapy, compliance with recommended therapy is limited by chemotherapy-induced acute toxicities like neutropenia. Neutropenia is one of the most common reasons for chemotherapy dose reductions, delays, or discontinuation.9-11 Chemotherapy dose reductions make it difficult to prevent tumor regrowth, reduce chemotherapy effectiveness, increase the probability of recurrence and mortality, and might also increase the overall treatment costs.9,10,12-19 The rate of neutropenia is especially high among the elderly (age >65 years).20

The primary prophylactic (PP) use of granulocyte-colony stimulating factor (G-CSF) helps prevent neutropenia and sustain chemotherapy dose intensity.9,16,21-24 Use of PPG-CSF can also reduce costs associated with neutropenia-related hospitalizations and systemic antibiotic administration used to treat neutropenia-associated infections and fever.

More than 60,000 neutropenia hospitalizations occur each year in the United States,25 and each hospitalization could cost $10,000 to $30,000 on average.26-29 Administration of PPG-CSF might offset neutropenia hospitalization costs by reducing both the probability21,24,30-32 and duration21,26,33 of hospitalization. Other factors that might reduce costs for patients receiving PPG-CSF during the first year after diagnosis (when bulk of the cancer therapy is administered) include reducing the need for second-course chemotherapy due to an ineffective first course, a shorter but more intense first course of chemotherapy, and recurrence prevention. Nevertheless, G-CSFs themselves are expensive. Hence, their administration should be justified both clinically and economically.

Evidence supporting the cost-effectiveness of PPG-CSF in cancer patients is ambiguous. Some clinical trials and observational studies have documented cost-savings from PPG-CSF administration due to reduced hospitalization costs or overall treatment costs.34-36 Other studies have demonstrated no change in costs because the cost savings are offset by G-CSF administration costs.26,37,38 Lyman et al determined that PPGCSF is cost-effective only for patients with >20% risk of chemotherapyinduced febrile neutropenia.21

However, studies based on clinical trials underreport the incidence and costs of treatmentrelated adverse events because they exclude high-risk patients like elderly, have short study durations, and have incomplete adverse event documentation.39-41 The American Society of Clinical Oncology guideline for GCSF use identifies the lack of evidence for G-CSF—related costs and calls for studies establishing the economic impact of G-CSF use.42 The guideline also states that current recommendations for G-CSF use are predominantly based on clinical benefits and not economic outcomes.

The lack of external validity and inadequacies of clinical trials in the elderly call for nationally representative population- based studies to establish the economic impact of PPGCSF use in the elderly.39 This study aims to address this gap and analyzes the effects of PPG-CSF administration on the very first neutropenia hospitalization costs (for patients with any neutropenia hospitalization) and overall first-year Medicare costs after chemotherapy initiation (for the entire study sample) in elderly breast cancer patients.

This paper focuses on the effectiveness of primary prophylaxis and not secondary prophylaxis or therapeutic G-CSF administration. This study also examines the effect of duration of PPG-CSF administration on overall first-year Medicare costs in elderly female breast cancer patients receiving chemotherapy. The commercially available PPG-CSF during the study period (predominantly filgrastim) had to be administered daily for up to 14 days until the absolute neutrophil count reached 10,000 cells per microliter.43 Empirical studies show that PPG-CSF is often administered for 4 to 7 days in practice, and inadequate G-CSF administration is a concern since it is associated with reduced drug effectiveness.30,44,45 Since each shot of PPG-CSF costs more than $250, the study explores costs associated with increased duration of PPG-CSF administration. The effect of PPG-CSF duration on neutropenia hospitalization costs were not estimated due to limited sample size.

Methods

Data Set

This study uses the Surveillance, Epidemiology and End Results (SEER) data from 16 registries containing newly diagnosed breast cancer cases from 1994 to 2002 linked to Medicare claims through 2003. The 16 SEER registries are representative of approximately 25% of the US population, of which the Medicare population is a subset.46 The data are valid, high quality, and complete in terms of cancer incidence and diagnosis reporting in the United States.46

Study Cohort

The study period begins in 1994 after the introduction of G-CSF Healthcare Common Procedure Coding System codes in Medicare claims. To allow for the assessment of prediagnostic comorbidity from 1 year of Medicare claims prior to diagnosis, the study sample was limited to women aged 66 years and older. Women with stage 0 breast cancer were excluded because they do not require chemotherapy. Women with stage IV disease were excluded because chemotherapy is used as palliative rather than curative therapy in these patients. To ensure complete claims, the sample was limited to women enrolled in both Medicare part A and B, and not enrolled in an HMO, for 1 year before and after their diagnosis. The sample was limited to women whose chemotherapy was initiated within the first 6 months of diagnosis because chemotherapy initiation after 6 months could be for recurrence and not for the primary tumor. The initial study sample thus included 10,441 patients aged 66 years and older who were diagnosed with stage I to III breast cancer during 1994 to 2002 and who received chemotherapy within 6 months of diagnosis. The study period was particularly chosen because the period offers a clean sample for exploring the impact of short-acting PPG-CSF on Medicare costs. The utilization of long-acting pegfilgrastim among Medicare patients began in 2003. Unlike short-acting PPG-CSF, pegfilgrastim is administered once per chemotherapy cycle (every 2-3 weeks); currently both shortand long-acting G-CSFs are used.

Study Measures and Framework

Dependent Variables. Two types of dependent variables were used in the analysis. (1) The hospitalization costs for the first neutropenia hospitalization within 3 months and within 6 months after first-course chemotherapy initiation were estimated using Medicare inpatient files. A hospitalization with International Classification of Diseases, Ninth Revision diagnosis code of 288.0X was considered a “neutropenia hospitalization.” The durations of 3 months and 6 months were chosen because 3 to 6 months is the typical duration of first-course chemotherapy during which a chemotherapy-induced neutropenia- hospitalization can occur. (2) Inpatient, outpatient, neutropephysician office, and durable medical equipment claims were used to construct the overall healthcare costs during the first year after chemotherapy initiation.

In this study, cost was measured from Medicare’s (payer’s) perspective; hence, cost is the actual amount paid by Medicare to the provider. Costs were reported in 2002 dollars using an inflation rate of 3%. Due to the right-skewed log-normal distribution of costs, the logarithm of costs were used based on the conclusions from the Box-Cox model specification test. The analysis of the effect of PPG-CSF on the first hospitalization costs was performed only for patients who were hospitalized. The analysis of the effect of PPG-CSF on overall Medicare costs was performed for all patients for a duration of 1 year after chemotherapy initiation (the point of initiation of the exposure), when the bulk of cancer treatment is administered.

Treatment Variables. Use of PPG-CSF was identified by procedure codes for all the commercially available GCSF drugs (ie, filgrastim [J1440 and J1441] and sargramostim [J2820]), from the Medicare claims. A dummy variable was created for whether or not PPG-CSF was administered.

Because G-CSF is administered both as a prophylactic and as a therapeutic drug for neutropenia, primary prophylactic use is hard to distinguish using claims data. However, neutropenic symptoms only begin a week after chemotherapy initiation and peak after 2 weeks.47,48 Consequently, any administration of G-CSF within 5 days of chemotherapy initiation cannot be therapeutic or secondary prophylactic. To prevent misclassification of therapeutic or secondary prophylactic G-CSF use as PPG-CSF, only G-CSF initiated within 5 days of first-course chemotherapy initiation was considered primary prophylaxis, which is consistent with the methods and findings of related studies.11,22,33,44,49 Studies also show that PPG-CSF initiation after the first 5 days of chemotherapy initiation is less effective in preventing neutropenia22,49; hence, the 5-day window is a crucial period to investigate.

Duration of PPG-CSF therapy was defined as a continuous variable measuring the number of consecutive days of PPGCSF administration. Duration was also defined as a dummy variable measuring whether or not PPG-CSF was administered for >5 days (adequate administration) based on previous literature.30,44,,45,49

Table 1

Confounders. The 3 categories of variables controlled for were: patient sociodemographic, clinical (overall and tumorrelated), and therapeutic characteristics.9,11,50-57 The details about each variable are provided in 2 previously published papers.24,56 Descriptive statistics for the confounders are provided in . Because administration of PPG-CSF and neutropenia hospitalizations occur as a result of chemotherapy initiation, all the clinical history variables including the modified Charlson Comorbidity Index score, past history of cancer, and history of infections, antibiotic use, and hospitalizations were measured with respect to chemotherapy initiation. In order to ensure that the confounders were pretreatment variables and not those influenced by PPG-CSF administration itself, therapy (eg, chemotherapy) characteristics included only those aspects that were decided upon before the administration of PPG-CSF (eg, characteristics of the first cycle of the first-course chemotherapy).

The variables in each of the categories (patient sociodemographic, clinical, and therapeutic characteristics) were included in the model based on past evidence from the literature.56 For the continuous variables, different specifications and higher-order terms were evaluated, resulting in the inclusion of the square of number of drugs in the first cycle and the square of the duration between the first and second cycle.

Statistical Analysis

Because this study used observational data, the estimate of average costs associated with PPG-CSF administration could be biased due to treatment selection; patients more vulnerable to the toxic effects of chemotherapy are more likely to receive PPG-CSF, so the treatment group might have a higher risk of neutropenia hospitalization and overall healthcare costs than the control group. Controlling for confounders in regression models would yield unbiased results only in the absence of omitted variable bias and model dependence. These biases are unavoidable in observational studies due to lack of complete data on confounders. Bias due to model dependence occurs when treated and untreated groups lack an area of common support. Untreated observations could be far outside the range of treated observations (or vice versa), thereby requiring extrapolations in ranges where there are no treated (or untreated) observations to compute the treatment effects. These extrapolations are dependent on the model specifications. Thus, model specification errors could bias the treatment effect estimates.58,59

Nonparametric preprocessing of the data by matching on the observed covariates reduces model dependence and facilitates analysis in the area of common support.58 Preprocessing might partially control for omitted variables if the unobserved heterogeneity is correlated with the observed variables. Parametric analysis after preprocessing controls for any residual imbalances, making the estimation doubly robust.60

In order to address concerns about bias, nonparametric matching was performed before estimating the treatment effect.61,62 The matching only discarded untreated observations that did not match with treated observations. Linear regression was used to estimate the effect of PPG-CSF on neutropephysiciania hospitalization costs and annual overall healthcare costs using the preprocessed/matched data and the unmatched data for comparison.

Results

After matching, 1760 observations were used for the analysis. The descriptive statistics for treated and untreated groups for the unmatched and matched samples are presented in Table 1. The statistically significant differences in covariates that existed before matching were no longer present after matching. However, 3 variables became significantly different. These variables were not statistically correlated with PPG-CSF administration before matching; hence, the nonparametric matching algorithm assigned smaller weights to these variables. Thus, their balance was compromised while ensuring good balance in variables that were significant determinants of PPG-CSF administration. Performing the matching multiple times with different specifications gave similar results. Because parametric analysis was performed to control for remaining imbalances, further analysis was performed on this matched pool.

Table 2

presents the descriptive statistics for the overall Medicare cost and the cost of the first neutropenia hospitalization in the first 3 months and 6 months in patients who received PPG-CSF (column 1), and in patients who did not receive PPG-CSF before matching (column 2) and after matching (column 3). Marginal effects from the log-ordinary least squares (OLS) model (in percentages) and predicted mean cost with and without PPG-CSF administration, after controlling for all the covariates, are also displayed.

The difference in the hospitalization costs with and without PPG-CSF administration was not statistically significant. However, the mean hospitalization cost was lower in magnitude for patients who received PPG-CSF, and the difference was more pronounced after matching (Table 2).

Table 3

The descriptive analysis of the overall Medicare costs for a year after chemotherapy initiation showed that the costs were on average higher for patients who received PPG-CSF than for patients who did not receive PPG-CSF both before and after matching (Table 2). This trend continued among patients who received a longer duration of PPG-CSF therapy (), such that the patients receiving 5 or more days of PPG-CSF had higher overall Medicare costs for a year after chemotherapy initiation compared with patients receiving fewer than 5 days of PPG-CSF.

The log-OLS analyses controlling for all covariates demonstrated similar statistically significant relationships. In the postmatching analysis, the overall costs were 57.25% higher in women who received PPG-CSF (Table 2). Receiving >5 days of PPG-CSF increased the overall cost by 19.62%, and increase in PPG-CSF administration by 1 day increased the overall cost by 1.74% (Table 3).

Discusion

The authors previously demonstrated that PPG-CSF administration reduced the incidence of neutropenia hospitalization.24 This study aimed to determine whether PPG-CSF administration lowered hospitalization expenditures once a patient was hospitalized and lowered the overall cost of breast cancer management in the first year after chemotherapy initiation. The study found that PPG-CSF therapy was not associated with significantly lower hospitalization costs among women hospitalized for neutropenia; however, overall Medicare costs during the first year after chemotherapy initiation were significantly higher.

Table 4

Figure 1

Figure 2

Multiple reasons could explain the higher overall costs in spite of effective primary prophylaxis. The administration of PPG-CSF increases the successful administration of firstcourse chemotherapy and radiation therapy.57 Because bulk of these therapies are administered in the first year, increased adherence to these therapies will increase the Medicare costs in the first year. It is also important to understand that G-CSF is expensive, and the cost of prophylactic G-CSF administration during the entire first-course chemotherapy could range from $5000 to $30,000. This considerable cost could offset any cost reductions in neutropenia management and hospitalizations during the first year of treatment. In order to understand the main driving factors behind higher first-year costs among women who received PPG-CSF, a descriptive analysis was performed to identify different components of these overall costs (). The components of overall first-year costs for women without and with PPG-CSF administration are presented in and .

In the matched data, the average G-CSF cost in the first year for women who received PPG-CSF was $7914 versus $1369 for women who did not receive PPG-CSF (but received it later in some cases). Chemotherapy costs were twice as high for women who received PPG-CSF, with a mean cost of $6444 among women who did not receive PPG-CSF and $11,242 among women who received PPG-CSF. Given the difference in costs due to G-CSF and chemotherapy, it is not surprising that women receiving PPG-CSF had higher costs in the first year in spite of cost reductions related to neutropenia hospitalization. On average, chemotherapy costs accounted for a 42% increase in the first-year Medicare costs and G-CSF administration costs accounted for a 57% increase in the firstyear Medicare costs.

Previous studies examining the economics of PPG-CSF administration can be broadly classified into 3 types: studies estimating costs associated with neutropenia hospitalization; studies using cost models to estimate the threshold above which prophylaxis of neutropenia becomes cost-effective; and studies estimating the reduction in hospitalization costs and other neutropenia-related costs due to G-CSF administration. Studies examining cost of neutropenia hospitalization have found that cost of care and inpatient care is around 1.5 to 2 times higher in women experiencing neutropenia.29,63 Neutropenia hospitalization costs could range from $10,000 to $30,000.26-29,64,65 Studies examining cost-effectiveness of G-CSF have demonstrated that PPG-CSF therapy is costreducing and cost-effective in patients with high risk of neutropenia, febrile neutropenia, or neutropenia hospitalization (>20% febrile neutropenia risk according to the currently accepted American Society of Clinical Oncology model).21,26,31,36,55 Studies looking at reductions in costs due to G-CSf administration are mostly clinical trials and retrospective chart reviews with low external validity and have mixed fi ndings.34,35,37,38 Some studies reveal a drop in costs, on average, in patients receiving G-CSf compared with those not receiving G-CSf,34-36 while some reveal no change in costs because the reduction in hospitalization costs is offset by the initial cost of G-CSf administration.37,38

In this study, neutropenia hospitalization costs were lower for patients receiving G-CSf, but the difference was not statistically signifi cant. Overall first-year costs were significantly higher for patients receiving G-CSf. However, the increase in costs was predominantly due to the cost of PPG-CSf and improved first-course chemotherapy administration. first-course chemotherapy administration among breast cancer patients is associated with improved disease prognosis.1-8 Hence, future studies should explore whether these short-term costs might translate into long-term benefi ts like improved survival, reduced recurrence, and lower disease progression—related costs.

This study has some limitations. first, the study examined elderly Medicare fee-for-service breast cancer patients. Hence, the generalizability of the effect of PPG-CSf on short-term costs is limited. Nevertheless, the effect of PPGCSf therapy on first-course chemotherapy administration in breast cancer patients should be similar for all ages and insurance categories. Moreover, because G-CSf is an expensive drug, PPG-CSf use can lead to short-term cost increases in most settings. Second, this study defi ned primary prophylactic administration as that initiated within the fi rst 5 days of chemotherapy to distinguish it from secondary prophylactic and therapeutic administrations. The restrictive definition of PPG-CSf ensured high specifi city but could have led to misclassification of patients receiving PPG-CSf at later chemotherapy cycles as nonrecipients. However, such misclassification would mainly lead to underestimation of the cost differences between the treated and untreated groups; therefore, the study provides conservative estimates of the treatment effects.

In conclusion, the increased short-term costs estimated in this study were chiefl y associated with improved short-term treatment administration. In chronic diseases like breast cancer, it is vital for cost analyses to account for long-term benefits and costs, indirect patient care costs, and quality-oflife aspects. future studies should look into long-term benefi ts and costs associated with PPG-CSf administration.Acknowledgments

This study used the linked Seer-Medicare database. The interpretation and reporting of these data are the sole responsibility of the authors. The authors acknowledge the efforts of the applied research Program, National Cancer Institute; the Office of research, Development and Information, Centers for Medicare & Medicaid Services; Information Management Services, Inc; and the Surveillance, epidemiology, and end results (Seer) Program tumor registries in the creation of the Seer-Medicare database. We appreciate the support provided by Paul Godley for providing us with data access and useful advice. We also appreciate George Pink’s advice as part of the first author’s (SSr) dissertation committee.

Author Affiliations: from Management, Policy and Community Health Division (SSr), School of Public Health, university of Texas Health Science Center, Houston, TX; Health Policy and Management (WrC, SCS), university of North Carolina, Chapel Hill, NC; Duke university School of Medicine (GHL), Duke Comprehensive Cancer Center, Durham, NC.

Funding Source: None.

Author Disclosures: The authors (SSr, WrC, SCS, GHL) report no relationship or financial interest with any entity that would pose a confl ict of interest with the subject matter of this article.

Authorship Information: Concept and design (SSr, WrC, SCS, GHL); acquisition of data (SSr, WrC); analysis and interpretation of data (SSr, SCS, GHL); drafting of the manuscript (SSr, SCS, GHL); critical revision of the manuscript for important intellectual content (SSr, WrC, SCS, GHL); statistical analysis (SSr, SCS); obtaining funding (WrC); administrative, technical, or logistic support (SSr); and supervision (SCS).

Address correspondence to: Suja S. rajan, MHa, MS, PhD, assistant Professor, Management, Policy and Community Health Division, School of Public Health, university of Texas Health Science Center, 1200 Herman Pressler St, Houston, TX 77030. e-mail: suja.s.rajan@uth.tmc.edu.1. Earle CC, Nattinger AB, Potosky AL, et al. Identifying cancer relapse using SEER-Medicare Data. Med Care. 2002;40(8)(suppl):IV-104- IV-117.

2. Early Breast Cancer Trialists’ Collaborative Group (EBCTCG). Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomized trials. Lancet. 2005;365(9472):1687-1717.

3. Breast Cancer Trialists’ Collaborative Group. Polychemotherapy for early breast cancer: an overview of the randomised trials. Lancet. 1998;352(9132):930-942.

4. Effects of adjuvant tamoxifen and of cytotoxic therapy on mortality in early breast cancer. An overview of 61 randomised trials among 28,896 women. Early Breast Cancer Trialists’ Collaborative Group. N Engl J Med. 1992;319(26):1681-1692.

5. Early Breast Cancer Trialists’ Collaborative Group (EBCTCG). Systemic Treatment of Early Breast Cancer by Hormonal, Cytotoxic, or Immune Therapy: 133 Randomised Trials Involving 31,000 Recurrences and 24,000 Deaths Among 75,000 Women. Lancet. 1988;339(8785):1- 15, 71-85.

6. Fisher B. Highlights from recent National Surgical Adjuvant Breast and Bowel Project studies in the treatment and prevention of breast cancer. CA Cancer J Clin. 1999;49(3):159-177.

7. Fisher B, Redmond C, Legault-Poisson S, et al. Postoperative chemotherapy and tamoxifen compared with tamoxifen alone in the treatment of positive-node breast cancer patients aged 50 years and older with tumors responsive to tamoxifen: results from the National Surgical Adjuvant Breast and Bowel Project B-16. J Clin Oncol. 1990;8(6):1005-1018.

8. Hortobagyi GN. Treatment of breast cancer. N Engl J Med. 1998; 339(14):974-984.

9. Shayne M, Crawford J, Dale DC, Culakova E, Lyman GH; ANC Study Group. Predictors of reduced dose intensity in patients with earlystage breast cancer receiving adjuvant chemotherapy. Breast Cancer Res Treat. 2006;100(3):255-262.

10. Ziegler J, Citron M. Dose-dense adjuvant chemotherapy for breast

cancer. Cancer Nurs. 2006;29(4):266-272.

11. Chrischilles EA, Link KB, Scott SD, Delgado DJ, Fridman M. Factors associated with early termination of CHOP therapy and the impact on survival among patients with chemosensitive intermediate grade non- Hodgkin’s lymphoma. Cancer Control. 2003;10(5):396-403.

12. Citron ML, Berry DA, Cirrincione C, et al. Randomized trial of dosedense versus conventionally scheduled and sequential versus concurrent combination chemotherapy as postoperative adjuvant treatment of node-positive primary breast cancer: First report of Intergroup Trial C9741/Cancer and Leukemia Group B Trial 9741. J Clin Oncol. 2003; 21(8):1431-1439.

13. Bonilla L, Ben-Aharon I, Vidal L, Gafter-Gvili A, Leibovici L, Stemmer SM. Dose-dense chemotherapy in nonmetastatic breast cancer: A systematic review and meta-analysis of randomized controlled trials. J Natl Cancer Inst. 2010;102(24):1845-1854.

14. Ozer H, Armitage JO, Bennett CL, et al; American Society of Clinical Oncology. 2000 update of recommendations for the use of hematopoietic colony-stimulating factors: evidence-based, clinical practice guidelines. American Society of Clinical Oncology Growth Factors Expert Panel. J Clin Oncol. 2000;18(20):3558-3585.

15. Baquiran DC. Biologic response modifiers. In: Baquiran DC, ed. Lippincott’s Cancer Chemotherapy Handbook. 2nd ed. Philadelphia, PA: Lippincott Williams & Wilkins; 2001:13-41.

16. Webster J, Lyman G. Use of G-CSF to sustain dose intensity in breast cancer patients receiving adjuvant chemotherapy: a pilot study. Cancer Control. 1996;3(6):519-523.

17. Welte K, Gabrilove J, Bronchud MH , Platzer E, Morstyn G. Filgrastim (r-metHuG-CSF): the first 10 years. Blood. 1996;88(6):1907-1929.

18. Hryniuk WM. The importance of dose intensity in the outcome of chemotherapy. In: DeVita VT, Hellman S, Rosenberg SA, eds. Important Advances in Oncology. Philadelphia, PA: JB Lippincott; 1988:121-141.

19. Budman DR, Berry DA, Cirrincion CT, et al. Dose and dose intensity as determinants of outcome in the adjuvant treatment of breast cancer. The Cancer and Leukemia Group B. J Natl Cancer Inst. 1998;90(16): 1205-1211.

20. Lyman GH, Kuderer N, Agboola O, Balducci L. Evidence-based use of colony-stimulating factors in elderly cancer patients. Cancer Control. 2003;10(6):487-499.

21. Lyman GH, Kuderer NM, Greene J, Balducci L. The economics of febrile neutropenia: implications for the use of colony-stimulating factors. Eur J Cancer. 1998;34(12):1857-1864.

22. Kuderer NM, Dale DC, Crawford J, Lyman GH. Impact of primary prophylaxis with granulocyte colony-stimulating factor on febrile neutropenia and mortality in adult cancer patients receiving chemotherapy: a systematic review. J Clin Oncol. 2007;25(21):3158-3167.

23. Lyman GH, Kuderer NM, Djulbegovic B. Prophylactic granulocyte colony-stimulating factor in patients receiving dose-intensive cancer chemotherapy: a meta-analysis. Am J Med. 2002;112(5):406-411.

24. Rajan SS, Lyman GH, Stearns SC, Carpenter WR. Effect of primary prophylactic granulocyte colony-stimulating factor use on incidence of neutropenia hospitalizations for elderly early-stage breast cancer patients receiving chemotherapy. Med Care. 2011;49(7):649-657.

25. Caggiano V, Weiss RV, Richert TS, Linde-Zwirble WT. Incidence, cost, and mortality of neutropenia hospitalization associated with chemotherapy. Cancer. 2005;103(9):1916-1924.

26. Eldar-Lissai A, Cosler LE, Culakova E, Lyman GH. Economic analysis of prophylactic pegfilgrastim in adult cancer patients receiving chemotherapy. Value Health. 2007;11(2):172-179.

27. Kuderer NM, Dale DC, Crawford J, Cosler LE, Lyman GH. Morbidity, mortality and cost associated with febrile neutropenia in adult cancer patients. Cancer. 2006;106(10):2258-2266.

28. Weycker D, Malin J, Glass A, Oster G. Economic burden of chemotherapy-related febrile neutropenia. J Support Oncol. 2007;5(4) (suppl 2):44-45.

29. Weycker D, Malin J, Glass A, Oster G. Cost of chemotherapy-related febrile neutropenia [abstract]. J Clin Oncol. 2006 ASCO Annual Meeting Proceedings (Post-Meeting Edition). 2006;24(18S) (June 20 suppl). http://meeting.ascopubs.org/cgi/content/abstract/24/18_suppl/6068. Published June 2006. Accessed January 11, 2013.

30. Weycker D, Hackett J, Edelsberg JS, Oster G, Glass AG. Duration of G-CSF therapy and risk of hospitalization for neutropenia or infection [abstract 6731]. J Clin Oncol. 2004 ASCO Annual Meeting Proceedings (Post-Meeting Edition). 2004;22(14S) (July 15 suppl).

31. Lyman GH, Lyman CG, Sanderson RA, Balducci L. Decision analysis of hematopoietic growth factor use in patients receiving cancer chemotherapy. J Natl Cancer Inst. 1993;85(6):488-493.

32. Crawford J, Ozer H, Stoller R, et al. Reduction by granulocyte colony-stimulating factor of fever and neutropenia induced by chemotherapy in patients with small-cell lung cancer. N Engl J Med. 1991;325(3):164-170.

33. Chrischilles E, Delgado DJ, Stolshek BS, Lawless G, Fridman M, Carter WB. Impact of age and colony-stimulating factor use on hospital length of stay for febrile neutropenia in CHOP-treated Non-Hodgkin’s lymphoma. Cancer Control. 2002;9(3):203-211.

34. Glaspy JA, Bleecker G, Crawford J, Stoller R, Strauss M. The impact of therapy with filgrastim (recombinant granulocyte colony-stimulating factor) on the health care costs associated with cancer chemotherapy. Eur J Cancer. 1993;29A(suppl 7):S23-S30.

35. Zagonel V, Babare R, Merola MC, et al. Cost-benefit of granulocyte colony-stimulating factor administration in older patients with non- Hodgkin’s lymphoma treated with combination chemotherapy. Ann Oncol. 1994;5(suppl 2)127-132.

36. Uyl-de Groot CA, Vellenga E, Rutten FF. An economic model to assess the savings from a clinical application of hematopoietic growth factors. Eur J Cancer. 1996;32A(1):57-62.

37. Dranitsaris G, Sutcliffe SB. Economic analysis of prophylactic G-CSF after mini-BEAM salvage chemotherapy for Hodgkin’s and non- Hodgkin’s lymphoma. Leuk Lymphoma. 1995;17(1-2):139-145.

38. Bassan R, Lerede T, Di Bona E, et al. Granulocyte colony-stimulating factor (G-CSF, filgrastim) after or during an intensive remission induction therapy for adult acute lymphoblastic leukaemia: effects, role of patient pretreatment characteristics, and costs. Leuk Lymphoma. 1997;26(1-2):153-161.

39. Hassett MJ, O’Malley AJ, Pakes JR, Newhouse JP, Earle CC. Frequency and cost of chemotherapy-related serious adverse effects in a population sample of women with breast cancer. J Natl Cancer Inst. 2006;98(16):1108-1117.

40. Russo A, Autelitano M, Bisanti L. Re: frequency and cost of chemotherapy- related serious adverse effects in a population sample of women with breast cancer. J Natl Cancer Inst. 2006;98(24):1826-1827.

41. Du XL, Osborne C, Goodwin JS. Population-based assessment of hospitalizations for toxicity from chemotherapy in older women with breast cancer. J Clin Oncol. 2002;20(24):4636-4642.

42. Smith TJ, Khatcheressian J, Lyman GH, et al. 2006 update of recommendations for the use of white blood cell growth factors: an evidence-based clinical practice guideline. J Clin Oncol. 2006;24(19): 3187-3205.

43. Neupogen (Filgrastim) [package insert]. Thousand Oaks, CA: Amgen;

2009.

44. Weycker D, Hackett J, Edelsberg JS, Oster G, Glass AG. Are shorter courses of filgrastim prophylaxis associated with increased risk of hospitalization? Ann Pharmacother. 2006;40(3):402-407.

45. Scott SD, Chrischilles EA, Link BK, et al. Days of prophylactic filgrastim use to reduce febrile neutropenia in patients with non-Hodgkin’s lymphoma treated with chemotherapy. J Manag Care Pharm. 2003;9(2 suppl):15-21.

46. Warren JL, Klabunde CN, Schrag Deborah, Bach PB, Riley GF. Overview of the SEER-Medicare data: content, research applications, and generalizability to the United States elderly population. Med Care. 2002;40(8)(suppl):IV-3-IV-18.

47. Perry MC, Yarbro JW. Toxicity of Chemotherapy. Orlando, FL: Grune & Stratton; 1984.

48. Shapiro CL, Recht A. Side effects of adjuvant treatment of breast cancer. N Engl J Med. 2001;344(26):1997-2008.

49. Crawford J, Kreisman H, Garewal H, et al. The impact of filgrastim schedule variation on hematopoietic recovery post-chemotherapy. Ann Oncol. 1997;8(11):1117-1124.

50. National Comprehensive Cancer Network. Myeloid Growth Factors. NCCN Guidelines for Supportive Care. http://www.nccn.org/ professionals/physician_gls/f_guidelines.asp#supportive. Published 2010. Accessed December 2011.

51. Lyman GH, Dale DC, Crawford J. Incidence and predictors of low dose-intensity in adjuvant breast cancer chemotherapy: a nationwide study of community practices. J Clin Oncol 2003;21(24):4524-4531.

52. Gómez H, Hidalgo M, Casanova L, et al. Risk factors for treatmentrelated death in elderly patients with aggressive non-Hodgkin’s lymphoma: results of a multivariate analysis. J Clin Oncol. 1998;16(6):

2065-2069.

53. Du XL, Lairson DR, Begley E, Fang S. Temporal and geographic variation in the use of hematopoietic growth factors in older women receiving breast cancer chemotherapy: findings from a large population-based cohort. J Clin Oncol. 2005;23(34):8620-8628.

54. Lyman GH. Risk assessment in oncology clinical practice: from risk factors to risk models. Oncology (Williston Park). 2003;17(11)(suppl 11):8-13.

55. Lyman GH, Kuderer NM. Economics of hematopoietic growth factors. In: Morstyn G, Foote M, Lieschke GJ, eds. Hematopoietic Growth Factors in Oncology: Basic Science and Clinical Therapeutics. Totowa, NJ: Humana Press Inc; 2004:409-443.

56. Rajan SS, Lyman GH, Carpenter WR, Stearns SC. Chemotherapy characteristics are important predictors of primary prophylactic CSF administration in older patients with breast cancer. Breast Cancer Res Treat. 2011;127(2):511-520.

57. Rajan SS, Stearns SC, Lyman GH, Carpenter WR. Effect of primary prophylactic G-CSF use on systemic therapy administration for elderly breast cancer patients. Breast Cancer Res Treat. 2011;130(1):255-266.

58. Ho DE, Imai IK, King G, Stuart EA. Matching as nonparametric preprocessing for reducing model dependence in parametric causal inference. Polit Anal. 2007;15:199-236.

59. King G, Zeng L. The dangers of extreme counterfactuals. Polit Anal. 2006;14:131-159.

60. Robins J, Rotnitzky A. Comment on “Inference for semiparametric models: some questions and an answer” by P. J. Bickel and J. Kwon. Statistica Sinica. 2001;11:920-936.

61. Sekhon JS, Mebane WRJ. Genetic optimization using derivatives: theory and application to nonlinear models. Polit Anal. 1998;7(1): 187-210.

62. Diamond A, Sekhon J. Genetic matching for estimating causal effects: a genetic multivariate matching method for achieving balance in observational studies. Rev Econ Stat. In press.

63. Gandhi SK, Arguelles L, Boyer JG. Economic impact of neutropenia and febrile neutropenia in breast cancer: estimates from two national databases. Pharmacotherapy. 2001;21(6):684-690.

64. Brooks JM, Klepser DG, Chrischilles EA, Voelker MD, Chen-Hardee SS, Scott SD, Link BK, Delgado DJ. Early termination of first course chemotherapy among Medicare beneficiaries newly diagnosed with non-Hodgkin’s lymphoma: National SEER-Medicare study. Proc Am J Clin Oncol. 2003;22: (abstr 2145).

65. Chrischilles EA, Klesper DG, Brooks JM, et al. Effect of clinical characteristics on neutropenia-related inpatient costs among newly diagnosed non-Hodgkin’s lymphoma cases during first course chemotherapy. Pharmacotherapy. 2005;25(5):668-675.