Abstract
Myeloid colony-stimulating factors (CSFs) decrease the risk of febrile neutropenia (FN) from high-risk chemotherapy regimens administered to patients at 20% or greater risk of FN, but little is known about their use in clinical practice. We evaluated CSF use in a multiregional population-based cohort of lung and colorectal cancer patients (N = 1849). Only 17% (95% confidence interval [CI] = 8% to 26%) patients treated with high-risk chemotherapy regimens received CSFs, compared with 18% (95% CI = 16% to 20%) and 10% (95% CI = 8% to 12%) of patients treated with intermediate- (10%–20% risk of FN) and low-risk (<10% risk of FN) chemotherapy regimens, respectively. Using a generalized estimating equation model, we found that enrollment in a health maintenance organization (HMO) was strongly associated with a lower adjusted odds of discretionary CSF use, compared with non-HMO patients (odds ratio = 0.44, 95% CI = 0.32 to 0.60, P < .001). All statistical tests were two-sided. Overall, 96% (95% CI = 93% to 98%) of CSFs were administered in scenarios where CSF therapy is not recommended by evidence-based guidelines. This finding suggests that policies to decrease CSF use in patients at lower or intermediate risk of FN may yield substantial cost savings without compromising patient outcomes.
CONTEXT AND CAVEATS
Prior knowledge
Use of colony-stimulating factors (CSFs) with chemotherapy regimens decreases the risk of febrile neutropenia (FN) in patients who are at high risk. CSFs are expensive ($2000 per cycle of chemotherapy), but not much is known about whether it is under- or overused in clinical practice.
Study design
Associations between CSF use outside of the American Society of Clinical Oncology and National Comprehensive Cancer Network guidelines and multiple sociodemographic and clinical factors, chemotherapy regimens (high-risk, intermediate-risk, and low-risk), and health maintenance organization (HMO) enrollment were assessed in 1785 lung or colorectal cancer patients enrolled by the Cancer Care Outcomes Research and Surveillance Consortium (CanCORS).
Contribution
Only 17% patients receiving high-risk chemotherapy regimens (≥20% risk of FN) received CSFs compared with 18% intermediate- (10–20% risk of FN) and 10% low-risk (<10% risk of FN) regimens. Factors that showed a strong association with CSF use included intermediate-risk regimens, severe comorbidity, having small-cell lung cancer, and non-HMO enrollment. Overall 96% of CSF use occurred outside the current evidence-based American Society of Clinical Oncology and National Comprehensive Cancer Network guidelines.
Implications
Lower discretionary use of CSFs in HMO plans (financial compensation has no incentives for physicians), suggests that financial interests could be driving the higher discretionary use of CSFs in non-HMO plans, and decreasing the unnecessary use of CSFs may reduce costs without compromising the quality of treatment.
Limitations
CSF use may have changed in patients since the time they were treated (2004–2006), and data on white blood cell counts or diagnoses of FN episodes were not obtained. The analysis may have underestimated the use of CSF.
From the Editors
Neutropenia is a potentially serious complication of chemotherapy that increases the risk of life-threatening infections because of an abnormally low number of neutrophils in the blood. Myeloid colony-stimulating factors (CSFs), when given prophylactically, substantially decrease the risk of febrile neutropenia (FN) (1,2) and are widely used in clinical practice since their approval by the Food and Drug Administration (FDA) in the early 1990s (2–5). CSFs now used in practice include granulocyte-CSF (G-CSF; filgrastim), pegylated G-CSF (pegfilgrastim), and granulocyte-macrophage-CSF (GM-CSF; sargramostim). The American Society of Clinical Oncology (ASCO) first introduced guidelines on the use of CSFs in 1994, which recommend primary prophylaxis (ie, with the first cycle of chemotherapy) with a CSF when the anticipated risk of FN associated with chemotherapy is 40% or higher (changed to ≥20% in 2006) (6,7). The National Comprehensive Cancer Network (NCCN) guidelines also recommend CSFs when FN risk is high (≥20%) and recommend consideration of CSFs when there is an intermediate risk (10%–20%) of chemotherapy-induced FN (8,9). The NCCN guidelines also suggest consideration of CSFs for secondary prophylaxis (ie, in patients who experience FN in a previous chemotherapy cycle) if dose reduction may compromise survival. None of the above guidelines recommend using CSFs to maintain chemotherapy dose and schedule.
Although CSFs are an important therapeutic advance for patients at high risk of neutropenic complications, little is known about their use in general clinical practice. The only population-based study conducted thus far examined the discretionary use of CSFs in the region of western Washington and suggested that they are most likely overused (10). Given the high costs of CSFs ($2000 per chemotherapy cycle), understanding the multiple factors associated with their use may have important implications for optimizing the use of CSFs in clinical practice.
In this study, we examined the patterns of CSF use in a population-based, observational, multiregional cohort of lung and colorectal cancer patients receiving care in diverse health-care settings and assessed the association of clinical factors and type of insurance with discretionary CSF use outside of the ASCO and NCCN guidelines. Study subjects were patients aged 21 years or older who were diagnosed with lung or colorectal cancer from September, 2003, through December, 2005, and enrolled by the Cancer Care Outcomes Research and Surveillance Consortium (CanCORS) (11). CanCORS patients have been shown to be representative of all cancer cases reported by National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) Program (12). Data collection included surveys of patients or their surrogates approximately 4 months after diagnosis. Data on chemotherapy regimens, CSF use, cancer stage and histology, and comorbidities (13) were abstracted from medical records through 15 months after diagnosis. We used information from four of the seven CanCORS data collection sites, which collected complete information on CSF use. Three sites were population-based cancer registries, and the fourth was a system of health maintenance organizations (HMOs). These sites obtained local institutional review board approval and patient consent to participate.
We classified chemotherapy regimens received by FN risk as follows: low (<10% risk), intermediate (10%–20% risk), or high (≥20%) using NCCN guidelines (14). Primary prophylaxis was defined as receiving CSF within the first 5 days of initiating a new chemotherapy regimen. Because each patient could have received CSFs over multiple chemotherapy regimens, a generalized estimating equation regression model was used with multiple observations (ie, chemotherapy regimens) per patient, thus accounting for autocorrelation effects (clustering of regimens within patients). The model was used to examine associations between CSF use (yes or no) and independent sociodemographic (age, race or ethnicity, sex), clinical (cancer type, stage, histology, comorbidities, FN), insurance type (HMO vs non-HMO), and chemotherapy regimen (based on risk of FN) covariates. We performed two-sided Wald χ2 tests to assess statistical significance of the association with covariates, and all P values less than .05 were considered to be statistically significant.
Overall, we identified 1849 patients who received chemotherapy that formed our study cohort. Only 64 of these patients received a “high-risk” chemotherapy regimen, identified by the inclusion of the drug topotecan, which is linked with a high risk (≥20%) of FN (14). Among these 64 high-risk patients, 11 received CSFs (17%; 95% confidence interval [CI] = 8% to 26%). This represented 4% (95% CI = 2% to 7%) of all patients who received CSFs (n = 268) in the study cohort. This finding suggested that given the strong evidence base and guidelines for CSF prophylaxis, interventions to increase CSF use was warranted in these patients.
Because so few patients received high-risk chemotherapy regimens, we focused our analyses on the factors that are associated with CSF use among patients receiving low- and intermediate-risk regimens, where all CSF use is discretionary. Table 1 shows the distribution of the sociodemographic, clinical, and health-care setting characteristics of these patients, after exclusion of 64 high-risk patients (n = 1785). A total of 982 of 1785 (55%; 95% CI = 53% to 57%) patients received at least one chemotherapy regimen classified as intermediate-risk of FN (10%–20%) based on NCCN guidelines. The unadjusted crude rates of the receipt of CSFs for each patient covariate along with odds ratios (ORs) of receipt of CSFs adjusted for all other covariates are also shown in Table 1.
Table 1.
Rates and associations of multiple factors with receipt of colony-stimulating factors (CSFs) in intermediate-risk and low-risk patients*
| Covariate | No. of patients (%)† | No. of CSF recipients (%)‡ | Adjusted OR (95% CI)§ | P‖ |
| Age at diagnosis, y | ||||
| <65 (reference) | 876 (49) | 113 (13) | 1.0 (referent) | |
| 65–74 | 548 (31) | 86 (16) | 1.21 (0.89 to 1.64) | .22 |
| >75–82 | 361 (20) | 58 (16) | 1.12 (0.79 to 1.58) | .53 |
| Race or Ethnicity | ||||
| White (reference) | 1295 (73) | 189 (15) | 1.0 (referent) | |
| Non-Hispanic Black | 168 (9) | 25 (15) | 1.21 (0.77 to 1.91) | .42 |
| Hispanic | 122 (7) | 18 (15) | 1.18 (0.70 to 1.99) | .54 |
| Other Race | 200 (11) | 25 (13) | 1.35 (0.86 to 2.11) | .19 |
| Sex | ||||
| Men (reference) | 768 (43) | 126 (16) | 1.0 (referent) | |
| Women | 1017 (57) | 131 (13) | 0.76 (0.58 to 0.99) | .04 |
| Cancer type | ||||
| Colorectal cancer (reference) | 654 (37) | 58 (9) | 1.0 (referent) | |
| NSCLC | 921 (52) | 129 (14) | 0.88 (0.59 to 1.31) | .53 |
| SCLC | 210 (12) | 70 (33) | 3.10 (2.05 to 4.69) | <.001 |
| Stage at diagnosis¶ | ||||
| Stage I–III (reference) | 1090 (61) | 137 (13) | 1.0 (referent) | |
| Stage IV | 695 (39) | 120 (17) | 1.23 (0.93 to 1.61) | .14 |
| Comorbidity (ACE-27)# | ||||
| None (reference) | 424 (24) | 50 (12) | 1.0 (referent) | |
| Mild | 743 (42) | 104 (14) | 1.24 (0.86 to 1.80) | .25 |
| Moderate | 330 (18) | 40 (12) | 1.25 (0.81 to 1.91) | .31 |
| Severe | 288 (16) | 63 (22) | 1.77 (1.16 to 2.70) | .01 |
| Regimen risk for FN** | ||||
| Low (reference) | 803 (45) | 81 (10) | 1.0 (referent) | |
| Intermediate | 982 (55) | 176 (18) | 1.44 (1.04 to 1.99) | .03 |
| FN | ||||
| No (reference) | 1661 (93) | 230 (14) | 1.0 (referent) | |
| Yes | 124 (7) | 27 (22) | 1.39 (0.88 to 2.20) | .16 |
| HMO enrollment | ||||
| No (reference) | 1174 (66) | 207 (18) | 1.0 (referent) | |
| Yes | 611 (34) | 50 (8) | 0.44 (0.32 to 0.60) | <.001 |
| Total No. (%) | 1785 (100) | 257 (17) |
All patients included in this study were recruited at the following CanCORS data collection centers that collected data on the use of CSF: HMO Cancer Research Network covering the states of Alabama and Iowa, and Los Angeles County of California. ACE= Adult Comorbidity Evaluation; AJCC = American Joint Committee on Cancer; CanCORS = Cancer Care Outcomes Research and Surveillance Consortium; CI = confidence interval; FN =Febrile Neutropenia; HMO = Health Maintenance Organization; NSCLC = non–small cell lung cancer; OR = odds ratio; SCLC = small cell lung cancer.
Numbers and percentages of patients in respective categories of patient characteristics (covariates). Percentage denominators were entire sample of 1785 patients receiving either low- or intermediate-risk chemotherapy regimens.
Numbers and percentages of CSF recipients in respective categories of patient covariates. Percentage denominators were the combined numbers of both recipients and nonrecipients of CSFs that belong to a particular covariate.
Adjusted odds ratio estimates were computed as the natural antilogarithm of estimates of the logarithm of the odds ratios (log odds). Similarly, 95% confidence intervals for odds ratio estimates were computed as the antilogarithm of confidence intervals of the log odds ratio estimates. Log odds estimates were computed using a generalized estimating equation model.
P values were calculated using a two-sided Wald χ2 test for testing the null hypothesis of no association (ie, the odds ratio = 1).
AJCC staging system (15) was used.
Comorbidity was measured using the ACE-27 index which categorizes each patient as having no, mild, moderate, or severe comorbidity based on medical chart review of the presence and severity of 27 specific medical conditions (16). Patients are assigned to one of the categories based on the highest level of severity detected across all 27 conditions.
Risk of febrile neutropenia can vary by regimen for patients receiving more than a single chemotherapy regimen.
Overall, 9% (95% CI = 7% to 11%) of patients diagnosed with colorectal cancer, 14% (95% CI = 12% to 16%) of patients diagnosed with non–small cell lung cancer, and 33% (95% CI = 27% to 40%) with small cell lung cancer ever received a CSF (Table 1). Also, 10% (95% CI = 8% to 12%) of the low-risk regimen patients received CSFs compared with 18% (95% CI = 16% to 20%) of the intermediate-risk regimen patients. Most CSF use was not for primary prophylaxis: 13% of small cell lung cancer (95% CI = 9% to 18%), 4% of non–small cell lung cancer (95% CI = 3% to 6%), and less than 1% (95% CI = 0.3% to 2%) of colorectal cancer patients received CSFs with the first cycle of a chemotherapy regimen (data not shown). Neither FN during a previous cycle (indicating secondary prophylaxis use) nor stage was associated with increased odds of receiving CSFs. Although the ASCO and NCCN guidelines suggest consideration of CSF, prophylaxis in patients older than 65 years was not associated with increased odds of receiving CSF. Factors that were statistically significantly associated with receipt of CSF included treatment with an intermediate-risk regimen (OR = 1.44, 95% CI = 1.04 to 1.99, P = .03), having small cell lung cancer vs colorectal cancer (OR = 3.10, 95% CI = 2.05 to 4.69; P < .001), and having severe vs no comorbidity (OR = 1.77, 95% CI = 1.16 to 2.70; P = .01). The odds of receiving CSFs were lower in women (OR = 0.76, 95% CI = 0.58 to 0.99; P = .04) and patients enrolled in group or staff HMO plans (OR = 0.44, 95% CI = 0.32 to 0.60; P < .001).
In this population-based, observational, multiregional cohort of lung and colorectal cancer patients, only 17% of patients at high risk of FN received a CSF, much lower than the percentage reported by previous studies conducted in either academic centers (17) or in selected community oncology practices (2,4,5). Most CSF use was neither for primary or secondary prophylaxis of FN as recommended by the guidelines but instead appeared to be reactive (ie, in response to neutropenia), most likely to maintain dose and schedule. Although we were unable to explicitly distinguish reasons for discretionary CSF use, the low incidence of FN among those receiving CSF in our sample of patients not treated with high-risk chemotherapy regimens suggested that only a small percentage of the discretionary use of CSF was in response to FN. However, cancer stage was not associated with CSF use, suggesting that the intent of therapy (curative vs palliative) did not appear to influence the decision to use CSFs. We did not find an association with age, despite recent evidence suggesting that older lung cancer patients may be at increased risk for FN (18). Discretionary CSF use was strongly associated with whether care was delivered within an HMO, where physician compensation is generally not directly affected by drug administration, vs outside an HMO. This suggests that financial incentives may be driving the largely discretionary use of CSFs, at least in part. Other factors, such as a more uniform, system-wide approach to care may help explain our finding of lower discretionary use of CSFs in HMO plans. The HMO plans included in our study did not place any specific restrictions on CSF use during the study period.
Our study has a few limitations. This included possible changes in CSF use since 2003–2005 when these patients were diagnosed and first treated, inability to collect data on white blood cell counts or diagnoses of FN episodes, and possible underestimation of CSF use because of errors in medical record abstraction or inability to access the treating oncologist’s chart.
Despite these limitations, our study had several strengths. It is among the first to describe variations in practice patterns for CSF use in a population-based cohort of patients of all age groups. Other strengths of our study were its population-based design, which reflects a range of health-care delivery settings; diverse patient population with respect to age, race or ethnicity, and socioeconomic status; and a standardized medical record abstraction component linked with registry data and patient surveys.
Overall, 96% (95% CI = 93% to 98%) of CSFs in lung and colorectal cancer patients were administered in clinical situations where CSF therapy is not recommended by current evidence-based guidelines. This finding suggests that policies to decrease CSF use in lower-risk patient subsets may yield substantial cost savings without compromising patient outcomes. Finally, research is needed to better guide patient selection, including determination of whether CSFs reduce FN risk among specific patient subgroups defined by age, comorbidities, treatments received, or other prognostic risk factors.
Funding
The work of the Cancer Care Outcomes Research and Surveillance Consortium (CanCORS) was supported by grants from the National Cancer Institute to the Statistical Coordinating Center (U01 CA093344); Primary Data Collection and Research Centers (Dana-Farber Cancer Institute and the Cancer Research Network) (U01 CA093332); Harvard Medical School and Northern California Cancer Center (U01 CA093324); RAND and University of California Los Angeles (U01 CA093348); University of Alabama at Birmingham (U01 CA093329); University of Iowa (U01 CA093339); University of North Carolina (U01 CA 093326); and by a grant from the Department of Veterans Affairs to the Durham Veterans Affairs Medical Center (CRS 02-164).
Footnotes
Dr J. L. Malin was an employee of Amgen Inc, the manufacturer of granulocyte-colony stimulating factor (G-CSF; filgrastim), from 2005–2007, and has served as a consultant to Amgen on research studies. All human investigations were performed after approval by a local Human Investigations Committee at each participating site and in accord with an assurance filed with and approved by the Department of Health and Human Services, where appropriate. The investigators obtained informed consent from each participant or each participant’s guardian. The authors are solely responsible for the design of the study, analysis or interpretation of the results, writing of the article, and decision to submit the article for publication.
References
- 1.Heuser M, Ganser A, Bokemeyer C American Society of Clinical Oncology; National Comprehensive CN; European Organization for Research and Treatment of Cancer. Use of colony-stimulating factors for chemotherapy-associated neutropenia: review of current guidelines. Semin Hematol. 2007;44(3):148–156. doi: 10.1053/j.seminhematol.2007.04.002. [DOI] [PubMed] [Google Scholar]
- 2.Hershman D, Hurley D, Wong M, Morrison VA, Malin JL. Impact of primary prophylaxis on febrile neutropenia within community practices in the US. J Med Econ. 2009;12(3):203–210. doi: 10.3111/13696990903238603. [DOI] [PubMed] [Google Scholar]
- 3.Bennett CL, Somerfield MR, Pfister DG, et al. Perspectives on the value of American Society of Clinical Oncology clinical guidelines as reported by oncologists and health maintenance organizations. J Clin Oncol. 2003;21(5):937–941. doi: 10.1200/JCO.2003.07.165. [DOI] [PubMed] [Google Scholar]
- 4.Swanson G, Bergstrom K, Stump E, Miyahara T, Herfindal ET. Growth factor usage patterns and outcomes in the community setting: collection through a practice-based computerized clinical information system. J Clin Oncol. 2000;18(8):1764–1770. doi: 10.1200/JCO.2000.18.8.1764. [DOI] [PubMed] [Google Scholar]
- 5.Naiem A, Friedman L, Pasta DJ, et al. Prophylaxis of chemotherapy-induced neutropenia: patterns of care in U.S. community oncology practices [2007 ASCO Annual Meeting Proceedings Part I] J Clin Oncol. 2007;25(18S):9123. [Google Scholar]
- 6.Bennett CL, Weeks JA, Somerfield MR, Feinglass J, Smith TJ. Use of hematopoietic colony-stimulating factors: comparison of the 1994 and 1997 American Society of Clinical Oncology surveys regarding ASCO clinical practice guidelines. Health Services Research Committee of the American Society of Clinical Oncology. J Clin Oncol. 1999;17(11):3676–3681. doi: 10.1200/JCO.1999.17.11.3676. [DOI] [PubMed] [Google Scholar]
- 7.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. doi: 10.1200/JCO.2006.06.4451. [DOI] [PubMed] [Google Scholar]
- 8.Crawford J, Althaus B, Armitage J, et al. Myeloid growth factors. Clinical practice guidelines in oncology. J Natl Compr Canc Netw. 2007;5(2):188–202. doi: 10.6004/jnccn.2007.0019. [DOI] [PubMed] [Google Scholar]
- 9.McNeil C. NCCN guidelines advocate wider use of colony-stimulating factor. J Natl Cancer Inst. 2005;97(10):710–711. doi: 10.1093/jnci/97.10.710. [DOI] [PubMed] [Google Scholar]
- 10.Ramsey SD, McCune JS, Blough DK, et al. Colony-stimulating factor prescribing patterns in patients receiving chemotherapy for cancer. Am J Manag Care. 2010;16(9):678–686. [PubMed] [Google Scholar]
- 11.Ayanian JZ, Chrischilles EA, Fletcher RH, et al. Understanding cancer treatment and outcomes: the Cancer Care Outcomes Research and Surveillance Consortium. J Clin Oncol. 2004;22(15):2992–2996. doi: 10.1200/JCO.2004.06.020. [DOI] [PubMed] [Google Scholar]
- 12.Catalano PJ, Ayanian JZ, Harrington DP, Kahn KL, Landrum MB, Zaslavsky AM. Paper presented at Academy Health Annual Research meeting. Washington, DC: 2008. Representativeness of the CanCORS participants relative to Surveillance, Epidemiology & End Results (SEER) cancer registries. ; June 8. [Google Scholar]
- 13.Piccirillo JF, Tierney RM, Costas I, Grove L, Spitznagel EL., Jr Prognostic importance of comorbidity in a hospital-based cancer registry. JAMA. 2004;291(20):2441–2447. doi: 10.1001/jama.291.20.2441. [DOI] [PubMed] [Google Scholar]
- 14.National Comprehensive Cancer Network. NCCN Clinical Practice Guidelines in Oncology™. Myeloid Growth Factors Guidelines, Version 1. Fort Washington, PA: 2009. [Google Scholar]
- 15.Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A. AJCC cancer staging manual. 7th ed. New York, NY: Springer: 2010. [Google Scholar]
- 16.Piccirillo JF, Lacy PD, Basu A, Spitznagel EL. Development of a new head and neck cancer-specific comorbidity index. Arch Otolaryngol Head Neck Surg. 2002;128(10):1172–1179. doi: 10.1001/archotol.128.10.1172. [DOI] [PubMed] [Google Scholar]
- 17.Yim JM, Matuszewski KA, Vermeulen LC, Jr., Ratko TA, Burnett DA, Vlasses PH. Surveillance of colony-stimulating factor use in US academic health centers. Ann Pharmacother. 1995;29(5):475–481. doi: 10.1177/106002809502900504. [DOI] [PubMed] [Google Scholar]
- 18.Chrischilles EA, Pendergast JF, Kahn KL, et al. Adverse events among the elderly receiving chemotherapy for advanced non-small-cell lung cancer. J Clin Oncol. 2010;28(4):620–627. doi: 10.1200/JCO.2009.23.8485. [DOI] [PMC free article] [PubMed] [Google Scholar]